i , , ‘
it 1 .2 ,x
1 if . . ‘ $5.?“

‘4. 5::
Jan

 

 

 

 

 

 

 

 

 

‘mvzn..um
an :32.

"hm-watfl

 

 

 

 

 

,- .i’

.vu‘hft)”. 3:!
5%! («.4
1...») x .3...
‘ “a A.

4

 

 

‘ PM. 9.2... ‘

a f rid}.

away? . , 2 1.1.. at, , . ‘ 1. .. ‘
L _ . | x. , 55.. *1; , .gaiummgyar: . .wmge

‘ ‘ ‘ . 1:
IA!

 

 

3.
Us

LIBRARIES
g) MICHIGAN STATE UNIVERSITY
EAST LANSING, MICH 48824-1048

am

This is to certify that the
dissertation entitled

THE VALIDITY OF A NON-INVASIVE METHOD OF MATURITY
ESTIMATION AND INTRINSIC RISK FACTORS FOR INJURY IN
YOUTH FOOTBALL PLAYERS: ANALYSIS OF THE 2002 AND
2003 SEASONS

presented by
Thomas Patrick Dompier

has been accepted towards fulfillment
of the requirements for the

PhD. degree in Kinesiology

 

 

MfiW

Major Professor’ 5 Signature

 

MSU is an Affirmative Action/Equal Opportunity Institution

 

 

PLACE IN RETURN Box to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2/05 0: lRC/Dat .Indd-p.15

 

THE VALIDITY OF A NON-INVASIVE METHOD OF MATURITY ESTIMATION
AND INTRINSIC RISK FACTORS FOR INJURY IN YOUTH FOOTBALL
PLAYERS: ANALYSIS OF THE 2002 AND 2003 SEASONS

By

Thomas Patrick Dompier

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Department of Kinesiology

2005

ABSTRACT
THE VALIDITY OF A NON-INVASIVE METHOD OF MATURITY ESTIMATION
AND INTRINSIC RISK FACTORS FOR INJURY IN YOUTH FOOTBALL
PLAYERS: ANALYSIS OF THE 2002 AND 2003 SEASONS
By

Thomas Patrick Dompier

Youth participation in tackle football is increasing each year. Late maturation has
been implicated as a risk factor for injury. Percentage of predicted adult stature derived
from the Khamis and Roche [KR](1994) non-invasive method of adult stature prediction
has been proposed as an alternative to invasive measures of maturity. Percentage of
predicted adult stature remains untested as a maturity indicator in youth football players.
The purpose of this study was to determine the validity of percent of predicted adult
stature in youth football players, and to examine maturity as a risk factor for injury in the
same population. There were 779 youth football players in grades fourth through eighth
involved in the injury analysis study and a subset of 64 participated in the validation of
the non-invasive method of maturity estimation. Partial correlations controlling for
chronological age revealed that the KR percent of predicted adult Stature was moderately,
but significantly related to skeletal age (partial I, adjusted for CA, = 0.54; p < .001 ).
Injury analysis revealed that 284 players accounted for 474 injuries and 26565 exposures.
Players were twice more likely to be injured in games than in practices and 1.4 times
more likely to suffer a non-time-loss injury. Risk factor analysis revealed that maturity
was not a risk factor for injury. Stature and previous injury were significant risk factors

for injury.

Copyright by
THOMAS PATRICK DOMPIER
2005

This dissertation is dedicated to my father who always told me to get a college degree; he
just never told me when I could stop.

iv

ACKNOWLEDGEMENTS

I would like to thank my wife Jennifer for her endless amounts of patience,
understanding, and love. Jennifer, you have provided the strength and motivation I
needed to finish this project. Thanks for sticking it out.

I would like to thank Dr. John Powell for his mentorship, but most of all, for his
friendship over the course of my journey. I would be remiss if I did not also thank
Noreen Powell for her caring and gourmet home cooked meals. I cannot thank you both
enough for everything you have done for Jennifer and I.

I would like to thank Dr. Sally Nogle for her inspiration, but most of all, I would
like to thank her for the 8—years of caring and understanding I received under her tutelage.

I would like to thank Dr. Crystal Branta for sparking my interest in the study of
growth and development. I would also like to thank Dr. Branta for posing challenging
questions when I thought I knew all the answers, but learned otherwise.

I would like to thank Dr. Jeff Kovan for teaching me bedside manner. Dr. Kovan
is not only one of the best physicians I know, but he is also the most down to earth. I
would also be thoughtless if I forgot to thank Dr. Kovan for the kitchen table sutures.

I would like to thank the athletic training and sports medicine staff at Michigan
State University for their friendship and the family like atmosphere they provide to all
students in the athletic training program. I would especially like to thank Dave Carrier
for his mentorship, friendship, and inspiration over the many years I have known him.
Dave provided the little push with the business end of a spot built that I needed to get my

act together and my career on track.

I would like to thank Bob and Eva Malina for allowing me to stay at their
amazing ranch in Texas, and for teaching me the Fels method of skeletal age estimation.

I would like to thank Dr. B. Fett of the Mandalorian Institute for his consultation
and friendship.

I would like to thank all the radiologists and all employees at the Michigan
Athletic Club Radiological Center for their assistance and professionalism. Without their
patience and professionalism, we would never have had such a good turnout and response
for the radiographs.

Most importantly, I would like to thank the staff, parents, but most importantly,
the players in the Mid-Michigan Pony Football League, specifically those in Holt and St.

Johns. None of this was possible without their help and participation.

vi

TABLE OF CONTENTS

 

 

 

LIST OF TABLES - - - - _ _ - X
CHAPTER ONE - - - -- _ _ _ - 1
INTRODUCTION ................................................................................................................. 1
Purpose of the Study ................................................................................................... 7
Hypotheses .................................................................................................................. 7
Research Design ......................................................................................................... 9
Limitations ........................................................... 9
Delimitations ............................................................................................................... .9
Operational Definitions .............................................................................................. 9
Athlete exposure .................................................................................................... 10
Body mass index .................................................................................................... 10
Incidence ............................................................................................................... 10
Incidence density ratio .......................................................................................... 10
Injury definition .................................................................................................... IO
Injury rate ............................................................................................................. 10
Injury severity ....................................................................................................... 1 1
Midparent stature .................................................................................................. 1 1
Odds ratio ............................................................................................................. ll
Predicted adult stature .......................................................................................... 1 1
Percent predicted adult stature ............................................................................. l 1
Relative risk .......................................................................................................... 11
Risk ........................................................................................................................ 12
Session ................................................................................................................... 12
CHAPTERTWO- - -- - _ - - - - -- - l3
LITERATURE REVIEW .................................................................................................... 13
Matching Youth by Maturity for Sports Participation .............................................. 13
Maturity and performance .................................................................................... 14

Age effect in sports ................................................................................................ 16
Risk of injury in late maturing football players .................................................... 19
Methods for Assessing Maturity Status ..................................................................... 21
Secondary sex characteristics ............................................................................... 22
Skeletal age assessment ........................................................................................ 25
Somatic maturity assessment ................................................................................ 32
Injury Incidence, Risk, and Rates in Football ........................................................... 35
The history of injury surveillance infootball ........................................................ 36
College football injury incidence, risk, and rates ................................................. 38
High school football injury incidence, risk, and rates .......................................... 39
Youth football injury incidence, risk, and rates. ................................................... 41
Risk F actors for Injury in Football ........................................................................... 43
Exposure time ........................................................................................................ 45
Playing surface and shoe interface ....................................................................... 46

vii

Fitness level .......................................................................................................... 47

 

 

 

 

 

 

 

 

Psychological variables ........................................................................................ 48
Player position and game situations ..................................................................... 49
Maturity status ...................................................................................................... 49
Summary of the Literature ........................................................................................ 51
CHAPTER THREE 53
METHODS ................... ' .................................................................................................... 53
Overview ................................................................................................................... 53
Research Design ....................................................................................................... 53
Subjects ..................................................................................................................... 54
Instrumentation ......................................................................................................... 55
Procedures ................................................................................................................ 56
Stage one, validation of percent of predicted adult stature .................................. 56
Stage two, injury analysis ..................................................................................... 59
CHAPTER FOUR 61
RESULTS ........................................................................................................................ 61
Stage One The Validity of Percent of Predicted Adult Stature ................................. 6]
Participant demographic data .............................................................................. 62
Partial correlations and t-tests ............................................................................. 63
Stage T wo Results: Injury Analysis ........................................................................... 65
Descriptive epidemiology ...................................................................................... 65
Player sport participation and injury history ....................................................... 66
Injury data analysis ............................................................................................... 72
Intrinsic risk factor analysis ................................................................................. 73
CHAPTER FIVE 83
DISCUSSION ................................................................................................................... 83
Stage One: The Validity of Percent of Predicted Adult Stature as a Maturity
Indicator .................................................................................................................... 83
Subsample baseline data ....................................................................................... 83
Subsample statistical analysis ............................................................................... 84
Stage Two: Injury Analysis ....................................................................................... 85
Baseline data and maturity estimation ................................................................. 85
Injury analysis ....................................................................................................... 86
Intrinsic risk factor analysis ................................................................................. 88
Conclusions ............................................................................................................... 92
Future Research ........................................................................................................ 92
APPENDIX A 94
APPENDIX B 96
APPENDIX C 97
APPENDIX D 99
APPENDIX E 100

viii

APPENDIX F

 

APPENDIX C

 

APPENDIX H

 

APPENDIX I

 

APPENDIX J

 

REFERENCES

 

ix

102
103
104
105
106
107

LIST OF TABLES
Table 1: Summary Of Injury Incidence, Risk, And Rates In High School, Junior High,

And Youth Football Players .............................................................................................. 44
Table 2: Grade Specific Mean Ages And Physical Characteristics For 64 Youth Football
Players .............................................................................................................................. 62
Table 3: Mean Predicted Adult Statures And Percents Of A dult Statures For 64 Youth
Football Players By Grade ......... , ...................................................................................... 63
Table 4: Partial Correlations Controlled For Chronological Age By Grade .................. 64
Table 5: Paired T- Test Results For Stature Prediction Methods By Grade ..................... 64
Table 6: Paired T -T est Results For Percent Of Predicted Adult Stature Methods By
Grade ................................................................................................................................ 65
Table 7: Mean Player And Parent Demographics By Grade ........................................... 68
Table 8: Proportion Of Returning Players By Grade And Year In Study ......................... 69
Table 9: Proportion Of Participants With Prior Football Experience By Grade ............ 69

Table 10: Proportion 0f Players Who Reported Playing Sports Other Than Football By
Grade ...................................................................................................................... 69

Table l 1: Proportion (H The Ages That Participants Reported Beginning T 0 Play
Organized Sports By Grade .............................................................................................. 70

Table 12: Proportion Of First Sports Played As Reported By Participants By Grade 70
Table 13: Proportion Of Participants Who Reported Having A Previous Injury By Grade

........................................................................................................................................... 71
Table 14: Proportion Who Reported Missing Practices And Games Due T 0 A Previous

Injury By Grade ................................................................................................................ 71
Table 15: Proportion 0f Reported Previous Injury Locations By Grade ......................... 71
Table 16: Frequency And Proportions Of Reported Previous Injury Type By Grade ...... 72
Table 17: Summary Of Injury Data By Grade .................................................................. 74
Table 18: Relative Risk For Injury By Grade ................................................................... 75
Table 19:Risk Factors For Injury In Youth Football Players: All Grades Combined ..... 78
Table 20:Risk Factors For Injury In Youth Football Players: 4”t5”' Grades .................. 79
Table 21 :Risk Factors For Injury In Youth Football Players: 6th Grade ........................ 80
Table 22:Risk Factors For Injury In Youth Football Players: 7th Grade ........................ 81
Table 23: Risk Factors For Injury In Youth Football Players: 8th Grade ....................... 8 2

Chapter One

Introduction

Participation in organized sports is an increasingly popular form of recreational
activity for children. One sport that is growing in popularity is American football. Over
2.8 million persons over the age of six participate in tackle football each year (Sporting
Goods Manufacturers Association, 2004). Two of the largest youth football
organizations, Pop Warner Little Scholars [PW] (2003) and American Youth Football
[AYF] (2004), boast over 240,000 and 200,000 annual participants respectively. Many
other local, regional and state organizations exist making the total number of participants
much higher.

The National Safe Kids Campaign [NSKC] (2004) reports that more than 3.5
million children between the ages of 5 and 14 are injured while participating in sports
each year. Of those 3.5 million, large proportions were seen in hospital emergency
rooms. Of those visits, 207,400 were attributed to basketball, 187,800 to football,
116,900 to baseball and softball, 76,200 to soccer, and 21,200 were attributed to
gymnastics. The NSKC also reports that 28 percent of youth football players between the
ages of 5 and 14 were hurt while playing their sport. While emergency room visits are
important when considering the burden of injury on society, it does not take into account
the numerous injuries that occur but are not treated in emergency rooms. Powell and
Dompier (2004) found that only 25 percent of the injuries reported by college football
players warranted medical attention or time loss from participation. If these proportions

are similar in youth football, then the actual number of children injured playing youth

football is much higher than previously reported. The relative increase in injury
incidence in youth sports, and the increased publicity of injured professional and
collegiate athletes, has heightened interest in the risk and prevention of injury among
youth football players.

Injury is an inevitable part of sport, but some proportion is probably preventable.
Adirim and Cheng (2003) describe a general injury prevention model that identifies the
“three E’s”, educational, environmental, and enforcement interventions. In the context of
youth football, education includes the mastering of football techniques and proper
training for coaches. Environmental interventions include the type or condition of the
playing surface, temperature conditions, or inclement weather policies. Lastly,
enforcement involves coaches following policies and procedures and officials enforcing
the rules and equipment standards. Hergenroeder (1998) further proposed six areas of
injury prevention, including preseason physicals, medical coverage at practices and
events, coach education, proper hydration, proper equipment and field maintenance, and
proper rule enforcement and modification. As technology and knowledge advances, the
properties of protective equipment improve and rule modifications must take place to
accommodate new products. To monitor the effects that changes to equipment or rules
have on sport, injury surveillance is used to identify and analyze the associations of risk
factors and or the efficacy of interventions.

Injury patterns in high school and collegiate football have been described, but
there remains a severe paucity in the literature describing injury patterns in youth football
(14 years of age and under). The major reason for this difference is the organizational

structure at lower levels of competition. High school and college athletic conferences

have league and or national injury surveillance programs to monitor injury patterns over
time. The advantage that high schools and colleges have over youth sports organizations
is the availability of trained medical professionals to document injury and exposure
information. Youth sports generally do not have the organizational structure, personnel,
or finances to conduct injury surveillance programs. The coaches and league officials,
although generous with time and effort, rarely have formal training in coaching or injury
prevention and are typically volunteers. This lack of formal training limits the majority
of available injury data regarding youth football to hospital registries, retrospective
surveys, or coach reports. Research examining specific risk factors for injury in youth
football is even more limited and is further complicated by the nonlinear growth of
adolescents as they progress through maturity.

Maturity is the process of progressing to the mature state of adulthood (Malina,
Bouchard and Bar-Or, 2004). Maturity status can vary greatly among individuals of the
same sex and within the same age group. Variations are also dependent on the biologic
system that is considered. The three most commonly considered systems are the Skeletal,
sexual and somatic systems. These three systems are related, but the status obtained from
each can differ Significantly from a child’s chronological age and from each of the other
systems. The disparity is most notable during the years surrounding puberty and the
onset of the adolescent growth spurt (9-1 1 for females and 10-12 for males).

The methods of assessing skeletal, secondary sexual characteristics, and somatic
maturity have strengths and weaknesses. Skeletal maturity measured from radiographs is
considered one of the most accurate methods of assessing biologic maturity (Bayley,

1946; Groell, Lindbichler, Riepl, Gherra. Roposch. and Potter, 1999; Roche and Davila,

1976; Roche, Davila, and Eyman, 1971; Tanner, 1962). This method can be used
throughout all periods of growth, but is costly, invasive, and exposes the subject to
radiation. Secondary sexual characteristics such as pubic hair development, the age at
menarche, testicular volume, and breast development have also been used to determine
sexual maturation. These processes are invasive and can be embarrassing for
adolescents, and are further limited to a small time period surrounding the ages of
puberty. The method of assessing secondary sexual characteristics is the most widely
used measure in clinical studies, and has been shown to be reliable (Bonat,
Pathomvanich, Keil, Field, and Yanovski, 2002; Brooksgunn, Warren, Rosso, and
Gargiulo, 1987; Demirjian, Buschang, Tanguay, and Patterson, 1985; Matsudo and
Matsudo, 1994; Schlossberger, Turner, and Irwin, 1992; Taylor, Whincup, Hindmarsh,
Larnpe, Odoki, et a1., 2001; Tanner, 1962). There are two common methods of assessing
somatic maturity. The first requires longitudinal data from which the onset of the
adolescent growth spurt and peak height velocity (PHV) is derived. This method
however, is limited by the need for longitudinal data and a narrow period during which it
is useful. Bayer and Bayley (1959) described a method of somatic maturity estimation
that expresses a child’s current stature as the percentage of their predicted adult stature
(PPAS). Other methods of estimating somatic maturity include taking other various
anthropometric measurements, but PPAS is regarded as the most versatile (Malina et al,
2004a)

Estimating a child’s PPAS is applicable to a variety of study designs and can be
used across a wide range of age groups. To determine a child’s PPAS, the child’s

predicted adult stature (PAS) must first be calculated. There are various methods to

estimate PAS, but most require the child’s current chronological age (CA), stature,
weight, skeletal age (SA), and mid parent stature (MPS). More recently, Khamis and
Roche (1994) developed a non-invasive method that does not require SA. The PPAS can
be used to estimate maturity because two children who are the same stature and age may
differ because one has already attained a greater percentage of predicted adult stature.
This method offers several advantages. First, it is easy to perform on large samples and
can be conducted at a time convenient to the individual, team, and investigators.
Secondly, the equipment requirements are minimal. Lastly, this method is non-invasive
and inexpensive making it practical for a variety of study designs.
Few studies have examined maturity as a risk factor for injury in youth football,
yet it has been argued that children should be matched based on maturity rather than
chronological age, ability, or grade in school (Baxter-J ones, 1995; Caine & Broekholl,
l 987; Gallagher, 1969; Goldberg & Boiardo, 1984; Hafner, Scott, Veras, Goldberg,
Rosenthal, Robertson, and Nicholas, 1988; Kreipe, 1985). This argument is centered on
vvhether or not a child who is less mature is at risk for injury if playing football with other
children of the same age who are more mature. Many systems currently exist for
matching children for competition. These include CA, sex, skill level, weight, and
biologic maturity (often measured in the form of sexual maturity). Pop Warner
developed a classification matrix based on statures and weights rather than age or grade
level. The AYF Organization classifies children according to grade level for tackle
Football and by age for touch and flag football. The Mid-Michigan Pony Football League
(the focus of the current study) classifies children by grade, but has further restrictions

that are based on birth dates. The most commonly used criteria are CA. grade in school,

and sex. These criteria are most often disputed around the age of puberty when sex
differences in timing and tempo of maturation can have a significant effect on skill,
strength, fitness, and size (Beunen, Ostyn, Simons, Renson, & van Gerven, 1980;
Katzmarzyk, Malina, and Beunen, 1997; Mota, Guerra, Leandro, Pinto, Ribiero et al.,
2002; Pratt, 1989; Rarick and Oyster, 1964). Three studies that have examined maturity
as a risk factor for injury in youth football are inconclusive (Linder, Towsend, Jones,
Balkcom, & Anthony, 1995; Malina, Morano, Barron, Miller, and Cumming, 2002;
Violette, 1976).

Research on the relationship of maturity and youth football injury is sparse.
There have been three studies to date that have examined maturity as a risk factor for
injury in youth football. Linder, et al. (1995) speculated that junior high school football
players who were more mature were at greater risk for injury. To determine this
relationship, the authors used the stages of secondary sex characteristics described by
Tanner (1962) to determine maturity status. A physician provided these evaluations at
the time of their preparticipation physicals. Players with higher levels of maturity were
found to be at greater risk of injury. In the second study, Malina, Morano, et al. (2002)
reported no relationship between injury risk and maturity in a group of youth football
players. Lastly, Violette (1976) studied stages of secondary sexual characteristics in
middle and high school football players, and found that amongst the younger age groups,
those less mature were at greater risk of injury. This disparity among the few available
studies makes it impossible to determine if matching children by maturity is an effective
means of reducing injury risk in youth football players. Opponents of matching by

maturity versus age cite that removing children from peer groups and placing them on

teams with older or younger children may have profound effects on the child’s self-
esteem and or self-concept (Baxter-Jones, 1995). To further investigate the efficacy of
matching youth football players by maturity, a valid method of estimating maturity most
be found.
Purpose of the Study

The recent advent of the Khamis and Roche [KR] (1994) method for estimating
PAS in absence of skeletal age (SA) provides a practical method for classifying children
by PPAS in a variety of study designs. Although the KR method has shown to have little
increase in error when compared to its correlate that includes SA, it remains untested in a
sample of youth football players. Youth football players may differ significantly from
the reference population from which these regression coefficients were derived.
Participants in sports such as football are prone to selection bias because characteristics
such as size, strength and speed are considered essential for success. Thus, this method
should be applied and validated in a sample of youth football players to determine if it is
a valid method of differentiating maturity status in a selection-biased sample.

Speculation exists that maturity is a risk factor for injury and that competition
levels should be arranged according to maturity status versus age or grade level. Few
studies have considered maturity as a risk factor for injury in youth football. The purpose
Of this study is to determine the validity of PPAS when derived from the non-invasive
KR method of predicting adult stature, and to use these estimates in an injury risk model.
Hypotheses

This study has two facets that will be examined. The first facet is to assess the

efficacy of the KR method for predicting adult stature. and to determine the validity of

PPAS as an estimate of maturity. This comparison will be accomplished by comparing
the KR method to the Roche, Wainer, & Thissen [RWT] (1975) method of PAS that was
updated by Khamis and Guo [RWT-KG] (1993). The RWT-KG method is generally
considered superior to the KR method because of the inclusion of SA (Khamis & Guo,
1993), but the KR method (derived from the same sample) has been shown to have little
increase in the 90% error bounds (Khamis & Roche, 1994). The second Stage of the
study will involve the analysis of injury risk, rates and risk factors. Maturity status has
not been adequately studied as a risk factor for injury in youth football and will be
analyzed in relation to other intrinsic player variables and injury using univariate and
logistic regression methods. The specific research questions and corresponding null

hypotheses that were examined include:

1. Is the KR method a valid estimator of predicted adult stature?
H01: No linear relationship exists between the predicted adult statures derived
from the KR and RWT methods.

2. Is KR method of predicted adult stature a valid measure of maturity when
expressed as a percentage of the predicted adult stature?
H02: No linear relationship exists between PPAS derived with the KR method
and skeletal age.

3. Is maturity a risk factor for injury in youth football players?

H03: No relationship exists between maturity status and injury.

Research Design

Two study designs were utilized to examine the research questions. The first
Stage consisted of a cross-sectional design, and the second consisted of an observational
cohort design. The cohort consisted of youth football players in the fourth through eighth
grades (8.5-14.5 years old) that were observed over two years in two central Michigan
communities that participate in the Mid-Michigan Pony Football League.
Limitations

Reported injuries are limited to those that were reported to the onsite investigator.
Some injuries may not have been brought to the attention of the investigator. The
samples in both stages are convenience samples. All youth football players who are
registered participants on one of the two teams, and their parents who complete the
necessary informed consents were included in the respective sample. The samples may
not be representative of other youth football players in the State of Michigan. ‘2» 7:
Del imitations . i

The sample population selected for this study is limited to youth football players
in two central Michigan communities in the 2002 and 2003 football seasons. The ability
to generalize to all youth football players beyond those of Similar communities in central
Michigan is limited. The sample does represent the suburban nature of central Michigan
and can be generalized to similar areas within the State.
Operational Definitions

This study uses a consistent set of definitions previously established by the
National Athletic Injury Reporting System (NAIRS) (Powell, 1980) and reported

elsewhere (Powell and Dompier, 2004).

Athlete exposure. Athlete exposures (AE) are defined as an opportunity for an
athlete to be injured. Exposures are calculated by tallying the number of active
participants in each coach directed session. Exposures are separated by session type and
participation level for analysis purposes. Each player active in a session is counted as
one exposure. Total daily exposures are the total number of active participants during
that session. These are then totaled across participation levels for the total season
exposures in games and practices.

Body mass index. Body mass index (BMI) is a measure of body weight relative to
stature and is defined as weight in kilograms divided by the square of stature in meters.

Incidence. Incidence is used synonymously with frequency. It is the count of
occurrences of injury that occurred during the study period.

Incidence density ratio. The incidence density ratio (IDR) is the proportion of
two injury rates and provides a basis of comparison.

Injury definition. Time-”loss (TL) injuries include all cases that require the
athlete’s removal from the current or subsequent sessions. Included are any suspected
concussions requiring medical referral or observation prior to returning to play, dental
injuries that require referral, any fracture, and any injury requiring medical evaluation.
Non-time-loss (NTL) injuries include any athlete and investigator contact that requires
evaluation and or treatment by the Certified Athletic Trainer (ATC) but does not require
removal from the current or subsequent sessions.

Injury rate. The injury rate (IR) is the proportion of injuries that occur per 1000

AB. The IR is calculated by dividing the number of injuries by the number of Ali.

10

Injury severity: Time-loss injuries are further broken down into minor, moderate
and major injuries. Minor injuries are those that require seven or fewer days lost from
participation. Moderate are those injuries that require eight to twenty days lost from
participation, and major are those that require greater than twenty-one days lost
participation.

Midparent stature. The midparent stature (MP8) is the average stature of the
biological mother and father. The MP8 was calculated by summing the statures of both
biological parents and dividing by two.

Odd ratio. The odds ratio (OR) is the ratio of two odds. The OR is calculated
with the following equation: (exposed cases * non-exposed non-cases)/(non-exposed
cases * exposed non-cases).

Predicted adult stature. Predicted adult stature (PAS) is an estimate of the stature
a subject will attain at the age of 18-years old. Predicted adult stature can be obtained by
using any number of regression equations. , . '

Percent predictedadult stature. The percent of predicted adult stature (PPAS) is
a somatic measure of maturity. The child‘s current stature is expressed as a percentage of
their PAS. Children of the same chronological age and stature can have attained various
percentages of their adult stature, thus differentiating them from one another based on
maturity.

Relative risk. The relative risk (R) is a ratio between two risks. The RR is
calculated with the following equation: (exposed cases * the total exposed)/(non-exposed

cases * the total non-exposed).

ll

Risk. Risk is the proportion of injured players among the total sample of players
during the study period and is expressed per 100 players. It is the number of injured
players divided by the number of players.

Session. Any coach directed practice or game.

Chapter Two

Literature Review

The purpose of this study was twofold. First, the validity of the Khamis and
Roche (1994) method of estimating predicted adult stature, and subsequent percent of
adult stature was examined in a group of youth football players in the Mid-Michigan
Pony Football League. Secondly, maturity was assessed as an injury risk factor in the
same population. In order to gain insights from previous research that has been
conducted in the field of maturity assessment and injury risk, this review of literature was
divided into five major sections: (1) Matching youth by maturity for sports participation,
(2) Methods for assessing maturity status, (3) Injury incidence in football, (4) Risk
factors for injury in football, and (5) Summary of the Literature. Special emphasis was
placed on locating studies that included youth and junior high school football players.
Matching Youth by Maturity for Sports Participation

The biologic maturity of children and adolescents of the same chronological age
(CA) can vary greatly. Chronological age can vary from biologic age or maturity status
by four or more years (Jones, Hitchen, and Stratton, 2000; Katzrnarzyk, et al., 1997;
Malina, Bouchard, et al., 2004; Roche et al., 1975). This difference is even more
apparent during adolescence around the time of puberty and the adolescent growth spurt.
This period coincides with significant alterations in growth, body composition,
cardiovascular endurance, and muscular strength (Malina, Bouchard, et al., 2004). For
those reasons, matching youth in football by maturity status has been proposed as an

alternative to matching by CA or grade in school (Baxter-Jones, 1995; Caine &

l3

Broekhofi, 1987; Gallagher, 1969; Goldberg & Boiardo, 1984). There are three main
reasons why matching by maturity was proposed: (1) Matching youth football players by
maturity status was thought to make competition fairer and more satisfactory by allowing
youth to participate against peers that are similar in size and strength, (2) Matching by
maturity will reduce the age bias that is thought to exist because of age cutoff dates, and
(3) Late maturing youth football players are thought at risk of injury when pitted against
average and early maturity adolescents of the same CA.

Maturity and performance. There is a relationship between maturity status and
performance parameters. In males, average and early maturing adolescents perform
better in skill tests, are stronger, taller, and heavier. Conversely, females who are late
maturing tend to perform better in some tasks, but are shorter and lighter than their more
mature counterparts. These observations are sport specific and can vary due to many
factors. Researchers have sought to determine if maturity and other personal
characteristics such as CA, stature, and weight affect fitness parameters and performance.

The research examining maturity status and performance is consistent. Early
studies examining skeletal maturity in males and females demonstrated that children who
are advanced in maturity are also advanced in motor skill performance and strength
(Rarick and Oyster, 1964). Similar results were reported by Katzmarzyk et a1. (1997)
using the Tanner-Whitehouse 11 method if SA assessment (Tanner, Whitehouse,
Marshall, & Carter, 1975). Katzmarzyk et al. found similar results when analyzing the
effect of maturity on strength and performance parameters in children between the ages
of 7 and 12 years. Parameters included grip strength, pushing and pulling strength, a 35-

yard dash, standing long jump, and a softball throw for distance. The combined effect of

14

skeletal age (SA) and CA was the best predictor of motor performance while strength was
best predicted by mass. The authors conceded the interaction of many parameters and
note that performance and strength variables are dependent on many influencing factors
and not exclusively on maturation or mass. Malina et al. (2000) found a performance
advantage in elite soccer players who were advanced in maturity. Malina et al. used SA
derived from the Fels Method described by Roche, Chumlea, and Thissen (1988). A
similar effect has been noted in other studies that used other methods to estimate
maturity.

An association between performance and maturity has been reported when
secondary sexual characteristics were used to assess maturity. Jones, et al., (2000)
examined maturity in relation to skill performance in a cross-sectional study of girls and
boys between the ages of 10 and 16 years. The investigators measured multiple
parameters, and maturity was graded by self-assessed sexual maturation. Sexual maturity
was significantly and positively correlated to vertical jump, shuttle run, and grip strength
scores in both boys and girls demonstrating a parallel relationship between increases in
maturity and performance. Mota et al., (2002) found that maturity had an effect on
running performance but not cardiorespiratory fitness. In another study examining
running performance, Eisenmann and Malina (2003) studied age and sex associated
variation in male and female distance runners. Males tended to increase task
performance in most skills with concurrent increases in age, but females tend to plateau
around the time of the adolescent growth spurt. This plateau effect has been
demonstrated in other studies (Jones et al., 2000; Mota et al., 2002). A recent study by

Malina, Eisenmann, Cumming, Ribeiro, and Aroso (2004) examined sexual maturity and

15

 

performance in elite soccer players and found a large proportion of the variance in the
performance parameters was related to maturity.

Maturity is an important factor that influences performance parameters and should
be considered in research involving youth. This consistency in findings supports the
assertion that children who are in the highest or lowest percentiles of maturity for a given
age group should be considered for matching by maturity. Hafner et al., (1982)
systematically matched athletes based on maturity and other player characteristics in an
attempt to make competition fairer and more rewarding for middle and high school
athletes. Adolescents who wished to be considered for moving up or down in
competition level had to undergo a thorough evaluation of personal characteristics such
as stature, weight, skill performance, and stage of sexual maturation. Throughout the
season, data were recorded regarding playing status, injuries, success, and Skill
improvement. Outcome variables included athlete satisfaction, performance, and injury.
Hafner et al. reported that the program of placement based on the listed criteria improved
player satisfaction and competitiveness, but data regarding injury was regarded as
unreliable and not adequately reported. Therefore, no inferences were made regarding
player safety.

Age effect in sports. Age is another influencing factor thought to affect
competitiveness in youth sports (Edwards, 1994; Simmons and Paull, 2001). Various
sports have age cut-off-dates that dictate the level of competition a child will participate
in and are solely based on the child‘s date of birth. Little League Baseball [LLB] (2004)
is an example of a sport with a specific cut-off date. Little League Baseball has an age

cut-off of l2-years-old and a year cut-off date of August 1‘“. A 12-year-old child born

16

after August lSt can play while a child who is born on August 1St or before of the same
year cannot play because they are too old. Youth football leagues use different
classification criteria.

Youth football leagues have different criteria than little league baseball, and often
differ from other youth football leagues. The PW (2004) youth football organization has
developed an extensive age and weight chart that allows older participants of the
specified weight to play at competition levels as many as three divisions lower.
Conversely, a participant who is younger but much heavier can play on a level up to three
divisions higher. The PW matrix was developed to equalize the competition levels,
reduce size mismatches, and maximize development opportunities for the players (PW,
2004). The approach taken by the AYF (2004) organization is much different. The
highest division within the AYF is only restricted to participants under the age of 16
years. The next division is comprised of 6th to 8‘h graders, and the lowest tackle football
division is comprised of 3rd to 5th graders. Both the PW and AYF have flag football
divisions that are determined solely by age. The Mid-Michigan Pony Football league (on
which this study was based) has a different system of matching. There are four divisions,
4‘h and 5th grades are combined into one division, while the 6m, 7m, and 8th grades each
make up separate divisions. Players who are 9 years of age on or before September 15‘, or
in 4th grade of the current year will play in the youngest division. Players who are 12
years old before September 1St must play in the 6th grade, and those who are 13, must play
in the 7th grade. Players who are 14 years old must play in the 8th grade. Potential players

who are 15 years old before September lSt are not permitted to play. These classification

17

systems were designed with the best intentions of making competition fair and providing
a safe environment for the athletes.

Few studies have examined the age effect in American football, and none have
examined the possible age effect among youth football players. Daniel and Janssen
(1987) surveyed professional football players in the Canadian Football League (CFL) and
National Football League (NFL) and found no age effect among the players. In that
study, 49% of the players in the CFL and 52% of the players in the NFL were born in the
first half of the competition year. In a review of birth dates among 167 NFL Hall of
Fame members, Stanaway and Hines (1995) also failed to find an age effect. That study
reported 57% of those included were born in the first half of the competition year.
Glamser and Marciani (1992) found an age effect among college football players at two
universities. They found that 66% of the respondents were born in the first half of the
competition year. The inconsistencies and paucity of literature surrounding the age effect
in football makes forming any conclusions impossible. Other sports such as baseball,
soccer and hockey have been more widely studied for age effect.

The age effect bias has been studied more extensively in sports such as baseball,
soccer, and hockey. Most sports with the exception of baseball show an age effect. Two
studies of baseball player birth dates revealed no age effect (Stanaway & Hines, 1995;
Thompson, Bamsley & Stebelsky, 1991). Studies involving soccer players have included
adult and youth, both at elite and at non-elite levels, and have all demonstrated that a
birth effect does exist (Dudink, 1994; Maffulli, King, and Helms, 1994; Verhulst, 1992).
Similarly, hockey has shown an age effect that has been widely reported (Daniel and

Janssen, 1987; Hurley, Lior, and Tracze, 2001). It is unclear if the age affect is the result

18

of athletes being born at the beginning of the competition season, or if it is from normal
variation. If being born at the beginning of a competition year is advantage, than those
who are earlier are at a competitive advantage over the younger members of their birth
cohort.

Risk of injury in late maturing football players. Matching by maturity status has
been proposed as a means of reducing injury risk in youth football. This
recommendation, however, is based on speculation and intuition rather than research that
demonstrates increased risk for less mature players. Three studies have examined
maturity as a risk factor for injury in youth football. These studies report conflicting
results further confounding the issue.

Only one of the three studies examining maturity in relation to injury risk found
an increased injury risk in less mature players. Violette (1976) studied stages of sexual
maturity and injury in middle through high school football players. Results demonstrate
slightly greater risk of injury amongst the less mature in the 13, 14, and 15-year-old age
groups, but no relationship was found in the 16 and 17 year old players. One of the
weaknesses noted by the authors was the limited utility of sexual maturation in the 16 and
17-year—old age groups. Nearly all the participants in the higher ages were sexually
mature or approaching sexual maturity. The index of sexual maturity may not have been
sensitive enough in these age groups to detect a difference. The strength of this study
was the prospective design, large sample, and thorough injury documentation. The other
two studies examining this issue found an inverse or no relationship between maturity

and injury risk in youth football players.

19

The two studies that failed to detect a positive relationship between maturity and
injury risk in youth football players were biased by exposure time. Players who are
heavier, taller, and stronger are likely to receive more individual playing time during
games and more repetitions during practices making them more likely to suffer an injury.
Linder et al., (1995) conducted a two-year prospective study of 340 junior high school
football players between the ages of 11 and 15. There were 55 injuries reported during
the duration of the study, and injury risk was higher in those athletes who were assessed
as more mature. Exposure time was not recorded in that study. Malina, et al., (2002)
reported no relationship between injury risk and maturity in a group of 9 to 14 year old
youth football players. Malina et al. did note that injury risk increased with age and
stature, but remained unrelated to maturity status. Team exposures were recorded, but
individual playing time was not. Both studies were likely biased because the exposure
definitions could not detect a difference between levels of maturity status. Children who
are more mature were probably exposed more because they received more playing time.

Matching schemes for youth Sports have been a source of much debate. The
relationship between maturity and performance is well established, but the age effect bias
is not clear in college and professional football and has not been studied in youth football.
There are few studies that have examined maturity as a risk factor for injury in youth
football players, and that have, are inconclusive. This disparity in the literature makes it
impossible to determine which method of matching is most appropriate for providing a
safe and fair competitive environment in youth football. Organizations such as PW have
constructed complex matching schemes that make allowances for those children who are

very small or large in relation to their CA peers, but others such as AYF continue to rely

20

on matching by age and grade. The focus of the current study, the Mid-Michigan Pony
League, uses grade and CA, but makes allowances for exceptionally large individuals to
move up divisions. None of these organizations considers biologic maturity status
assessed in any form as a component in the matching process. The difficulty associated
with assessing maturity is the main reason that maturity assessments are not routinely
conducted. It has been recommended that an assessment of maturity be conducted with
every preparticipation physical exam (Caine and Broekhoff, 1987; Goldberg and Boiardo,
1984). Assuring consistent evaluation procedures amongst different doctors in different
populations would be difficult. Alternative maturity assessment methods need to be
explored.
Methods for Assessing Maturity Status

Maturity is a difficult concept to conceive and is often confused with the state of
being mature. Maturity is the degree to which a person has progressed to a fully mature
state or adulthood (Malina, Bouchard, et al., 2004). Maturation is the process of moving
toward a mature State. Maturity is mistakenly considered as a total body process but
should be considered separately for each biological system. Skeletal, somatic, and sexual
maturity are related, but one or more of these processes will progress at a different rate
than the others. In addition, biologic maturity does not coincide with CA. Children of the
same sex and CA may differ greatly in their level of maturity. It is common for same sex
children of the same CA to differ in biologic maturity by two or more years. The sex of
the child is also important and inter-sex differences can be as high as four or more years
of difference. Females, on average, are two years in advance of their male counter parts

when nearing the age of puberty (9-11 for females and 10-12 for males). This sex, age

and intersystem variability makes assessment of biologic maturity difficult, but
nonetheless important.

The ability to estimate maturity is dependent upon the ability to measure progress
of maturity at a given time (Malina, Bouchard, et al., 2004). Once this measure has been
taken, criteria for assessing the level of maturity can be assigned for a Specific sample.
To assess maturity in youth athletes, the method must be applicable to the population of
interest and practical for the restrictions of the study design. There are multiple methods
of assessing maturity and each has strengths and weakness.

The common biological systems used to estimate biological maturity are the
sexual, skeletal, and somatic systems. Sexual maturity is estimated using a number of
different indicators. Secondary sex characteristics often include pubic hair development,
testicular volume, breast development, and age at menarche. Skeletal maturity is
estimated by determining a child’s SA. Skeletal maturity is important because it can be
measured during the entire grth period starting at birth and ending in young adulthood
when growth is complete. The third most common method of assessing maturity
involves anthropometric measurements. Several methods exist for estimating somatic
maturity, but the most widely used is the age at take off (TO) and the age at PHV of the
adolescent growth spurt. These measures, as with sexual maturation, are limited to the
short time frame surrounding the event and require serial measurements. The less used
somatic maturity measurement method is determining the PPAS a child has attained at a
give age.

Secondary sex characteristics. Sexual maturation is the most widely used index

of biologic maturity in the clinical and research settings. Sexual maturation is a process

to
IQ

that ends Shortly after the onset of puberty. Puberty is a period of transition between

adolescence and adulthood (Malina, Bouchard, et al., 2004). Tanner (1962) described
stages of secondary sexual characteristics development. These include pubic and axillary
hair development, genital development, and breast development. Tanner defined five
stages for each indicator. Stage five is fully mature while stage one indicates that the

individual is pre-pubertal. Stages two, three, and four indicate that sexual maturity has

begun but not yet finished and represent progressively increased development. This

method has been widely used but has limitations.

The evaluation of secondary sexual characteristics can only be used at and around
the ages of puberty (9-11 for females and 10-12 for males). Besides being limited to a
short time frame around puberty, assessment of secondary sex characteristics is made
difficult by the very nature of the evaluation. This form of evaluation by a clinician
req uires the adolescent to disrobe and the examiner having direct observation of the
genitals, pubic area, and breasts. This procedure can embarrass some adolescents and can
make them reluctant to participate. Allowing subjects to perform self-assessments can
miti gate the invasive and embarrassing nature of the exam, but at a cost to sensitivity.
Two studies have reported good concordance between physician and subject self-
asSessment of secondary sexual characteristics. Duke, Litt, and Gross, (1980) provided
f€31‘nales with five photographs each of breast and pubic hair development. Similarly, the
l”Tlales were provided with five photographs each of genital and pubic hair development.
The subjects were instructed to select the photograph that best represented their own
Characteristics. These were then compared to physician ratings using the same

photographs. The investigators report kappa coefficients of 0.81 for breast stage, 0.91 for

23

female pubic hair, and 0.88 for genital and male pubic hair development. Similarly,
Matsudo and Matsudo (1994) compared the ability of subjects to perform a self-
evaluation using Tanner (1962) stage photographs and mirror to that of a physician.
Concordance for pubic hair was over 69%, and for breasts and genitals it was over 60%.
This evidence is compelling, but other studies have failed to produce as high
concordance.

Although physician assessment seems a reliable method of estimating sex
characteristics, not all studies report good agreement between physician and subject
assessment. Hergenroeder et al., (1999) conducted a study that not only examined
physician and subject agreement, but also interobserver agreement between physicians.
Participants were provided with drawings and written descriptions to aid in the self-
assessment. The results demonstrated a 76% agreement rating between physicians, but
kappa coefficients for physician subject agreement were poor (0.34-breast; 0.3 7-pubic
hair) - Similarly, Taylor et al., (2001) and Bonat et al., (2002) used simple drawn pictures
based on Tanner’s photographs but included simple text descriptions. Both studies found
a 10W to moderate agreement between the physicians and subjects. The use of drawings
in these studies versus the use of pictures used in the former studies may have made it

more difficult for the subjects to determine stage. The inclusion of text descriptions did
inCrease the accuracy of subject ratings.
Physician examination of secondary sexual characteristics, although a reliable
method, is not always practical and decreases subject participation. Subject self-
aSsessment can be used when physician assessment is unavailable. The Studies that used

aCtual copies of Tanner’s photographs of stages of development produced better

 

agreement. Those studies that used drawings and text descriptions were not as accurate.
Matsudo and Matsudo (1994) recommended that if self-assessments must be used, then
the subjects should be provided with color photographs, text descriptions, and a mirror to
aid in the assessment process. There are situations however, when even self-assessment
of secondary sex characteristics is not indicated and other methods must be sought. One
such method that can be used in females is the age at menarche.

Age at menarche can be assessed separately or in conjunction with other sexual
maturity indicators. In prospective and cross-sectional designs, questioning of the girls
and their mothers is reliable in determining if the girl is pre or postrnenarcheal.
Retrospectively, women can be asked to recall their age at menarche with reasonable
accuracy. Koprowski, Coates, and Bernstein (2001) reported a recall correlation of 0.83
in teenagers who were an average of 17 years old. Must et al., (2002) found similar
findings (r = 0.79) in women asked to recall age at menarche in a 30-year follow-up
stud y. A third method of estimating age at menarche is called the status quo method.
TTlis method provides a population estimate from the sample. Using the exact age of the
adolescents and whether or not they have achieved menarche, an estimated percentage of
girls who attained menarche in each age group can be derived (Malina, Bouchard et al.,

20 ()4). This method is not useful for application to individual females, but does provide
311 alternative if individual data are not available.
Skeletal age assessment. There are a number of reasons why SA age assessment
Inight be performed. Clinically, SA assessments can be used to determine a child’s
present developmental status and to estimate or predict the adult stature of the child.

Estimating adult stature is most often used when children and adolescents are thought to

 

hav e a growth or endocrine disorder. From a research perspective, SA can be used to
classify children according to maturity level. Numerous methods have been developed to
assess SA from radiographs of various areas of the body including the hand, hip, knee.

311d ankle. Methods utilizing the left hand have persisted because little radiation is .

required and the ease of positioning of the subject.

Each of the three common methods used to assess SA from left-hand radiographs
have strengths and weaknesses. In the order of the most commonly used are the
Greulich-Pyle [GP] (Greulich & Pyle, 1959), Tanner-Whitehouse [TW] (Tanner, 1962;
Tanner et al., 1975; Tanner, Landt, Cameron, Carter, and Patel, 1983; Tanner et al.,

2 001), and the F els methods (Roche et al. 1988). The GP and TW methods are the most
commonly used by pediatricians and endocrinologists in the assessment of Skeletal
maturity. More recently, the Fels method has become popular in growth studies because
it also provides a standard error of measure for the SA it estimates. All three methods

Provide regression equations for predicting adult stature using the skeletal age obtained

from the radiograph.

The assessment of SA using the GP method has evolved from the early works of
Todd (193 7). In those early studies, anteroposterior radiographs of the left hand were
e){amined for similarities in bone development based on age groups. The GP method was
developed using the same radiographs obtained in Todd’s original Brush Foundation
Study conducted in Ohio between 1931 and 1942. Greulich and Pyle (1959) published a
text on the assessment of skeletal radiographs in what is now called the “atlas”. The atlas

is the most commonly used method of clinical SA analysis in the United States (Zerin

and Hernandez, 1991).

26

 

The GP method has comparisons and ratings for all 28 bones in the hand. To use
the atlas, a current radiograph is compared to the pictures of radiographs in the atlas. This
method is commonly applied incorrectly by comparing the current radiograph with a
composite slide in the atlas. For example, if the x-ray of a 5-year old boy matches the
SI ide of a 7-year old, then the boy’s SA is 7-years. As described in the original text
however, to correctly assess SA, the SA of each bone should be assessed and then
(1 ivided by the total number of bones used in the analysis (Greulich and Pyle, 195 9).

U sing the correct method, the SA might be assessed much differently from the incorrect
method that estimated the age at 7-years. This method of visual comparison was thought
very subjective and inaccurate by Tanner (1962), thus encouraging further research.
The subjectivity of the GP method was the basis for Tanner (1962) to develop a
m ethod that used written descriptions, ratings, and scores for each common criterion
C b One description). This method included criteria for 20 bones in the left hand and wrist
Vtarsus the entire hand used in the GP. The TW method defined specific criteria that had
tO be met in order for a bone to receive a particular rating that was then provided a score.
The scores are then totaled and compared against a table that provides an estimated SA
for a particular score. This method has been refined several times. The first revision
called the TW2 Method was published 1975, and the most recent called the TW3 in 2001
(Tanner et al., 1975; Tanner et al., 2001). The TW2 and TW3 methods provide an
alternative to the 20-bone (TW-20) method by providing scoring systems for just carpals
and one for just long bones called the Radius Ulna Short Bone (RUS) method. The RUS
method uses the radius, ulna and some of the metacarpals and phalanges, but no carpals.

Throughout the revisions, the criteria were not changed, but the scoring system was

chaJIged. The TW method is commonly used by physicians due to the speed and ease of
interpreting the radiographs, but has been criticized for subjectivity and lack of
quantifiable measurements.

The F elS method is the newest method to assess SA. The F els method was
developed from serial radiographs Collected during the Fels Longitudinal Study (Roche et
a1 . 1988). This method uses 22 bones that may include up to seven additional indicators
(bone descriptions) for each. The rating criteria provided for this method included
written descriptions, pictures, and graphic examples. This method also uses quantifiable
measurements to 0.5 millimeters for some indicators. The ratings for each indicator for a
gi ven age group are then entered into a computer program that calculates the SA and
standard error of measurement (SE). This method is more difficult and time intensive to

perform and not widely used by clinicians, but it is superior for research because it
provides a SE with the SA.

Although the GP, TW, and Fels methods are commonly used, other methods of
as sessing SA have been developed for other locations on the body. One are that has been
StUdied is the knee (Aicardi et al., 2000 Roche et al., 1975). Roche et al. (1975)

developed a method for analysis of the knee, but later continued to focus on the left hand.
In addition to the knee, hip and pelvis, attempts were made to assess SA age using
radiographs of the foot, but none of these earlier methods have endured. Radiographic
analysis of the left hand has remained popular because imaging of the hand allows for
consistent positioning and it requires minimal radiological exposure. Positioning the
foot, knee, hip and pelvis is much more difficult and the exposure to radiation has to be

greater because of the thicker soft and bony tissue. Additionally, the bones of the hand

and wrist provide multiple indicators that have been repeatedly documented across the
maj ority of the skeletal maturational period. For these reasons, SA assessment has and
will continue to require the anteroposterior view of the left hand.
The GP, TW, and Fels methods are similar in that they all use the anterOposterior
View of the left hand for SA analyses. Despite this similarity, they all have very different
methods and reference populations. The reference population is an important
consideration because each was derived from very different populations. The Brush
Foundation Study that provided the sample for the GP method was conducted at Case
Western Reserve in Northern Ohio with a sample of upper socioeconomic white children.
The TW method was derived from a sample of the British population, and the Fels
m ethod used a sample from the Fels Longitudinal Study conducted in Southern Ohio with
mi ddle-income children. In a comparison of the GP and TW methods, Acheson, Vicinus.
and Fowler (1966) found that the radiographs judged using the GP were about one year
less on average than those same radiographs judged using the TW method. The
investigators also found that interobserver error was smaller with the GP method, but that
the confidence limits within a single reading were narrower with the TW method. Since
tllat study the TW method has evolved twice into what is now called the TW3 method.
More recently, the GP, TW, and Fels methods have been compared.
Researchers have sought to determine the interrelationship of the GP, TW2, and
I:els methods of SA assessment. Vignolo et al. (1992) compared the three methods in a
Sample of male and female Italian adolescents. Their ages ranged from one to 17 years of
age. Two independent investigators performed each method of assessment, and one

observer reassessed all radiographs after six months again using all three methods. Both

29

the 'TW-20 and TW—RUS methods were performed. Analysis indicated that all methods
were adequate for assessing skeletal maturity in Italian adolescents. In a more recent
study, van Lenth, Kemper, and van Mechelen (1998) used multiple investigators to
compare the TW2 and Fels methods in a group of boys and girls with ages between 12
and 16 years. This sample was derived from the Amsterdam Growth and Health Study
(AGHS) and produced four radiographs for each adolescent. The authors concluded that
each method has acceptable intra-observer agreement, but no agreement exists between
the SA ages derived from the two methods. The latter could be the result of different
investigators performing each method. The differences between assessments reported

Vi gnolo et al. (1992) were not as great which was likely due to the same observers

co liducting all three methods.

Comparison of the GP, TW, and F els methods of SA assessment reveals good
reproducibility within each method, but little agreement between methods. Consideration
mLIst be given to the reference population from which each method was derived when
Considering reference to future samples. The reference population should also be taken
into consideration when using SA to predict adult stature. Predicting adult stature from

S A is desirable in growth studies or clinically when a growth impediment is suspected
arid longitudinal data collection is impractical. All three, the GP, TW, and Fels methods
of SA assessment have corresponding methods of adult stature prediction.
The development of the three main methods of SA prediction has also yielded
methods of predicting adult stature. The GP method was the first and is the most
Commonly used method of SA assessment. It was also the first to have a corresponding

method of adult stature prediction. Bayley (1946) was the first to publish a method to

30

 

predict adult stature and it was based on tables derived from the GP standards. Bayley
and Pinneau (1952) later revised this method to what is now commonly referred to as the
Bayley-Pinneau (BP) method. This method is considered the most simple because it only
requires current stature and current GP SA. Tanner et al. (1975) reported the first of a
series of models developed to predict adult stature using the TW method of SA
assessment. This method (TW) is a prediction model based on multiple linear regression
equations. This method predicts adult stature based on three variables: present stature,
chronological age, and TW skeletal age. The three variables are then multiplied by their
corresponding correlation coefficients to predict adult stature. The latest version was
reported in conjunction with the most recent TW3 method of SA assessment (Tanner et
a1 - , 2001). Similarly, Roche, Wainer, and Thissen [RWT] (1975) developed a multiple
linear regression model, but used SA derived (from GP method of assessment, recumbent
length, and MPS. The RWT method was further improved by Khamis and Guo [RWT-
KG] (1993), and requires the present stature of the child, body weight, mid-parent stature,
atld the Fels skeletal age. The greatest disadvantage of the RWT and RWT-KG methods
is mat they require the midparent stature of the biological parents. This is problematic
because the midparent stature of the biologic parents for one or both is not always
av ailable. Missing parental statures can be replaced with national means, but this
iIlereases the individual error bound (Khamis and Roche, 1994). Since introduction of
these three methods, researchers have sought to compare predictions made by these three
methods to determine the relative accuracy of each.
The F els method is the most versatile method of adult stature prediction. but

requires many parameters in the calculation. Harris, Weinstein, Weinstein, and Poole

 

(1 980) compared predicted adult statures using the BP, TW2HP, and RWT methods.
Included in the study were the records of 22 male and 24 female adolescents fi'om the
Denver Child Research Council. Radiographs were taken from the ages of five years to
16 Years. Also collected were the statures and weights of the children and the statures of
the Parents. The investigators were also able to follow up and collect the mature stature
of 33C}: child. The investigators found that each of the prediction models underestimated
mat‘JI‘e stature. This underestimation is likely due to the fact that the investigators
m€°va-S\:rred adult statures when the subjects were in their twenties, and the prediction
models were based in subjects 18 years of age at the time of the last measurement. The
reSIJItS also demonstrated that the Fels method was the most accurate in predicting adult
S{Eitlire in both males and females. The investigators speculate that this is due to more
in f0 nnation being used in the equation. Of the three methods, the TW2 was the least
ace 1.1rate. The investigators hypothesized that these differences were due to the different
methodologies of SA assessment and the variables used in the prediction model. Since
this study, revisions have been made to both SA, assessment methods. Despite these
im11>I‘()vements, the underlying differences in reference populations, methodology, and
parallfieters exist. If simplicity is required and information is limited to the stature and SA
0f the child than the BP method is preferred. If the stature, weight, SA, and midparent
Stan-11%: of the child are available then the Fels method should be used. Nevertheless. one
diff} Qulty that these three methods have in common is the dependence on skeletal age.
Somatic maturity assessment. Assessment of secondary sexual characteristics and
Slm'le‘tal age are not always available or preferable due to their cost and the negative

Perception of undressing or exposing the children to radiation. Because of these negative

perceptions, somatic maturity indicators have been studied as a method to estimate
maturity. Somatic assessment involves measurements of the body dimensions and size.
Body- size itself is not an indicator of maturity and cannot be used directly to assess
matLlrity (Malina, Bouchard, et al., 2004). Even with this limitation, somatic parameters
can be used to identify the age-at-take-off, the age at peak height velocity, and percent of
pfedi Qted adult stature. Both, peak height velocity and percent of predicted adult stature
can be used as maturity indicators (Malina, Bouchard, et al., 2004).
All adolescents go through an adolescent growth spurt. The age-at-take-off refers
‘0 tlike age of the adolescent when they begin their growth spurt. The age-at-take-off in
feIllales occur at nine to 10 years of age and in males 10 to 11 years. The peak height
V31 Ocity refers to the maximum rate of growth of the child. This occurs about the age of
12 in girls and 16 in boys. Girls tend to stop growing in stature about the age of 16 and
boys about the age of 18. Both girls and boys can continue to grow beyond those ages.
To ascertain peak height velocity or age-at-take-off, serial measurements must be
col 1 ected over a wide range of years. This method befalls the same limitation as
see Ondary sex characteristics because it is limited to a narrow time frame during the
gro Wth period and requires serial measurements. Percent of predicted adult stature
ho""‘v’ever, offers an alternative for estimating somatic maturity and does not require serial
mea~Surements.
The percent of predicted adult stature is an expression of the child’s current
Statllre in relation to their predicted adult stature. A child who is closer to their adult
smmre is more mature than a child of the same CA and stature who is further from their

adult stature. A child who is 80% of their adult stature is more mature than a child who is

75% of their adult stature when age and current stature are equal. Predicted adult stature
can be derived from SA as described in the BP, TW3l-IP, and Fels methods. However,
less invasive methods of predicting adult stature have been developed.

Assuming that SA would be impractical to acquire in most study designs, Wainer,
Roche, and Bell (1978) developed a method of predicting adult stature without using SA.
The investigators used the same data and regression model that was used to develop the

RWT method. The investigators replaced SA with CA in the multiple regression models.
The results yielded predictions that had only a small increase in error versus those
Obtained using SA. The major weakness of this method was the use of recumbent length
bee«ause recumbent length is not always practical to measure. The KR method alleviates
the need for recumbent length, and only requires current CA, weight, stature and
Iniclparent stature. The authors compared predictions using this new method to those
118 ing the RWT-KG method, and noted only small increases in imprecision. The 90%
eI‘I‘czn bounds were 2.1cm for males and 1.7cm for females. Few studies have tested the
V211 i dity of the KR method.

The validity of the percent of predicted adult stature has been compared to
3°C elated methods of maturity estimation. Roche, Tyleshevski, and Rogers (1983)
deseribed the procedure for calculating the percent of predicted adult stature that a child
has attained. To determine the usefulness of this measure as an indicator for maturity the
auttlors correlated their predictions to known indicators of maturity. These included:
peak height velocity, RWT skeletal age, GP skeletal age, and TW skeletal age. Predicted

adult stature was calculated using both the RWT invasive method and the non-invasive

metlf‘nod described by Wainer et al. (1978). Both methods of adult stature prediction

yielded percents that were significantly correlated to all measures of maturity between the

ages of 5 and 15 years in boys and 3 and 13 years in girls. This study demonstrates the

applicability of percent of predicted adult stature as a maturity indicator in children and
adole scents within the prescribed age ranges.

In summary, there are three common methods of assessing maturity. These
indude secondary sex characteristics, skeletal maturity, and somatic maturity. Secondary
sexual characteristics typically include Tanner (1962) stages of genital, breast, testicular,
and pubic hair development. In women, menarche can be used as another indicator. The
limitation of secondary sex characteristics is the invasiveness and limited applicability to
field studies. They are also limited by the narrow timeframe around the ages of puberty.

Skeletal maturity is the best method to estimate biologic maturity because it encompasses
growth from birth to full skeletal maturity. There are three commonly used methods to
assess SA, but all require exposure to radiation and are expensive to conduct. Somatic
maturity estimates involving age-at-take-off and peak height velocity require serial
measures and are limited to a narrow timeframe similar to secondary sex characteristics.
Cal Culating the percent of predicted adult stature shows great promise as an indicator for
sort'lv'iuic maturity in adolescents. The KR method of predicting adult stature is

nor-1i hvasive and it does not require SA. It can be readily applied in field studies, and has
been shown to correlate well with the RWT-KG method. Further studies need to validate

the application of the KR method in a variety of study designs and sample populations.

Injum Incidence, Risk, and Rates in Football

Injury Surveillance programs are generally conducted to describe and mitigate

injury occurrence in a population. Injury surveillance can be applied to many different

35

settings including automobile use, drug use, gangs, recreational sports, and more
specifically football. The purpose of a surveillance program is to first describe the nature
of the problem, identify interventions, and then monitor the affect the intervention has on
the previously identified problem. Surveillance of injury in football is nearly as old as
the gaJne itself.
The history of injury surveillance in football. Concerns about injuries in football
begatl shortly after the first game was played. The first recorded game of American
fOOtb all was played between Princeton and Rutgers universities in 1869. The 1905
season was terminated mid-season after what could be considered the first injury
surveillance report was published in the Chicago Tribune (Cantu & Mueller, 2003). The
3111i Cle reported that 18 deaths and 159 serious injuries had occurred in the previous
f0()‘I:ball season. That article prompted, not only rule and equipment changes, but it also
higlulighted the need for continued surveillance and further research describing the nature
0F i njuries in football. The most serious and most studied injuries are those that include
the head and spine.
Brain and spine related injuries have always been a major concern in football.
Carl 1:11 and Mueller (2003a) recently examined brain inj ury-related deaths that have
occuhed in football in the years from 1945 to 1999. In a similar report, Cantu and
Muel ler (2003b) reported catastrophic spine injuries that occurred between 1977 and
200 1 - Both reports indicate a recent decrease in the frequency and rates of incidence in
both types of injury. Catastrophic spinal injuries have decreased by 270% in last 10 years
and deaths from football related brain injuries have decreased significantly since 1969.

Both of these trends began following rule changes and educational campaigns. The

importance of these two studies is that they highlight the need for continued injury
surve i llance, without which, these trends would not be detected and the effect of the rule
changes and educational campaigns would remain unknown.
Many of the early and some current studies in football epidemiology were and are
limite (1 to cross-sectional and registry data derived from questionnaires, hospital records,
and insurance claims. Those types of studies provide useful data regarding incidence of
injury and the burden those injuries have on the healthcare system, but they provide little
information about player specific injury data or risk factors. Registries also
underestimate the true incidence of football injury because they will only include those
inj Dries significant enough to warrant hospital visits, insurance claims, or were significant
€110 11gb to be recalled from memory. More recently, with the availability of trained
health care providers and the increased emphasis on injury definitions and exposure
documentation, the quality of epidemiological data has significantly improved.

The defining of athlete exposures (AE) was an important step in the improvement
or injury surveillance. Time of exposure has been identified as an important confounder
that Warrants consideration (Cahill and Griffith, 1978; Dagiau, Dillman, and Milner,
198 0). Powell (1980) defined AE as any opportunity for an athlete to become injured in
a C0 ach directed session. Each player present at each practice or competition was
Courlted as one exposure. Powell’s research was conducted as part of the National
Athl é‘tic Injury/Illness Reporting System (NAIRS) during the 1975 to 1978 seasons. The

NAIRS study can be considered the beginning of modern sports epidemiology. The
defillitions of injury and AE used in NAIRS continue to be used. The study reported by

Garl‘i ck and Requa (1978) and the NAIRS study were the first to use trained healthcare

37

providers in the form of Certified Athletic Trainers (ATC) to document injuries and
exposure information. The ATC reported injury data were a significant improvement in
quality and quantity of information. The NAIRS study was also one of the first to

calcul ate injury rates that were based on the total AB.

The risk of injury increases with age. It is unknown if this is a function of

intrinS ic factors related to aging or if it is related to the increased level of competition as

Competitors progress through the ranks. Because of this relationship, college, high school,
and Youth football players should be considered separately. Maturity should be
considered in any study that includes youth football players younger and older than the
average age of puberty (10-12 years-of- age for males). For this reason, this review
fOcLised on college, high school, and youth football injury incidence separately with
erI‘113hasis placed on junior high school and youth players due to their proximity to the age
OF puberty.

College football injury incidence, risk, and rates. Injury incidence, risk, and rates
have been thoroughly reported at the college level of competition. This is due to the
avai 1 ability of data collectors and the cooperative nature of university medical
professionals in the production of research. Examining college football players as part of
the NAIRS study, Powell (1980) reported injury rates per 1000 AE. Injuries were
defined as any incident that required cessation of activities the following day or
subsequent days following the event, concussions, dental injuries, or any injury requiring
SUb S1lantive medical attention. The injury rates were 7.0 and 63.0 per 1000 AE for

prae"lices and games respectively. These rates were based on over 1.4 million AB, seven

percent of which were attributed to game exposures. Sixty percent of the injuries

reported occurred to the lower extremity and 20% occurred to the upper extremity. The

knee was the most commonly injured body region (20.5%). More recently, Powell and

Dornp ier (2004) reported injury risk and rates that were similar, but also included a

definition of injury that included a non-time loss component. Powell and Dompier found

that the previous NAIRS definition accounted for about 25% of the total injury picture in
football, College level data provides a good reference point, but because the vast
malot‘ity of college age athletes have stopped linear growth, further review was limited to

high School, junior high school, and youth football studies.

High school football injury incidence, risk, and rates. The injury risk and rates of
high school football players have been extensively described, but maturity is rarely
considered. Maturity should be considered in this demographic because most males
begin puberty between the ages of 10-12 and complete their adolescent growth spurt at
the age of 16 on average (Malina, Bouchard, et al., 2004). Few studies involving high
sch (30] age football players consider maturity as a possible confounder. Maturity has
been difiicult to assess accurately and logistically, and has therefore, not been included in
many studies involving football injury. Because maturity is rarely considered, injury risk
for those who are less mature may not be adequately represented. Players who are larger,
stron ger, and more mature receive the majority of playing time in football.

The majority of available research regarding high school football is descriptive in
natLJJI‘e. Prior to the NAIRS study, Bylth and Mueller (1974) sought to describe injury in
high school football players in North Carolina. Injury was defined as any event occurring

durin g football that restricts participation one day following the day that the injury

0cQ'llrs. Player demographics were collected by interview at the beginning of the seasons

39

and inj uries were reported to the investigators on a weekly basis by the coaches. The risk
(number of injuries divided by number of players) was 48%. Players who were older,
heavier, taller, and had more experience had higher risk of injury. The greatest limiting
factor to this early study was the reporting of injuries by coaches and lack of exposure
infortrntion.
The study of injury in youth and high school football was greatly improved with
the in clusion of ATCs as data collectors. Garrick and Requa (1978) conducted one of the
early studies utilizing ATCs as the primary injury data collectors. Injury was defined as
any event that required removal from practice or caused the absence of subsequent
seSSions. There were 506 injuries reported for an injury rate of 81 injuries per 100
P1 ayers. That injury rate was nearly twice the rate reported by Blyth and Mueller (1974)
and is likely due to the increased sensitivity of the injury definition and trained data
rttccnders. Other studies have further described injury incidence in high school football
P1 a yers and have reported similar results (Beachy, Akau: Martinson and Olderi', 1997;
Cu 1 pepper and Niemann, 1983; Delee and Famey, 1992; Powell and Barber-Foss; 1999).
The inclusion of AE further improved the analysis of injury data. The bulk of
r ese arch involving high school football players is limited to small demographic areas and
cmltlot be widely generalized. Powell and Barber-Foss (1999) conducted one of the
lal'gést and most comprehensive studies involving high school football players. There
were 246 high schools included in the study, and athletic trainers collected data. The
sam ple included a normal distribution of schools with small to large enrollments. The
. inj l1I‘y definition included that previously described by Powell (1980). Exposures were

dag Sified as team-seasons, player-seasons, and AE. There were 400 team-seasons, 6831

40

p1ayer-seaSons, and 1.3 million athlete exposures reported. There were 10557 reported
football injuries amongst 7310 injured players. The player rate of injury was 34.6, the
case rate was 50.0, and the injury rate per 1000 AE was 8.1. Practices accounted for 56.4
percent of the injuries. The case rate per 1000 AE for practices was 5.3, but for games it
was 26.4. This indicates that players were injured fives times more often in games than
in practices. Powell and Barber-Foss further described injury as minor, moderate and
major. These were based on the number of days lost from participation due to injury.
Minor injuries were those that lost less than eight days, moderate were between eight and
21 days, and severe were greater than 21 days. Minor injuries accounted for 73% of the
injuries while moderate and major accounted for 16% and 11% of the injuries
respectively. These data provide a good overview of injury to high school football
players but to more accurately understand the influence of maturity on injury,
examination of injury in youth football players is needed.

Youth football injury incidence, risk, and rates. Incidence of injury increases with
each succeeding level of football from youth to college. For the purposes of this
dissertation, youth football players were considered those in primary and junior high
school, or those below the ages of 16 years. It is unknown if the age affect is confounded
or mediated by maturity. The injury incidence, risk, and rates of high school, junior high
school, and youth footballers are summarized in Table 1.

The differences between high school and younger players are apparent. The first
study to focus on junior high players found a risk of 37% (Violette, 1976) as compared to
the nearly 49% (Bylth and Mueller, 1974) reported for high school players using the

same study model and cohort. Other studies have reported high school injury risks that

41

range between 18.4% (Turbeville et al., 2003b) and 81% (Garrick and Requa, 1978).
Comparatively, Stuart et al. (2002) reported risk as low as 3% in fourth graders, but
Radelet et al. (2002) reported injury risk in youth as high as 51%. Injury rates follow a
similar pattern with the lowest reported by Radelet et al. (0.43 per 1000 AE) for the
youngest age groups while Turbeville et al. (2003a) reported the highest for middle
school players (9.9 per 1000 AE). Comparatively, Powell and Barber-Foss (1999)
reported an average injury rate of 8.1 per 1000 AE for high school players and Powell
and Dompier (2004) reported an injury rate of 40 per 1000 AE for college players. This
age effect is further demonstrated within subgroups at the youth level.

An increasing risk of injury is seen in youth football players within succeeding
groups. Stuart et al. (2002) reported that 3% of the fourth graders were injured while
11% of the eighth graders were injured. Malina et a1. (2002) found similar proportions
with the fourth grades having the lowest and the eight graders having the highest rates.
Turbeville et al. (2003a) found a 10%,incidence in middle school players, and 18% in
high school players (Turbeville et al., 2003b). Combined injury rates have ranged from
0.43 (games only) (Radelet et al., 2002) to 8.84 (Turbeville et al., 2003a), with the
highest at 10.4 (Malina et al., 2002) per 1000 AE. Turbeville et al. (2003a, b) directly
compared participants from middle school and from high school. Injury rates per 1000
A13 for high school players were 1.31 for practices, 13.12 for games, and 3.20 overall. In
middle school players, the injury rates per 1000 AE were 0.97 for practices and 8.84 for
games. It is unknown if this age affect is confounded or mediated by maturity or other

factors.

42

Descriptive epidemiological studies have described injury risk in high school and
youth football. From these data, testable hypotheses regarding risk factors for injury can
be derived. Few studies have investigated risk factors for injury in football and even less
have examined risk factors in youth players. The identification of risk factors is
important for two main reasons. First, identification of injurious conditions, rules, or
equipment aids decision makers in the development of policies, rules, and equipment
changes. As in all analytic epidemiology, special care must be taken to identify and
control possible confounders. In youth football, one such confounder rarely considered is
maturity status.

Risk F actors for Injury in Football

Few studies have thoroughly examined both extrinsic and intrinsic variables and
their relationship to injury. Extrinsic variables are those that are not directly affecting the
subject, and changing the variable in some way does not-require modificationof the
athlete. Intrinsic or player related variables are those that are directly related to the
player. These would include exposure time, stature, weight, fitness level, age,
psychological profile, and maturity to name a few. Most player-related variables cannot
be changed, but some can be modified. Age, maturity status, or stature for example,
cannot be changed, but factors such as their weight and fitness level can be changed.
Intrinsic variables require player level analysis that is labor intensive and requires diligent
assessment and frequent follow up. Only recently, have there been studies that have
considered these variables in football. but even less have considered them in youth

football.

8:898 S 3336 3:35 83:52 a
$933 he 35:5: :38 2: .3 36233 8:33 do 83:52 ...

when» A. $3 32580 n
23:..an 32 u .32

85898 25% 32:2 u O<

35% 23 8282a :53 .8 $5398 2256. n m<

 

 

mz .32 9. we 2333:: E 68: 2333

m? 82 an 83 3.3.2 3: :5 P? m: 383 .3 3 nasefi

9482 33 3 3.3.3 E 93 P? E 3893 .3 3 3:33.33.

o< 82 an 3 .33 mm ma 333.93 33> £88 .3 3 333m

m< 82 33 32 3:. am. N2 386 53> 38$ .3 3 333m

m< 82 33 S .33 R2: me a 6.2 m: 38: 38,3338 3 330,.

«2 .32 mm 2% 335 m5 3%: .33 33.:

32 .3; 8m :3 u? m: 38: 9333 3 3:30

93 e: 82 3a m .3: $2 an. 6.2 m: 323: 35$ 3 33o

mz .333 a? at” .338 m: 2.3: .3 3 ii

32 .335 83 $2 a: a: 3 m: :3: .3 3 223m

23m ,0??— w. 8522:... more? magma mousom wcgom 83%:ng
Ema

 

.238: 233$» See» 323 £me 3.55. .Neeaem «NE E 323% 3:3 33% 36233.8: @335? «92.52%.

_ 033,—.

44

Exposure time. An early study by Dagiau et al. (1980) demonstrated the
importance of exposure time to injury rates in football. More specifically, the authors
wanted to test if there was an optimal time that a player should be exposed to a specific
practice or game condition. The investigators followed the University of Illinois varsity
football team for two seasons. Injury risk decreased with an increased number of plays.
For practices however, there was an increased risk of injury with increased time of
exposure. Cahill and Griffith (1979) conducted a more comprehensive study of
exposure, injury, and other variables in football. This study was important because it
identified exposure time as a risk factor for injury in football, but as the authors indicate,
there is likely a systematic bias that affected player exposure time. If a player does not
sustain an injury, he will continue to participate in more plays during the game where an
injured player will have systematically less because he was removed.

Exposures'can be measured at three progressive levels of specificity. The most
general form, or sport level, would be to simply multiply the number of players on the
team roster by the number of games and practices. This provides a crude exposure
figure, but over estimates the denominator that in turn leads to an underestimation of
injury rate. The underestimation occurs because players who do not participate in a
practice or play in a game for any reason are still counted as an exposure. The second
level of exposure calculation is at the team level. This method tallies daily exposure as
number of participants present at each session or the number that participate in each
game. If a player is not present for any type of session or do not play in a game, they are
not counted. This is more sensitive than the first and is likely to produce higher rates

because the denominator will be smaller. The third level of exposure analysis is at the

individual player level. These exposures are expressed as either player hours of exposure
or the number of repetitions of exposure. Measuring the player level of exposure is time
intensive and difficult to achieve, but will control the exposure bias described by Cahill
and Gri :ffith (1979).

[laying surface and shoe interface. The shoe surface interface was one of the
first extrinsic variables to be examined and demonstrated to be a risk factor in football.
One of the first studies to report this relationship was that of Torg and Quedenfeld
(1 971} Their hypothesis was that the traditional seven by 3-quarter inch cleat shoe was

respons i ble for a disproportionate amount of knee injuries. During the first year of the
study, 3.1 l athletes wore the traditional shoe, but in subsequent years, one league wore a
soccer S‘Lj'le shoe and another league wore a short molded 14-cleat shoe. The number of
knee inj IJries significantly decreased in both leagues during the two intervention seasons.
Similar 13, Blyth, Mueller and Frederic (1974) reported significantreduction (30.5%) of
knee and ankle injuries when playing surfaces and a 48% reduction when provided with
both the resurfaced field and soccer shoes.

Artificial turf has been implicated as a risk factor for injury in football because of
higher levels of friction between the shoe and surface when on it. Bramwell, Requa, and
Garrick (1972) conducted a study to determine if grass or artificial playing surfaces were
risk fr=1ctors for injury during football games. Also included in the analysis were the
sulfate, e conditions (wet or dry) during each event. Bramwell et a1. (1972) found that
injury rates were higher on artificial turf, but injury rates were lower on both artificial and
natural surfaces when they were wet. Powell and Schootman (1993) provided further

S . . . . .
upI)Qrt that artificial surfaces are related to injury. Ankle injury rates on artrficral

46

surfaces were significantly higher than on natural grass. Multivariate analysis revealed
that thi 5 difference was only present in specific combinations of player position, play
type, and type of surface. Orchard and Powell (2003) found similar finding, but noted
that lower ambient temperatures also contributed to lower injury rates. The investigators
hypoth esize that this is due to the reduced shoe-surface friction.

The associations noted in the studies above may have been affected by league rule
changes- An important point regarding the timeframe of some of these studies was that in
1974 National Football League [NFL] (2004) began making drastic rule changes in an

attempt to reduce the severity and incidence of lower extremity injuries. These included
moving the goal posts to the back of the end zone, eliminating roll-blocking, cut blocking
0f Wide receivers, and wide receivers were no longer allowed to block below the waist
(NF L, 2 (304). The infamous crackback block was not outlawed in the NFL until 1979,
but it i S unclear to what extent these concerns or rule changes were in focus at any level
0f football prior to the NFL rule changes.

The data support that injury risk is associated with the shoe and playing surface
interface. This has been accepted to such an extent that specialized shoes and even new
types Of artificial turf have been developed to mimic natural playing surface
Characteristics. However, the many risk factors were not controlled in the early studies
and 0my the studies by Powell and Schootman (1993) and Orchard and Powell (2003)
attempted to systematically control for other factors that may have confounded earlier
studié 8.

Fitness level. Fitness level is one of the most studied intrinsic variables associated

wlth football. Cahill and Griffith (1978) reported the effects of an intervention consisting

47

of a six—week preseason-conditioning program on knee injuries. The authors found a
significant reduction in knee injuries during the intervention period. The greatest injury
reducti on occurred among the linemen (61%) followed by the backs (20%). Gomez et al.
(1998 ) examined the relationship of body fatness and lower extremity injury rates in
junior high and high school linemen. Gomez et al. found that lower extremity injury risk
increas ed as BMI increased. The most comprehensive and well-controlled study of
player fitness to date was that of Turbeville et al. (2003a, b). Player variables measured
included : experience, position, injury history, BMI, weight, stature, and grip strength.
Logisti C regression was used to determine odds ratios associated with each parameter.
Injured players were on average older, had a higher BMI, were stronger, had more
CXPCI’iel’l ce, had a history of previous injury, and used optional equipment. When only
those cOmsidered were first-string players (controlled exposure), the only parameters that
remained significant were BMI, grip strength, years of experience, and injury history.
When eXposure was uncontrolled, injury history and experience demonstrated the
gr eate St risk of injury. When exposure was controlled, the lineman position and
experi ence remained significant.

Psychological variables. Thompson and Morris (1994) took a different approach
and elaiarnined the relationship of injury with psychological variables. Analyses of the
relatiQnship of stressful life events, anger, and attention were examined in 120 high
SChOQ 1 football players. Players were followed for the duration of the 1987 football
SeaS‘Q In. The psychological instruments were administered at the beginning of the season.
LogiS‘tic regression revealed that players with high levels of anger directed outward, those

th elevated recent stress, and those With low focused attention were at the greatest risk

48

of inj 1.117. The authors of this study propose that children who are distracted by stressful
events in their life or are angry are less focused on playing the game and are therefore at
greater risk.

Player position and game situations. Specific player positions are at greater risk
for in j ury than others. Blyth et al. (1974) demonstrated that the halfback position was the
most commonly injured in high school football players. Powell and Schootman (1993)
SUppor-ted the findings of Blyth et al. by comparing NFL ankle injury rates with the type
0f surface, player position, and type of play ran by the offense. Multivariate analysis
revealed that injury was related to specific combinations of player position, play type,
and type of surface. Powell and Schootman also found that player position also dictates
WhiCh itI‘juries specific player positions are most susceptible. As indicated by these
StUdieS a jalayer position and type of play contribute to the variability of the injury model.

Maturity status. Few studies have considered maturity as a risk factor or
confounding variable in the study of youth football injury, and those that have are
inconsistent. As part of the North Carolina Football Study, Violette (1976) was one of
the first to attempt to answer this question by examining football injuries in junior high
SChOO 1 players. Using the same design as the original study, Violette used secondary
sexual characteristics to group junior high school football players by maturity status.
Violette found that those who were less mature were at significantly greater risk of injury
than t‘tieir more mature teammates. Comparatively, Linder et al. (1995) found that those
Who V\>'vere most sexually mature were more likely to be injured. The most recent study by

Malina et al. (2002) found no relation between maturity status and injury when PPAS

49

was used to group players by maturity status. Many differences exist between all three
studies comri buting to the lack of consistency.

“R three studies that examined maturity as a risk factor for injury in youth
football Players had different definitions of exposure, but used similar definitions of
injury. Linder et al. (1995) did not consider exposure, but Violette (1976) used a sport
level of exposure and Malina et al. (2002) used a team level of exposure. Injury
definitions were similar and included any football related injury that required removal

from the current or subsequent sessions. The method by which injuries were
documented was the most significant difference between the three studies. Violette used
Coaches to report injuries to an investigator who would then complete an interview.
Linder et al. asked coaches to report injuries, and Malina et al. had ATCs onsite
couccting, evaluating, and documenting injuries. It is difficult to determine if the method
used by Violette or Linder et al. was more sensitive, but the inclusion of ATCs onsite
would definitively make the method of data collection used by Malina et al. the most
561151. ti Ve. Increased sensitively would cause a greater number of injuries to be
k%mnented. Analysis of data also varied, and the analyses by Malina et al. were the

only to include multivariate logistic regression.

Multivariate analysis of injury and the suspected associated variables is needed to
rule out confounding variables. Malina et al. (2002) used logistic regression to examine
injury and maturity in context of other suspected risk factors. Risk factors included
previous injury, stature, weight, chronological age, previous sport experience, previous
football experience, and psychological variables. The only significant factors related to

injury reported by Malina et al. were stature and previous injury.

50

Injury in football is multifactorial. Both extrinsic and intrinsic factors contribute
to the injury model. Extrinsic factors include the sport played, level of play, weather,
player position, playing surface. surface condition, equipment, and rules. Intrinsic factors
include age, sex, stature, weight, BMI, fitness, psychological status, and injury history.
Modifiable extrinsic factors include rules, exposure time, surface condition or type, and
equipment. Modifiable intrinsic factors include fitness level, weight, BMI, and some
psychological parameters. In children, maturity status, even though it is not modifiable,
should be considered a confounder. No studies to date have examined all potential risk
factors simultaneously. To do so would be difficult, and not all risk factors are relevant
to every research question or situation.

Summary of the Literature

The main purpose of this review was to examine whether maturity status should
have any bearing on sport classification in youth football. To answer that question, two
areas of research were reviewed. First, applicable measures of maturity status were
compared and contrasted. These included sexual maturity status, skeletal maturity status,
and somatic maturity status. Assessment of each form of maturity status has strengths
and weaknesses based on the study characteristics and limitations. Skeletal age is the
single best estimate of maturity status, but is costly and exposes children to radiation.
Secondary sexual characteristics are the most widely used method of estimating maturity
status, but this method potentially limits participant willingness to be included because of
the embarrassing nature of the examination. Somatic maturity estimation using the

percent of predicted adult stature, although relatively untested in youth football players,

51

shows the most promise for field studies where invasive methods are not applicable or
desirable.

The second stage of this review examined injury incidence and risk factors for
injury in football with special emphasis on studies involving high school age players, or
when available, youth players. Injury mechanisms in football are multifactorial and
involve the interaction of numerous risk factors. Risk factors include both controllable
and uncontrollable extrinsic and intrinsic variables. Maturity status is an uncontrollable
risk factor that is of particular interest to the current study. This factor has also been the
least studied in the youth football population. Maturity status must be determined or
refitted to be a risk factor for injury before recommendations can be made regarding sport

classification using maturity status as a factor.

52

Chapter Three

Methods

A child’s level of maturity may be a risk factor for injury. If so, competition
levels should be arranged according to maturity status versus age or grade level. Few
studies have systematically studied risk factors for injury in football and even fewer have
considered maturity in the injury model. The purpose of this study was to determine the
validity of the Khamis and Roche [KR](l 994) method for predicting adult stature (used
to determine maturity status) in youth football players, and to analyze maturity status in
univariate and multivariate models of injury risk.

Overview

The current study was a 22-year subset from a 4-year observational cohort of youth
football players that began at the start of the 2000 football season and continued through
2003. The data used in the current study were obtained during the 2002 and 2003
seasons. The original intent of the study was to examine suspected risk factors for injury
in youth football players. Independent variables such as maturity, anthropometric
measures, injury history, participation history, psychological variables, player position.
and surface conditions were included. The overall study has used percent of predicted
adult statures derived from adult stature predictions using the KR method.

Research Design

The study consisted of two separate designs. The first consisted of a cross-

sectional design, and the second consisted of an observational cohort design. The

convenience samme used in the first leg of the study was a subset of the larger cohort

53

observed as part of the second leg of the study. The Specific research questions and null

hypotheses were:

1. Is the Khamis and Roche (1994) method a valid estimator of predicted adult
stature?

H01: No linear relationship exists between the predicted adult statures derived
from the Khamis and Roche (1994) and Khamis and Guo (1993) methods.

2. Is Khamis and Roche (1994) method of predicted adult stature a valid measure
of maturity when expressed as a percentage of the predicted adult stature?
H02: No linear relationship exists between percent of predicted adult stature
derived with the Khamis and Roche (1994) method of adult stature prediction
and skeletal age.

3. Is maturity a risk factor for injury in youth football players?

H03: No relationship exists between maturity status and injury.

Subjects

The subjects consisted of youth football players in grades 4m-8th from two
communities that participated in the Mid-Michigan Pony Football League in south-
central Michigan during the 2002 and 2003 seasons. By league rules, 4th and 5th graders
were grouped together on the same teams and were therefore considered as one group (4-
5th) during data collection and analysis. The subjects in this study were a convenience
sample. The criteria for inclusion were registration in the youth football league and

informed consent from both the parents and participants (Appendices A, B).

54

An additional convenience sample of 64 children volunteered to participate in the
validity component of the study. An equal distribution across grade levels was sought.
Additional criteria for inclusion in this component included no history of fracture to the
lefi upper extremity, the completion of an additional informed consent (Appendices C,
D), and no medical conditions that would preclude radiographic examination.
Instrumentation

Player demographic information and previous sport experience, previous injury,
and parental statures were obtained from surveys distributed with the parental informed
consents (Appendix E). Player stature was measured to the nearest 0.1 cm using a field
anthropometer (GPM Anthropological Instruments). Weight was measured to the nearest
0.2 kg using a digital scale (Taylor Precision Products LP). Stature and weight measures
were taken following the procedures outlined by Malina et al (2002). Previously reported
standard error of measure using the same procedure was 0.22 cm (Malina et al. 2002).
The standard error is within the range of measurement variability in surveys of children
(Malina and Bielicki, 1996). The participants in the validation study also had their
statures and weights taken at the time that the radiograph was taken as required by
procedures outlined by Roche et al. (1988).

Certified athletic trainers served as data collectors and documented injuries and
exposures. A standardized reporting form (Appendix F) was used to maintain
consistency between the two communities. Daily exposures were tallied by counting the
number of participants present at each session and confirmed with the count reported by
each coach. Data recorders listed exposures by town, grade, and type of session (game or

practice). A licensed and experienced radiological technician took the radiographs during

55

scheduled dates and at the convenience of the participants. The F els SA with standard
error of measure (SE) was determined using the Fels software (FELShw version 1.0).
Procedures

This study had two distinct stages. The first stage consisted of the validation of
the PPAS as an estimate of maturity in a group of youth football players. The second
stage consisted of the injury risk analysis. Informed consent was obtained prior to all
data collection from the parents and children using procedures outline by the University
Committee on Research Involving Human Subjects (UCRIHS).

Stage one, validation of percent of predicted adult stature. Volunteers for
participation in the validation study were sought at the time of equipment handout.
Those who volunteered for the validation study were asked to provide contact
information and were informed that they would be contacted at a later time. Originally,
random selection of volunteers for each age group was planned, but due to an insufficient
number of volunteers, the final group of volunteers was a convenience sample. Some
volunteers who were contacted chose to not participate which resulted in all volunteers
being contacted negating any random selection. There were 78 parents and children who
volunteered, of those, 64 agreed to participate and completed the x-ray. Investigators
were present at the radiology center at all scheduled times and obtained informed
consent, the parents’ reported statures, each child’s current stature, current weight, and
DOB.

The specific guidelines for radiographic analysis outlined by Roche et al. (1988)
were followed. The specific guidelines included a posteroanterior view of the left hand

that includes 3 cm of the distal radius and ulna. The forearm, palm, and fingers were in

contact with the cassette. The fingers were fully extended with the 3rd inline with the
forearm, and the distance of the central tube-to~film was 91.4 cm (36 inches). Lastly, the
central ray was directed at a right angle to the distal end of metacarpal three.

An expert with years of experience estimating SA using the Fels method
examined the radiographs. The F els method uses a set of criteria as maturity indicators
that are based primarily on a variety of shape changes and ratios derived from several
linear measurements of long bones (Roche et al., 1988). Measurements are made to 0.5
cm. Grades are then assigned to each indicator, and are then entered into the FELShw 1.0
computer program (Roche et al., 1988). The software program then produces a SA with
SE. The SA can then be used to predict adult stature following the RWT-KG method.

Prediction of adult stature using SA was performed using the RWT-KG method
(Khamis and Guo, 1993). The parameters included in the RWT-KG method are the
child’s current stature, weight, SA, andthe MP8 of the biological parents. These
variables were entered into a regression equation and factored with age specific
coefficients described by Khamis and Guo. Tables provided by Khamis and Guo list
coefficients for all chronolOgical ages from three to 17.5 years. The corresponding CA in
the table indicates which coefficients to use in the equation. The equation is as follows,

where B is the coefficient:

RWT-KG PAS = [30 + (BStature * stature) - (Bwe.gm * weight) + (BMPS * MP3) - (BSA * SA)

Prediction of adult stature using the KR non-invasive method is similar (Khamis

and Roche, 1994) to the RWT-KG method. The variables used in this method include

57

current CA, current stature, current weight, and MPS. The non-invasive KR method has
been shown to have only a slight increase in the 90% error bounds when compared to the
invasive RWT-KG method. The 90% error bounds for males are 1.8 inches and 2.1
inches for the RWT-KG and KR methods respectively (Khamis and Roche, 1994). For
females, the 90% error bounds are tighter (RWT-KG = 1.5 inches; KR = 1.7 inches). The

equation for the KR method, where B is the coefficient, follows:

KR PAS = B0 + (BStature * Stature) + (Bweight * weight) + (BMPS * MP8)

It was impractical to obtain measurements from all parents in the current study so
reported parental statures were obtained. The parent self-reported statures were corrected
for over-estimation. Wing, Epstein, and Neff (1980) have shown that adults over-
estimate stature by an average 1.7’inches. Himes and Roche (1982) and Himes and
F aricy (2001) have shown reported statures to be useful proxies when measured parental
statures are not available. The correction used in the current study is the same method
reported by Epstein, Valoski, Kalarchian, and McCurley (1995) and later used by

Roemmich et al. (1996).

Corrected MPSma.e, = 2.316 + (0.955 * stature/inch)

Corrected MPSfema.es= 2.803 + (0.953 * stature/inch)

The predicted adult stature is used to calculate the PPAS. The 'children’s PPAS
were calculated by dividing the children’s current stature by their PAS. Statistical

analyses of the stage one data included calculation of partial correlations between the SA

58

and PPAS for both the total sample and within each grade level while controlling for CA.
Additionally, a two-tailed t-test was used to compare the means of PAS using both the
RWT-KG and KR methods. The Statistical Package for Social Sciences (SPSS 10.0) was
used to perform all statistical procedures. This KR method of determining PPAS was
also used to estimate maturity in the injury analysis stage of the study.

Stage two, injury analysis. The second stage of this study involved the
continuation of the original injury surveillance model that began in August 2000. An
ATC was present at every practice and every home game to document injuries (Appendix
F), athlete exposures, and to provide first aid and basic athletic training services when
appropriate. Detailed information regarding specific injury variables was collected for
each injury (Appendix F). The coaches of teams that had away games were queried at
the first practice of the following week regarding injuries and exposures that occurred
during the away game. All reported injuries were followed up with the player and or
parents for accuracy and completeness. Athlete exposures were counted on a daily basis
by the ATC and each player present at each session was counted as one exposure (team
level).

Injury was defined as any incident that required ATC evaluation and assessment.
or those that were reported by coaches during away games. Injury was further classified
as time-loss (TL) or non-time-loss (NTL) based on whether or not the athlete was
returned to sport. If the athlete was evaluated and returned to participation the same day,
the injury was classified as NTL. If the athlete was removed from participation that day,
missed subsequent sessions, or sought medical attention then the injury was classified as

TL. This definition is more sensitive than that reported by Powell and Dompier (2004),

59

but was necessary to accommodate the sporadic practice schedules of many of the teams
where daily follow-up was impossible.

Injury analysis included descriptive and analytic procedures. Using the injury and
exposure incidence data, injury risk and injury rates were calculated for each age group.
Univariate analyses for specific intrinsic variables was performed with the data as a
whole and stratified by grade. Backwards-stepwise logistic regression was used to
control for all significant variables simultaneously. Odds ratios and 95% confidence
intervals (CI) were reported for the final model (Motulsky, 1995). The Statistical

Package for Social Sciences (SPSS 10.0) was used to perform all statistical procedures.

60

Chapter Four

Results

The purpose of this study was to determine the validity of the Khamis and Roche
[KR](1994) method for predicting adult stature as an estimator of maturity in youth
football players, and to analyze maturity as a risk factor for injury using epidemiologic
methods of analysis. Two distinct stages of data collection and analysis were performed.
Stage One: The Validity of Percent of Predicted Adult Stature

Stage one was conducted to determine the validity of percent of predicted adult
stature (PPAS) as an estimate of maturity. Stage one data collection consisted of
collecting x-rays, current stature, current weight, date of birth, and the midparent stature
(MPS) from a convenience sample of youth football players during the 2003 season. The
stage one data were used to calculate adult statures using the KR and Khamis and Guo
[RWT-KG](1994) methods. Current statures were then divided by adult stature estimates
to produce the percent of predicted adult stature (PAS). Inferential statistics included
partial correlations correcting for chronological age (CA), and two-tailed t-tests. The
specific stage one research questions and null hypotheses included:

Q1: Is the KR method a valid estimator of predicted adult stature?

H01: No linear relationship exists between the predicted adult statures derived

from the KR and RWT methods.

Q2: Is KR method of predicted adult stature a valid measure of maturity when

expressed as a percentage of the predicted adult stature?

61

H02: No linear relationship exists between the percent of predicted adult stature

derived with the KR method of adult stature prediction and skeletal age.

Participant demographic data. A total of 64 (85% of the target) youth football
players participated in stage one of this study. The means for chronological age, skeletal
age, stature, and weight are presented in Table 2. This sample represents 16% of the
2003 season population from each age group with the exception of the 4-5th graders
(13%). The sample proportions were 31%, 28%, 25%, and 16% for 4-5‘“ 6‘“, 7‘“, and 81h
grades respectively. For the purpose of comparison, the mean difference between CA
and SA was calculated by subtracting each subject’s CA from their SA. The mean
difference between SA and CA was 0.7 years with the 8‘h grade having the highest mean
difference (1.4) while the 4-5th graders had the smallest difference (0.2). The skeletal
ages were higher than the chronological ages for all grades indicating that the sample is
on average advanced in skeletal maturity. The percent of adult stature calculated using
SA (RWT-KG) is higher than the KR percent adult statures for all grades except 4-5‘h as

shown in Table 3.

Table 2

Grade Specific Mean Ages and Physical C haracteristicsfor 64 Youth Football Players

 

 

CA (yrs) SA (yrs) Stature (cm) Weight (kg)
Variable n M so M so M so M so
46m 20 10.50 0.72 10.74 1.55 144.35 7.94 45.56 12.83
6th 18 12.03 0.45 13.08 1.24 156.17 8.46 54.96 14.49
7tln 16 12.81 0.78 13.44 1.51 157.39 8.31 60.55 24.63
8h 10 13.93 0.16 15.36 1.28 169.75 8.13 69.52 15.78
AllGrades 64 12.05 1.36 12.80 2.10 154.90 11.69 55.69 18.85

 

Partial correlations and t-tests. Statistical analysis consisted of partial
correlations and two-tailed t-tests. Partial correlations were used to control the
covariance associated with CA. Percents of predicted adult statures were moderately,
but significantly related to SA (partial r, adjusted for CA, = 0.54; p < .001). Grade
specific partial correlations for percents of predicted adult statures and SA are presented
in Table 4. The grade specific partial correlations that were significant were limited to
the 7th graders (partial r, adjusted for CA, = 0.78; p < .001), and the 5th graders (partial r,

adjusted for CA, = 0.57; p < .05).

Table 3

Mean Predicted Adult Statures and Percents of Adult Statures for 64 Youth Football
Players by Grade

 

 

 

Predicted Adult Stature Percent of Adult Stature

KR RWT-KG KR RWT-KG

Variable N M(cm) SD M(cm) SD M% SD M% SD
45th 20 178.18 5.90 178.77 5.71 “ 80.99 3.05 80.73 3.24
6th 18 183.35 7.18 181.34 6.12 85.17 2.82 86.09 2.75
7th 16 177.97 6.35 177.56 5.19 88.45 3.83 88.66 4.37
8th 10 181.29 7.00 178.83 5.88 93.63 2.53 94.91 2.95
All Grades 64 180.07 6.83 179.20 5.77 86.01 5.33 86.44 5.83

 

The RWT-KG and KR methods of adult stature prediction were strongly related
(partial r, adjusted for CA, = 0.88; p < .001) Table 4. Partial correlation also revealed
that percents derived from the KR and RWT methods were strongly related (partial r,

adjusted for CA, = 0.85; p < .001).

Table 4

Partial Correlations Controlled for Chronological Age by Grade

 

 

 

 

Maturity Predicted Stature Percent Stature
Group N SA / KR KR / RWT-KG KR / RWT-KG
46th 20 057* 0.90‘ 085“
6th 18 0.47 0.87“ 0.831
7th 16 0.781 096* 0.90*
8h 10 0.66 0.94T 0.86“
All Grades 64 0.541 0.88’r 0.85T

 

*p<.05 **p<.01 1‘p<.001

A two-tailed t-test was used to compare the sample means derived from the KR
and RWT-KG methods of adult stature prediction. There was a significant effect for
PAS, t(63) = 2.29, p < .05. Two-tailed t-test results are reported for adult statures by
grade in Table 5. All group differences for adult statures were significantly different
except for the 4-5th (PAS, t(19) = -l .27,p > .05) and 7‘11 grades (PAS, t(15) = 0.59,p >
.05). Similar results were found when the sample means of the percents of adult statures

were compared using a two-tailed t-test (Table 6). There was a significant effect for

PPAS, t(63) = -2.35,p < .05.

Table 5

Paired T - Test Results ior Stature Prediction Methods by Grade

 

 

M 95%c1

Group df difference SD Lower Upper t p
45th 19 -059 2.09 -1.57 0.39 -1.27 >.05
6th 17 2.00 3.36 0.34 3.68 2.54 < .05
7‘h 15 0.41 2.79 -1.08 1.90 0.59 >05
8th 9 2.45 3.19 0.17 4.74 2.43 < .05
All Grades 63 0.86 3.03 0.11 1.62 2.29 < .05

 

64

Table 6

Paired T-Test Results for Percent of Predicted Adult Stature Methods by Grade

 

 

M 95% CI
Group df difference SD Lower Upper t p
45th 19 0.27 0.98 -0.19 0.72 1.22 >.05
6th 17 -092 1.55 -1.70 -0.15 -2.52 <.05
7th 15 -021 1.37 -094 0.52 -0.62 >.05
8h 9 -1.28 1.68 -2.48 -0.08 -2.41 <.05
All Grades 63 -043 1.46 -079 -0.06 -235 <.05

 

Stage Two Results.“ Injury Analysis

Stage two of the study consisted of injury analysis and sought to describe and
examine intrinsic player risk factors. Analysis of risk factors included both univariate
comparisons of relative risk and odds ratios derived from backwards-stepwise logistic

regression. The specific research question and null hypothesis addressed was:

Q3: 1s maturity a risk factor for injury in youth football players?

H03: No relationship exists between maturity status and injury.

Descriptive epidemiology. During the 2-year study period, there were a total of
779 youth football players in grades 4 through 8 who participated in stage two of the
study. Fourth and 5th graders were combined on the same teams as per league rules and
we therefore included them in the analysis as one group. The 4-5‘h grade teams had the
highest participation (296) while the 8‘h grade represented the lowest (92.). Player
demographic data are presented in Table 7 by grade. The mean ages were 10.1 , 11.4.

12.5, and 13.4 for grades 4-5‘“, 6‘”, 7m. and 8th respectively. The PAS estimates

65

approximated 179 cm consistently across all grades. The estimates of PPAS consistently
increased from 80% (4-5th grade) to 91% (8th grade).

Player sport participation and injury history. Questionnaires distributed to the
parents at the beginning of the season (Appendix E) solicited information regarding each
player’s past sport participation and injury histories. Tables 8 through 12 report player
sport participation data. Participants who have participated in the study for multiple
years are presented in Table 8. Nearly half (45%) of the players were first year
participants. Only 47 (6%) players have participated during all four years of observation.
The low number of four-year participants is partially due to player attrition and only one
of the two towns having 8th grade teams.

Over 75% of the players reported prior football experience (Table 9). Within the
4-5‘h grade group, over 64% reported having some form of football experience (includes
flag football). Over 60% of the players reported playing between three and five sports
annually, and 2% reported participating in seven sports annually (Table 10). The most
common age at which the players reported beginning organized sports participation
(Table l 1) was five-years-old (43%). Soccer and tee ball were the most fiequent first
sport (32%, 33% respectively) as shown in Table 12.

Tables 13 through 16 summarize player reported previous injury data. The
proportion of players who reported having a previous injury increased with each
successive grade (Table 13). The 4-5th grade group had the lowest proportion of those
reporting previous injury (18%) while the 8th grade had the highest (57%). The
proportions of previously injured players who reported missing practices or games due to

a previous injury decreased with each successive level of participation (Table 14). Of the

66

4-5‘h graders, 51% reported missing a practice or game due to injury while this proportion
decreased to 40% for 8‘h graders. The ankle/foot (26%) followed by the wrist and hand
(11%) were most common sites of reported previous injury (Table 15). Sprains and
strains were the most frequent injury types accounting for 53% while general trauma such
as contusions made up another 22% (Table 16). Fractures accounted for 13% of the

injuries previously reported

67

 

mo N4: omo age mg: mm age _.:._ w: No ~42 mi 9w _.~: mom 9.3mm:

 

vo odo. coo o.o N.mo_ mm o.o :o. m: vo v.vo_ w: mo o.vo_ own 38:: :52

368.50
m8 3w: wwo mo ix: mm wo as: m: ms ow: w: o.\. Wm: mmm 35:2 own

38880
no mhofi coo mo o.mo_ mm mo o.mo_ m: no o.mo_ w: m.\. m.m_o_ vmm 35:: :52

3.83%
NS mow— nwo .8 52: mm fin mom: m: Nix Now. w: as v.2: Em 95:: can

3.53%
we 5mm wmo 5m woo mm wd o.ow mo. m.m fimw oofi VN Non wmm Ev mod;
No odt mmo no wow— mw o.o o.o: of _.o 0.x: of mo Wm: mmm A63 m<m
5v NAN an 5v 0mm ww o.m m.mm w: m6 odm mm: NV New Em ANEPEHEm
5.2 vow NNN. o.: o.mo mm _.o_ Wow w: Wm. adv mm: ma: o.ov Km Awe: Emma?
w.o_ v.9; 2K. m6 o.vo_ mm Wm oom. w: NS 0%: mm: o.o mil chm A83 238m
m._ v.2 owo Yo v.2 mm Yo w.m_ m: to v.2 m: no 2: com Amos <U
Gm E Z Gm E : Om .2 : Om E : Om 2 : oEmtw>

8930 =< ooEO Em 8ch ex. £50 5o 25ch 59v

 

$3.6 3 eroEREMQEmQ BEEN Etc ESE Eco:

N. Boer—r

68

Table 8

Proportion of Returning Players by Grade and Year in Study

 

 

Grade

4-5th 6‘h 7th 8th All Grades
Year n P n P n P n P N P
lst 204 68.9 78 38.4 45 23.9 21 22.8 348 44.7
2nd 92 31.1 57 28.1 51 27.1 19 20.7 219 28.1
3rd 0 0.0 68 33.5 69 36.7 28 30.4 165 21.2
4th 0 0.0 0 0.0 23 12.2 24 26.1 I 47 6.0
Total 296 100.0 203 100.0 188 100.0 92 100.0 779 100.0

 

Table 9

Proportion of Participants with Prior Football Experience by Grade

 

 

Grade
Previous 4-5th 6th 7th 8th All Grades
Experience n P n P n P n P N P
No 90 35.7 37 21.1 28 16.7 13 15.3 168 24.7
Yes 162 64.3 138 78.9 140 83.3 72 84.7 512 75.3
Total 252 100.0 175 100.0 168 100.0 85 100.0 680 100.0

 

Table 10

Proportion of Players who Reported Playing Sports Other than Football by Grade

 

Grade
Number 4-5th 6‘h 7th 8th All Grades
of Sports 11 P n P n P n P N P

 

0 9 3.6 7 4.0 4 2.4 0 0.0 20 2.9

l 34 13.5 17 9.7 16 9.5 5 5.9 72 10.6
2 40 15.9 28 16.0 21 12.5 8 9.4 97 14.3
3 61 24.2 32 18.3 37 22.0 21 24.7 151 22.2
4 49 19.4 42 24.0 44 26.2 22 25.9 157 23.1
5 43 17.1 36 20.6 37 22.0 19 22.4 135 19.9
6 10 4.0 12 6.9 8 4.8 7 8.2 37 5.4

7 6 2.4 1 0.6 1 0.6 3 3.5 11 1.6
Total 252 100.0 175 100.0 168 100.0 85 100.0 680 100.0

 

69

Table 1 1

Proportion of the Ages that Participants Reported Beginning to Play Organized Sports by

 

 

 

 

 

Grade
Grade
Age at 4-5tln 6th 7th 8th All Grades
1St Sport n P n P n P n P N P
3 2 0.8 4 2.3 1 0.6 0 0 7 1.0
4 31 12.3 18 10.3 14 8.3 5 5.9 68 10.0
5 111 44.0 71 40.6 66 39.3 44 51.8 292 42.9
6 58 23.0 37 21.1 30 17.9 18 21.2 143 21.0
7 17 6.7 17 9.7 22 13.1 12 14.1 68 10.0
8 6 2.4 10 5.7 13 7.7 3 3.5 32 4.7
9 22 8.7 6 3.4 8 4.8 2 2.4 38 5.6
10 5 2.0 4 2.3 5 3.0 1 1.2 15 2.2
11 0 0 6 3.4 4 2.4 0 0 10 1.5
12 0 0 1 0.6 3 1.8 0 0 4 0.6
13 0 0 1 0.6 2 1.2 0 0 3 0.4
Total 252 100.0 175 100.0 168 100.0 85 100.0 680 100.0
Table 12
Proportion of First Sports Played as Reported by Participants by Grade
Grade

4-5‘h . 6th 7th 8th All Grades
Sport n P n P n P n‘ P N P
Baseball 7 2.8 7 4.0 9 5.4 2 2.4 25 3.7
Basketball 6 2.4 10 5.7 14 8.3 2 2.4 32 4.7
Flag football 11 4.4 5 2.9 3 1.8 2 2.4 21 3.1
Floor hockey 16 6.3 6 3.4 6 3.6 7 8.2 35 5.1
Football 17 6.7 13 7.4 11 6.5 3 3.5 44 6.5
Hockey 9 3.6 5 2.9 3 1.8 1 1.2 18 2.6
Soccer 81 32.1 55 31.4 52 31.0 32 37.6 220 32.4
Sofiball 1 0.4 2 1.1 0 0.0 0 0.0 3 0.4
Swimming 3 1.2 3 1.7 1 0.6 0 0.0 7 1.0
Wrestling 15 6.0 9 5.1 10 6.0 1 1.2 35 5.1
Tee ball 84 33.3 58 33.1 53 31.5 32 37.6 227 33.4
Other 1 0.4 0 0.0 0 0.0 0 0.0 1 0.1
Gymnastics 1 0.4 1 0.6 2 1.2 1 1.2 5 0.7
Karate 0 0.0 1 0.6 3 1.8 2 2.4 6 0.9
Bowling 0 0.0 0 0.0 1 0.6 0 0.0 1 0.1
Total 252 100.0 175 100.0 168 100.0 85 100.0 680 100.0

 

70

Table 13

Proportion of Participants who reported having a Previous Injury by Grade

 

 

Grade
Previous 4-5th 6‘h 7th 8th All Grades
Injury n P n P n P n P N P
NO 204 81.9 108 62.8 96 58.5 32 42.7 440 66.7
Yes 45 18.1 64 37.2 68 41.5 43 57.3 220 33.3
Total 249 100.0 172 100.0 164 100.0 75 100.0 660 100.0

 

Table 14

Proportion who Reported Missing Practices and Games Due to a Previous Injury by
Grade

 

 

Grade
Previous 4-5th 6th 7th 8th All Grades
Injury n P n P n P n P N P
No 22 48.9 33 52.4 36 52.9 26 60.5 117 53.4
Yes 23 51.1 30 47.6 32 47.1 17 39.5 102 46.6
Total 45 100.0 63 100.0 68 100.0 43 100.0 219 100.0

 

Table 15

Proportion of Reported Previous Injury Locations by Grade

 

 

Grade

Previous 4-5th 6th 7th 8th All Grades
Injury n P n P n P n P N P
Head/neck 2 4.4 8 12.5 4 5.9 O 0.0 14 6.3
Face 2 4.4 3 4.7 0 0.0 0 0.0 5 2.3
Shoulderarm 5 11.1 7 10.9 7 10.3 5 11.4 24 10.9
Wrist/hand 6 13.3 10 15.6 13 19.1 8 18.2 37 16.7
Trunk 3 6.7 l 1.6 1 1.5 0 0.0 5 2.3
Hip/thigh/leg 4 8.9 3 4.7 2 2.9 2 4.5 11 5.0
Knee 2 4.4 7 10.9 7 10.3 3 6.8 19 8.6
Ankle/foot 11 24.4 17 26.6 19 27.9 11 25.0 58 26.2
Other 0 0.0 0 0.0 3 4.4 l 2.3 4 1.8
Multiple 10 22.2 7 10.9 12 17.6 14 31.8 43 19.5
Not specified 0 0.0 1 1.6 0 0.0 0 0.0 1 0.5
Total 45 100.0 64 100.0 68 100.0 44 100.0 221 100.0

 

71

Table 16

Frequency and Proportions of Reported Previous Injury Type by Grade

 

Grade
Previous 4-5‘h 6th 7th ‘ 8"1 All Grades
Injury n P n P n P n P N P
Sprain/strain 20 44.4 33 52.4 38 55.9 25 58.1 116 53.0
Fracture 6 13.3 8 12.7 9 13.2 5 11.6 28 12.8
Laceration 2 4.4 1 1.6 0 0.0 0 0.0 3 1.4
General 9 20.0 16 25.4 13 19.1 10 23.3 48 21.9
Combined 8 17.8 4 6.3 5 7.4 2 4.7 19 8.7
Other 0 0.0 0 0.0 2 2.9 1 2.3 3 1.4
Not specified 0 0.0 1 1.6 1 1.5 0 0.0 2 0.9
Total 45 100.0 63 100.0 68 100.0 43 100.0 219 100.0

 

Injury data analysis. Injury data were analyzed cumulatively and across grades.
The player frequencies, injury incidence, athlete exposures, injury rates, and incidence
density ratios are reported in Table 17. There were 779 players for all grades with 37%
sustaining an injury. The 4-5‘h grade players had the least risk of injury (29%) while the
7‘h grade players had the highest (58%). The risks for 6th and 8th graders were 37% and
46% respectively.

There were 474 injuries and 26565 AE. Practices accounted for 69% of the
injuries and 82% of the AE. Of the total injuries, 59% were classified as NTL. Injury
rates were reported with 95% CI. Injury rates were calculated by dividing the number of
injuries by the number of exposures and were expressed per 1000 AE. The overall injury
rate was 17.8 (95% CI: 16.3, 19.4) per 1000 AE. The game injury rate was 30.5 (95%
CI: 25.6, 35.4) and the practice injury rate was 15.1 (95% CI: 13.5, 16.7) per 1000 AE.

The overall and practice injury rates increased with each succeeding grade, but the game

injury rate was highest in the 6th grade players (35.3, 95% CI: 24.7, 45.9). Time-loss
injury rates also increased with each succeeding grade but NTL injury rates were more
variable.

Incidence density ratios (IDR) and 95% confidence intervals were calculated as
described by Powell and Dompier (2004) to provide comparison between the injury rates
for games and practices and between NTL and TL injuries. The data reveal that players
were twice as likely to be injured in a game versus practice and that they were 1.4 times
more likely to suffer a NTL injury than a TL injury. The incidence density ratios were
similar across all grades for both comparisons except for the 8th graders. The IDR for 8th
grade NTL versus TL injuries was 0.5 (95% CI: 0.2, 1.5). In summary, players had twice
the risk of suffering a game related injury, and were 1.4 times more likely to suffer a
NTL injury.

,Intrinsic risk factor analysis. Player related intrinsic variables were analyzed by
calculating relative risks for PPAS, stature, weight, BMI, previous injury, previous
football experience, and prior injuries that were serious enough to cause time loss from
participation. Variables identified as significant through univariate analysis were further
scrutinized using backwards—stepwise logistic regression to control for confounding.
Univariate and logistic regression analyses were performed and reported with the data
both stratified and not stratified by grade to control for grade as a confounding variable.
Stratification by grade was thought necessary because practices and games only occur
between teams of the same grade level. Univariate analysis of grade when the 4-5th
graders are the referent reveals that relative risk increases with each grade and 7th and 8th

grade levels are close to being significantly more at risk than the 4-5th graders (Table 18).

73

comm bacon 8522: u MA: 35895 22:2 :2: can 33. ASP. .1. y:

mmOJrv:—_-—.r:oz H 1_—Z mmOi—rv—Er—r H J—

 

36. 3.. 3.3 S 2.2 S 2%.: 3 in goeeome
3.2 E 2.3 2 3.2 f 5.2 2 mm 5.5+sz
4.2.92 2: 0.2.2: 2m 363.8 9% 3:3 02 250
$3.: a: defies we 33.: 9: 33.2 on 855
21$ 2: 212 2 33:: 92 6.2.3 2: :12
2.3 3 6.2.3: 3: 3.2 E 52.8 3 .:
2&in a: 33%. EN 9862 S: 32.2 of drapes
S; 3 mm :4 250

62 mm as ex 8:er

”5 mm 5 E :2

ed. a. a, a d.

E. E e: E 8:6: .85

$2. o3 2.2 NS _ seed

:5: EN 2% 2mm seeded

memem 83 one :8 8.38%.

3a a, c: on a was: ESE

a: 8 n: 8m .8er

Bag: 332 :28. 8%.: $32 5w .25: .526— 55 .53: 532 so coma: $32 smrv Eu:

a £63

5 $3

6 £ng

. 5 food

633% 3 BBQ 35.3?5 «935554

S 033.

74

Table 18

Relative Risk for Injury by Grade

 

 

Grade
Grade 11 cases RR 95%CI
W 346 137 Referent
6th 254 127 1.3 0.9, 1.2
7th 246 136 1.4 1.0, 1.9
8‘“ 123 74 1.5 1.0, 2.3

 

All intrinsic player-related variables were categorized. Within-grade z-scores
were calculated for statute, weight, and BMI for half-year age groups. The z-scores for
maturity were calculated using the means and standard deviations for PPAS provided by
Bayer and Bayley (1959). Players were categorized into terciles based on their respective
z-score within their respective grade level (Malina et al., 2002). A player who had a z-
score less than —1.00 was considered in the lowest tercile while a player with a z-score of
greater than 1.00 was in the highest tercile. All z-scores ranging from —1 to 1 were listed
in the middle tercile. As an example, maturity was categorized as late (2 < -l.00),
average {-0.1 5 z 5 1), and early (2 > 1.00). Once categorized, referents for'each variable
were selected based on hypotheses and previously reported risks.

To calculate relative risks within each category one of the tercile groups had to be
selected as a referent. The maturity referent was the early maturity group because it was
hypothesized that late maturing players were at greater risk of injury. Similar
conventions were used for all variables. Univariate relative risks and logistic regression
odds ratios with 95% CI for all grades combined are summarized in Table 19, and grade
stratified results are summarized in Tables 20-23.

The variables found significant through univariate analysis for all grades

combined were the average and tall stature groups (R = 1.4, 95%CI: 1.0, 2.1 and RR =

75

1.7 95%CI: 1.0, 2.8 respectively) and previous injury (RR = 1.5, 95%CI: 1.1, 2.0).
There is also a gradient effect for stature. Risk increased from the average (1.4) to the
tall (1.7) groups. Although insignificant, this gradient effect is also present in the
maturity, weight, and BMI relative risks (Table 19).

Univariate analysis after stratification by grade (Tables 20-23) revealed that only
the 4-5th grade group had significant relative risks for average stature players (RR = 2.2,
95%CI: 1.0, 4.7) and those with previous injury (RR = 1.9, 95%CI: 1.1, 3.4). The
gradient effect was still present when stratified by grade except in the stature variable for
the 4-5th grade group. Variables found significant through univariate analysis, and the
maturity variable were included in the logistic regression analysis. Maturity was not
found to be significant through univariate analysis. but it was included during the logistic
regression analysis because it was the variable of interest in this study.

The results of the backwards-stepwise logistic regression were consistent with the
univariate analysis of risk factors. The results of a logistic regression are reported as an
odds ratio, or the odds of one group developing the outcome versus another. Average
and tall stature and previous injury remained significant in the final model for all grades
combined (Table 19). Previous injury was not significant for grades 7th and 8th when
stratified by grade (Tables 20-23). Only the 4-5th graders had significant odds ratios for
both average (OR = 3.5, 95%CI: 1.5, 8.6) and tall (OR = 2.88, 95%CI: 1.0, 8.0) stature
groups. Grades 6th and 7th had significant odds ratios for the tall group of the stature
variable. The 81h grade group had no significant odds ratios in the final model. The 7th
and 6th grades Show a gradient effect of increasing risk for lower levels of maturity, but

remain non-significant. The risk of late maturing 7th graders is 4.5 times higher (OR =

76

4.5, 95%CI: 0.92, 22.28) than that of an early maturing 7th grader, and is nearly
significant.

In summary, stature and previous injury were significant risk factors for injury
across all grades. There was also a gradient effect present for maturity, stature, weight,
and BMI. Previous injury remained significant for 4-5‘h and 6th graders but not for 7th
and 8th graders when stratified. None of the factors were significant for 8th graders during 9
univariate or logistic regression analysis. Stature remained significant for 4-5th graders in

both the average and tall groups, but only the tall group in grades 6‘h and 7th.

77

Table 19

Risk F actors for Injury in Youth Football Players: All Grades Combined

 

 

Risk No. of No. of Estimated Estimated
factor players cases RR 95% C1 OR 95% C1
Maturity
Late 48 25 0.87 0.47, 1.62 1.61 0.55, 2.46
Average 519 252 0.81 0.59, 1.11 0.88 0.61, 1.26
Early 235 141 Referent Referent
Stature
Short 1 27 45 Referent Referent
Average 623 315 1.43 0.96, 2.12 1.93 1.24, 3.01
Tall 147 88 1.69 1.03, 2.76 3.46 1.98, 6.04
Weight
Low 97 47 Referent
Average 660 318 0.99 0.65, 1.52
Heavy 141 82 1.20 0.71, 2.02
BMI
Low 90 41 Referent
Average 643 313 1.07 0.69, 1.66
Heavy 162 92 1.25 0.74, 2.09
Prev. injury
Yes 302 195 1.51 1.13, 2.03 2.69 1.97, 3.69
No . 525 224 Referent Referent
Prev. experience
Yes 637 325 Referent
No 213 104 0.96 0.70. 1.31
Prev. injury caused time-loss
Yes 146 102 1.19 0.74, 1.92
No 1 53 90 Referent

 

RR = Relative Risk OR = Odds Ratio

78

Table 20

Risk F actors for Injury in Youth Football Players: 4’h-5'h grades

 

Risk No. of No. of Estimated Estimated
factor players cases ’ RR 95% CI OR 95% C1
Maturity
Late 15 3 0.45 0.11, 1.78 0.53 0.11, 2.49
Average 202 93 1.04 0.57, 1.89 1.60 0.79, 3.24
Early 54 24 Referent Referent
Stature
Short 44 9 Referent Referent
Average 216 96 2.17 1.00, 4.74 3.53 1.45, 8.59
Tall . 54 23 2.08 0.84, 5.17 2.88 1.03, 8.04
Weight
Low 27 6 Referent
Average 243 100 1.85 0.72, 4.75
Heavy 48 24 2.25 0.77, 6.55
BMI
Low 27 6 Referent
Average 241 101 1.89 0.73, 4.84
Heavy 47 22 2.11 0.72, 6.16
Prev. injury
Yes 70 48 1.93 1.09, 3.41 4.19 2.21, 7.96
No 22 8 8 1 Referent Referent
Prev. experience
Yes 196 87 Referent
No 105 43 0.92 0.57, 1.49
Prev. injury caused time-loss
Yes 42 33 1.47 0.52, 4.18
No 28 15 Referent

 

RR = Relative Risk OR = Odds Ratio

79

Table 21

Risk F actors for Injury in Youth Football Players: 6th grade

 

Risk No. of No. of Estimated Estimated
factor players cases RR 95% CI OR 95% Cl
Maturity
Late 18 1 1 0.89 0.29, 2.70 0.95 0.24, 3.75
Average 133 59 0.65 0.33, 1.26 0.46 0.22, 0.96
Early 54 37 Referent Referent
Stature
Short 3 3 1 3 Referent Referent
Average 166 81 1.24 0.58, 2.65 1.82 0.65, 5.11
Tall 37 27 1.85 0.68, 5.07 6.07 1.63, 22.59
Weight
Low 22 9 Referent
Average 180 93 1.26 0.51, 3.10
Heavy 33 18 1.33 0.45, 3.97
BM]
Low 25 1 2 Referent
Average 166 85 1.07 0.46, 2.47
Heavy 42 22 1.09 0.41, 2.94
Prev. injury
Yes 51 1.38 0.79. 2.41 2.24 1.17, 4.29
No 132 58 Referent Referent
Prev. experience
Yes 1 72 88 Referent
No 47 22 0.92 0.57, 1.49
Prev. injury caused time-loss
Yes 42 29 1.29 0.52, 3.16
No 22 41 Referent

 

RR = Relative Risk OR = Odds Ratio

80

Table 22

Risk F actors for Injury in Youth Football Players: 7th grade

 

Risk No. of No. of Estimated Estimated
factor players cases RR 95% Cl OR 95% Cl
Maturity
Late 15 11 1.24 0.36, 4.25 4.45 0.92, 22.28
Average 120 61 0.86 0.48, 1.53 0.95 0.49, 1.85
Early 78 46 Referent Referent
Stature
Short 33 1 5 Referent Referent
Average 160 87 1.20 0.56. 2.54 2.25 0.78, 6.43
Tall 35 23 1.45 0.54, 3.84 5.58 1.36, 22.86
Weight
Low 34 22 Referent
Average 160 82 0.79 0.37, 1.71
Heavy 35 21 0.93 0.35, 2.46
BMI
Low 24 1 5 Referent
Average 162 84 0.83 0.34, 2.00
Heavy 42 25 0.95 0.34, 2.67
Prev. injury ‘
Yes 89 54 1.38 0.79, 2.41 1.73 0.95, 3.14
No 120 58 Referent Referent
Prev. experience
Yes 1 75 93 Referent
No 39 22 0.92 0.57, 1.49
Prev. injury caused time-loss
Yes 40 23 1.29 0.52, 3.16
No 49 31 Referent

 

RR = Relative Risk OR = Odds Ratio

81

Table 23

Risk F actors for Injury in Youth Football Players: 8th grade

 

 

Risk No. of No. of Estimated Estimated
factor players cases RR 95% C1 OR 95% CI
Maturity
Late 0 0
Average 64 39 0.88 0.40, 1.93 1.06 0.38, 3.00
Early 49 34 Referent Referent
Stature
Short 1 7 8 Referent Referent
Average 81 51 1.34 0.47, 3.84 1.62 0.49, 5.41
Tall 21 15 1.52 0.40, 5.81 3.34 0.70, 16.00
Weight
Low 14 1 0 Referent
Average 77 43 0.78 0.23, 2.71
Heavy 25 19 1.06 0.24, 4.67
BMI
Low 14 8 Referent
Average 74 43 1.02 0.32, 3.23
Heavy 31 23 1.30 0.34, 4.91
Prev. injury
Yes 59 42 1.19 0.52, 2.70 1.62 0.70, 3.71
No 45 27 Referent Referent
Prev. experience
Yes 94 57 Referent
No 22 17 1.27 0.43, 3.75
Prev. injury caused time-loss
Yes 22 17 1.23 0.37, 4.12
No 35 22 Referent

 

RR = Relative Risk OR = Odds Ratio

Chapter Five

Discussion

The study consisted of two distinct stages of data collection and analysis. The
first stage was designed to test the validity of the Khamis and Roche [KR](1994) method
of adult stature prediction and subsequent percent of predicted adult stature as a method
of maturity estimation. Stage two consisted of analyses of injury incidence, risk, rates,
and intrinsic risk factors.

Stage One: The Validity of Percent of Predicted Adult Stature as a Maturity Indicator

Subsample baseline data. It was hypothesized that the noninvasive KR method of
adult stature prediction is a valid method of estimating maturity when expressed as a
percent of the child’s predicted adult stature. The KR estimates were compared to those
derived using the Khamis and Guo [RWT-KG](1993) method in a sample of youth
football players between the ages of 9 and 14 years. The RWT-KG method is invasive
because it requires an x-ray to determine skeletal age (SA), but is considered more
accurate than estimates lacking SA (Khamis and Roche, 1994). To date, no reported
studies have examined this relationship in a group of youth football players.

Overall, the subsample (n = 64) was slightly advanced in skeletal maturity. The
average deviation between SA and chronological age (CA) was 0.7 years. Mean statures
and weights were compared to the CDC Growth Charts for boys 2 to 20 for stature-for-
age and weight-for-age (Ogden, Kuczmarski, F legal, Mei, Guo, et al., 2002). In
reference to stature, mean statures were between the 50‘h and 75th percentiles for all

grades. The mean weights of all groups were between the 75th and the 90th percentiles for

83

the corresponding mean age. These findings are consistent with the sample being
advanced in skeletal maturity.

Maturity and size variation among football players has been previously reported.
Malina, Meleski, and Shoup (1982) reported on anthropometric measurements and
maturity estimates in youth football players taken in 1970. Malina et al. (1982) found
that stature was near the United States median, but weight was just below the 75th
percentile. Although the more weight-for-stature relationship is consistent between] this
study and that reported by Malina et al., the results of the current study show a drastic
increase in the percentiles in the 33 years between studies. This difference could be the
result of many factors such as taller statures, increased obesity, or increased muscle mass
as the result of early ages at which children begin weight training. Body fat percentages
were not considered in this study.

Subsample statistical analysis. Analysis of the subsample data demonstrates a
moderate but statistically significant correlation between the KR percent of predicted
adult stature and SA (partial r, adjusted for CA, = 0.54; p < .001). The mean estimates of
predicted adult statures derived from the RWT-KG and KR methods were statistically
different (t(63) = 2.29, p < .05), but highly correlated (partial r, adjusted for CA, = 0.88; p
< .001). This finding is consistent with the findings reported by Khamis and Roche
(1994) and demonstrates that the non-invasive KR method is a useful proxy when SA is
unavailable. Thus, the use of percent of predicted adult stature was a useful tool in
differentiating between those of varying maturity status in the current study. The

application of the KR percent of predicted adult stature as a maturity estimate is

84

recommended when skeletal age is contraindicated or unavailable, and sample ages range
between 3 and 17.5 years of age.

The strong correlations between the RWT-KG and KR methods of adult stature
prediction found in the current study are consistent with other published comparisons.
Khamis and Roche (1994) compared estimates for males and females using both
methods. Comparison of estimates in males showed a slightly larger 90% error bound
(2.1) when SA is omitted (KR method) than when it is included (1.8 cm) as in the RWT-
KG model. Estimates in females were 1.7 cm when SA was omitted (KR), and 1.5 cm
with SA (RWT-KG). Zarow (1997) found similar results, but both methods produced
higher mean errors than those reported by Khamis and Roche (1994). This difference is
likely the result of the difference between the reference population and the population
sampled by Zarow. The results of the current study indicate that the percentage of
predicted adult stature derived from the KR method of adult stature prediction is a valid
estimate of maturity and can be a useful tool in field studies where invasive methods of
maturity estimation are impractical or contraindicated. These recommendations are
limited to youth football players between the ages of 9 and 14 years of age.

Stage Two: Injury Analysis

Baseline data and maturity estimation. Measurements of stature, weight, and BMI
for the entire study population were reported for each grade in Table 7. The current study
means were compared to national averages (Appendices G-J) using the CDC Grth
Charts for boys 2 to 20 for stature-for-age, weight-for—age, and BMI-for-age percentiles
(Ogden et al., 2002). Mean ages that were compared to the growth charts were 10, 11.4,

12.5, and 13.5 for the 4-5‘“, 6‘“, 7th, and 8‘h grades respectively. For all a es, stature was
g

85

between the 50th and 75th percentiles with the 4-5th and 7th grades nearly equal to the 75‘h
percentile. Weight and BMI were closer to the 90‘h percentile for all grades. These
comparisons would indicate that the current sample is on average, slightly taller, but
much heavier than the majority of children in the United States of the same age. National
averages for PPAS do not exist, but comparisons can be made with means of PPAS
reported by Bayer and Bayley (1959).

Bayer and Bayley (1959) produced means of percents of predicted adult statures
(PPAS) for ages 3 to 18 years with data from the Berkley Growth Study (Bayley and
Pinneau, 1953). On average, the children of the current study had higher mean PPAS as
compared to those reported by Bayer and Bayley, indicating that the current sample is on
average more mature (Appendix J). This finding may indicate that the assignment of
terciles of maturity status was biased. Assignment to terciles of maturity in the current
study followed the methods reported by Malina et a1. (2002). Because the current
population was on average more mature, the z-scores would also drift toward a tercile of
higher maturity. It is possible that this effect reduced the number of participants listed in
the lower two terciles of maturity and increased the likelihood of not finding a significant
relative risk for maturity in the average and late maturing.

Injury analysis. The incidence of injury reported in the current study is consistent
with previous reports involving junior high school and youth football players. Table 1
summarizes previous studies that report injury risks that range from 6% (Stuart et al.,
2002) to 51% (Radelet et al., 2002). The injury risk in the current study was 37%, and is
equal to the risk reported by. Violette (1976). Injury risk varies according to injury

definition and the data source. The ATC data collector and injury definition used in the

86

 

current study is more sensitive than that used by Stuart el al. because the current
definition included every injury evaluated by an ATC. The definition used by Stuart et

a1. required that the player be removed from play or required advanced medical
procedures or evaluation. In addition, Stuart utilized physicians as data collectors who
were only present at competitions. Tuberville et al. (2003a) used a similar time-loss
definition, and utilized ATCs to collect injury data. Even with the increased sensitivity of
an ATC data collector, Tuberville et al. reported a 10% incidence which is lower than
those found in the current study. Similar trends are evident when injury rates are
compared.

The current overall injury rate was over twice as high as those previously reported
for youth football players. The difference in the current study is due to the increased
sensitivity of the injury definition that included NTL injuries. When the injury rates were
dichotomized into TL and NTL injuries, the TL injury rate (7.4, 95%CI: 6.3, 8.4 per 1000
AE) was similar to that reported by Powell and Barber-Foss (1999) for high school
football players (8.1 per 1000 AE). In addition, the current TL injury rate of 7.4 is also
similar to that reported by Powell and Dompier (2004) (9.8) who examined both TL and
NTL injuries in college football players. The NTL injury rates were not as similar
however. The NTL injury rate was 10.5 (95%CI: 9.2, 11.7) or about one third of that
reported by Powell and Dompier (30.8).

Injuries that are defined as non-time-loss are important to consider because of the
impact that they have on players, coaches, and parents. The results of the current study
demonstrate that only 40% of the injuries that require a decision regarding the playing

status of injured players have been previously reported. This suggests that the coach or a

87

parent has to make a decision regarding the playing status of that child. This is
concerning because previous injury has been reported as a risk factor for injury. If
appropriate decisions are not made regarding the disposition of an injured athlete, they
could be potentially put at risk by being returned to play while recovering from a
previous injury. Providing coaches and or parents with basic first aid skills may help to
mitigate this risk.

Intrinsic risk factor analysis. Risk factor analysis was limited to player intrinsic
factors. This limitation was primarily due to inadequacies in study design, but also
partially due to uncontrollable factors. Specific player positions have been shown to be
risk factors for injury in previous studies (Beachy et al., 1996; Blyth and Mueller, 1974;
Powell, 1980; Powell and Schootman, 1992; Turbeville eta1., 2003a), but were
untraceable in the current study. Players were not specialized to any one specific position
and often played multiple different positions during any single session. Although players
were asked what positions they were playing at the time of inj ury,.positions of uninjured
players were not known. Therefore, no comparison can be made between injured and
uninjured in respect to player position. Additionally, environmental conditions were not
systematically reported therefore no comparisons could be made between surface
conditions. Previous studies have implicated weather related playing surface conditions
and the shoe-surface interface as risk factors (Bramwell, etal., 1972; Orchard, 2002;

' Orchard and Powell, 2003; Powell and Schootman, 1992; Torg and Quedenfeld, 1971 ).
These limitations restricted the risk factor analysis in the current study to player intrinsic

variables.

88

Player intrinsic variables have been implicated as risk factors for injury in youth
and junior high school football players (Caine and Broekhoff, 1987; Gomez et al., 1998;
Linder et al., 1995; Malina etal., 2002; Turbeville, 2003a, Violette, 1979). Variables that
have been studied include stature, weight, BMI, previous injury, previous experience,
grip strength, and maturity. Findings among the different variables are inconsistent and
comparisons of maturity are made difficult because of inconsistent methods. The current
study examined maturity, stature, weight, BMI, previous injury, previous experience, and
previous injuries that caused time-loss. Of those, only stature and previous injury were
found to be significant even though maturity was the variable of interest.

Maturity was the intrinsic variable of interest in the current study, but found to be
non-significant. Late maturity status has been implicated as a risk factor for injury in
children who play sports with children of the same age or grade, but are of higher
maturity status (Linder et al., 1995; Malina et al., 2002; Violette, 1979). Linder et al.
found that those in higher Tanner stages were more at risk of injury, but Malina et al.
found no relationship. Conversely, Violette found that those who were late maturing
were at greater risk. The results of the current study cannot clarify this disparity because
no significant relationships were found. However, even though no significant relative I
risks were found for maturity status and injury, there is an increasing gradient effect with
higher maturity status. This gradient effect would support the findings by Linder et al. in
that as maturity status increased, a concurrent increase was seen in the relative risk
(Tables 19-22).

Many plausible explanations exist for the differences found between studies

examining maturity status as a risk factor for injury in youth football players. In the

89

current study, exposure was not adequately controlled. A team level of exposure was
collected, meaning that those present at each session were counted as one AE. Dagiau et
a1. (1980) demonstrated that time of exposure can bias risk factor analysis in football
because not all players receive equal playing time in competitions or equal repetitions in
practice. To determine if maturity is a risk factor, a player level exposure is necessary.
The player level of exposure is the specific number of repetitions or playing minutes each
player receives during practice or competitions. Analyses would then stratify or
otherwise control for playing time between those who are injured and those who are not.
If late maturity were a risk factor for injury in the current study, then the exposure bias
could have prevented it from being detected.

The nature of the population could also be a reason why maturity was not detected
as a risk factor. Because this sample of youth football players was, on average, more
mature than the reference population (Bayer and Bayley, 1959), fewer subjects would be
classified in the lower terciles of maturity therefore the groupings may be biased. This
bias would reduce the frequencies in the lower terciles and would prevent detection of a I
difference. Another reason might simply be that there really is no difference in injury
risk between those of later and those of earlier maturity status. Finally, there may have
been systematic flaw in the prediction of maturity or in the method used to categorize
maturity groups. There are similar disparities when comparing other intrinsic variables.

There is no consensus on which intrinsic risk factors for injury are the most
important in youth football. Gomez et al. (1998) found that body fatness and BMI were
significant risk factors for injury. Malina et al. (2002) found that stature and previous

injury were significant risk factors. Turbeville et al. (2003a) found that previous

90

experience was a risk factor. The current findings that stature and previous injury are
significant risk factors are in agreement with Malina et al. (2002). The agreement with
Malina et al. is expected however, because the current study is a continuation of the
original. Although non-significant, previous experience was in agreement with
Turbeville et al. (2003 a), and weight was in agreement with Gomez et al. Weight
demonstrated an increasing gradient effect on the relative risk for the average and heavy
groups (Tables 19-22). The only similarity between these studies and the current was
that the results of each were likely confounded by player exposure bias.

Player exposure bias may not be the only bias to affect the stature, weight, and
BMI variables. As noted above, the study sample was on average more mature, taller
(75th percentile), and heaver (90th percentile) than national means. This difference likely
caused less frequencies in the lower two terciles in each of those variables. If the current
means were closer to national estimates, a greater difference may have been found. This
bias should be tempered for stature, weight, and BMI because z-scores were based on the
sample mean, not a national mean. Intrinsic variables are not the only variables that may
have been biased and Malina et al. (2002) have reported that there are inherent
differences between the 4-5th—6, and 7m-8th grade teams.

The classification of players by grade in which they play may not have adequately
represented the sport. In grades 4-5th and 6th, the rules do not allow kicking plays (punts,
kick-offs, and field goals). The 7111 and 8th grade teams follow the same rules as the local
high school teams do, and allow all types of play situations. Statistical analyses of the
data by kicking exposure would have added power and more adequately represented the

type of game played. Univariate analysis between no-kicking and kicking dichotomies

91

reveals that the kicking group (7th-8th grades) is at greater risk of injury (RR = 1.29,
95%CI: 1.0-1.7). Future analyses should seek to control for kicking as a possible
confounder.
Conclusions

The non-invasive KR method of predicted adult stature is a valid estimate of
maturity when current stature is expressed as a percentage of predicted adult stature.
Injury incidence, risk, and injury rates are within previously reported ranges. Non-time-
loss injuries accounted for 5 9% of the total injury picture, and these findings are
consistent with previously reported NTL injury risk. Maturity is not a risk factor for
injury, but stature and previous injury are. All intrinsic variables may be biased by
exposure time and the anthropometric characteristics of the samme. Additional research
should focus on the player level of exposure versus the team or sport levels of exposure.
Without adequately controlling for exposure, it will not be possible to determine if those
who are late maturing are at greater risk of injury when playing with children their same
age but of higher maturity status.
Future Research

The issue of matching competitors by maturity, age, or skill remains unresolved.
Before recommendations to match competitors by maturity can be made, research must
definitively demonstrate that maturity is related to injury risk. Maturity and other
intrinsic variables can only be analyzed if exposures are reported at the player level.
Controlling for the player level of exposure could be accomplished by reporting the

number of plays each player participates in during games, or the number of minutes each

92

is on the field. Practice exposures are more difficult to control however, and would
require careful scrutiny of what type of drill and how much each player is involved.

The case control design is another method to control for exposure. When using a
control design, at the time that an injury occurs, random selection of up to four other
players who were on the field at the same would allow comparison of both sport and
player level variables. A case control design would also mitigate the difficulty tracking
practice activities because it would only be necessary to document the activities of the
injured player and the four controls at the time of injury. One drawback of the case
control design is that time order cannot be determined in most studies. This weakness is
also mitigated by documenting the cases and controls at the time of injury because time
order will be known. The case control design would be a powerful tool if applied to
sports injury research.

This study has shown that maturity can be estimated by determining the percent
of adult stature that a child has attained. There are multiple options choose from when
determining how to apply the estimates derived from this procedure. The percent of
predicted adult stature is a continuous variable and can be analyzed using numerical
methods, or it can be converted into a categorical variable and analyzed using categorical
procedures as demonstrated in this study. Although the external validity of the current
results are limited to Caucasian youth football players from rural and suburban
communities, it is the opinion of this author that percent of predicted adult stature could
be applied in a variety of setting where other methods of maturity estimation are

unavailable.

93

Appendix A

Relationships Among Player Risk Factors and Injuries in Youth Football
Parental Informed Consent Form
(Injury Surveillance and Surveys)

For questions regarding this stuay. For questions regarding your rights as a research
please contact: participant, please contact:
John W. Powell, PhD, ATC Ashir Kumar, MD
Principle Investigator Chair Person
Department of Kinesiology Committee on Research Involving Humans
Michigan State University Michigan State University
105 IM Circle 202 Old Hall
East Lansing, MI 48824 East Lansing, MI 48824
517-432-5018 517-355-2180

Dear Parents & Guardians:

Hello! My Name is John W. Powell, PhD, ATC, Assistant Professor of Kinesiology and Certified Athletic
Trainer at Michigan State University. Thomas Dompier, ATC and Mary Barron, ATC, and I are working
on a research study entitled, “Relationships Among Player Risk Factors and Injuries in Youth Football.”
This year will be the 4th year of the project, and the first year that the project will be funded with a grant
from the National Football League Charities (NFL). The continuation of this study allows us to provide
athletic training services for the junior football team your child is participating on. Dr. Jeff Kovan, The
Director of Sports Medicine at MSU, is also a consultant on the project. The study will continue to monitor
injury patterns in youth football and the relationship between maturity status and players’ perception of
risk.

The study will involve your child‘s participation throughout the football season. At the beginning of the
season we will measure your child’s height and weight. Height and weight will be measured as part of
equipment handout process and will take less than 10 minutes. At the end of the season we will ask them
to complete a questionnaire designed to learn more about their thoughts regarding injury risk in football.
This questionnaire usually takes about 10 minutes to complete, and will be conducted during practice time.
Included with this consent form is a questionnaire regarding your child’s previous experience in youth
sports and if and what type of injuries they might have had. This questionnaire should also takes about 10
minutes or less to complete. Additionally, at the end of the questionnaire, we ask that you provide the
heights of both biological parents. In total, we ask for about 20-30 minutes of your child’s and your time to
complete this form and the questionnaires.

The height of your child plus the heights of both biologic parents allows us to estimate your child‘s
maturity status. We can then compare the maturity status of players to the injury rates for each age group
to determine if maturity is a factor for injury.

Throughout the season, the Certified Athletic Trainer assigned to your child’s team will document
information concerning injuries that occur during practices and games. This information will include the
severity, type of injury, the position played, activity performing when hurt, etc. Additionally, with your
permission, we may discuss the injury with your child, and or contact you by phone to obtain additional
information about the injury.

All identities and recorded information collected during this study will remain confidential and will be
replaced and analyzed with individual identification numbers. Participants will remain anonymous in any

reporting of the data from this study. and your privacy will be protected to the maximum extent allowable
by law.

94

In order for us to allow your child to participate in the study, we will need your written consent in the
spaces provided below. Your child’s participation is voluntary and you or your child may discontinue their
participation at any time. If your child’s participation is discontinued, their data will not be used in our
study. ‘

Any questions concerning participation in this study should be directed to John W. Powell, Assistant
Professor of Kinesiology (517) 432-5018. If you have any additional questions concerning your child’s
rights in this research study. please feel free to contact Ashir Kumar, MD, Michigan State University’s
chair of the Committee on Research Involving Human Subjects at (517) 355-2180.

INFORMED CONSENT:
This section indicates that you are giving your informed consent.

1 have read and agree to allow my child,

 

Please Print Your Child ’5 Name
to participate in this study as described above.

 

Please Print Your Name

 

 

Your Signature Date

Appendix B

Relationships Among Player Risk Factors and Injuries in Youth Football
Participant Informed Consent Form
(Injury Surveillance and Surveys)

For questions regarding this studv, For questions regarding your rights as a
please contact: research participant, please contact:
John W. Powell, PhD, ATC Ashir Kumar, MD
Principle Investigator Chair Person

Department of Kinesiology Committee on Research Involving Humans

Michigan State University Michigan State University
105 IM Circle 202 Old Hall

East Lansing. MI 48824 East Lansing, MI 48824

517-432-5018 517-355-2180

This study is designed to assess the thoughts you have concerning being injured when playing sports. This
study will help us understand the things that might lead to injury in youth football.

For this study, you will be asked to complete a questionnaire regarding your thoughts on being injured in
youth football. This questionnaire will take about 10 minutes to complete and you will have time during
practice to complete it. We will also measure your height and weight at the beginning of the season during
equipment handout. Also, a certified athletic trainer will record information about injuries you may have
throughout the season. If you are injured, we will ask you additional questions like how it happened and
what position you were playing.

All the information you provide, and the results of the study will be confidential and anonymously
reported. You will be assigned a coded identification number that will be used on all information you
provide. All the questionnaires and information you provide will be stored in a locked file cabinet inside a
locked office that is accessible only to the investigators of this project. Only group data will be reported
and group data will be provided to you at your request. Your participation in this study is voluntary. You
may choose to quit and refuse to answer any questions at any time without penalty. Your information will
remain anonymous in any reporting of the data from this study, and your privacy will be protected to the
maximum extent allowable by law.

Any questions you may have concerning your participation in this study should be directed to Dr. John W.
Powell at the Department of Kinesiology at Michigan State University, 517-432—5018. If you have
additional questions or concerns about your rights in this research study. please feel free to contact Ashir
Kumar, MD, Michigan State University’s Chair of the Committee on Research Involving Human Subjects
at 517-355-2180.

Thank you for you time and cooperation.

I have read or have had read to me, the above description of the study and agree to participate.

Please Print Your Name:

 

First Name Middle Initial First Name

 

 

Sign Name Here Date

96

Appendix C

Relationships Among Player Risk Factors and Injuries in Youth Football

Parental lnfonned Consent Form
(X-ray Maturity Analysis)

For questions regarding this study, For questions regarding your rights as 0
please contact: research participant, please contact:
John W. Powell, PhD, ATC Ashir Kumar, MD
Principle Investigator Chair Person
Department of Kinesiology Committee on Research Involving Humans
Michigan State University Michigan State University
105 IM Circle 202 Old Hall
East Lansing, MI 48824 East Lansing, MI 48824
517-432-5018 517-355-2180

Dear Parents & Guardians:

Hello! My Name is John W. Powell, PhD, ATC, Assistant Professor of Kinesiology and Certified Athletic
Trainer at Michigan State University. Thomas Dompier, ATC and Mary Barron, ATC, and I are working
on a research study entitled, “Relationships Among Player Risk Factors and Injuries in Youth Football.”
This year will be the 4th year of the project, and the first year that the project will be funded with a grant
from the National Football League Charities (NFL). The continuation of this study allows us to provide
athletic training services for the junior football team your son is participating on. Dr. Jeff Kovan, The
Director of Sports Medicine at MSU, is also a consultant on the x-ray portion of the project. The study will
continue to monitor injury patterns in youth football and the relationship between maturity status and
players’ perception of risk.

You have received this informed consent if you have volunteered to participate in the x-ray portion of the
study. This portion of the study will involve taking one x-ray of your son’s left hand. The single left hand
x-ray is the most accurate method of estimating your son’s skeletal age. This estimate allows us to estimate
your son’s predicted adult height. We will compare this information to the non-invasive method of
estimating predicted adult height that is based on your son’s current height, age, and heights of the
biological parents. It is important that we validate our non-invasive method of estimating the maturity of
children so in future studies researchers can use the non-invasive method with a reasonable degree of
certainty. You will be provided with the individual results of your son’s estimate.

The x-ray may require that your son miss an evening practice. You will be asked to transport your son to
the medical facility nearest your city that has volunteered to assist in the study. You will be scheduled for a
specific time, and given a week notice before your scheduled x-ray date. You will not be charged for the x-
ray. The grant provided by the NFL Charities will pay for the single left hand x-ray for each participant.
This process may take up to 1 hour typically and in rare situations more than an hour depending on
emergencies and the like that may be brought into that medical facility. This hour includes driving time to
and from the facility.

Your son will be minimally exposed to radiation, but all customary safeguards will be used to limit this
exposure. This is less exposure than most diagnostic x-ray visits because we are asking for only one x-ray
versus the multiple x-rays that are often taken for diagnostic purposes.

All identities and recorded information collected during this study will remain confidential and will be
replaced and analyzed with individual identification numbers.

Participants will remain anonymous in any reporting of the data from this study. and your privacy will be
protected to the maximum extent allowable by law.

97

 

In order for us to allow your son to participate in the study, we will need your written consent in the spaces
provided below. Your son’s participation is voluntary and you or your son may discontinue their
participation at any time. If your son’s participation is discontinued, their data will not be used in our
study.

Any questions concerning participation in this study should be directed to John W. Powell, Assistant
Professor of Kinesiology (517) 432—5018. If you have any additional questions concerning your son’s
rights in this research study, please feel free to contact Ashir Kumar, MD, Michigan State University’s
chair of the Committee on Research Involving Human Subjects at (517) 355—2180.

INFORMED CONSENT:
This section indicates that you are giving your informed consent.

1 have read and agree to allow my child,

 

Please Print Your Child '3 Name
to participate in this study as described above.

 

Please Print Your Name

 

 

Your Signature Date

98

Appendix D

Relationships Among Player Risk Factors and Injuries in Youth Football

Participant Informed Consent Form
(X-ray Estimate)

For questions regarding this study, For questions regarding your rights as a
please contact: research participant, please contact:
John W. Powell, PhD, ATC Ashir Kumar, MD
Principle Investigator Chair Person
Department of Kinesiology Committee on Research Involving Humans
Michigan State University Michigan State University
105 IM Circle 202 Old Hall
East Lansing, MI 48824 East Lansing, MI 48824
517-432-5018 517-355-2180

This study is designed to compare to two methods that determine how tall you may become as an adult.
Your participation in this study will allow future researchers to use the non-invasive method of comparing
your current height with the average of your parents’ heights.

For this study, you will be asked to provide a single left hand x-ray. This will require that you possibly
miss one day of practice. You will be exposed to a very small amount of radiation, but all efforts will be
made to reduce this amount. The amount you will be exposed to will be less than is typically required if
getting an x-ray to find broken bones. This process may take up to a hour or longer to complete.

Your information will remain anonymous in any reporting of the data from this study, and your privacy
will be protected to the maximum extent allowable by law. You will be assigned a coded identification
number that will be used on all information you provide. Only group data will be reported and group data
will be provided to you at your request.

Your participation in this study is voluntary. You may choose to quit and refuse to answer any questions at
any time without penalty. Any questions you may have concerning your participation in this study should
be directed to Dr. John W. Powell at the Department of Kinesiology at Michigan State University, 517-432-
5018. If you have additional questions or concerns about your rights in this research study, please feel free
to contact Ashir Kumar, MD, Michigan State University’s Chair of the Committee on Research Involving
Human Subjects at 517-355-2180.

Thank you for you time and cooperation.
I have read or have had read to me, the above description of the study and agree to participate.

Please Print Your Name:

 

First Name Middle Initial First Name

 

 

Sign Name Here Date

99

 

Appendix E

Background in Sports Information 2002-2003

 

 

th th th th th
Child’s Name: Team? 4 -5 6 7 8
First Last
Date of Birth: / /
Today’s Date: / /

 

How old was your child when he/she began to play on an organized sports team that practiced and played a
regular schedule of games or competitions? Organized sports means that there was an assigned coach for
the team. Examples include swimming, t-ball, football, basketball, etc.

years old.

 

What was the first organized sport that your child played? Years played?

 

What other organized sports has your son/daughter played and how many years has he played each?

SPORT? YEARS PLAYED?

 

 

 

 

 

 

In evaluating the height and weight of your child, it is important to know the size of the biological parents.
Please report the height of both biological parents to the nearest V4 inch without shoes.

Father’s Height

 

Mother’s Height

 

100

Has your child ever been injured during a sport practice or during a game/competition?

 

YES NO (please circle) If YES, please list the one or two most serious injuries and answer
the questions:

INJURY ONE
0. What specific body part was injured?
Head/Neck Face Shoulder/Arm F orearm/ wrist/hand
Trunk Hip/thigh/leg Knee Ankle/foot Other

b. What type of injury was it?

Sprain/strain Fracture Laceration General Trauma (bruise etc)

c. Did your child receive treatment? YES NO If yes, was he treated at:
An Emergency Room YES NO
A Doctor’s Office YES NO
At Home YES NO

c. Did your child miss any games, competitions or practices due to this injury?

 

. YES NO
INJURY TWO
a. What specific body part was injured?
Head/Neck Face Shoulder/Arm Forearm/wrist/hand
Trunk Hip/thigh/leg Knee Ankle/ foot Other

b. What type of injury was it?

Sprain/strain Fracture Laceration General Trauma (bruise etc)

c. Did your child receive treatment? YES NO If yes, was he treated at:
An Emergency Room YES NO
A Doctor’s Office YES NO
At Home YES NO

c. Did your child miss any games. competitions or practices due to this injury?

YES NO

101

 

Appendix F

 

 

 

 

 

 

Injury Report Form

NAME: Date:

Athletic Session
Game: Warm-up 1“ Quarter 2nd Quarter 3rd Quarter .4th Quarter
Practice:

Position of Injured Player: Offense: Defense

Type of Surface Natural Artificial

Surface Condition Dry ______Wet ___Muddy _____Frozen

Weather Conditions: ____Hot ____Warm ____Cool ___Cold _______Rain ___Snow

Point in the Season

 

Action Taken: Removed from participation and returned immediately
Returned from participation after resting
Removed from remainder of participation

Taken to hospital by parent

Taken to hospital by ambulance

Clinical Impression:

Injured Part of Body: .

Head Neck Shoulder Upper arm Elbow Forearm Hand Wrist Fingers
Hip Thigh Knee Shin Calf Ankle Foot Toes
Back Abdomen Chest Other

Type of Injury:

Sprain Strain Fracture General Trauma Neurotrauma

Laceration Overuse Other

Perceived Severity of Injury: Mild Moderate Severe
Summary of

Evaluation:

 

 

 

 

 

102

(cm)

Appendix G

Study Mean Statures vs the National Center for Health
Statistics Medians by Grade

 

 

170

 

[I] Study I 2000 Growth Charts

 

160

 

140 -

 

 

 

 

 

 

 

130

5th (10.0) 6th (11.5) 7th (12.5) 8th (13.5)
N531 Grade and Mean Age

103

(cm)

Appendix H

Study Mean Statutes vs the National Center for Health
Statistics Medians by Grade

 

  

 

 

170

[I Study I 2000 Growth Charts
160
150 %

 

 

5th (10.11)
N='21

140 —:j:-__*
130 — 4 '

 

 

 

6th (11.5)

104

 

7th (12.5) 8th (13.5)
Irade and Mean Age

B1\ II

Appendix I

Study Body Mass Index Means vs National Center for Health
Statistics 85th Percentile by Grade

[:1 I 2 000 Growth Charts

 

5th (10.0) 6th (11.5) 7th (12.5) 8th (13.5)
N=‘2 1 Grade and Mean Age

105

Percent

Appendix I

Study Percent of Predicted Adult Stature vs Bayer and Bayley
(1959) Means by Grade

 

5th (10.0) 6th (11.5) 7th (12.5)

N=63 5 Grade and Mean Age

8th (13.5)

106

References

Acheson, R.M., Vicinus, J .H., Fowler, GB. (1966). Studies in the reliability of assessing
skeletal maturity from x-ray. Part III: Greulich-Pyle atlas and Tanner-Whitehouse method
contrasted. Human Biology. 38, 211-228.

Aicardi, G., Vignolo, M., Milani, S., Naselli, A., Magliano, P., Garzia, P. (2000).
Assessment of skeletal maturity of the hand-wrist and knee: a comparison among
methods. American Journal of Human Biology. 12, 610-615.

Adirim, T.A., Cheng, TL. (2003). Overview of injuries in the young athlete. Sports
Medicine. 33, 75-81.

American Youth Football. (2004). This is American Youth Football. Retrieved April 7,
2004 from http://www.americanyouthfootball.com/about.html

Baxter—J ones, A.D.G. (1995). Growth and development of young athletes - should
competition levels be age-related. Sports Medicine. 20, 59-64.

Bayley, N. (1946). Tables for predicting adult height from skeletal age and present
height. Journal of Pediatrics. 28, 49-64.

Bayley, N., Pinneau, 8.8., (1952). Tables of predicting adult height from skeletal age:
Revised for use with the Greulich-Pyle hand standards. Journal of Pediatrics. 40, 423-
441. -

Bayer, L.M., Bayley, N. (1959). Growth Diagnoses. Berkley, California. University of
California Press.

Beachy, G., Akau, C.K., Martinson, M., Olderr, T.F. (1996). High school sports injuries -
A longitudinal study at Punahou School: 1988 to 1996. American Journal of Sports
Medicine. 25, 675-681.

Beunen, G.P., Ostyn, M., Simons, J ., Renson, R., Van Gerven, D. (1981). Chronological

and biological age as related to physical fitness in boys 12 to 19 years. Annals of Human
Biology. 8, 321-331.

Blyth, C.S., Mueller, FD. (1974). An epidemiological study of high school football
injuries in North Carolina: 1968-72, final report, final report funded and distributed by
U.S. Consumer Product Safety Commission.

Bonat, S., Pathomvanich, A., Keil, M.F., Field, A.E., Yanovski, J .A. (2002). Self-
assessment of pubertal stage in overweight children. Pediatrics. 110, 743-747.

Bramswig, J .H., Fasse, M., Holthoff, M.L., vonLengerke, H.J., von Petrykowski, W.,
Schellong, G. (1990). Adult height in boys and girls with untreated short stature and

i 107

 

constitutional delay of growth and puberty: accuracy of five different methods of height lv
prediction. Journal of Pediatrics. 11 7, 423-441.

Bramwell, S.T., Garrick, J .G., Requa, R.K. (1972). High-school football injuries - a pilot
comparison of playing surfaces. Medicine and Science in Sports and Exercise. 4, 166-
169.

Brooksgunn, J ., Warren, M.F., Rosso, J ., Gargiulo, J. (1987). Validity of self-report
measures of girls pubertal status. Child Development. 58, 829-841.

Cahill, B.R., Griffith, EH. (1978). Effect of preseason conditioning on the incidence
and severity of high school football knee injuries. American Journal of Sports Medicine.
6, 180-184.

Cahill, B.R., Griffith, EH. (1979). Exposure to injury in major college football: a
preliminary report of data collection to determine injury exposure rates and activity risk
factors. American Journal of Sports Medicine. 7, 183-185.

Caine, D.J., Broekhoff, J. (1987). Maturity assessment - a viable preventive measure
against physical and psychological insult to the young athlete. Physician and
Sportsmedicine. 15, 67-80.

Cantu, R.C., Mueller, F.O. (2003a). Brain injury-related fatalities in American football,
1945-1999. Neurosurgery. 52, 846-852.

Cantu, R.C., Mueller, F.O. (2003b). Catastrophic spine injuries in American football,
1977-2001. Neurosurgery. 53, 358-362.

Culpepper, M.I., Niemann, K.M.W. (1983). High-school football injuries in Birmingham,
Alabama. Southern Medical Journal. 76, 873-878.

Dagiau, R.F., Dillman, C.J., Milner, E.K. (1980). Relationship between exposure time
and injury in football. American Journal of Sports Medicine. 8, 25 7-260.

Daniel, T.E., Janssen, C.T.L. (1987). More on relative age effect. Canadian Association
for Health, Physical Education, and Recreation. 53, 21-24.

Delee, J .C., Farney, WC. (1992). Incidence of injury in Texas high-school football.
American Journal of Sports Medicine. 20, 575-5 80.

Demirjian, A., Buschang, P.H., Tanguay, R., Patterson, D.K. (1985). Interrelationships
among measures of somatic, skeletal, dental, and sexual maturity. American Journal of

Orthodontics and Dentofacial Orthopedics. 88, 433-43 8.

Dudink, A. (1994). Birth date and sporting success. Nature. 368, 592-592.

108

Duke,'P.M., Litt, I.F., Gross, R.T. (1980). Adolescents self-assessment of sexual-
maturation. Pediatrics. 66, 918-920.

Edwards, S. (1994). Born too late to win. Nature. 3 70, 186-186.

Eisenman, J .C., Malina, RM. (2003). Age and sex associated variation in neuromuscular
capabilities of adolescent distance runners. Journal of Sport Sciences. 21, 551-557. .

Emery, C .A. (2003). Risk factors for injury in child and adolescent sport: A systematic
review of the literature. Clinical Journal of Sport Medicine. 13, 256-268.

Epstein, L.H., Valoski, A.M., Kalarchian, M.A., McCurley, J. (1995). Do children lose
and maintain weight easier than adults - a comparison of child and parent weight changes
from 6 months to 10 years. Obesity Research 3 (5), 411-417.

Gallagher, J.R. (1969). Problems in matching competitors. Adolescence athletics and
competitive sports. Clinical Pediatrics. 8, 434-436.

Garrick, J .G., Requa, R.K. (1978). Injuries in high-school sports. Pediatrics. 61, 465-
469.

Glamser, F., Marciani, L. (1992). The birth date effect and college athletic participation:
some comparisons. Journal of Sport Behavior. 15, 227-238.

Goldberg, B., Boiardo, R. (1984). Profiling children for sports participation. Clinics in
Sports Medicine. 3, 153-159.

Goldberg, B., Rosenthal, P.P., Robertson, L.S., Nicholas, J .A. (1988). Injuries in youth
football. Pediatrics. 81, 255-261.

Gomez, J .E., Ross, S.K., Calmbach, W.L., Kimmel, R.B., Schmidt, D.R., Dhanda, R.
(1998). Body fatness and increased injury rates in high school football linemen. Clinical
Journal of Sport Medicine. 8, 115-120.

Greulich, W.W., Pyle, 8.1. (1950). Radiographic atlas of skeletal development of the
hand and wrist. Stanford: Stanford University Press.

Groell, R., Lindbichler, F., Riepl, T., Gherra, L., Roposch, A., Fotter, R. (1999). The
reliability of bone age determination in central European children using the Greulich and
Pyle method. British Journal of Radiology. 72, 461-464.

Hafner, J .K., Scott, S.E., Veras, G., Goldberg, B., Nicholas, J .A., Shaffer, T.E. (1982).

Interscholastic athletics - method for selection and classification of athletes. New York
State Journal of Medicine. 82, 1449-1459.

109

Harris, E.F., Weinstein, S., Weinstein, L., Poole, A.E. (1980) Predicting adult stature - a
comparison of methodologies. Annals of Human Biology. 7, 225-234.

Hergenroeder, A.C., Hill, R.B., Wong, W.W., Sangi-Haghpeykar, H., Taylor, W. (1999).
Validity of self-assessment of pubertal maturation in African American and European
American adolescents. Journal of Adolescent Health. 24, 201 -205.

Himes, J .H., Faricy, A. (2001). Validity and reliability of self-reported stature and
weight of US adolescents. American Journal of Human Biology. 13, 255-260.

Himes, J .H., Roche, AF (1982). Reported versus measured adult statures. American
Journal of Physical Anthropology, 58, 335-341.

Hurley, W., Lior, D., Tracze, S. (2001). A proposal to reduce the age discrimination in
Canadian minor hockey. Canadian Public Policy-Analyse De Politiques. 27, 65-75.

Jones, M.A., Hitchen, P.J., Stratton, G. (2000). The importance of considering biological

maturity when assessing physical fitness measures in girls and boys aged 10 to 16 years.
Annals of Human Biology. 27, 57-65.

Katzmarzyk, P.T., Malina, R.M., Beunen, GP. (1997). The contribution of biologic
maturation to the strength and motor fitness of children. Annals of Human Biology. 24,
493-505.

Koprowski, C ., Coates, R.J., Bernstein, L. (2001). Ability of young women to recall past
body size and age at menarche. Obesity Research. 9, 478-485.

Kreipe, R.E., Gewanter, H.L. (1985). Physical maturity screening for participation in
sports. Pediatrics. 75, 1076-1080.

Khamis, H.J., Guo, SM. (1993). Improvement in the Roche—Wainer-Thissen stature

prediction model - a comparative-study. American Journal of Human Biology. 5, 669-
679.

Khamis, H.J., Roche, A.F. (1994). Predicting adult stature without using skeletal age - the
Khamis-Roche method. Pediatrics. 94, 504-507.

Linder, M.M., Townsend, D.J ., Jones, J .C., Balkcom, I.L., Anthony, CR. (1995).
Incidence of adolescent injuries in junior—high-school football and its relationship to

sexual maturity. Clinical Journal of Sport Medicine. 5, 167-170.

Little League Baseball. (2004). 2004 Little League age chart. Retrieved April 7, 2004
from http://www.littleleague.org/commonlforms/agechartpdf

110

Maffulli, N., King, J .B., Helms, P. (1994). Training in elite young athletes (the training
of young athletes (toya) study) - injuries, flexibility and isometric strength. British
Journal of Sports Medicine. 28, 123-136.

Malina, R. M., Bouchard, C. Bar-Or, O. (2004a). Growth, maturation, and physzcal
activity. (2nd ed.). Champaign: Human Kinetics.

Malina, R.M., Bielicki, T. (1996). Retrospective longitudinal growth study of boys and
girls active in sport. Acta Paediatr. 85 (5):570-576.

Malina, R.M., Eisenmann, J .C., Cumming, S., Ribeiro, B., Aroso, J. (2004b). Maturity-
associated variation in the growth and functional capabilities of youth football (soccer)
players 13-15 years. European Journal of Applied Physiology. 91, 555-562.

Malina, R.M., Meleski, B.W., Shoup, RF. (1982). Anthropometric, body-composition,
and maturity characteristics of selected school-age athletes. Pediatric Clinics of North
America 29 (6).“ 1305-1323.

Malina, R.M., Morano, P., Barron, M., Miller. S., Cumming, S. (2002). Incidence and
player riskfactors for injury in youth football. Symposium conducted at the National
Athletic Trainers’ Association, Dallas, Texas.

Malina, R.M., Pena Reyes, M.E., Eisenmann, J .C., Horta, L., Rodrigues, J ., Miller, R.
(2000). Height, mass and skeletal maturity of elite Portuguese soccer players aged 11-16
years. Journal of Sports Sciences. 18, 685-693.

Matsudo, S.M.M., Matsudo, V.K.R. (1994). Self-assessment and physician assessment of
sexual-maturation in Brazilian boys and girls - concordance and reproducibility.
American Journal of Human Biology. 6, 451-455

Mclain, L.G., Reynolds, S. (1989). Sports injuries in a high-school. Pediatrics. 84, 446-
450.

Mota, J ., Guerra, S., Leandro, C ., Pinto, A., Ribeiro, J .C., Duarte, J .A. (2002).
Association of maturation, sex, and body fat in cardiorespiratory fitness. American
Journal of Human Biology. 14, 707-712.

Motulsky, H. (1995). Intuitive biostatistics. (1St ed). Oxford University Press.

Must, A., Phillips, S.M., Naumova, E.N., Blum, M., Harris, 8., Dawson-Hughes, B.,
Rand, W.M. (2002). Recall of early menstrual history and menarcheal body size: after 30
years, how well do women remember. American Journal of Epidemiology. I55, 672-

679.

National Football League (NFL). (2004). NFL history 1971-1980. Retrieved April 7,
2004 from http: //www. nfl. com/history/chronology/l 971- 1980.

111

National SAFE KIDS Campaign (N SKC). Sports Injury Fact Sheet. Washington (DC):
NSKC, 2004.

Ogden, C.L., Kuczmarski, R.J., F legal, K.M., Mei, Z., Guo, S., Wie, R., Grummer-
Strawn, L.M., Curtin, L.R., Roche, A.F., Johnson, CL (2002). Centers for Disease
Control and Prevention 2000 growth charts for the United States: improvements to the
1977 National Centers for Health Statistics Version. Pediatrics, 109(1). 45-60.

Orchard, J. (2002). Is there a relationship between ground and climatic conditions and
injuries in football? Sports Medicine. 32, 419-432.

Orchard, J .W., Powell, J .W. (2003). Risk of knee and ankle: Sprains under various
weather conditions in American football. Medicine and Science in Sports and Exercise.
35,1118-1123.

Pop Warner Little Scholars. (2004). Age and weight structure chart. Retrieved April 7,
2004 from http://www.popwarner.com/football/footballstructure.asp?lable=structure

Powell, J .W. (1980) A multivariate epidemiological model applied to injuries to the knee
in college football: 1975-1978. Doctoral Dissertation, The Pennsylvania State
University. State College.

Powell, J .W., Barber-Foss, K.D. (1999a). Injury patterns in selected high school sports: A
review of the 1995-1997 seasons. Journal of Athletic Training. 34, 277-284.

Powell, J .W., Dompier, T.P. (2004). Analysis of injury rates and treatment patterns for
time-loss and non-time-loss injuries among collegiate student-athletes. Journal ofAthletic
Training. 39, 56-70.

Powell, J .W., Schootman, M. (1992) a multivariate risk analysis of selected playing
surfaces in the National Football League - 1980 to 1989 - an epidemiologic-study of knee
injuries. American Journal of Sports Medicine. 20, 686-694.

Pratt, M. (1989). Strength, flexibility, and maturity in adolescent athletes. American
Journal of Diseases of Children. 143, 560-563.

Radelet, M.A., Lephart, S.M., Rubinstein, E.N., Myers, J .B. (2002). Survey of the injury
rate for children in community sports. Pediatrics. 110, Art. No. e28.

Rarick, G.L., Oyster, N. (1964). Physical maturity, muscular strength, and motor
performance 0 young school-age boys. Research Quarterly. 35, 523-531.

Roche, A.F., Chumlea, W.C., Thissen, D. (1988). Assessing the Skeletal Maturity of the
Hand- Wrist: F els Method. C.C. Thomas, Springfield, Illinois. ~

112

Roche, A.F., Davila, G.H. (1976). Reliability of assessments of maturity of individual
hand-wrist bones. Human Biology. 48, 585-597.

Roche, A.F., Davila, G.H., Eyman, S.L. (1971). Comparison between Greulich-Pyle and
Tanner-Whitehouse assessments of skeletal maturity. Radiology. 98, 273-278.

Roche, A.F., Tyleshevski, F., Rogers, E. (1983). Non-invasive measurements of physical
maturity in children. Research Quarterly For Exercise And Sport. 54, 364-371.

Roche, A.F., Wainer, H., Thissen, D. (1975). RWT method for prediction of adult stature.
Pediatrics. 56, 1026-1033.

Roemmich, J .N., Blizzard, R.M., Peddada, S.D., Malina, R.M., Roche, A.F., Tanner,
J.M., Rogol, AD. (1996). Prediction accuracy of adult height from bone age in boys.
Pediatric Research. 3 9, 566-566.

Schlossberger, N.M., Turner, R.A., Irwin, CE. (1992). Validity of self-report of pubertal
maturation in early adolescents. Journal of Adolescent Health. 13, 109-113.

Simmons, C., Paull, GO (2001). Season-of-birth bias in association football. Journal of
Sports Sciences. 19, 677-686.

Sporting Goods Manufacturers Association. (2004). Organized Sports Attract Millions of
Americans. Retrieved April 7, 2004 from
http://www.sgma.com/press/2004/press1088003422-22941.htm1

Stanaway, K.B., Hines, T. (1995). Lack of a season of birth effect among American
athletes. Perceptual and Motor Skills. 81, 952-954.

Stuart, M.I., Morrey, M.A., Smith, A.M., Meis, J .K., Ortiguera, C .J . (2002). Injuries in
youth football: A prospective observational cohort analysis among players aged 9 to 13
years. Mayo Clinic Proceedings. 77, 317-322.

Tanner, J .M. (1962). Growth and adolescence. (2nd ed.). Blackwell, Oxford.

Tanner, J .M., Healy, M.J R, Goldstein, H., Cameron, N. (2001). Assessment of skeletal
maturity and prediction of adult height (TW3 Method). (3rd ed.). Saunders.

Tanner, J .M., Landt, K.W., Cameron, N., Carter, B.S., Patel, J. (1983). Prediction of adult
height from height and bone-age in childhood. Archives of Disease in Childhood. 58,
767-776.

Tanner, J .M., Whitehouse, R.H., Marshall, W.A., Carter, BS. (1975). Prediction of adult

height from height, bone age, and occurrence of menarche, at ages 4 to 16 with allowance
for midparent height. Archives of Disease in Childhood. 50, 14-26.

113

Taylor, S.J.C., Whincup, P.H., Hindmarsh, P.C., Larnpe, F ., Odoki, K., Cook, DO.
(2001). Performance of a new pubertal self-assessment questionnaire: a preliminary
study. Paediatric and Perinatal Epidemiology. 15, 88-94.

Thompson, A., Barnsley, R., Stebelsky, G. (1991). “Born to play ball”: The relative age
effect and major league baseball. Sociology of Sport Journal. 8, 146-151.

Thompson, N., Halpem, B., Curl, W.W., Andrews, J .R., Hunter, S.C., Mcleod, W.D.
(1987). High-school football injuries — evaluation. American Journal of Sports Medicine.
15, 117-124.

Thompson, N.J., Morris, RD. (1994). Predicting injury risk in adolescent football players
- the importance of psychological variables. Journal of Pediatric Psychology. 19, 415-
429.

Torg, J.S., Quedenfeld, T. (1971). Effect of shoe type and cleat length on incidence and

severity of knee injuries among high school football players. Research Quarterly. 42,
203-211.

Turbeville, S.D., Cowan, L.D., Asal, N.R., Owen, W.L., Anderson, M.A. (2003a). Risk
factors for injury in middle school football players. American Journal of Sports Medicine.
31, 276-281.

Turbeville, S.D., Cowan, L.D., Owen, W.L., Asal, N.R., Anderson, M.A. (2003b). Risk

factors for injury in high school football players. American Journal of Sports Medicine.
31, 974-980. -

‘

Van Lenthe, F.J., Kemper, H.C.G., Van Mechelen, W. (1998). Skeletal maturation in
adolescence: a comparison between the Tanner-Whitehouse ii and the Fels method.
European Journal of Pediatrics. 15 7, 798-801.

Vignolo, M., Milani, S., Cerbello, G., Coroli, P., Dibattista, E., Aicardi, G. (1992). Fels,
Greulich-Pyle, And Tanner-Whitehouse bone-age assessments in a group of Italian
children and adolescents. American Journal of Human Biology. 4, 493-500.

Violette, R.W. (1976). An epidemiologic investigation of junior high school football
injury and its relationship to certain physical and maturational characteristics of the

players. Doctoral Dissertation, University of North Carolina, Chapel Hill.

Verhulst, J. (1992). Seasonal birth distribution of West European soccer players: a
possible explanation. Medical Hypotheses. 38, 346-348.

Wainer, H., Roche, A.F., Bell, S. (1978). Predicting adult stature without skeletal age and
without paternal data. Pediatrics. 61, 569-572.

114

 

Wing, R.R., Epstein, L.H., Neff, D. (1980). Accuracy of parents reports of height and
weight

Journal of Behavioral Assessment. 2 (2): 105-110.

Zachman, M., Sobradillo, B., Frank, M., Frisch, H., Prader, A. (1978). Bayley-Pinneau,
Roche-Wainer-Thissen, and Tanner height predictions in normal children and in patients
with various pathologic conditions. Journal of Pediatrics. 93, 749-755.

Zarow, R. (1997). Adult height prediction in boys by various methods. Biology of Sport.
14,193-198.

Zerin, J .M., Hernandez, R.J. Approach to skeletal maturation. Hand Clinics. 7, 53-62.

115

   

1111111111111111