Published Jul 3, 2014

Brisbane, Australia - In burden of disease studies, decisions about the pooling of estimates, such as prevalence across countries, are largely made using conventional meta-analysis models. However, when there is true diversity and dissimilarity of effects, meta-analysis is not recommended. If the ultimate aim is to generate an average estimate of the disease measure across populations, there is a need to conduct sub-population-standardization, to account for differences in sub-population distribution within the populations from which primary disease measures are drawn. In the article “An Updated Method for Risk Adjustment in Outcomes Research,” published in Value in Health, researchers from theUniversity of Queensland in Brisbane describe how meta-analytic methods can be adapted to synthesize descriptive epidemiological data from different populations to arrive at pooled estimates for specific disease measures. In addition, internal and external standardization are distinguished for the first time. This approach enables a more accurate estimation of the standardized effect, and also simplifies the process of estimation of pooled measures to assess global or regional burden of disease. “Using meta-analytic methods in standardization makes it easier to carry out such pooled analyses of measures of incidence and prevalence, and this will lead to a more realistic reporting of the magnitude of global burden from different conditions, as well as differentials in disease burden across populations” said Suhail Doi, PhD who is from the University of Queensland. Having more realistic results will enhance decision making for intervention programs, for example, more equitable allocation of resources for such interventions.

---