Meta-analysis on continuous outcomes in minimal important difference units: an application with appropriate variance calculations

Ian Shrier; Robin Christensen; Carsten Juhl; Joseph Beyene

doi:10.1016/j.jclinepi.2016.07.012

Meta-analysis on continuous outcomes in minimal important difference units: an application with appropriate variance calculations

J Clin Epidemiol. 2016 Dec:80:57-67. doi: 10.1016/j.jclinepi.2016.07.012. Epub 2016 Jul 30.

Authors

Ian Shrier¹, Robin Christensen², Carsten Juhl³, Joseph Beyene⁴

Affiliations

¹ Centre for Clinical Epidemiology, Lady Davis Institute, Jewish General Hospital, McGill University, 3755 Cote Ste-Catherine Road, Montreal, Quebec H3T 1E2, Canada. Electronic address: ian.shrier@mcgill.ca.
² The Parker Institute, Bispebjerg and Frederiksberg Hospital, Nordre Fasanvej 57, DK-2000 Copenhagen F, Denmark.
³ Research Unit for Musculoskeletal Function and Physiotherapy, Department of Sports Science and Clinical Biomechanics, University of Southern Denmark, Campusvej 55, DK-5230 Odense M, Denmark; Department of Physiotherapy and Occupational Therapy, Copenhagen University Hospital, Herlev and Gentofte, Kildegaardsvej 28, 2900 Hellerup, Copenhagen, Denmark.
⁴ Department of Clinical Epidemiology & Biostatistics, Faculty of Health Sciences, McMaster University, 1280 Main Street West, MDCL 3211, Hamilton, ON L8S 4K1, Canada; Department of Mathematics & Statistics, McMaster University, 1280 Main Street West, Hamilton, Ontario L8S 4K1, Canada.

PMID: 27480962
DOI: 10.1016/j.jclinepi.2016.07.012

Abstract

Objective: To compare results from meta-analyses for mean differences in minimal important difference (MID) units (MD_MID), when MID is treated as a random variable vs. a constant.

Study design and setting: Meta-analyses of published data. We calculated the variance of MD_MID as a random variable using the delta method and as a constant. We assessed performance under different assumptions. We compare meta-analysis results from data originally used to present the MD_MID and data from osteoarthritis studies using different domain instruments.

Results: Depending on the data set and depending on the values of rho and coefficient of variation of the MID (CoV_MID), estimates of treatment effect and P-values between an approach considering the MID as a constant vs. as a random variable may differ appreciably. Using our data sets, we provide examples of the potential magnitude. When rho = 0.5 and CoV_MID = 0.8, considering MID as a constant overestimated the treatment effect by 33-110% and decreased the P-value for heterogeneity from above 0.95 to below 0.08. When rho = 0.8 and CoV_MID = 0.5, the magnitude of the effects was similar.

Conclusions: Considering MID as a random variable avoids unrealistic assumptions and provides more appropriate treatment effect estimates.

Keywords: Continuous outcomes; Meta-analysis; Methods; Minimal important difference; Ratio of means; Standardized mean difference; Variance.

Publication types

Meta-Analysis
Research Support, Non-U.S. Gov't

MeSH terms

Data Interpretation, Statistical*
Epidemiologic Studies*
Humans
Meta-Analysis as Topic*

Grants and funding

Canadian Institutes of Health Research/International