Article type
Year
Abstract
Background: Several studies with a meta-epidemiological approach have been published on effect of quality measures in clinical research. The results from these studies show some variations.
Objectives: The goal of this systematic review of meta-epidemiological studies was to systematically search and evaluate the evidence for effect of quality measures in clinical research and to determine whether differences in effect of quality measures could be attributed to type of intervention.
Methods: Systematic search of Medline, EMBASE Web of Science, Cochrane methodology register, and reference lists for meta-epidemiological studies with estimates of effect of quality measures. References were selected by predefined criteria and appraised with the Amstar tool adjusted for the purpose. Selection and quality evaluation were performed in duplicate. We anticipated, but were not limited to, effects of study size, study type, randomization, concealment, blinding (care-provider, patient and observer), and sponsoring. We made a distinction between surgical, pharmaceutical, and treatment programs (psycho-therapeutical, physio-therapeutical) as separate types of interventions. Heterogeneity was calculated to identify differences between types of intervention.
Results: We found 23 meta-epidemiological studies from which we were able to extract 50 effect estimates of 11 different quality measures. Eleven of these estimates were statistically significant. All of these confirmed the common assumptions of smaller effect estimates in studies with 'better’ quality, defined as larger studies, use of randomised allocation, proper allocation concealment, and double blinding. The only exception is a surgical study that found an increased effect estimate in randomised studies.
Conclusions: Study size and allocation concealment were more consistently found to have an influence on effect size than other quality measures. For randomization and allocation concealment there are indications that the effect size might be different between different types of interventions.
Objectives: The goal of this systematic review of meta-epidemiological studies was to systematically search and evaluate the evidence for effect of quality measures in clinical research and to determine whether differences in effect of quality measures could be attributed to type of intervention.
Methods: Systematic search of Medline, EMBASE Web of Science, Cochrane methodology register, and reference lists for meta-epidemiological studies with estimates of effect of quality measures. References were selected by predefined criteria and appraised with the Amstar tool adjusted for the purpose. Selection and quality evaluation were performed in duplicate. We anticipated, but were not limited to, effects of study size, study type, randomization, concealment, blinding (care-provider, patient and observer), and sponsoring. We made a distinction between surgical, pharmaceutical, and treatment programs (psycho-therapeutical, physio-therapeutical) as separate types of interventions. Heterogeneity was calculated to identify differences between types of intervention.
Results: We found 23 meta-epidemiological studies from which we were able to extract 50 effect estimates of 11 different quality measures. Eleven of these estimates were statistically significant. All of these confirmed the common assumptions of smaller effect estimates in studies with 'better’ quality, defined as larger studies, use of randomised allocation, proper allocation concealment, and double blinding. The only exception is a surgical study that found an increased effect estimate in randomised studies.
Conclusions: Study size and allocation concealment were more consistently found to have an influence on effect size than other quality measures. For randomization and allocation concealment there are indications that the effect size might be different between different types of interventions.