English 
Français

Using routinely collected data to estimate quality of care: evidence from tuberculosis diagnostic testing in South Africa

Zoe McLaren, University of Michigan
Alana Sharp, University of Michigan
Jifang Zhou, University of Michigan
Margaret Triyana, Nanyang Technological University
Ananta Nanoo, National Institute for Communicable Diseases. South Africa

The quality of health care provided varies across health facilities and affects patient health outcomes. Identifying poorly performing health facilities enables the targeting of additional resources to improve the quality of care. We use data from the National Health Laboratory Service database on tuberculosis tests performed on patients in public health facilities between 2004 - 2011, which includes over 30 million test records from over 10 million patients. We examine compliance with clinical guidelines for TB diagnosis and follow up testing of patients with suspected TB. Overall compliance was lower than expected, but facility-level measures of quality of care were correlated across indicators. Compliance was higher in hospitals than in clinics and varied considerably by district. Within a facility, however, compliance varied little by patient gender or age. Routinely collected TB care indicators are useful as a standardized measure of quality of care and a source of continuous facility monitoring.

  See extended abstract

Presented in Session 21: Big Data and Data Revolution