Improving the interpretation of quality of life evidence in meta-analyses: the application of minimal important differen
- PDF / 788,771 Bytes
- 5 Pages / 595.276 x 793.701 pts Page_size
- 72 Downloads / 188 Views
RESEARCH
Open Access
Improving the interpretation of quality of life evidence in meta-analyses: the application of minimal important difference units Bradley C Johnston1, Kristian Thorlund1, Holger J Schünemann1,2, Feng Xie1,3, Mohammad Hassan Murad4, Victor M Montori4, Gordon H Guyatt1,2*
Abstract Systematic reviews of randomized trials that include measurements of health-related quality of life potentially provide critical information for patient and clinicians facing challenging health care decisions. When, as is most often the case, individual randomized trials use different measurement instruments for the same construct (such as physical or emotional function), authors typically report differences between intervention and control in standard deviation units (so-called “standardized mean difference” or “effect size”). This approach has statistical limitations (it is influenced by the heterogeneity of the population) and is non-intuitive for decision makers. We suggest an alternative approach: reporting results in minimal important difference units (the smallest difference patients experience as important). This approach provides a potential solution to both the statistical and interpretational problems of existing methods. Introduction Health-related quality of life (HRQL) is increasingly recognized as an important outcome in randomized trials. Disease-specific HRQL instruments provide critical information because of their ability to detect small but important treatment effects [1,2]. Typically, for specific conditions, a number of disease-specific instruments are available. For example, there are at least five instruments available to measure HRQL in patients with chronic obstructive respiratory disease (COPD) (Chronic Respiratory Questionnaire, Clinical COPD Questionnaire, Pulmonary Functional Status & Dyspnea Questionnaire, Seattle Obstructive Lung Disease Questionnaire, St Georges Respiratory Questionnaire)[3]. Clinical trial investigators use different HRQL instruments for various reasons, including their familiarity with an instrument. This creates challenges for metaanalysts seeking summary estimates in systematic reviews of trials addressing the same or similar HRQL constructs. Choices include reporting summary estimates for each separate measurement instrument, or * Correspondence: [email protected] 1 Department of Clinical Epidemiology & Biostatistics, McMaster University, Hamilton, Ontario, Canada Full list of author information is available at the end of the article
pooling across instruments. The former approach is less appealing in that it leaves the clinician with multiple imprecise estimates of effect. A widely used approach to providing summary estimates across instruments - an approach endorsed by the Cochrane Collaboration - involves dividing mean differences between intervention and control in each study by the study’s standard deviation (SD) and calculating what are called “standardized mean differences” (SMDs) or “effect sizes”. Ultimately, systematic reviews using this approach will
Data Loading...
