|Home | About | Journals | Submit | Contact Us | Français|
Stress fractures can be debilitating in athletes and military personnel. Insoles may lower stress fracture rates by improving biomechanics, lessening fatigue, and attenuating impact. The objective of this study was to systematically review the best evidence on the use of insoles as a method of stress fracture prevention in a high-risk population. Using MEDLINE, Cochrane, Current Controlled Trials, UK National Research Register, ScienceDirect, CINAHL, and EMBASE, a review of randomized (level I) and quasi-randomized (level II) controlled trials was performed using an insole as the intervention and stress fracture incidence as the primary outcome measure. Five trials were included, and a random effects model was used to generate a summary estimate and an overall odds ratio. One study found a significant reduction in overall stress fracture incidence using a semirigid insole, while four studies found no overall reduction in military personnel. However, when the data are pooled, orthotic use was beneficial. When stratified by site, there was a reduction in femoral and tibial stress fracture incidence. Shoe insoles may reduce the overall femoral and tibial stress fracture incidence during military training. It is unclear if the use of insoles would prevent stress fractures in athletes. Additional studies are necessary to determine the efficacy of insoles in an athletic population.
Stress fractures can be debilitating injuries that negatively impact both competitive athletes and active military personnel. The overall incidence of stress fractures among National Collegiate Athletic Association athletes is 1.0–2.6%, and, in the military, stress fractures affect 1.0–1.7% of the soldiers [1–7]. Stress fractures can occur in a variety of anatomic locations and with varying frequency. In the lower extremity, femoral, tibial, and metatarsal stress fractures are most common. In a review of 370 athletes with stress fractures, the tibia was the most commonly involved bone (49.1% of cases), followed by the tarsals (25.3%) and the metatarsals (8.8%) . Bilateral stress fractures occurred in 16.6% of cases. Current prevention strategies include calcium supplementation, participation in a graduated training or running program prior to the onset of military training or athletic participation, cyclical rather than progressive training strategies, introduction of a rest period after the onset of physical training, and improved shock attenuation of the foot, via either shoe modification or use of an insole [9–13].
Shoe inserts, or orthotics, are available in a variety of preparations and can be constructed from many different materials depending on their function and the desired rigidity or support. Similarly, both generic (over-the-counter or off-the-shelf) and custom-molded (prescription) preparations can be used. The size of these insoles ranges from a simple heel cup to a full-length insert that extends to support the forefoot and toes. A variety of devices accommodate for any number of different foot types or limb alignments and can be constructed with individualized posting or recesses to address specific pathology. In this way, all foot types, from planovalgus to cavovarus, can be fitted with an appropriate accommodative orthotic insert.
Previous studies have investigation whether insoles lower stress fracture incidence by improving foot biomechanics, lessening fatigue of the lower extremities, and attenuating impact better than a conventional military boot or training shoe [14–16]. Because the military boot is designed to protect the foot from blunt trauma and to protect the ankle from inversion injury, the standard boot is particularly amenable to the insertion of insoles to prevent lower extremity stress fractures in military recruits. However, despite considerable interest, this belief has not been uniformly supported. At the same time, no systematic review or meta-analysis has made use of the considerable body of evidence on this topic to critically assess the role of insoles in stress fracture prevention.
For this reason, a systematic review was performed to evaluate the best evidence (levels I and II studies) on the use of shock-absorbent insoles as a method of stress fracture prevention in a high-risk population. The objective of this review was to determine the effect of insole use on the overall and site-specific incidence of stress fractures.
To identify relevant publications on the use of shock-absorbent insoles for the prevention of stress fractures, an exhaustive review of the English-language literature was conducted using MEDLINE (January 1966 through November 2007), Cochrane databases, Current Controlled Trials (http://www.controlled-trials.com), UK National Research Register (http://www.update-software.com/national), ScienceDirect, CINAHL, and EMBASE. Initially, the search was restricted to randomized controlled (level I) trials. The results of this search were limited, however, and the search was expanded to include quasi-randomized controlled (level II) studies evidence as well  (Table 1).
Using the search term “stress fracture” (without quotation marks) with the limit “English language”, articles were retrieved from MEDLINE. The same search terms with the addition of the term “prevention” were used to search EMBASE. From these two searches, 17 original full-text manuscripts were identified. No additional studies were identified through similar searches of the other scientific databases. The inclusion criteria for the review were (1) level I or II evidence, (2) use a shock-absorbent insole as the intervention, and (3) incidence of lower extremity stress fractures as the primary outcome measure. Each of the 17 articles was reviewed independently by two authors using a published worksheet adapted from evidence-based guidelines . The search processes and trial flow are summarized in Fig. 1, while information regarding exclusion of specific articles is presented in “Appendix” [10, 15, 16, 19–25].
From the literature review, five trials met the inclusion criteria: three level I and two level II studies . Data for each of the studies were collected and were presented in Table 2. Subjects are male active duty military personnel serving in the Army, Naval Academy, and Marines. Age was generally consistent in the five studies, with a range of 18.5–20.0 years [14, 26, 27]. The median duration of follow-up was 13 weeks in the four studies that included this information. The study of Gardner et al. was the only study to report race [26, 27]. The material composition of the shock-absorbent insoles in each study varied significantly from custom made orthoses to commercially available neoprene insoles, and little information was provided regarding the rationale for particular insole selection.
To evaluate the effect of insoles use on stress fracture prevention, we performed an analysis of overall stress fracture incidence. Stress fracture incidence was defined as the total number of stress fractures divided by the total number of subjects per group at the beginning of the study. In order to account for study variation (heterogeneity), a random effects model was developed using Review Manager 5.0 (Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration, 2008). This model was used to generate a summary estimate and to calculate an overall odds ratio (OR) using the method of DerSimonian–Laird.
One study only reported the overall incidence of tibial stress fractures . Because the methodology was unclear, whether or not they evaluated their subjects for fractures at other locations is unknown. For this reason, the data from this study were included only in the tibia subgroup analysis .
In this systematic review, only one of the five studies reviewed found a statistically significant reduction in overall stress fracture incidence following the introduction of an insole. In this study, subjects were allocated to three groups: a custom-made semirigid insole, a custom-made soft insole, or a control group. The authors reported a 14% reduction in overall stress fracture incidence (p=0.013) when the use of a custom-made semirigid insole was compared to control . The other four studies reviewed found no reduction in overall stress fracture incidence in military personnel using insoles compared to controls [9, 26–28].
While the study by Milgrom et al. did not find a reduction in the overall incidence of stress fractures, a significant reduction (p<0.05) in the incidence of femoral stress fractures was present in their subgroup analysis . This benefit did not, however, result in a reduction in the number of tibial or metatarsal stress fractures.
When the data from the five-included studies were pooled for analysis, a beneficial trend in favor of the use of an orthosis was identified. These data, comparing the use of an insole to control, have an OR of 0.54 (95% confidence interval (CI) 0.29–1.01; Fig. 2).
When the data set is stratified by anatomic location, there is a more significant reduction in the incidence of femoral (OR=0.44, 95% CI 0.25–0.78) and tibial (OR=0.52, 95% CI 0.33–0.83) stress fractures with the use of insoles (Figs. 3 and and4).4). For metatarsal stress fractures, the trend suggests that insoles prevent stress fractures, but it is not conclusive (OR=0.29, 95% CI 0.08–1.05; Fig. 5).
In considering the variation in the type of insole used in the individual studies, a separate analysis of the data was performed excluding the trial by Andrish et al. This investigation enrolled some 1,797 participants and made use of a heel cup, rather than a full-length insole. This separate effort to control for the potential confounding effect of this one study revealed an overall OR of 0.52 (95% CI 0.27–1.01). The site-specific analysis of the pooled data without this larger study yielded a femoral OR of 0.42 (95% CI 0.24–0.75), a tibial OR of 0.51 (95% CI 0.32–0.82), and a metatarsal OR of 0.29 (95% CI 0.08–1.05).
Additionally, as part of the systematic review, each study was evaluated for potential sources of bias. The trial by Gardner et al. is limited in its generalizability due to a selection bias. In this study, a quasi-random allocation structure was used; the participants’ intervention was assigned by platoon number . Similarly, for this review, the method of diagnosis for the stress fracture is central to the quality of investigation. The use of different diagnostic methods by separate studies may influence the number of stress fractures detected in each trial. Consequently, this disparity may introduce a detection bias to the systematic review. Both Andrish et al. and Schwellnus et al. used plain radiographs to diagnose stress fractures, while Finestone et al. used a bone scan [9, 14, 27]. Bone scans can detect stress fractures at an earlier stage than radiographs; thus, a study utilizing bone scans will likely detect a higher incidence of stress fractures than one utilizing only plain radiographs [14, 28]. Ideally, each study would have blinded the radiologist and diagnosing physician to the subjects’ assigned group to prevent any observer bias. Only Gardner et al. reported blinding of the radiologists or clinicians to the participants’ assigned group .
Generally, to limit the effect of an exclusion bias, the preferred percentage of follow-up for a level I study is 80% [17, 29]. In this series, one study did not meet this criterion, provided follow-up on less than 50% of participants, and, therefore, may be affected by exclusion bias .
Each of the five studies provided limited information on the method of statistical analysis. A sample size calculation for a given power was not reported in any of the studies. Milgrom et al. reported that the average number of stress fractures per recruit in the insole and noninsole groups was the same, concluding that insoles did not affect overall stress fracture incidence. In a subgroup analysis, however, this study also reported that the use of insoles resulted in a significant reduction in femoral stress fracture incidence. After contacting the authors of this study, the data regarding the total number of recruits with stress fractures in each group were requested. Unfortunately, this information was not provided.
In order to avoid undermining the randomization process, it is important that studies use an intent-to-treat analysis. With this type of evaluation, the outcomes of patients who did not finish the study, or who changed groups during the study, are reviewed according to their original group assignment. None of the five studies in this review present the results of an intent-to-treat analysis. While the study by Gardner et al. would not qualify as they experienced no withdrawals or crossover, both Milgrom et al. and Schwellnus et al. excluded the outcomes of subjects who withdrew from the study in their analysis [26–28]. While Andrish et al. and Finestone et al. did not exclude withdrawals, they also did not perform an intent-to-treat analysis for those participants that deviated from their group assignment [9, 14].
Stress fractures are overuse injuries that result from repetitive activity and occur most often in athletes and military personnel. In this systematic review, we evaluated the best evidence available to determine the effect of insole use on the overall and site-specific incidence of stress fractures in a high-risk population. When considered individually, only one of the five studies that meet the inclusion criteria found a significant reduction in overall stress fracture incidence using a semirigid insole . Similarly, only the study by Milgrom et al. identified a significant site-specific reduction in the incidence of stress fractures within their femoral subgroup analysis . However, drawing on the power of a systematic review, when the data from all five studies were pooled, a trend in favor of orthotic use was identified. This benefit included a reduction in the overall incidence of stress fractures (OR=0.54) as well as a more significant site-specific reduction in the incidence of femoral (OR=0.44) and tibial (OR=0.52) stress fractures.
Considering the quality of the pooled data, it is important to acknowledge that the quantity and quality of physical activity of the study participants is paramount to the stress fracture incidence. In this way, to accurately compare separate groups in different studies, injury exposure may be calculated to normalize an individual’s risk by recording the total number of training hours or it can be expressed as training per unit of time. However, even this measure does not account for all characteristics of the exposure, such as the intensity. Gardner et al. reported that recruits underwent 92 h of physical training and 41.5 h of drill or ceremony over the 12-week course, while Schwellnus et al. reported that insole and control groups underwent “identical training” throughout the study [26, 27]. None of the remaining studies provided information regarding time or intensity of physical training, and, as a result, the incidence of stress fractures is not corrected per unit exposure.
Each study reported the cumulative incidence of stress fractures, or the incidence of injury, is calculated over a specific time period in which all individuals are at risk . Because the five studies evaluated military trainees over varying lengths of time (9 to 14 weeks), studies with a longer duration are more likely to report a higher number of stress fractures.
Additionally, the effective sample size for each study is determined by how many outcomes of interest are observed, i.e., stress fractures. Given the small number of stress fractures in these studies, the statistical power to detect significant differences among treatment groups is limited.
As with all clinical trials, it is important to consider the clinical relevance of these findings and whether or not they justify a change in clinical practice. Because standard insoles are easily used, inexpensive, and pose little risk, any reduction in stress fractures might justify their use. In subgroup analyses, the use of insoles was associated with a lower risk of femoral (OR=0.44, 95% CI 0.25–0.78) and tibial (OR=0.52, 95% CI 0.33–0.83) stress fractures. However, when the 95% confidence intervals are considered, the overall effect of insoles preventing stress fractures (OR=0.54, 95% CI 0.29–1.01) is less conclusive.
The Cochrane Database of Systematic Reviews has published reviews that address the use of insoles for preventing and treating stress fractures [31, 32]. Gillespie et al. evaluated four of the five studies included in this review and an additional study performed by Smith et al. [9, 16, 26–28]. The study by Smith et al. was excluded from this study because “tibial stress” was a primary outcome measure. Gillespie et al. pooled the data from Andrish et al., Milgrom et al., and Schwellnus et al. using a fixed effects model to perform a meta-analysis. They concluded that insoles significantly reduce the number of overall stress fractures in military recruits .
Rome et al. assessed six studies: four of which are included in this review, an additional unpublished manuscript, and a study by Mundermann et al. that was excluded from this analysis because “stress fracture or pain” was the primary outcome [14, 15, 27, 28, 31]. In this review, data were not pooled, but relative risks with 95% confidence intervals were calculated for each study. The authors indicate that they used a random effects model, but tabulated data are presented as “relative risk (fixed)”, suggesting a fixed, not random, effects model. Noting these concerns, they concluded that the evidence from randomized trials, although limited, suggests that the use of shock-absorbent insoles may reduce the overall incidence of stress fractures in military recruits. Both reviews agree that the available evidence is poor in quality, but neither review commented on how study bias or the variation in exposure may have affected the results [31, 32].
In this way, it is important to note that none of the investigation included in this review performed an a priori sample size calculation. Thus, those studies finding no significant difference between insole and control groups cannot discount the possibility of type II error. This methodological fault is more concerning when the pooled data are reviewed closely. In Figs. 3 and and4,4, the aggregated data indicate that insoles may indeed lower the incidence of femoral and tibial stress fractures where the overall incidence was not significantly different between groups. However, without an appropriately powered sample size calculation, such a post hoc subgroup analysis should be interpreted with caution. In order to stratify the stress fracture incidence by anatomic site, additional well-designed research would be necessary using a larger sample size with adequate enrolment.
Other methodological concerns warrant consideration. In the study by Milgrom et al., 30 subjects assigned to the insole group discontinued using the insoles due to discomfort in the first 14 days of the trial. Rather than complete an intent-to-treat-analysis, the authors elected to exclude these participants from the final analyses . Similarly, Andrish et al. did not provide an intent-to-treat analysis for participants that crossed over between groups. Instead, outcomes from platoons that did not adhere to their group assignment were combined with the control group for analysis .
Finestone et al. assigned 18 soldiers to the insole groups at the onset of the study. When they did not wear the insoles as directed, their data were treated as a second “control group” and pooled with the original controls in the statistical analysis . A secondary analysis reported by Finestone et al. based on intent-to-treat manner found no significant difference between groups (p=0.14). This lack of significance persisted after excluding withdrawals (p=0.07).
Recognizing that the size of the study by Andrish et al. might influence the conclusions from the pooled data, a separate analysis was completed without these data. When the same random effects model was applied to the abbreviated data set, no substantial change was present in the resulting odds ratios. The overall and metatarsal estimates continued to trend toward a reduction in the incidence of stress fractures, while the significant reduction of femoral and tibial stress fractures persisted.
Based upon the best available evidence, shoe insole modification may reduce overall stress fracture incidence during military training, but the findings of these studies are heterogeneous and inconclusive at the present time. From this systematic review, there is compelling evidence that insoles may reduce femoral and tibial stress fractures; however, additional studies are necessary in order to stratify the analysis by anatomic site. Given the current evidence, it is unclear whether the use of insoles would prevent stress fractures in a high risk athletic population. Well-designed studies are necessary to determine the efficacy of insoles as a preventive intervention in an athletic population.
The authors would like to acknowledge Tara Holmes, Lynn Cain, and Teresa Chipps for their editorial assistance.
Level of evidence: Level II systematic review.
The project described was supported by Grant # 5K23AR052392-02 from the National Institute of Arthritis and Musculoskeletal and Skin Diseases (Dunn, Warren R.) and a Pfizer Scholars Grant in Clinical Epidemiology (Dunn, Warren R.).