Menu
Menu
Menu
Menu
Introduction Despite access to free antiretroviral therapy (ART), many HIV-positive Zambians disengage from HIV care. We sought to understand how Zambian health system ‘hardware’ (tangible components) and ‘software’ (work practices and behaviour) influenced decisions to disengage from care among ‘lost-to-follow-up’ patients traced by a larger study on their current health status. Methods We purposively selected 12 facilities, from 4 provinces. Indepth interviews were conducted with 69 patients across four categories: engaged in HIV care, disengaged from care, transferred to another facility and next of kin if deceased. We also conducted 24 focus group discussions with 158 lay and professional healthcare workers (HCWs). These data were triangulated against two consecutive days of observation conducted in each facility. We conducted iterative multilevel analysis using inductive and deductive reasoning. results Health system ‘hardware’ factors influencing patients’ disengagement included inadequate infrastructure to protect privacy; distance to health facilities which costs patients time and money; and chronic understaffing which increased wait times. Health system ‘software’ factors related to HCWs’ work practices and clinical decisions, including delayed opening times, file mismanagement, drug rationing and inflexibility in visit schedules, increased wait times, number of clinic visits, and frustrated access to care. While patients considered HCWs as ‘mentors’ and trusted sources of information, many also described them as rude, tardy, careless with details and confidentiality, and favouring relatives. Nonetheless, unlike previously reported, many patients preferred ART over alternative treatment (eg, traditional medicine) for its perceived efficacy, cost-free availability and accompanying clinical monitoring. Conclusion Findings demonstrate the dynamic effect of health system ‘hardware’ and ‘software’ factors on decisions to disengage. Our findings suggest a need for improved: physical resourcing and structuring of HIV services, preservice and inservice HCWs and management training and mentorship programmes to encourage HCWs to provide ‘patient-centered’ care and exercise ‘flexibility’ to meet patients’ varying needs and circumstances.
This qualitative study was conducted in 2015 and was nested within a larger quantitative study exploring rates and reasons for disengagement.39 It aimed to understand, from Zambian patients’ and HCWs’ perspectives, how care within the health system influenced engagement and disengagement from long-term HIV care and treatment. In our framework, the health system includes the formal system, as well as the patients and the larger community. The qualitative study was conducted in 12 clinics, selected from 4 Zambian provinces—Lusaka, Southern, Eastern and Western provinces. These settings comprise multilingual ethnic groups, with Bemba and Nyanja the most widely spoken local languages in Lusaka Province, Tonga in Southern Province, chi-Nyanga in Eastern Province and Lozi in Western Province. The socioeconomic status and housing conditions of residents in these different locations are mixed, but predominantly poor. Site selection was achieved in three phases. First, a random sample of 31 health facilities stratified to ensure rural and urban facilities from each of the four provinces were selected. Second, from this larger sample, eight health centres—one urban, and one rural for each of the four provinces—were purposively selected based on facility characteristics, location and patient load. We considered urban health centres and level 1 hospitals interchangeable for the purpose of the qualitative study since they are often of similar size and operating capacity and are located in more comparable socioeconomic and geographical environments. After completing the first round of data collection at these eight facilities, an additional four facilities (one per province) were purposively selected to conduct follow-up focus group discussions (FGDs) that tested the initial findings and probed emerging and unclear issues. Selection was made based on the nature of the issues we sought clarification about—for example, in two provinces rural facilities were selected as clarification and testing of findings related predominantly to these aspects; in the other two provinces, urban facilities were selected. The 12 clinics all provided outpatient health services, and all the urban facilities additionally provided some inpatient services. Other services shared by the 12 facilities included maternal and child health department (MCH), tuberculosis treatment department (TB corner), HIV or antiretroviral department (ART clinic), and laboratory and environmental health team (EHT) depending on location and resourcing. The professional HCWs working in the clinics predominantly included clinical officers, nurses, pharmacy technologists, data associates, and environmental and health technologists. In the ART department, the professional HCWs provided counselling, testing, clinical consultations and drug distribution. By comparison, lay HCWs were responsible for HIV testing and counselling, health education, patient navigation and file retrieval, but may also sometimes substitute for professional HCWs and carry out designated clinical tasks such as blood pressure measurements. We conducted 69 indepth interviews (IDIs) with patients from the following categories: (1) currently in HIV care at the clinic where they initiated HIV care (‘in care’); (2) disengaged from care; (3) in HIV care after transfer to a different clinic; and (4) next of kin for deceased patients. Among the in care, transferred, disengaged and dead categories, male and female representation was almost equal (male=29/female=31). Geographical representation was also similar, although with slightly higher participation among rural residents (n=38, 55%) (table 1). Sex and location of interview and FGD participants FGD, focus group discussion; NoK, next of kin. Twenty-four FGDs with a total of 158 participants were conducted with urban and rural lay and professional HCWs to understand their perceived role in patients’ care engagement decisions (table 2). FGDs were separated between lay and professional health workers but were mixed sex. Number of FGD participants by cadre and location FGD, focus group discussion; HCW, healthcare worker. Direct observations at health facilities were undertaken to clarify the operational context of care. The ‘in-care’ patients were recruited from the files of patients present on a study-visit day using a simple random sampling method. The categories of ‘lost’ patients were recruited during the study tracing exercise with the help of peer educators who were engaged as data collectors of patients lost from HIV care. From each of the categories of ‘lost’ patient (disengaged, transferred and next of kin to dead) traced by the main study, research assistants (RAs) asked participants if they would be willing to take part in a follow-up interview until two participants were recruited from each facility. A balance between male and female participants was sought, although due to pragmatic considerations not always achieved. No patient sampled from the ‘in-care’ patients refused to participate. No patient sampled from the ‘lost’ traced category declined to participate in the interviews after full information was provided by the tracers. Four Zambian RAs with competence in local languages spoken in the study sites were recruited to collect data. Their recruitment considered previous experience in health-related research. They underwent a 5-day training covering human subjects’ protection, familiarisation with the study’s aim and the study tools, and best-practice approaches to qualitative research data collection. The IDI questionnaire guide was designed in English and translated into the four main local languages used in the study sites, namely Nyanja, Lozi, Bemba and Tonga. IDIs lasted between 40 and 120 min and were conducted in the participant’s choice of language. We asked patients about their personal experiences while accessing care. We included questions on caregiver attitudes, information availability and sociocultural aspects, and how they affected the interviewee’s perceptions and choices in seeking care. Interview questions were all open-ended to enable RAs to probe for causal mechanisms influencing patients’ engagement in care. The RAs took down summarised field notes that included any non-verbal expressions they observed. For the deceased, we interviewed a close family member or friend to understand both the sequence of events that led up to death, as well as the families’ perception of the dead patient’s experiences in care and their own role in this process. In each of the 12 facilities, the recruitment of FGD participants was achieved by issuing open invitations to all HCWs at the facility to attend one of two FGD sessions. Participants were then enrolled on a first come first served basis. The FGDs for lay and professional staff were separated to enable lay staff to speak freely without the fear of interference from their supervisors. The FGD guide explored the patient–HCW interaction to understand the relationship between HCWs’ perceptions and patients’ own description of experiences accessing care. The questions were open-ended to enable the facilitator to probe emerging themes, and field notes containing contextual details and non-verbal observations were taken by the RAs to aid in the interpretation and analysis of the data. FGDs took between 1 hour and 3.5 hours. The main language used in the FGDs with professional HCWs was English, but facilitators allowed the use of local language in discussions. All the FGDs were conducted at the health facilities. The RAs documented direct observations of healthcare facilities’ operations as field notes that were then formalised into research memos. Direct observations took place in the original eight health facilities for two consecutive ART clinic days and lasted between 8 and 10 hours. The process began with the RA introducing themselves to the person in charge of the facility. The RA would then proceed to sit in a certain department (eg, TB corner, laboratory, pharmacy (where one existed) and clinician rooms) and write down their observations on the research guide. The guide included sections on operational features, intraprovider relations, patient–provider relations and their context. Data from observations helped build a picture of typical workflows and human interactions that drive health centre operations and that influence patients’ experience and decisions related to care-seeking. Audio recordings, transcribed scripts and observation memos were saved using a unique identification method and saved on password-protected computers. All audio-recorded interviews were transcribed verbatim and simultaneously translated into English (for interviews in local language). A two-stage approach to quality checking was undertaken. In the first stage, the RAs quality-checked their own scripts and in the second stage the first author quality-checked all the scripts. Anonymised transcriptions, observation memos and notes were imported into QSR NVivo. Inductive methods were used to code the data40 41 by the first and last author. Coding was an iterative process that categorised related codes and subcodes and stratified them by study sites and participant engagement status for further exploration and interpretation of the findings. Arrangement of themes according to a conceptual framework (the hardware/software model) was subsequently used to help further refine, organise and reflect on the findings. Draft findings and interpretations were circulated for review to all the study investigators and the Study Advisory Committee. Written informed consent for participating in the study was sought and granted by all the interviewees who agreed to participate in the study. To ensure privacy and confidentiality, the interviews were conducted in a private environment either at a home or health facility with only the RA and the interviewee present. Permission to record interviews was sought from all the study participants.
N/A
