Methodological challenges in international performance measurement using patient-level administrative data

Abstract

Aim and methodsWe conducted this case study in order to test how health system performance could be compared using the existing national administrative health databases containing individual data. In this comparative analysis we used national data set from three countries, Estonia, Israel and Finland to follow the medical history, treatment outcome and resource use of patients with a chronic disease (diabetes) for 8 years after medical treatment was initiated. ResultsThis study showed that several clinically important aspects of quality of care as well as health policy issues of cost-effectiveness and efficiency of health systems can be assessed by using the national administrative health data systems, in case those collecting person-level health service data. We developed a structured study protocol and detailed data specifications to generate standardized data sets, in each country, for long-term follow up of incident cohort of diabetic persons as well as shared analyzing programs to produce performance measures from the standardized data sets. This stepwise decentralized approach and use of anonymous person-level data allowed us to mitigate any legal, ownership, confidentiality and privacy concerns and to create internationally comparative data with the extent of detail that is seldom seen before. For example, our preliminary performance comparisons indicate that higher mortality among relatively young diabetes patients in Estonia may be related to considerably higher rates of cardiovascular complications and lower use of statins. ConclusionsModern administrative person-level health service databases contain sufficiently rich data in details to assess the performance of health systems in the management of chronic diseases. This paper presents and discusses the methodological challenges and the way the problems were solved or avoided to enhance the representativeness and comparability of results.