Interrogating and Reflecting on Disability Prevalence Data Collected Using the Washington Group Tools: Results from Population-Based Surveys in Cameroon, Guatemala, India, Maldives, Nepal, Turkey and Vanuatu

The Washington Group (WG) tools capture self-reported functional limitations, ranging from 6 domains in the Short Set (SS) to 11 in the Extended Set (ESF). Prevalence estimates can vary considerably on account of differences between modules and the different applications of them. We compare prevalence estimates by WG module, threshold, application and domain to explore these nuances and consider whether alternative combinations of questions may be valuable in reduced sets. We conducted secondary analyses of seven population-based surveys (analyses restricted to adults 18+) in Low- and Middle-Income Countries that used the WG tools. The prevalence estimates using the SS standard threshold (a lot of difficulty or higher in one or more domain) varied between 3.2% (95% Confidence Interval 2.9-3.6) in Vanuatu to 14.1% (12.2-16.2) in Turkey. The prevalence was higher using the ESF than the SS, and much higher (5 to 10-fold) using a wider threshold of some or greater difficulty. Two of the SS domains (communication, self-care) identified few additional individuals with functional limitations. An alternative SS replacing these domains with the psychosocial domains of anxiety and depression would identify more participants with functional limitations for the same number of items. The WG tools are valuable for collecting harmonised population data on disability. It is important that the impact on prevalence of use of different modules, thresholds and applications is recognised. An alternative SS may capture a greater proportion of people with functional domains without increasing the number of items.

Automated summary

The study conducted secondary analyses of seven population-based surveys in Low- and Middle-Income Countries that used the WG tools, finding that different modules, thresholds, and applications have a significant impact on prevalence estimates. In some cases, using a broader threshold may result in higher prevalence rates.