No evidence for intervention-dependent influence of methodological features on treatment effect

• Published in: Journal of clinical epidemiology 2013-12, Vol.66 (12), p.1347-1355.e3
• Main Authors: Jacobs, Wilco C.H, Kruyt, Moyo C, Moojen, Wouter A, Verbout, Ab J, Oner, F. Cumhur
• Format: Article
• Language: English
• Subjects: Allocation concealment; Analysis. Health state; Bias; Biological and medical sciences; Bone surgery; Clinical trials; Clinical Trials as Topic - methods; Double-Blind Method; Drug therapy; Effect modifier; Epidemiologic; Epidemiology; General aspects; Heterogeneity; Humans; Influence; Internal Medicine; Intervention; Medical sciences; Meta-analysis; Methods; Mortality; Pharmaceuticals; Public health. Hygiene; Public health. Hygiene-occupational medicine; Random Allocation; Randomization; Randomized Controlled Trials as Topic - methods; Research Design; Risk of bias; Studies; Treatment Outcome
• ISSN: 0895-4356; 1878-5921
• DOI: 10.1016/j.jclinepi.2013.06.007

Abstract Objectives The goal of this systematic review was to evaluate if the influence of methodological features on treatment effect differs between types of intervention. Study Design and Setting MEDLINE, Embase, Web of Science, Cochrane methodology register, and reference lists were searched for meta-epidemiologic studies on the influence of methodological features on treatment effect. Studies analyzing influence of methodological features related to internal validity were included. We made a distinction among surgical, pharmaceutical, and therapeutical as separate types of intervention. Heterogeneity was calculated to identify differences among these types. Results Fourteen meta-epidemiologic studies were found with 51 estimates of influence of methodological features on treatment effect. Heterogeneity was observed among the intervention types for randomization. Surgical intervention studies showed a larger treatment effect when randomized; this was in contrast to pharmaceutical studies that found the opposite. For allocation concealment and double blinding, the influence of methodological features on the treatment effect was comparable across different types of intervention. For the remaining methodological features, there were insufficient observations. Conclusion The influence of allocation concealment and double blinding on the treatment effect is consistent across studies of different interventional types. The influence of randomization although, may be different between surgical and nonsurgical studies.