Describing Documentation of Electronic Health Records (EHR) in Anti-Retroviral Therapy (ART) Clinics to Improve Data Quality for Healthcare Processes in Malawi

Abstract

In healthcare, documentation deals with recording of significant patient encounters within a healthcare information system. Documentation has evolved from paper-based tools to electronic means such as Electronic Health Record (EHR) systems. The EHR may optimize healthcare processes by facilitating communication, offering decision support, and allocating and distributing resources to health facilities for patient care. Faced with the Human Immunodeficiency Viruses that leads to the Acquired ImmunoDeficiency Syndrome (HIV/AIDS) pandemic, Malawi has embraced an EHR for its Anti-Retroviral Therapy (ART) clinics since 2006. The 90-90-90 HIV/AIDS eradication strategy which Malawi adopted with the leadership of the World Health Organization (WHO) United Nations against AIDS (UNAIDS) may benefit from this EHR. In this initiative, identification and enrollment of HIV positive clients into ART clinics is essential for viral load suppression. The EHR benefits rely on good data quality, which has been lacking in Low-Income Countries (LICs) including Malawi. Although data quality has many definitions, a widely accepted concept is “data quality is fit for purpose.” This concept has not been used in many data quality assessments, particularly in LICs. I asked the following research question: How can the varying purposes of EHR data use for different stakeholders inform interventions for EHR data quality improvement? I had the following research aims: • To understand the importance of data quality characteristics in the context of EHR stakeholders’ purposes of data use • To conduct a data quality assessment and understand the drivers of the observed data quality I conducted this research at primary, secondary and tertiary level ART clinics that manage over 200 patients a day in Malawi. My study used an observational study design with mixed methods. I conducted semi-structured interviews with 34 stakeholders comprised of nurses and clinicians as well as public health officials and donors who support the EHR. Additionally, I assessed 160,647 patient records that had 549,826 visits across 10 different health facilities, extracted from existing EHR data. My work established that stakeholders have clinical facing purposes or administrative purposes of data use. Stakeholders with clinical facing purposes of data use expected their data to be plausible. On the other hand, stakeholders with administrative purposes of data use expected completeness first. I found variation in the observed proportions of the data quality characteristics of completeness and plausibility. Completeness ranged from 5% to 99% while plausibility ranged from 40% to 99% across variables essential to the 90-90-90 initiative. After integrating my qualitative and quantitative results, some results indicated agreement or convergence between different sources of information about data quality. Data quality was high for the preferred characteristic of plausibility for patient health tracking. Conversely, there were divergent results between findings in my first and second aims for data management and use. For this purpose of data use, I found that plausibility was highly prioritized by participants, but in the data assessment, I found that only 40% of the records were plausible. The recommendations that follow from my work regarding interventions to improve data quality include: • Consider priorities based on importance of specific data quality characteristics for different stakeholder purposes of data use. • Assess observed data quality proportions after conducting a quantitative data quality assessment to identify where interventions should first be implemented in relation with expected stakeholder data characteristics and purposes of data use.