Identifying priorities for data quality improvement within Haiti׳s iSanté EMR system: Comparing two methods

Abstract

ObjectivesThe aim of this case study was to compare two alternative strategies for prioritizing data elements for data quality assessment (DQA) in a routine health management information system. The study used data from iSanté, a multi-site electronic medical record implemented by the Haitian Ministry of Health. We described and compared two prioritization strategies: (1) a Delphi process drawing iterative feedback from clinicians and stakeholders responsible for monitoring and evaluation (M&E) of health programs to identify consensus priorities for data on HIV patients; and (2) a process using burden of disease estimates from Haiti to establish priorities for data on primary care patients. MethodsThe Delphi process included 26 individuals across 6 institutions, including clinicians and M&E specialists. Through three rounds of questionnaires, the stakeholders provided input for prioritization of 13 indicators for completeness, accuracy and timeliness of HIV data. The burden of disease prioritization process revealed that cardiovascular disease contributed to the greatest number of disability-adjusted life-years (DALYs). This resulted in the selection of 16 data quality indicators for primary care data. ResultsBoth methods informed the definition of a set of automated data quality queries to assess internal validity, completeness, and timeliness using logic and clinical plausibility. The Delphi process benefited from stakeholder input, but was lengthy in process. The burden of disease prioritization process was objective and easier to implement, but lacked stakeholder buy-in. ConclusionsA hybrid approach guided by both disease burden and stakeholder input may be most beneficial for prioritizing data elements for DQA.