Key words soccer - growth and development - epidemiology - sports medicine
Introduction
Elite football (soccer) academies guide players through structured and intensive
training programs to optimize long-term development. Injuries impact these
opportunities and identification of risk factors is an important step towards
reducing injury occurrence and severity by informing targeted injury risk reduction
strategies [1 ]. In youth football, emerging
evidence suggests an association between growth, maturation, and injury risk [2 ]
[3 ].
However, further research is needed to better understand these relationships and how
the impact of injuries may be reduced during periods of rapid changes to an
athlete’s body.
Growth represents an increase in the size of the body as a whole or of a specific
body part, assessed using anthropometric measures [4 ]. An accelerated period of somatic growth is observed during the
adolescent years, with peak height velocity (PHV) and peak weight velocity (PWV) for
an average boy occurring around the age of 13 to 14 years and 14 to 15 years,
respectively [4 ]. There is, however, a wide
range in both timing (age at PHV from 12 to 17 years) and intensity (PHV from 5 to
12 cm/year) [5 ]
[6 ]. Studies in high-level football have
indicated that phases characterized by rapid growth are associated with greater
overall, overuse, acute, and non-contact injury risk [7 ]
[8 ]
[9 ]
[10 ]
[11 ]
[12 ]
[13 ]
[14 ],
and that players with faster PHV have a greater overall and growth-related injury
burden compared to players with average or slower peak growth rates [15 ]. Still, it is difficult to provide clear
recommendations as the associations are inconsistent and methodological differences
make direct comparisons or pooling of results impossible.
Maturation is defined by Malina et al. [4 ] as
the process of becoming mature or progressing towards a mature (adult) state
(e. g., a fully ossified skeleton, adult height, or a functional
reproductive system). Timing and tempo vary greatly between individuals (onset of
puberty from 9 to 14 years in boys) [4 ], and
the maturity status (where an individual is in the process at a given point) of two
players who train and compete in the same age group can therefore differ
substantially [4 ]. Only a few studies have
related skeletal maturity – considered the single best marker of biological
maturity [16 ] – to injury risk in
high-level youth football [17 ]
[18 ]
[19 ]
[20 ]. While suggesting that
maturity plays a role in the occurrence of certain injury types (e. g.,
apophyseal or osteochondral injuries, muscle injuries, joint/ligament
injuries) [18 ]
[19 ]
[20 ], results are not
consistent. Observed age-related injury patterns do, however, indicate that older
players are at greater risk and that different pathology types are more prominent in
younger, compared to older, age groups [21 ]
[22 ]
[23 ].
Since firm conclusions cannot be drawn from the current pool of literature, the aim
of this study was to explore growth velocity, skeletal maturity, and chronological
age as injury risk factors in male academy players. More specifically, our main
research question was whether changes in height or body mass between assessments at
the start, middle, and end of a football season were related to the occurrence of
specific injury types when also taking age (chronological or skeletal),
growth×age/maturity interaction effects, and football exposure into
consideration. As an exploratory study, no a priori hypothesis was stated.
Materials and Methods
Study design and participants
We used injury, individual training and match exposure, anthropometric and
maturity data collected prospectively over three seasons (2016/17
through 2018/19) in the U13, U14, and U15 age groups at one elite
national football academy in Qatar (2016/17: 64 players in the program,
2017/18: 77 players, 2018/19: 91 players). Participants were
boys aged 11 to 15 years and full-time players typically participated in eight
sessions during the school week, in addition to local club games on weekends.
Part-time players participated in five sessions in addition to weekend club
games. Written informed consent to use routinely collected monitoring data for
research purposes was obtained from the players’ guardians and ethics
approval was granted from the Anti-Doping Lab Qatar Institutional Review Board
(IRB Application #20140000012).
Recording of injuries and football exposure
Training and match injuries were recorded by the designated team physiotherapist
(i. e., one physiotherapist per team) who was present at all team
sessions, supervised by two researchers. Recording procedures followed the
recommendations from Fuller et al. [24 ],
including only time-loss injuries (to reduce bias associated with using several
clinical recorders over multiple seasons [25 ]), i. e., any physical complaint leading to the medical
staff partially or fully restricting participation in future football
activities. Diagnoses were confirmed by a sports medicine physician (employed
full-time at the academy) in collaboration with the treating physiotherapist and
reported based on the Sports Medicine Diagnostic Coding System (SMDCS) [26 ], alongside details about the date of
injury and mechanism. No inter-rater reliability data is available regarding the
diagnosis or classification of injuries during the observation period. Each
team’s designated sports scientist recorded individual training and
match exposure.
Following the completion of the data collection, a researcher converted injury
diagnoses to the updated SMDCS categories for tissue and pathology types [27 ]
[28 ], and retrospectively allocated onset based on the reported
mechanism and diagnosis following the consensus recommended definitions by Bahr
et al. (“sudden”: resulting from a specific identifiable event,
or “gradual”: lack of definable sudden, precipitating event)
[28 ]. Only index injuries were
considered for this study while recurrent injuries were excluded; these were
defined as a time-loss injury to the same location of the same type as a
previous injury recorded during the observation period [28 ].
Anthropometric and maturity assessments
Measures of standing height and body mass were obtained at the start, middle, and
end of each academy season by trained sports scientists following the
recommendations outlined by Stewart et al. [29 ]. Measures were taken in the morning, prior to any activities to
minimize diurnal variations. Standing height was measured to the nearest 0.1 cm
applying the stretch stature method using a wall-mounted stadiometer (Holtain
Ltd, Crymych, UK) and body mass was measured to the nearest 0.1 kg using digital
scales (Adam Equipment, Milton Keynes, UK). Previously published test-retest
data in a subsample of 17 academy players revealed a standard error of
measurement (SEM) of 0.34 cm (95% confidence interval (CI): 0.25 to 0.52
cm) for standing height [30 ]. This
corresponds to a minimal detectable change (MDC) of approximately 1 cm.
Skeletal maturity was assessed at the beginning of the season using x-ray images
of the player’s left hand/wrist complex taken at Aspetar
Orthopaedic and Sports Medicine Hospital. Skeletal age was determined using the
Fels method [31 ] by one trained assessor.
Intra-rater reliability for this method has previously been reported (intraclass
correlation coefficient (ICC): 0.998, 95% CI: 0.996 to 0.999) [17 ].
Inclusion and exclusion of growth intervals
It has been suggested that researchers examine growth over shorter periods of
time to better account for non-linear growth patterns related to saltatory
(episodic) growth [32 ]. To minimize the
impact of measurement error on estimations and allow for detection of meaningful
changes [32 ]
[33 ], we calculated growth velocity per
academy semester, defined as the two intervals from season start
(August/September) to mid-season (January) and from mid-season to season
end (May/June). The absolute change (cm or kg) was divided by the number
of days between measurements and converted to expressions equivalent to
cm/year and kg/year, respectively [12 ]
[13 ]. For a growth interval to be included in the final analyses, a
skeletal maturity assessment had to be available for the given player and season
(assessed maximum 91 days within the start of the season).
Statistical analyses
Descriptive statistics are presented as means with standard deviation (SD). Four
separate mixed-effects logistic regression models (xtmelogit command)
estimated associations for the effects of changes in height and body mass on the
occurrence of overall, gradual onset, sudden onset, bone tissue and physis
injury. Growth velocity for height and body mass were specified as distinct
growth-related predictor variables (fixed effects). Models were adjusted for
chronological or skeletal age, and growth×age/maturity
interaction, with player specified as a random effect plus a random intercept.
The average weekly load (hours of training/match exposure per week)
during the growth interval (or until the event if an injury occurred) was added
as a covariate.
The Akaike Information Criterion (AIC) assessed the relative quality of each
mixed-effects logistic regression model in the set of candidate models. The
Akaike difference (ΔAIC) from the estimated best model (i. e.,
the model with the lowest AIC value; ΔAIC=0) was evaluated
according to the following scale: 0–2, essentially equivalent;
>2–7, plausible alternative; >7–14, weak
support; >14, no empirical support [34 ]. Akaike weights (w
i ) provided a scaled
interpretation about the relative quality of each competing model as the
probability that a given model is the best in the set of four candidate models
per endpoint. Thresholds for the adjusted odds ratios (OR) of 0.9, 0.7, 0.5,
0.3, and 0.1 and their reciprocals 1.11, 1.43, 2.0, 3.3, and 10 defined small,
moderate, large, very large, and extremely large beneficial and harmful effects,
respectively [35 ]. In the absence of an
established anchor defining practically relevant associations between growth
velocity and injury occurrence, we considered OR=0.90 or OR=1.11
to define substantially beneficial and substantially harmful effects,
respectively [35 ]. Associations were
declared practically relevant based on the location of the confidence interval
for the estimated ORs to these thresholds. Outcome statistics are reported as
point estimates and 95% CI. Statistical analyses were performed using
Stata (StataBE v17.0; StataCorp, College Station, TX, USA).
Results
Inclusion of players and growth intervals
The inclusion of player-seasons and growth intervals are shown in [Fig. 1 ], with an overview of exclusions
due to incomplete assessments. The final sample included 95 unique players
contributing to 223 growth intervals (17 players with one growth interval, 48
with two, 14 with three, 12 with four, and 4 with five intervals), with a mean
duration of 113 days (SD 24). A total of 161 index injuries (93 from training
sessions, 68 from matches) and 21 712 exposure hours (18 642 training hours and
3070 match hours) were recorded within the growth intervals included (overall
incidence: 7.4 injuries per 1000 h, training incidence: 5.0 per 1000 h; match
incidence: 22.1 per 1000 h). The most common injury locations were the thigh
(22%), hip/groin (16%), and ankle (16%), while
the most common pathology types were superficial contusions (20%),
physis injuries (19%), and muscle injuries (16%). A detailed
injury overview is included in [Table
1 ].
Fig. 1 Inclusion of player seasons and growth intervals, with an
overview of exclusions due to missing data. Player season: One player
taking part in one season. Eligible player season: Player season with
minimum two anthropometric measures. Eligible growth interval: An
interval spanning from either the start to the middle of the season or
the middle to the end of the season, with complete anthropometric
assessments on both sides. Included growth intervals: An interval
(Start-Mid or Mid-End) with a skeletal age assessment for the given
season for a player.
Table 1 Overview of the 161 index injuries sustained within the
growth periods included in the analyses, structured by location,
onset, and pathology type.
Location Onset Pathology
type
Injuries (count)
Incidence (Inj. per 1000 h, 95% CI)
Head and neck
6
0.28
(0.10 to 0.60)
Sudden onset
6
0.28
(0.10 to 0.60)
Brain/spinal cord
5
0.23
(0.07 to 0.54)
Laceration
1
0.05
(0.00 to 0.26)
Upper limb
18
0.83
(0.49 to 1.31)
Sudden onset
18
0.83
(0.49 to 1.31)
Fracture
13
0.60
(0.32 to 1.02)
Contusion (superficial)
4
0.18
(0.05 to 0.47)
Joint sprain
1
0.05
(0.00 to 0.26)
Trunk
7
0.32
(0.13 to 0.66)
Sudden onset
4
0.18
(0.05 to 0.47)
Contusion (superficial)
2
0.09
(0.01 to 0.33)
Fracture
1
0.05
(0.00 to 0.26)
Non-specific
1
0.05
(0.00 to 0.26)
Gradual onset
3
0.14
(0.03 to 0.40)
Bone stress injury
2
0.09
(0.01 to 0.33)
Non-specific
1
0.05
(0.00 to 0.26)
Hip/groin
26
1.20
(0.78 to 1.75)
Sudden onset
4
0.18
(0.05 to 0.47)
Contusion (superficial)
2
0.09
(0.01 to 0.33)
Muscle injury
1
0.05
(0.00 to 0.26)
Non-specific
1
0.05
(0.00 to 0.26)
Gradual onset
22
1.01
(0.64 to 1.53)
Physis injury
20
0.92
(0.56 to 1.42)
Bone stress injury
1
0.05
(0.00 to 0.26)
Bursitis
1
0.05
(0.00 to 0.26)
Thigh
35
1.61
(1.12 to 2.24)
Sudden onset
31
1.43
(0.97 to 2.03)
Muscle injury
20
0.92
(0.56 to 1.42)
Muscle contusion
5
0.23
(0.07 to 0.54)
Non-specific
5
0.23
(0.07 to 0.54)
Cartilage
1
0.05
(0.00 to 0.26)
Gradual onset
4
0.18
(0.05 to 0.47)
Physis injury
3
0.14
(0.03 to 0.40)
Non-specific
1
0.05
(0.00 to 0.26)
Knee
19
0.88
(0.53 to 1.37)
Sudden onset
11
0.51
(0.25 to 0.91)
Contusion (superficial)
7
0.32
(0.13 to 0.66)
Joint sprain
2
0.09
(0.01 to 0.33)
Fracture
1
0.05
(0.00 to 0.26)
Non-specific
1
0.05
(0.00 to 0.26)
Gradual onset
8
0.37
(0.16 to 0.73)
Physis injury
5
0.23
(0.07 to 0.54)
Synovitis/capsulitis
2
0.09
(0.01 to 0.33)
Cartilage
1
0.05
(0.00 to 0.26)
Lower leg
12
0.55
(0.29 to 0.97)
Sudden onset
11
0.51
(0.25 to 0.91)
Muscle injury
5
0.23
(0.07 to 0.54)
Contusion (superficial)
3
0.14
(0.03 to 0.40)
Non-specific
2
0.09
(0.01 to 0.33)
Fracture
1
0.05
(0.00 to 0.26)
Gradual onset
1
0.05
(0.00 to 0.26)
Bone stress injury
1
0.05
(0.00 to 0.26)
Ankle
25
1.15
(0.75 to 1.70)
Sudden onset
20
0.92
(0.56 to 1.42)
Joint sprain
11
0.51
(0.25 to 0.91)
Contusion (superficial)
9
0.41
(0.19 to 0.79)
Gradual onset
5
0.23
(0.07 to 0.54)
Synovitis / capsulitis
5
0.23
(0.07 to 0.54)
Foot
13
0.60
(0.32 to 1.02)
Sudden onset
9
0.41
(0.19 to 0.79)
Contusion (superficial)
5
0.23
(0.07 to 0.54)
Fracture
2
0.09
(0.01 to 0.33)
Joint sprain
2
0.09
(0.01 to 0.33)
Gradual onset
4
0.18
(0.05 to 0.47)
Physis injury
2
0.09
(0.01 to 0.33)
Non-specific
2
0.09
(0.01 to 0.33)
Age, skeletal maturity, and growth velocity
The mean age at the start of a growth interval was 13.5 years (SD 0.8; range 11.9
to 15.0). Considering each player-season only once (a player could have two
growth intervals per season but only one maturity assessment), the mean skeletal
age at the start of the season was 14.4 years (SD 1.6) with skeletal ages
ranging from 10.7 to 14.9 years in the U13 age group, 11.8 to 18.0 in U14, and
13.7 to 17.8 in U15. On average, players were 1.0 year (1.1; –1.5 to
4.7) advanced in skeletal age relative to chronological age. One player was
skeletally mature (skeletal age 18 years), while 62 (45%) could be
considered early maturing (skeletal age minimum one year in advance of
chronological age), 70 (51%) as on time (skeletal age within one year)
and four (3%) as late maturing (skeletal age minimum one year delayed)
[36 ]. The mean semester growth
velocity was 6.3 cm/year (3.5; 0.0 to 17.8) for height and 5.3
kg/year (5.5; –14.6 to 19.3) for body mass.
Relative model quality
The relative model quality of the four model combinations within the five injury
categories is presented in [Table 2 ].
Growth velocity for body mass combined with skeletal age best explained the
overall and gradual onset injury risk, while change in height combined with
skeletal age best explained the risk of sudden onset, bone tissue, and physis
injuries. Other model combinations were, however, considered equivalent or
plausible alternatives.
Table 2 Relative model quality for each injury category.
Model
AIC
Δ AIC
w
i
Inference
Overall (119 events)
Δ Body mass & skeletal age
306.8
0.0
0.59
Best
Δ Height & skeletal age
308.0
1.2
0.33
Essentially equivalent
Δ Height & chronological age
311.6
4.8
0.05
Plausible alternative
Δ Body mass & chronological age
312.9
6.1
0.03
Plausible alternative
Sudden onset (90 events)
Δ Height & skeletal age
300.3
0.0
0.52
Best
Δ Body mass & skeletal age
300.8
0.5
0.41
Essentially equivalent
Δ Body mass & chronological age
305.2
4.8
0.05
Plausible alternative
Δ Height & chronological age
306.5
6.1
0.02
Plausible alternative
Gradual onset (42 events)
Δ Body mass & skeletal age
216.4
0.0
0.43
Best
Δ Height & skeletal age
216.7
0.3
0.37
Essentially equivalent
Δ Height & chronological age
219.3
2.9
0.10
Plausible alternative
Δ Body mass & chronological age
219.4
3.0
0.10
Plausible alternative
Bone tissue (49 events)
Δ Height & skeletal age
238.1
0.0
0.51
Best
Δ Height & chronological age
239.7
1.7
0.22
Essentially equivalent
Δ Body mass & skeletal age
240.4
2.3
0.16
Plausible alternative
Δ Body mass & chronological age
241.1
3.0
0.11
Plausible alternative
Physis injury (27 events)
Δ Height & skeletal age
166.2
0.0
0.51
Best
Δ Height & chronological age
168.2
2.0
0.19
Plausible alternative
Δ Body mass & skeletal age
168.6
2.4
0.15
Plausible alternative
Δ Body mass & chronological age
168.7
2.5
0.15
Plausible alternative
AIC, Akaike Information Criteria; Δ AIC, Akaike difference;
w
i
, Akaike weights.
Effects of growth velocity, age, and skeletal maturity
[Table 3 ] gives a complete overview of all
model outcomes. Effects for load and growth×age/maturity
interaction were not practically relevant for any injury categories. Practically
relevant harmful effects of older age were observed for overall and sudden onset
injury risk in the models adjusting for changes in height and body mass,
respectively. Significant associations (p<0.05) between greater change
in body mass (in the model with chronological age) and more advanced maturity
(in the model with body mass change) were seen for sudden onset injury risk.
These results were not practically relevant given our predefined thresholds
(95% CI for OR <0.9 or >1.1). Significant, but not
practically relevant, associations were also found between higher football load
and risk of gradual onset (all model combinations), bone tissue (all model
combinations), and physis injuries (only for the model including body mass
change and chronological age).
Table 3 Odds ratios for the four model combinations within each
of the five injury outcomes. Numbers in italics indicate significant
associations (p<0.05), while asterisks indicate practically
relevant findings based on our predefined thresholds (95% CI
for OR <0.9 or >1.1).
Outcome
Model
Odds ratio (95% CI)
p-value
Overall
Δ Height (cm/year)
3.78 (0.79 to 18.03)
0.10
(119 events)
Chronological age
2.61 (1.15 to 5.91)
0.022*
Δ Height×chronological age
0.91 (0.81 to 1.02)
0.10
Hours per week
1.06 (0.94 to 1.21)
0.35
Δ Height (cm/year)
1.43 (0.59 to 3.43)
0.43
Skeletal age
1.40 (0.97 to 2.03)
0.07
Δ Height×skeletal age
0.98 (0.92 to 1.04)
0.44
Hours per week
1.06 (0.93 to 1.20)
0.38
Δ Body mass (kg/year)
1.82 (0.67 to 4.92)
0.24
Chronological age
1.65 (0.99 to 2.73)
0.05
Δ Body mass×chronological age
0.96 (0.89 to 1.03)
0.29
Hours per week
1.08 (0.95 to 1.22)
0.26
Δ Body mass (kg/year)
0.92 (0.54 to 1.58)
0.77
Skeletal age
1.21 (0.97 to 1.51)
0.09
Δ Body mass×skeletal age
1.01 (0.97 to 1.05)
0.59
Hours per week
1.06 (0.94 to 1.20)
0.36
Sudden onset
Δ Height (cm/year)
2.26 (0.53 to 9.58)
0.27
(90 events)
Chronological age
1.95 (0.93 to 4.10)
0.08
Δ Height×chronological age
0.94 (0.85 to 1.05)
0.26
Hours per week
1.06 (0.94 to 1.21)
0.33
Δ Height (cm/year)
1.36 (0.59 to 3.13)
0.47
Skeletal age
1.38 (0.98 to 1.94)
0.07
Δ Height×skeletal age
0.98 (0.92 to 1.04)
0.45
Hours per week
1.06 (0.93 to 1.20)
0.40
Δ Body mass (kg/year)
2.81 (1.02 to 7.79)
0.046
Chronological age
1.98 (1.17 to 3.37)
0.011*
Δ Body mass×chronological age
0.93 (0.86 to 1.00)
0.06
Hours per week
1.07 (0.94 to 1.22)
0.29
Δ Body mass (kg/year)
1.18 (0.69 to 2.02)
0.54
Skeletal age
1.30 (1.04 to 1.63)
0.021
Δ Body mass×skeletal age
0.99 (0.96 to 1.03)
0.69
Hours per week
1.06 (0.94 to 1.21)
0.35
Gradual onset
Δ Height (cm/year)
1.93 (0.26 to 14.11)
0.52
(42 events)
Chronological age
1.38 (0.48 to 4.00)
0.55
Δ Height×chronological age
0.96 (0.83 to 1.11)
0.56
Hours per week
1.22 (1.01 to 1.48)
0.035
Δ Height (cm/year)
1.35 (0.44 to 4.16)
0.60
Skeletal age
1.13 (0.70 to 1.84)
0.61
Δ Height×skeletal age
0.98 (0.91 to 1.06)
0.69
Hours per week
1.23 (1.01 to 1.48)
0.035
Δ Body mass (kg/year)
1.13 (0.30 to 4.24)
0.85
Chronological age
0.91 (0.44 to 1.88)
0.79
Δ Body mass×chronological age
1.00 (0.90 to 1.10)
0.94
Hours per week
1.24 (1.02 to 1.50)
0.027
Δ Body mass (kg/year)
0.79 (0.34 to 1.81)
0.57
Skeletal age
0.89 (0.62 to 1.29)
0.55
Δ Body mass×skeletal age
1.02 (0.96 to 1.08)
0.46
Hours per week
1.23 (1.02 to 1.49)
0.031
Bone tissue
Δ Height (cm/year)
0.72 (0.13 to 4.05)
0.71
(49 events)
Chronological age
0.63 (0.25 to 1.57)
0.32
Δ Height×chronological age
1.03 (0.90 to 1.17)
0.69
Hours per week
1.17 (1.00 to 1.36)
0.048
Δ Height (cm/year)
1.17 (0.45 to 3.01)
0.75
Skeletal age
0.96 (0.64 to 1.43)
0.83
Δ Height×skeletal age
0.99 (0.93 to 1.06)
0.80
Hours per week
1.17 (1.01 to 1.37)
0.042
Δ Body mass (kg/year)
1.50 (0.49 to 4.57)
0.48
Chronological age
0.81 (0.45 to 1.46)
0.48
Δ Body mass×chronological age
0.97 (0.90 to 1.06)
0.52
Hours per week
1.18 (1.01 to 1.37)
0.040
Δ Body mass (kg/year)
1.06 (0.56 to 2.01)
0.86
Skeletal age
0.89 (0.68 to 1.17)
0.41
Δ Body mass×skeletal age
1.00 (0.95 to 1.04)
0.94
Hours per week
1.18 (1.01 to 1.37)
0.040
Physis injury
Δ Height (cm/year)
1.76 (0.22 to 14.37)
0.60
(27 events)
Chronological age
0.95 (0.29 to 3.07)
0.93
Δ Height×chronological age
0.97 (0.83 to 1.13)
0.66
Hours per week
1.20 (0.98 to 1.47)
0.08
Δ Height (cm/year)
1.75 (0.55 to 5.59)
0.35
Skeletal age
1.12 (0.66 to 1.89)
0.68
Δ Height×skeletal age
0.97 (0.89 to 1.05)
0.44
Hours per week
1.21 (0.98 to 1.48)
0.07
Δ Body mass (kg/year)
1.74 (0.40 to 7.65)
0.46
Chronological age
0.75 (0.32 to 1.75)
0.51
Δ Body mass×chronological age
0.97 (0.86 to 1.08)
0.53
Hours per week
1.23 (1.01 to 1.51)
0.044
Δ Body mass (kg/year)
0.82 (0.32 to 2.11)
0.68
Skeletal age
0.78 (0.50 to 1.20)
0.25
Δ Body mass×skeletal age
1.02 (0.95 to 1.09)
0.58
Hours per week
1.23 (1.00 to 1.52)
0.05
Discussion
This study explored growth velocity and age (skeletal and chronological) as injury
risk factors, accounting for growth×age/maturity interaction effects
and individual training and match exposure. Based on prospective data from 95 unique
players between 11 and 15 years over three seasons, we observed harmful effects of
older age on overall and sudden onset injury risk. Significant associations were
also found for greater change in body mass and more advanced maturity on sudden
onset injury risk, and for greater football load on gradual onset, bone tissue, and
physis injuries; however, these were not considered practically relevant based on
our pre-defined thresholds. No significant growth×age/maturity
interaction effects were seen.
Associations between growth velocity and injury risk remain unclear
A potential link between growth velocity and injury risk is typically attributed
to underlying mechanisms such as tissues adapting at different rates, increased
tension on apophyses, or decreased neuromuscular control [3 ]
[37 ]
[38 ]
[39 ]. In support of such a link,
associations between changes in height and injuries have been reported in Dutch,
Belgian, and English high-level football players [11 ]
[12 ]
[13 ]
[14 ]. Although these studies suggest a
growth-injury relationship, different analytical approaches and broad injury
categories make definite conclusions difficult. To improve our understanding of
growth as an injury risk factor, we included bone tissue and physis injuries as
specific outcomes, based on the assumption that they are more closely aligned
with the suggested underlying mechanisms. Importantly, these injuries are among
the most common and burdensome in U13-U15 age group players [21 ]
[22 ]
[23 ], making them a priority
for targeted injury reduction programs. The mean age of our sample (13.5 years)
was also close to the expected age at PHV (13.6 years) in this specific academy
population [30 ], which has been
highlighted as a vulnerable phase for football players [7 ]
[8 ]
[9 ]
[10 ]. Still, we did not find any practically
relevant effects of changes in height or body mass over an academy semester,
with only one significant association suggesting an increased risk of sudden
onset injuries with greater changes in body mass in the model adjusting for
chronological age.
The absence of observed effects in our study may be explained by the
methodological approach, using isolated and pre-determined growth intervals
(i. e., Start-Mid and Mid-End of a season). The duration of growth
intervals reflected recommendations for assessing growth velocity
(i. e., every 3 to 4 months [33 ]),
but they do not necessarily capture the periods of most rapid growth within an
individual’s growth process. They may also not be frequent enough to
identify shorter bursts of growth, which could be of interest [32 ]. Capturing data on a more frequent
basis is, however, associated with greater variance in the estimated growth
velocities [40 ], and recommendations to
focus on long-term tracking of anthropometric data therefore seem reasonable
[33 ]. This approach was taken by
Monasterio et al. [10 ]
[15 ], who calculated full growth curves.
They did, unfortunately, not have individual training and match data available
and growth curves could only be calculated for around 10% of their
initial sample, highlighting the logistical challenges associated with
accurately tracking growth, injuries, and individual exposure over a sufficient
duration in applied academy settings.
Practically relevant effects of age, but not maturity, on injury risk
Studies have indicated a changing injury pattern with age, where physeal or
“growth-related” injuries are more common in younger age groups
[21 ]
[22 ]
[23 ]. This may be
attributed, in part, to maturity status [18 ]
[19 ]
[41 ], with the immature skeleton
representing a relatively weak link in the muscle-tendon-bone chain prior to
reaching its mature state [42 ].
Consequently, similar injury mechanisms would result in different pathology
types (e. g., apophysitis or avulsions as opposed to tendinopathies or
muscle strains) depending on a player’s maturity status [42 ]. These patterns provide a rationale for
regular maturity assessments, which can be used to better accommodate for early,
on time, and late maturing players within an age group, who may be prone to
different injury types at different locations.
In our sample, higher chronological age was related to a practically relevant
increased risk of overall and sudden onset injuries, with point estimates
suggesting moderate to large harmful effects on these injury outcomes. This is
in line with our earlier study, which demonstrated increased injury incidence in
older age groups in this academy [21 ].
Although large variations in skeletal age were observed within age groups
(e. g., a six-year range in skeletal age between players in the U14 age
group), we did not detect any practically relevant effects of maturity on injury
risk. A significant association was found between older skeletal age and sudden
onset injury risk; still, we urge caution when interpreting this, as we could
not demonstrate consistent associations across models. This may be due to a
relatively small number of events for the number of variables included in our
models, and studies including larger samples are needed to better understand
these relationships.
Weekly load may affect the risk of gradual onset and bone tissue
injuries
Football load (operationalized as hours per week in this study) was a significant
covariate in all model combinations for gradual onset and bone injuries, and in
one model for physis injuries. Again, these associations could not be classified
as practically relevant, and a causal relationship cannot be established based
on our study design. Youth-specific consensus statements have previously
highlighted training load as a risk factor for these injury types [43 ]
[44 ], and studies in high-level youth football have related weekly
duration, cumulative absolute training loads over three and four weeks, and
greater week-to-week changes in exposure to injury risk [14 ]
[45 ]
[46 ]. While their findings
are inconsistent and our results do not provide definite answers, it does appear
sensible to focus on careful progression, diverse and variable movement
exposure, careful scheduling of total load, and allowing for sufficient rest and
recovery to ensure positive adaptations of training while minimizing the risk of
injury in youth settings [3 ]
[43 ]
[44 ]. Future studies may be able to incorporate more detailed measures
of load in their analyses to determine whether training load or athlete
responses (e. g., physiological, perceptual) are associated with injury
occurrence during specific phases of the growth and maturation process.
Methodological considerations
We addressed several limitations from previous studies on growth, maturation, and
injury risk with our models accounting for growth×age/maturity
interaction, repeated player-seasons, and daily individual football exposure,
including more detailed injury outcomes recorded and verified by on-site medical
staff. It is indeed rare that youth football teams have designated medical staff
present at each session, and that the detailed injury data they captured can be
merged with accurate individual football exposure and skeletal age assessments.
Still, there are limitations with our study design that should be considered
when interpreting our results.
First, our growth and maturity data were primarily collected for clinical and
applied purposes, which led to a substantial number of incomplete assessments
([Fig. 1 ]), presumably at random and
not related to injury occurrence. Using data collected for ongoing monitoring
purposes also introduces the possibility that training content could have been
adjusted based on the data collected. However, we do not believe that these
assessments impacted significantly on training content during the data
collection period, as systematic and contextualized reports were not available
to coaches at the time. The exploratory nature of our study, involving a
retrospective analysis, also precluded any formal a priori sample size
estimation relevant to a mixed-effects modeling framework. Future studies could
improve in this area by employing strict data collection routines to avoid
missing data and should attempt to include a larger and younger sample of
players than we were able to in our study, to cover the entire growth process
and allow for within-subject analyses.
Second, including skeletal age as our main indicator of biological maturity is
arguably a strength as it is considered a precise and reliable measure and can
be used across the maturation process [4 ]
[16 ]. However, we recognize
its limitations. The assessment requires equipment and expertise (and is
therefore costly), and importantly involves low-dose radiation (although this
has been described as almost negligible [47 ]). It is therefore considered an invasive measure that is not
feasible to include in most academy settings [3 ], reducing the direct applicability of our findings to practice.
Furthermore, the ossification of the hand-wrist complex does not necessarily
represent the maturity status of bones in other locations. Future studies may be
able to use radiation-free alternatives, such as magnetic resonance imaging or
ultrasound [47 ], to relate skeletal
maturity of multiple body parts to injuries in surrounding tissues.
Finally, although we could expand on findings from previous studies by using
injury outcomes that are more closely aligned with the proposed mechanisms for a
growth-maturity-injury relationship, the use of a time-loss definition limited
our ability to capture injuries that did not affect participation [28 ]. Future studies could improve in this
area by using athlete-reported measures with a broader, any-complaint
definition. Readers should also be cognizant that a study like ours may not be
representative for other contexts (e. g., different geographical
regions, training environments, or match schedules/formats) and
addresses but a few of many potential risk factors for injury, which are thought
to be both multifactorial, dynamic, and complex [48 ].
Conclusion
Our main research question was whether changes in height or body mass between
assessments at the start, middle, and end of a football season affect the risk of
injury in male academy football players. After accounting for age or skeletal
maturity, growth×age/maturity interaction effects, and football
exposure, no such associations were considered practically relevant based on our
pre-defined thresholds. We do, however, report practically relevant harmful effects
of older age on overall and sudden-onset injury risk. Additionally, significant
– but not practically relevant – associations were observed between
greater changes in body mass and older skeletal age on sudden onset injury risk, and
higher weekly load on gradual onset, bone tissue, and physis injuries. Researchers
should strive to establish robust surveillance systems that can include larger and
more diverse samples (athletes of both genders from multiple settings) and capture
the whole growth and maturity process (longitudinal tracking starting from
pre-adolescence) alongside reliable collection of injury and exposure data
(following consensus procedures and universal classification systems), to better
understand the relationship between growth, maturation, and injury risk.
Bibliographical Record
