Big Routinely Collected Clinical Data- A Treasure Hunt for Best DM Care

Abstract

Computerization of medical record and digitalization of investigation results and drug prescriptions have enabled structured collection and linkage of big longitudinal data on the process and outcomes of care of very large numbers of people with different characteristics. Big routinely collected clinical data reflect real-world practice that can provide empirical evidence on quality, cost and outcomes of care. Such data are most valuable in research on chronic diseases such as diabetes mellitus. This presentation will describe the use of routinely collected data of 316,869 patients with diabetes mellitus (DM) managed in 73 Hospital Authority primary care clinics from 2009 to 2015 to answer important research questions on risk of complication and mortality, predictors of complications and deaths, service burden and costs, quality of care of DM, and the cost-effectiveness of a structured risk assessment and management (RAMP) intervention for DM. The big data analyses showed that for an average DM patient aged 63 under usual care the risks of any DM complication, CVD and death were 31%, 23% and 21%, respectively, and RAMP was associated with a relative risk reduction by 38%, 49% and 56%, respectively. We found a Jshaped relationship between HbA1c/ blood pressure and risks of CVD/mortality, with HbA1c 7-7.5% and SBP 130-134mmHg and DBP 65-69mmHg associated with the lowest risks. The average annual public service cost for each DM patient without complication was HKD11,015 in 2013, which was increased by 1.2 to 2.6 by each complication, and RAMP saved significant cost from reduction of complications.