Identifying contextual determinants of problems in tuberculosis care provision in South Africa: a theory-generating case study | Infectious Diseases of Poverty
The study aimed to develop hypothetical propositions regarding the contextual determinants of problems in TB care in South Africa. To do so we devised a theory-building case study design  using mixed methods, comprising stakeholder interviews, observations of TB care, documentary review of national TB guidelines and policies and routinely available data collected between February and November 2019. To ensure coherence and wider theoretical generalisablity of findings across pre-implementation, intervention development and evaluation phases, we adopted the Context and Implementation of Complex Interventions (CICI) framework . CICI is a determinant and evaluation framework comprising three dimensions—context, implementation and setting—which interact with one another and with the intervention. For the pre-implementation phase we focused on CICI’s seven contextual domains as a means for developing hypothetical propositions on the contextual determinants of problems in the delivery of TB care, including: geographical, epidemiological, socio-cultural, socio-economic, ethical, legal and the political domain.
The setting was six public-sector primary healthcare (PHC) facilities or clinics and one public sector hospital serving impoverished urban and rural communities in Amajuba district of KwaZulu-Natal province, South Africa (see Table 1). In 2017, 7.4% of all deaths in KwaZulu-Natal were due to TB , with a high case fatality rate of 11% , and in Amajuba TB represented the highest cause of death . Amajuba district municipality is a geographically small (7102 km2) district in North-Eastern KwaZulu-Natal, comprised of three local municipalities (Newcastle, eMadlangeni and Dannhauser), eight towns, with a mix of urban, peri-urban and rural areas. The population is mainly isiZulu-speaking. The District has a total population of 556 580 (0.9% of South Africa) and 127 000 households, 12.3% of which live in informal dwellings. In 2019, 416 000 people lived in poverty, an increase of 11.3% from 2009. The largest economic sectors are community services, manufacturing, and financing .
TB care provision in Amajuba is governed by National TB guidelines and protocols (see Additional file 1 for description) [26, 27] and in 2019 the Amajuba district health department highlighted the need to address TB mortality as a key priority. Three PHC facilities and the hospital outpatient department were initially selected for inclusion and data collection was conducted in February 2019. We completed additional data collection in November 2019 in two of these PHC facilities, as well as an additional three facilities, at the request of the health department to assess screening processes and practices in greater detail.
Study population, sampling and recruitment
Primary healthcare facility staff
We recruited facility managers to inform our understanding of the organisation of TB care; and purposively sampled nurses, doctors, counsellors and community caregivers who were treating patients at each of the primary care facilities in the selected district to be interviewed and/or observed. We also recruited nurses who were not routinely seeing TB patients as well as a private sector general practitioner and one traditional healer to understand their perspectives about the organisation of TB care, and of TB and the management of it.
To be eligible for interview, patients diagnosed with TB needed to have taken treatment for at least one month, in order to be able to inform us about their experience of care. Eligible patients who arrived at the facility on each day of data collection were consecutively sampled. Nurses identified eligible patients and informed the research team who then approached the patient about participation in individual interviews.
In the second stage of data collection, patients who screened positive for TB symptoms (‘presumptive TB’) were also interviewed after vital signs’ assessment. Nurses informed patients about the research, and if the patient was willing, notified the researcher who approached the individual about participating in a short interview. Informed consent was taken and interviews were audio-recorded.
Other key stakeholders were identified and purposively sampled to obtain a broader range of perspectives of service delivery, including researchers working for the Desmond Tutu Tuberculosis Centre (https://blogs.sun.ac.za/dttc/), TB managers at district and provincial level, and members of TB Proof, a NGO which started representing health workers with occupational TB and which now represents the views of people living with and surviving TB more generally (http://www.tbproof.org/who-we-are/).
Interviews (see Table 2) were semi-structured and carried out in the language most appropriate to each participant (isiZulu = 29; English = 19), audio recorded, translated and transcribed in English. Interviews in English were conducted by RC (clinician and social scientist), AvR (social scientist), JM (social scientist) and AD (social scientist). Interviews in isiZulu were conducted by AD, supported by two fieldworkers trained in qualitative interview methods who lived locally and were known to the participating clinics. AD worked with the fieldworkers to translate the interview topic guides, identifying how best to adapt questions to retain the intended meaning when asked in isiZulu. The interview team carefully considered who was best placed to carry out different interviews so that participants would feel comfortable to disclose their experience and perceptions of TB care provision but also potentially sensitive topics such as stigma and psychological distress. The local fieldworkers were critical in this regard, facilitating insights into patients’ experiences which were perhaps less likely to be disclosed if conducted by another team member. Following national guidance on the ethics of payment, incentives to participate were not offered as interviews took place following the patient’s consultation and did not require additional travel or time off work to be interviewed .
For individual interviews with patients, semi structured questions were avoided to minimise the researcher imposing their own assumptions on participants’ experiences during the interview. Instead the researcher elicited stories  of patients’ journeys through the TB care pathway from the point where they first noticed symptoms through to treatment and followed up topics as they arose. Patients who screened positive for TB were briefly interviewed to elicit their understanding of the screening questions, instructions for providing sputum testing and next steps for their care.
Interviews with nurses, counsellors, community caregivers and doctors explored the provision of TB care, implementation of infection control measures, and solutions for strengthening TB care. Interviews with provincial managers and stakeholders explored current services and interventions to support people with TB, interventions to reduce TB infection and address psychological needs of patients with TB, and perspectives on required interventions to improve service delivery at a system wide, organisational and individual facility level.
Interviews conducted in isiZulu were translated and transcribed by one of the fieldworkers to help ensure the meaning of participants’ responses were retained rather than a direct literal translation. RC and AD then reviewed all isiZulu recordings and transcripts to identify and resolve any potential misrepresentation from the original meaning.
Healthcare facility observations
Within all six primary healthcare facility we carried out periods of direct, non-participant observations within non-clinical areas to understand the organisation and process of TB care including patient flows, TB screening and testing, infection control measures and data capturing processes. We recorded contemporaneous written field notes of their observations using a semi-structured observation guide (Additional File 2).
Documentary and routine data review
Relevant policy and guidelines (Additional File 1), district TB mortality reports and routine data  were reviewed to identify best practices for TB screening, testing, diagnosis, treatment initiation, infection control and follow-up of patients, mortality burden and data system bottlenecks.
In order to generate hypothetical propositions on the contextual determinants of problems in TB care delivery, we identified relationships between CICI domains and across macro-contextual features (e.g. national and international healthcare policy, discourses, infra-structural relations, socio-economic factors), meso-contextual features (i.e. organisation of TB care at a primary healthcare facility level), and micro-contextual features (i.e. patients’ and clinician’s behaviour). We drew on Braun and Clarke’s thematic analysis as a ‘contextualist’ method , examining how macro-contextual features shaped meso and micro (or vice versa), thereby tracing a thread between specific perspectives or observations to the broader social historical context in which they were manifested (see Fig. 1). Rather than necessarily developing higher-order themes within the discrete datasets, this approach required treating each participant report or observation as a potential contextual feature which we then explored within and across contextual levels and across data types to develop and test emerging theories , for example how reported implementation of infection control measures matched recommended practice within TB guidelines as well as our observations within the facility. This iterative approach enabled us to transition from the particularities of Amajuba as a single case to theoretical explanations of how different contextual determinants applicable in other South Africa settings may shape the patterns we observed, facilitating generalisable inferences and predictions on what kinds of intervention components are needed to tackle different contextual determinants of problems in TB care.
All interviews were inductively coded for features using the qualitative data analysis software NVivo 12 (QSR International, https://www.qsrinternational.com/nvivo/home). This provided detailed staff, manager and stakeholder perspectives of the process and content of TB care in facilities; and for patients, pathways to care, and experiences of living with and managing TB. Initially four researchers (RC, AD, JM, AvR) coded two of the same interviews and compared these to identify and resolve differences. The facilities were then divided between these research members, with the coding of features checked by JM. A constant comparison approach was adopted, working iteratively between data obtained from different interviewees within and between facilities to test out categories, including searching for disconfirming cases . First order codes were then analysed to consider if they could be developed into higher order codes to better facilitate understanding of emerging relationships between contextual features. Field notes were analysed to provide a detailed description of the process and content involved in provision of TB care, including screening, testing, data capture and infection control measures.
Data synthesis within CICI framework
As the analysis developed we mapped contextual features onto the seven contextual domains of the CICI Framework. Any feature which did not readily map onto a domain was discussed and assigned to a domain or an additional domain added. We then analysed the mapped domains in light of emerging theories to generate hypothetical propositions which specified the contextual determinants of problems in delivering effective person-centred TB care. Finally, we hypothesised which intervention components and implementation strategies would logically tackle those determinants. Throughout analysis, we held regular meetings between all project team members and the district health department to review findings, discuss emerging theories on the relationship between features and contextual domains, and later develop hypothetical propositions and intervention components.
The key ethical principles of voluntary and informed participation, confidentiality and safety of participants were used in all researcher and participant interactions. Written consent for interviews was obtained from all stakeholders, facility managers, clinicians and patients. Facility managers provided consent for observations of non-clinical areas. All participants were provided with written information about the research, informed that their participation was voluntary and that they could withdraw from participation at any time. Patients were typically approached after their consultation and interviews were conducted in outdoor areas with researchers wearing a mask where possible.