Page 124 - 2022 Spring JSOM

Methods Assessment Tool for Observational Cohort and Cross-Sectional
18
The Preferred Reporting Items for Systematic Reviews and Studies. The instrument has 14 questions that address subject
Meta-Analyses (PRISMA) guidelines were used to guide this selection, sources of bias, confounding, statistical power, statis-
16
study. Specific details of the review protocol are described tical analysis, and other factors. Each study was rated on the
below. 14 questions in the instrument as either “yes,” “no,” “cannot
determine,” “not reported,” or “not applicable.” The poten-
Information Sources and Search tial risk of bias was determined for each “no,” “cannot deter-
PubMed, Web of Science, and Cumulative Index to Nursing mine,” and “not reported” response. A rating of “good” was
and Allied Health Literature (CINAHL) were searched to find given if there was low risk of bias and results appeared valid,
articles relating to injuries during HIFT. Keywords used in the “moderate” if there was some risk of bias but results still ap-
search were injury AND CrossFit, OR “High Intensity Func- peared valid, or “poor” if there was significant risk of bias that
tional Training” OR “Extreme Exercise Programs” OR “Gym appeared to reduce confidence in the results.
Jones” OR “Insanity Training.” The reference lists of the ob-
tained articles and reviews found in the search 7–15 were also Meta-Analyses
examined for other articles that were missed by the formal The Comprehensive Meta-Analysis Statistical Package, Ver-
search. The final search was completed in November 2021. sion 3.2 (Biostat, Englewood, New Jersey, US) was used to per-
Four authors were contacted to clarify data in their study, but form the meta-analyses. For injury prevalence, data obtained
17
only one responded to requests. Figure 1 shows the results of from each study included the total number of participants and
the search and selection process as a PRISMA flow diagram. 16 the number who experienced injuries of any type. For injury
rates, the data include the numbers of injuries and the hours
FIGURE 1 Flow diagram showing articles identified, screened, of training (injuries/1000 h of training). For injury locations,
assessed, for eligibility, and included in this review.
data include the total number of injuries and the number of
injuries at each anatomical location. Although some authors
defined the anatomical locations slightly differently, in most
cases it was possible to determine the number and proportion
(%) of injuries at the shoulder, back/spine, knee, arm/elbow,
wrist/hand/fingers, and ankle/foot.
The meta-analyses produced a summary injury prevalence,
injury rate, and injury proportion at specific anatomical loca-
tions (called “effect sizes”) with their 95% confidence inter-
vals (95% CIs). These effect sizes and 95% CIs represented the
weighted and pooled results from all individual investigations.
A random effects model was used for all analyses. Heterogene-
ity in the individual studies was assessed using the Q-statistic
and I . Heterogeneity was the degree of variability among the
2 19
individual studies. For I , values could range from 0 to 100,
2
with higher numbers indicating more variability.
Study Selection and Data Extraction To assess publication bias, the Duval and Tweedie Trim and
21
20
Articles were included if they (1) reported quantitative data Fill Procedure and the Begg and Mazumdar Test were used.
20
on overall injury prevalence or injury rates experienced during The Trim and Fill Procedure imputes hypothetical “missing”
HIFT of any type, (2) involved adults ≥18 years of age, (3) were studies based on a plot of the standard error and individual
written in English, and (4) were published in peer-reviewed study effect sizes. The procedure estimates a new effect size
journals. Articles were excluded if they (1) were injury case and 95% CI if these hypothetical missing studies were in-
21
studies or case series, (2) reported on physiological, psycho- cluded in the plot. The Begg and Mazumdar Test calculates
logical, or performance changes during HIFT, (3) were limited the rank order correlation between the effect size and standard
to injuries at specific anatomical locations without providing error. A significant correlation suggests that publication bias
overall injury rates, (4) were reviews of injuries during HIFT, exists, although it does not address the causes of the bias.
or (5) involved individuals <18 years of age.
Several moderator variables were examined. These included
To guide the data extraction, a spreadsheet was constructed the study design (retrospective cross-sectional and prospective
that contained the study name; study design; participant num- cohort), the type of HIFT (CrossFit and other types of HIFT),
ber, age, and sex; number and proportion (%) of injuries; training in CrossFit-affiliated gyms, studies using similar injury
injury case definition; how injuries were recorded; injury re- definitions, and injury reporting periods (6 months, 1 year,
porting period; injury rate (injuries/1000 h of training); and since start of HIFT). Training in CrossFit-affiliated gyms might
number of injuries by anatomical location. In some cases, result in lower injury risk because of the closer supervision that
these data were not available in particular investigations, but might be provided by the gym staff, especially in monitoring
as long as injury prevalence or injury rates were reported the proper exercise execution and noting excessive fatigue. Injury
study was included in the review. definitions varied widely in the selected studies, but a num-
ber 22–29 used a very similar one. This was physical damage to
the body experienced during HIFT resulting in inability to train
Methodological Quality Assessment
≥1 week, modified training for >2 weeks, or a visit to a health
Methodological quality of the selected studies was assessed us- professional. For all moderator variables, meta-analyses and
ing the United States National Institute of Health Study Quality publication bias statistics were performed as described above.

122 | JSOM Volume 22, Edition 1 / Sping 2022