Customize your JAMA Network experience by selecting one or more topics from the list below.
Marshall SW, Mueller FO, Kirby DP, Yang J. Evaluation of Safety Balls and Faceguards for Prevention of Injuries in Youth Baseball. JAMA. 2003;289(5):568–574. doi:10.1001/jama.289.5.568
Context Safety balls and faceguards are widely used in youth baseball, but their
effectiveness in reducing injury is unknown.
Objective To evaluate the association of the use of faceguards and safety balls
and injuries in youth baseball.
Design, Setting, and Participants Ecological study using a national database of compensated insurance
claims maintained by Little League Baseball Incorporated, combined with data
on the number of participants in Little League and data from a census of protective
equipment usage for youth aged 5 to 18 years participating in Little League
Baseball in the United States during 1997-1999.
Main Outcome Measures Rate of injury and injury rate ratio comparing users with nonusers of
Results A total of 6 744 240 player-seasons of follow-up and 4233
compensated injury claims were available for analysis. The absolute incidence
of compensated injury per 100 000 player-seasons was 28.02 (95% confidence
interval [CI], 26.76-29.29) for ball-related injury and 2.71 (95% CI, 2.32-3.11)
for facial injury. Overall, use of safety balls was associated with a reduced
risk of ball-related injury (adjusted rate ratio, 0.77; 95% CI, 0.64-0.93).
This reduction was essentially due to 1 type of safety ball, known as the
reduced-impact ball (adjusted rate ratio, 0.72; 95% CI, 0.57-0.91). Use of
faceguards reduced the risk of facial injury (adjusted rate ratio, 0.65; 95%
CI, 0.43-0.98). Metal and plastic guards appeared to be equally effective.
Safety balls appeared to be more effective in the minor division (ages 7-12
years) than in the regular division (ages 9-12 years).
Conclusions Reduced-impact balls and faceguards were associated with a reduced risk
of injury in youth baseball. These findings support increased usage of these
items; however, it should be noted that the absolute incidence of injury in
youth baseball is low and that these equipment items do not prevent all injuries.
Baseball is an integral part of the cultural fabric of the United States,
in part because it provides a point of initiation for youth into organized
team sport.1,2 Nearly two thirds
of baseball participants are younger than 18 years.3 There
is concern, however, about injuries to youth baseball participants, and protective
equipment has been widely endorsed and studied as a means of preventing these
Biomechanical studies indicate that safety balls and protective faceguards
have the potential to reduce the risk of injury in youth baseball.15-17 Using data from a
survey of injured children seeking emergency department care for baseball
injury (n = 348), the Consumer Product Safety Commission estimated that up
to one third of emergency department visits for youth baseball injury could
be prevented if safety balls, faceguards, and safety bases were used universally.9 However, this estimate assumes that these devices
are 100% effective in preventing injury. The effectiveness of the 2 major
types of safety bases, the quick release (breakaway) and Hollywood-style,
have been previously described,10,11 but
no epidemiologic study has examined the effectiveness of safety balls and
The purpose of this study was to evaluate the use of faceguards and
safety balls for preventing injury in youth baseball. This study used a national
database of compensated insurance claims maintained by Little League Baseball
Incorporated, combined with data on the number of participants in Little League
and data from a census of protective equipment usage.
We used an ecological design combining data from 3 sources: compensated
injury claims filed with Little League Baseball Incorporated (hereafter referred
to as Little League), the participation database assembled by Little League,
and a national census of the safety equipment used in Little League, which
was specifically conducted for the purposes of this study. The study covered
the 1997-1999 regular seasons.
The term little league is often used as a generic
label for youth baseball; however, only leagues that were chartered affiliates
of Little League during the study period were included in this study. There
are approximately 5 million participants in organized youth baseball leagues
in the United States, and approximately 50% of participants compete in leagues
affiliated with Little League.3 This study
was commissioned by USA Baseball's Medical and Safety Advisory Committee and
approved by the institutional review board of the School of Public Health,
University of North Carolina at Chapel Hill.
We examined the association of faceguards and safety balls with injury.
Usage of this protective equipment was at the discretion of the individual
leagues in the study. It was not assigned at random. A wide variety of safety
balls are currently used in youth baseball. These include tennis balls, rubber
balls, cloth balls, and a special type of ball generically known as the reduced-impact
ball (also known as the Reduced Impact Factor or RIF ball). The reduced-impact
ball is designed to look and play like a regular baseball but has greater
deformation on impact than traditional balls, lowering the force transmitted
to the child. Reduced-impact balls generally have a polyurethane core, in
contrast with the wool yarn wound around a cork core in the traditional baseball.
There are 3 main types of reduced-impact balls: reduced-impact 10, reduced-impact
5, and reduced-impact 1. Type 10 is stiffer (harder) than type 5, which is
stiffer than type 1.
Protective faceguards are worn when the child is at bat and when running
the bases. The faceguards studied largely comprised the metal mesh guards
and the clear plastic protective visors. Both faceguards are designed to provide
minimal interference with the batter's field of vision.
To provide data on exposure to safety balls and faceguards, we conducted
a Census of Safety Equipment Usage in Little League in the 1997-1999 seasons.
The census was conducted under the auspices of Little League Baseball and
was conducted from Little League's national office in Williamsport, Pa. To
be eligible for inclusion in the census, leagues had to take up the optional
Little League insurance coverage (97% of leagues). They also had to have a
designated safety officer (87% of leagues accepting the insurance option).
Equipment data were collected by directly contacting every eligible
league. A representative from each league (typically either the league safety
officer or the league president) completed a baseline equipment usage questionnaire
for his/her league in 1997. Leagues were followed up for 2 additional years
(1998 and 1999) to collect data on any changes in their usage of safety equipment.
The study population was dynamic; leagues that affiliated with Little League
in 1998 or 1999 were enrolled in the study, administered a baseline survey,
and followed up for the remainder of the study period. Leagues that deaffiliated
with Little League in 1998 or 1999 were followed up only during the period
they were affiliated.
The survey instrument collected information on the usage of safety balls
and faceguards in each of the divisions offered by each league. Only protective
equipment usage pertaining to baseball play during the regular season was
collected (ie, postseason play and softball were excluded).
In 1997, the questionnaire was initially mailed with telephone follow-up
of the nonresponders. In 1998 and 1999, telephone calls were used exclusively
to contact the leagues and update the database to reflect any changes in equipment
usage from the previous year. The response rates were 94% in 1997, 98% in
1998, and 99% in 1999. The final census population numbered 5029 leagues in
1997, 5085 leagues in 1998, and 5055 leagues in 1999, representing 79% of
all Little League affiliates in the United States.
Little League's participation database was accessed to obtain data on
the number of participants in each division in each league. Little League
maintains detailed information on all affiliated leagues. This includes the
divisions operated by each league and the number of players competing in each
division in each league.
Division (level of play) was a priori considered likely to be a strong
confounder of any association between protective equipment and injury. Assignment
to level of play was based on age and ability. Any confounding effect of age
was assumed to be subsumed within the stronger confounding effect of level
of play. We did not control for potential residual confounding effects of
age within level of division. The playing population above the T-ball level
was overwhelmingly male, preventing us from investigating sex as a covariate.
We did not collect data on the type of bats used in each league. Little League
regulations constrain the physical characteristics of the bats approved for
use in affiliated leagues.
Challenger division, which offers competition for children with special
needs, was excluded because there were too few injuries to permit meaningful
analysis. To avoid overstratification of the data, we grouped the 8 remaining
divisions into 4 categories, as follows: T-ball (ages 5-8 years), Little League
minor (ages 7-12 years), Little League regular (ages 9-12 years), and upper
leagues (junior, senior-minor, senior, big league minor, and big league; ages
The outcome of interest was defined as a baseball injury resulting in
compensation under Little League's insurance policy. These data were obtained
from Little League's compensated claims database. Injuries were restricted
to those in scheduled games or practices during the regular season for 1997-1999.
Postseason injuries were excluded. The data were limited to baseball and T-ball
(ie, softball was excluded). A ball-related injury was defined as an injury
due to contact with a batted, pitched, or thrown ball, and a facial injury
was defined as an injury to the eye, face, or nose, while a player was either
at bat or running the bases. The descriptive epidemiology of injury in Little
League has been previously reported.13
We also analyzed separately the subgroup of higher-severity injuries
that were considered likely to result in long-term temporary disability. Higher-severity
injuries were defined as dismemberments, fractures, and dislocations, and
accounted for 38% of all compensated claims in the study.
Little League provides, on a national basis, the option of insurance
coverage for all affiliates. This insurance includes coverage for medical
and related expenses for injuries to participants. There is a $50 deductible
per claim and all claims are lodged through Little League. Affiliates may
accept or decline the coverage; 97% chose to accept it during the 3-year period.
In 1997 and 1998, claims that did not result in compensation were not
routinely electronically recorded by Little League; however, starting in 1999,
claims were electronically recorded irrespective of whether they were ultimately
compensated or not. In 1999, 54% of all registered claims were compensated.
Claims were not compensated if they were judged to be invalid under the policy
or if the losses did not reach the $50 deductible.
The main outcome of interest was the injury rate, defined as the number
of compensated injury claims divided by the number of player-seasons at risk
(× 100 000 for convenience). To calculate injury rates, the compensated
claims, participation, and safety equipment usage databases were linked together.
The resulting data were summarized to give the number of compensated injuries
and player-seasons of follow-up for each division in every study league. Rate
ratios (RRs) were calculated to assess the effect of each protective equipment
item on the injury rate. Traditional (nonsafety) equipment was the referent
category for all RRs, and 95% confidence intervals (CIs) were used to assess
the relative precision of the RRs.
Initially, a comprehensive stratified analysis of the data, using Mantel-Haenszel–type
estimators to control for division, was performed.18-20 Based
on the results of the stratified analysis, a multivariate analysis was conducted.
The multivariate analysis used a multilevel, or hierarchical, model.21,22 A multilevel model was used because
protective equipment usage is an ecological decision made by the league, not
by the individual player.22,23 Furthermore,
this model accounts for the lack of statistical independence of teams within
leagues, which could have arisen if some study leagues tended to be more safety-conscious
Because the rate of injury was the outcome of interest, we fit the multilevel
form of the Poisson regression model to the data.24 The
multilevel Poisson regression model belongs to a general family of models
known as generalized linear mixed models.25 Multilevel
Poisson regression is implemented using a log-link function and an overdispersed
Poisson model for residual variation, with the rate denominator (number of
player-seasons at risk) included in the model as an offset term.24,25 A
random effect was used to model a separate intercept for each league in the
study, under the constraint that these league-specific intercepts are normally
distributed.24,25 The model was
fit using SAS's PROC NLMIXED version 8.02 (SAS Institute Inc, Cary, NC), which
uses Gauss-Hermite quadrature to integrate an approximate likelihood function.25
To assess division as an effect-measure modifier (modifier), we examined
the effect of safety equipment stratified by division. We also used the likelihood
ratio test to compare models with and without the interaction term for protective
equipment by division. Limitations of this type of test include limited ability
to assess departures from jointly additive (as opposed to jointly multiplicative)
effects, and poor power to distinguish between additive and multiplicative
joint effects.26,27 Division was
assessed as a confounder by comparing RRs unadjusted and adjusted for division.
It is possible that an item of protective equipment could decrease the
risk of certain types of injuries but increase the risk for other types of
injury. For example, it might be possible that facemasks reduce the risk of
facial injuries but increase the risk of neck injuries. For this reason, in
addition to modeling the rate for ball-related and facial injury, we fit multilevel
Poisson regression models to estimate the effect of safety balls and faceguards
on the overall injury rate. We also fit multilevel Poisson regression models
restricted to high-severity injuries (dismemberments, fractures, and dislocations).
To determine whether nuances of insurance compensation could be a potential
source of bias in this study, uncompensated injury claims for 1999 were accessed,
combined with compensated injury claims for 1999, and all analyses were repeated.
Uncompensated injury claims were defined in an analogous fashion to compensated
injury claims (ie, injuries to players in regularly scheduled games or practices
during the regular baseball or T-ball season).
A total of 6 744 240 player-seasons of follow-up and 4233
compensated injury claims were available for analysis. Missing data resulted
in smaller study size for some analyses. Compensated injury claims will hereafter
be referred to as injuries.
Descriptive information on the distribution of protective equipment
and injuries by division is presented in Table 1 (these data have been annualized to facilitate interpretation).
Nearly one half of all the compensated injuries were in the Little League
regular division. Risk of injury increased with level of competition, but
usage of safety balls and faceguards deceased with increasing level of competition,
creating the potential for confounding by division. A mean 73.0% of leagues
used safety balls in at least 1 division, and 34.3% used faceguards in at
least 1 division.
The majority of injuries (44.6%) were ball-related (n = 1890). The absolute
incidence of compensated injury claims per 100 000 player-seasons was
28.02 (95% CI, 26.76-29.29) for ball-related injury. Overall, safety balls
were associated with a reduced risk of ball-related injury. The RR for all
types of safety balls combined vs traditional balls after adjusting for division
was 0.77 (95% CI, 0.64-0.93), indicating a protective association.
There was an indication that division modified the effect of safety
balls (Table 2). The association
appeared to be stronger in the minor division than in the regular division,
and the likelihood ratio test P value for statistical
interaction was .05. However, considerable overlap was observed in the CIs
for the other divisions, complicating the assessment of modification. Division
was a strong confounder of the safety-ball association. The RR for all types
of safety balls combined, unadjusted for division, was 0.20 (95% CI, 0.17-0.23).
All safety-ball RRs reported were adjusted for division.
When the various subcategories of safety balls were examined, the reduced-impact
ball was consistently associated with a reduced risk of injury (Table 3). The adjusted RR for all types of reduced-impact ball combined
vs traditional balls was 0.72 (95% CI, 0.57-0.91). The other main types of
safety balls (rubber, tennis, and cloth balls) did not appear to be associated
with a reduction in injury risk (adjusted RR, 1.04; 95% CI, 0.68-1.57). Interpretation
of the RR for the mixed/other category is problematic, because this group
included leagues using reduced-impact balls in combination with other types
of safety balls (27.7% of mixed/other) and safety balls in combination with
traditional balls (20.3% of mixed/other).
Infielders (34.7%), batters (27.9%), and outfielders (12.1%) were the
playing positions most commonly injured by balls. We theorized a priori that
safety balls would have a greater protective effect in higher velocity impacts
(eg, when at bat); however, the ball injury RR did not appear to vary when
stratified by playing position. The adjusted RR for safety balls vs traditional
balls was 0.81 (95% CI, 0.58-1.12) for batters, 0.76 (95% CI, 0.57-1.02) for
infielders, and 0.78 (95% CI, 0.47-1.29) for outfielders.
Ball-related injuries were evenly split between batted balls (36.0%),
thrown balls (31.4%), and pitched balls (28.6%; unspecified balls accounted
for 4.1%). How the ball acquired its velocity appeared to make almost no difference
to the protective association of safety balls. The adjusted RR for safety
balls vs traditional balls was 0.79 (95% CI, 0.59-1.05) for batted balls,
0.84 (95% CI, 0.61-1.15) for pitched balls, and 0.67 (95% CI, 0.49-0.93) for
For all types of injuries, ball-related and non–ball-related combined,
the adjusted RR for safety balls vs traditional balls was 0.84 (95% CI, 0.73-0.97).
The analogous RR for the higher severity ball-related injuries (n = 701) was
0.91 (95% CI, 0.66-1.24).
In contrast with ball-related injuries, facial injuries accounted for
only 4.3% of all injuries (n = 183). The absolute incidence of compensated
injury claims per 100 000 player-seasons was 2.71 (95% CI, 2.32-3.11)
for facial injury. Overall, faceguards were associated with a reduced risk
of facial injury. The RR for both types of faceguards combined vs no faceguard
was 0.65 (95% CI, 0.43-0.98), after adjusting for division. There was no strong
evidence of modification by the RR by division, considering the overlap in
the stratum-specific confidence intervals (Table 2). Division was a weak confounder of the faceguard association.
The RR for all types of faceguard combined unadjusted for division was 0.60
(95% CI, 0.40-0.89). In the interests of comparability between the safety
ball and faceguard results, all faceguard RRs reported are adjusted for division.
When the 2 types of faceguard were examined, there was an indication
that plastic faceguards might be more effective than metal, but there were
too few injury claims to determine if this difference was real or an artifact
of random variation (Table 3).
The mixed/other category for faceguards included situations where faceguards
were not required but were available for voluntary use (51.1%), and where
leagues used both metal and plastic faceguards in combination (25.0%).
Facial injuries occurred when at bat (80.0%) and running the bases (20.0%).
Facial injuries in other playing positions were excluded from analysis, because
they would not have been prevented through the use of a faceguard. The protective
association of faceguards appeared to be similar for batters (adjusted RR,
0.64; 95% CI, 0.41-0.99) and base runners (adjusted RR, 0.72; 95% CI, 0.28-1.89).
For all types of injuries, facial and nonfacial combined, the adjusted RR
for faceguards vs no faceguard was 0.86 (95% CI, 0.78-0.96). The analogous
RR for higher-severity facial injuries (n = 97) was 0.85 (95% CI, 0.47-1.53).
Uncompensated injury claims for 1999 were combined with the 1999 compensated
claims and all analyses were repeated to determine whether nuances of insurance
compensation could be responsible for the observed protective associations.
For safety balls vs traditional balls, the adjusted RR was 0.77 (95% CI, 0.62-0.97).
For faceguards vs no faceguards, the adjusted RR was 0.69 (95% CI, 0.37-1.30).
This 3-year study of Little League Baseball found that the use of safety
balls and faceguards in this population was associated with a reduced risk
of injury. The use of safety balls was associated with a 23% (RR, 0.77) reduced
risk of ball-related injury, and faceguards with a 35% (RR, 0.65) reduced
risk of facial injury. Reduced-impact balls appeared to be the most effective
type of safety ball (28% reduction; RR, 0.72). There was no compelling evidence
of any difference between plastic and metal faceguards. The protective benefits
of safety balls may be stronger in the minor division than in regular division.
This was a large study (more than 6.7 million player-seasons) with a
high rate of follow-up. All 50 US states were represented in the study. As
with any observational study, however, the observed reductions in injury risk
are potentially subject to confounding bias (ie, they may be due to differences
between the leagues other than the equipment used).
Another limitation of this study is the use of insurance records to
identify injuries. It is possible that some Little League injuries during
the study period did not result in compensated claims and therefore were not
included in this study. Legitimate injuries that were registered with Little
League may not have been compensated (undercompensation bias). In addition,
some injuries may not even have been registered with Little League (underregistration
Undercompensation bias would only arise if the probability of compensation
was associated with usage of safety equipment. This seems unlikely, because
data on the type of equipment used is not routinely available to the claim
assessors. Furthermore, when uncompensated and compensated claims for 1 year
(1999) were combined together and analyzed, the results were similar to those
based on only compensated claims.
To assist in quantifying to what extent underregistration could have
introduced bias, a small sensitivity analysis was conducted. Simple deterministic
methods were used to produce corrected RRs under various hypothetical underregistration
scenarios.28 Our working assumptions for the
sensitivity analysis were that specificity was 100% (ie, all registered injuries
represented true injuries) and that division of play was measured without
error. An important and well-known28 first
result from the sensitivity analysis was that no underregistration bias exists
in the study findings if the hypothetical underreporting was nondifferential
with respect to use of safety equipment. In other words, underregistration
would not have biased the study findings so long as the probability that an
injury registered with Little League was independent of use of safety balls
or faceguards in that league.
The second result from the sensitivity analysis was that differential
underregistration could have created bias in the study findings; however,
this bias was small as long as underreporting and equipment use were not strongly
associated. For example, if the difference in the underregistration rates
between users and nonusers of safety equipment was 10% or less and the proportion
of unregistered injuries was no more than 30%, only a modest degree of bias
was present in the study findings (RR, 0.77 for safety balls and 0.65 for
faceguards). In this scenario, the corrected RRs ranged from 0.67 to 0.88
for safety balls and 0.57 to 0.74 for faceguards. More bias was present if
the difference in underregistration rates between users and nonusers was 15%
or less, with corrected RRs from 0.63 to 0.90 for safety balls and 0.53 to
0.79 for faceguards. A very high proportion of unregistered injuries of 50%
or less resulted in only modest bias if the difference in the underregistration
rate between users and nonusers was small, 5% or less. The corrected RRs ranged
from 0.70 to 0.84 for safety balls and 0.59 to 0.72 for faceguards in this
scenario. In summary, the sensitivity analysis results suggest that underregistration
is not a threat to study validity as long as it is nondifferential or weakly
differential with respect to usage of safety equipment.
The findings are generally consistent with previous research. The overall
distribution and incidence of injuries observed are similar to those reported
elsewhere.9,12 Reanalysis of data
from a previous, much smaller, epidemiologic study of safety balls yields
an RR of 0.87 for reduced-impact balls, similar to that observed in this study.12 A recent epidemiologic faceguard study indicated
a reduction (RR, 0.77) in facial injuries following use of the faceguard.29
Theoretical biomechanical studies have indicated that baseballs with
lower mass and less stiffness have a reduced potential for injury.30 Laboratory studies have shown that reduced-impact
balls are less likely to result in head injury and skull fracture.16,17 Some experimental models suggest
that safety balls also reduce the risk of sudden death from chest-wall impacts
(commotio cordis),31,32 although
this claim has been debated.9,33 Although
there is concern about the potential for a softer baseball to penetrate the
orbit of the eye more deeply, this hazard is only present in extremely soft
balls.17 This concern can be reduced if safety
balls and faceguards that conform to current safety standards are used.14
The most important barrier to increased acceptance of the reduced-impact
ball may be dissatisfaction with the perceived performance characteristics
or play of the ball. In the first year of our study (1997), 63 leagues reported
that they had tried reduced-impact balls in previous years but had discontinued
their use, frequently citing problems with the bounce of the ball.34 On the other hand, a laboratory study reported that
when pitching, throwing, and batting under blinded conditions (with no identifying
notation on the ball), adults cannot distinguish between a traditional ball
and a safety ball, which has as little as 20% of the traditional ball's hardness,
and children (11-14 years) cannot distinguish between a traditional ball and
a safety ball with as little as 15% hardness.17 Thus,
the concerns about the play of safety balls possibly revolve around the perception
of the ball's play, rather than the actual performance of the ball.
Although our findings indicate that safety balls and faceguards are
associated with a reduced risk of ball-related and facial injuries, it is
clear that they are not 100% protective. Thus, the proportion of injuries
that would be prevented if organized youth baseball were to universally mandate
the use of safety balls and faceguards would be less than the 32% estimated
by the Consumer Product Safety Commission.9
These findings support the expanded use of reduced-impact balls and
faceguards in youth baseball. It should be noted, however, that the absolute
incidence of compensated injury claims in youth baseball is low and that these
equipment items do not prevent all injuries. Given the greater incidence of
ball-related injury relative to facial injury, we suggest that leagues with
limited resources consider pursuing the implementation of safety balls initially,
followed by implementation of faceguards.
Create a personal account or sign in to: