Are survey measures of medical careutilisation misleading?

Abstract

found when estimates from self-report data from anepidemiolgical study were compared to actual cost data extracted from administrative records. Even though the fewsubjects who were actually provided with two or more services in the two-week self-report period substantially under-reportedtheir medical care consumption, a large net over-estimate of medical care consumption was produced by theself-report data. This finding has important implications for use of self-report data from surveys such as the AustralianBureau of Statistics (ABS) National Health Survey for estimating health service consumption.By combining epidemiological survey data from the Australian Vietnam Veterans Health Study (AVVHS), with dataon actual medical care for which the Health Insurance Commission (HIC) or the Department of Veterans' Affairs (DVA)paid benefits, we were able to directly compare self-reported medical care consumption with actual medical careutilisation. The comparison revealed that veterans' self-reports were a valid measure of relative medical careconsumption because those who reported care over the past two weeks were much more likely to have been recentconsumers than those who did not. This relationship became even stronger if the comparison of self-report was extendedto data on benefits paid beyond the two-week self-report period. However, the HIC and DVA data confirmed only 51%of veterans self-reporting medical care consumption during the past two weeks actually received a service.