A realist evaluation of a clinic based lay health worker intervention to improve the management of Hypertension in rural South Africa

A thesis submitted for the degree: Doctor of Philosophy School of Public Health Faculty of Health Sciences University of the Witwatersrand, Johannesburg. 30 October 2017.1. Background Hypertension prevalence is high in Africa and is one of the commonest cardiovascular ailments. A cluster randomized control trial (RCT) was run in the Bushbuckridge sub-district, Mpumalanga, in South Africa, to test whether lay health workers (LHW), working alongside nurses in rural clinics can improve management of hypertension. The trial’s programme theory was thus management of hypertension would improve since LHWs would free up nurses by taking up some of their tasks. Nurses would then focus on clinical management of the patients. In this area, nearly half of adults are hypertensive, but only 9% have the blood pressure well controlled. In my PhD, I have used realist evaluation approach to understand the impact of the LHWs and explain “what worked for whom, under what conditions and how”. I have also discussed the practicality of combining realist evaluations and RCTs, contributing to an ongoing debate. 2. Aims To understand under what context and through what mechanisms a clinic based lay health worker intervention will enhance integrated chronic care for hypertensive patients and will modify patient outcomes in a cluster randomized trial in primary health care clinics. 3. Methods This study was a theory driven realist evaluation. It was based in realism approach which focused on explaining “why” and “how” improvements happened (or not). I used Medical Research Council’s (MRC) framework for process evaluation of complex interventions to understand and present how the different constituents of the intervention, implementation, context, mechanisms and outcomes are interconnected. My programme theory was adapted from Pawson and Tilley’s realist approach that considered outcomes from the intervention, as a configuration of the context and the mechanisms through which the intervention was implemented (context + mechanisms = outcomes). I also used other theories that describe factors for ideal chronic care (Wagner model) and effects of complexity in organizations (theory of complex adaptive system). I used a case study approach to compare and contrast experiences in the eight case clinics. The intervention and operation of the clinics were explored over time during the pre-trial period, during the preparation and development phase of the intervention, mid-way through the implementation of the intervention and towards the closure of the trial. Data collected was largely qualitative using detailed, observation of clinic activities and patient pathway, focus group discussions with community health workers and community members, semi-structured interviews Clinic Managers, Clinic Supervisors and sub-District Manager, in-depth interviews with LHWs and the Implementation Manager, semi-structured interviews with three cohorts of purposively selected hypertensive patients in their homes, patient exit structured interviews, and Implementation Manager’s and researcher diaries. Qualitative data was analyzed using Nvivo and data extraction sheets that pulled together data from different sources. Quantitative data from patient exit structured interviews was analyzed descriptively using simple statistical tests. 4. Findings At the time of the study, implementation of a government initiative called Integrated Chronic Disease Management (ICDM) model was underway in all clinics. There was rapidly increasing demand for chronic disease care as HIV management and management of stable chronic patients was referred down from hospitals to clinics. The trial clinics were swamped by HIV and hypertensive patients with 53% of the clinic visits by patients with chronic diseases done by HIV patients and 47% done by hypertensive patients. More support is available for HIV patients as compared with hypertensive patients such as tracing of patients that default treatment, counselling and testing by lay counsellors and data capturing. Clinics were affected by constant break down of BP machines and cuffs that were torn. There was limited maintenance of equipment and supply of materials i.e. patient files and packs for prepacking medication. Supply of hypertension drugs increasingly became erratic in all clinics. There was perceived shortage of nurses with some clinics being better off than others. Limited space and dilapidated Infrastructure affected chronic pathways in some clinics. Clinic management differed from clinic to clinic which affected relations among staff, relationship between staff and patients, and day to day operation of the clinic. Performance and motivation among LHWs varied across clinics and largely depended on support from other clinic staff. LHWs had background in community health work, were residents of villages served by respective clinics and had attained grade 12 (Matric). LHWs supported the nurses with appointment booking, pre-retrieval of files and filing back, measuring blood pressure, health education and prepacking of medication. They also reminded hypertensive patients prior to their appointment and followed up with those that missed appointment. The LHWs were supervised by an Implementation Manager who was a Professional Nurse by training. During the intervention, LHWs played an important role of identifying and following up with acute and other chronic patients with raised blood pressure. I placed the clinics into well, medium and poor functioning categories, although there was no clear cut difference between well and medium functioning clinics, and between medium and poor functioning clinics. However, my analysis showed that clinics require at least one of the following: strong management, teamwork, or a committed chronic care nurse, to get reasonable outcomes. If none of these exist, clinics perform poorly. 5. Discussion and conclusion The LHW programme theory partially worked as expected. The intervention was not successful in improving population levels of BP but successfully changed the functioning of clinics and delivery of care to patients with chronic diseases. The success in improving functioning of clinics varied across the intervention clinics. The LHW programme theory has explained the causal pathways that led to these differences in the programme outcomes and effects. These were mainly as a result of differences in context, mechanisms and implementation process. Using the MRC framework for process evaluation of complex interventions, the following configuration of intervention, context and mechanisms explains the study: Clinics with observed better contextual factors i.e. infrastructure, equipment, good clinic management, nurse levels, low patient loads; were clinics with positive effects in the work of the LHWs i.e. appointment booking, reminding and following up with patients, prepacking medication and filing. These were also clinics where staff related well among themselves and with patients, supported the work of LHWs and had motivated and skilled LHWs. Such clinics had positive clinic level proximal outcomes (collected through clinic link) that included patients adhering to their appointment dates and identifying patient with raised BP. Use of theories in this study has helped me to understand that health care facilities are complex organizations and are always evolving and changing. A complex mix of different factors i.e. relations, management, resources, resulted in no linear path of implementation and outcomes. Chronic care depends on positive interaction between the health system, the providers and the users. When carrying out health care interventions, implementers should consider the unique nature of facilities and strengthen the interactions between the health system, the providers and the users. I support the notion that realist evaluations can be used with RCTs and can be used to explain and strengthen findings from the trial. Trials should routinely include a process evaluation which should describe the context in detail and review how the contexts of the trial affect the implementation and outcomes, while understanding the mechanisms by which the intervention works. LHWs provided useful support to nurses in providing integrated care for chronic patients compared to usual clinics. However, the effectiveness of LHWs was affected by limited resources, increasing patient load and poor clinic management. The realist evaluation has reflected on policy and practice implications for effective chronic disease management. Such issues include, (a) lay persons can take up socially and medically oriented tasks of nurses with proper selection, training and supervision, (b) measuring vital signs for every patient that comes to the clinic has left the BP machines overwhelmed and often broken down, (c) despite introduction of integrated chronic disease management, programmes are still implemented vertically at clinic level with special attention given to HIV. The innovative methodological contribution in this PhD has been this additional level of information about the causal pathway in implementing the LHW intervention which otherwise could not have been identified just with a randomized controlled trial.LG201