Factors explaining heterogeneity in studies comparing surgical and nonsurgical treatment of midshaft clavicle fractures: a meta-regression analysis of randomized controlled trials and high-quality observational studies.

BACKGROUND

Previous meta-analyses comparing the surgical and nonsurgical treatment of midshaft clavicle fractures have demonstrated extensive heterogeneity, that is, treatment effect variation, in different pooled outcomes. We aimed to investigate the amount of heterogeneity seen in pooled treatment effects and to explore which moderator variables serve to explain this heterogeneity.

METHODS

A follow-up literature search, based a previous study, was conducted. All randomized controlled trials and high-quality observational studies with suitable treatment cohorts were identified and included in this systematic review and meta-regression analysis. The proportions of male patients, patients with the dominant hand injured, and smokers, as well as fracture type and mean age, were included as covariates in meta-regression analyses investigating the effect on the pooled estimate of treatment effect heterogeneity. The pooled treatment effects assessed were nonunion rate, malunion rate, Disabilities of the Arm, Shoulder and Hand score, and Constant score at 1 year, as well as revision surgery rate.

RESULTS

High heterogeneity was observed in 4 of 8 pooled treatment effects and moderate, in 2 of 8. An association between any of the covariates, including smoking, with the pooled treatment effect for the nonunion rate could not be established. Regarding malunion, the proportion of patients with the dominant hand injured showed linear dependency with the risk ratio. The proportion of male patients correlated with the pooled mean difference in both the Disabilities of the Arm, Shoulder and Hand score and Constant score. Mean age and fracture type correlated with the pooled mean difference in the Constant score.

CONCLUSION

On the basis of our results, several potential moderators influence the treatment effect when comparing surgical and nonsurgical treatment of midshaft clavicle fractures. These findings have implications for shared decision making and when making treatment recommendations.