Causes of death and strategies to improve linkage to HIV and TB care in adults being investigated for TB in South Africa

Abstract

Introduction There is high mortality amongst adults being investigated for tuberculosis (TB) in primary health care (PHC) facilities in South Africa, but the causes of death (CoDs) are not clearly defined. Patient and health system-related factors such as diagnostic delays, delays in delivery of laboratory results and treatment start delays exist and prevent this population from linking into care. Innovative interventions are needed to assist with linkage to HIV and TB care. This is important as the first two goals of the South African National Strategic Plan are to hasten the development of interventions to prevent new infections, and to reduce morbidity as well as mortality by providing treatment and adherence support. The first target of the global STOP TB plan is to reach 90% of all people with TB and place them on appropriate treatment. The national tuberculosis guidelines also recommend that everyone undergoing TB investigation should be tested for HIV, a recommendation that links to the Joint United Nations Programme on HIV/AIDS (UNAIDS) targets to ensure that 90% of people living with HIV know their status. The aim of this PhD was to determine the causes of death amongst adults being investigated for TB, and pilot two interventions in PHC facilities to improve linkage into HIV and TB care, a case manager and mobile health (mHealth) intervention. A study investigating causes of death in people being investigated for TB and two studies piloting interventions to improve linkage to care form the basis of this thesis: (i) Determining causes of death amongst people undergoing TB investigation using verbal autopsy in South Africa (ii) A pilot study to evaluate the feasibility and acceptability of using a case manager to improve linkage to TB and HIV care amongst people investigated for TB in primary health care facilities in South Africa (iii) Development and implementation of mHealth-based clinical decision support messaging for health care workers and patients in the initial investigation of tuberculosis in primary health care facilities in South Africa. This thesis comprises of these three studies and is presented in two parts: an integrative narrative which synthesises the studies, followed by three published papers and one unpublished manuscript presenting the studies. Methods The first study determined the CoDs in people being investigated for TB and was a sub-study of the XTEND Trial, a cluster-randomized trial comparing Xpert MTB/RIF versus smear microscopy as the initial test for diagnosing TB. The study was done amongst caregivers of the 231 deceased participants who were enrolled in the XTEND study. The caregivers were aged 18 years and above and lived in four provinces (Gauteng, Free State, Eastern Cape and Mpumalanga) of South Africa. For the sub-study, all contactable caregivers of deceased XTEND participants were interviewed using the World Health Organization (WHO) approved verbal autopsy (VA) tool. The VA tool has a quantitative section with questions requiring ‘Yes’ or ‘No’ responses and a narrative section which allows for detailed documentation of events leading to death. Cause of death (CoD) was determined using physician certified verbal autopsy (PCVA) and InterVA 4.3 software with comparison of the CoD profile between the two methods. Responses from the narrative section of VA were used to ascertain perceived barriers of care as described by caregivers of the decedents. The second study piloted a case manager intervention, where a lay counsellor supported patients undergoing TB investigations for three months by encouraging HIV testing, calling them to return for results, following up on their laboratory results, and assisting with treatment initiation if confirmed to have TB. The study was done amongst people undergoing TB investigation aged 18 years and above, in six PHC facilities in Mpumalanga and Gauteng provinces of South Africa. Linkage to TB care was defined as starting TB treatment within 28 days of having sputum taken for TB investigation. HIV linkage to care for HIV positive people was defined as having blood taken for CD4 count test and, for those eligible, starting antiretroviral therapy within three months. The feasibility of implementing the intervention was assessed by attempts made through telephonic calls or participants visiting the clinic. Qualitative interviews were conducted to assess the acceptability of the intervention and explore barriers and facilitators of linkage to HIV care and TB treatment initiation. The third study developed and piloted an mHealth application to capture patients’ TB investigation data and deliver laboratory test results directly to patients via text messaging.

The study was done amongst people undergoing TB investigation aged 18 years and above in two clinics in the Gauteng Province of South Africa. Data were collected over two periods: pre-implementation period and implementation period. The pre-implementation period involved facility health care workers recording TB investigation data in the paper-based TB identification register, and the implementation period involved the facility health care workers recording TB investigation data in an electronic register while research staff recorded the data in paper-based TB investigation registers. The feasibility of using the mHealth application was assessed in the implementation period by comparing proportions of patients with complete personal details and documented results using mHealth application versus paper. TB indicators (turn-around time from sample collection to when a patient gets sputum results, time to TB treatment initiation and proportion on treatment within a specific period) were compared in the pre-implementation and implementation periods to assess the potential effectiveness of the mHealth application. Qualitative interviews were conducted to assess the acceptability of the intervention. Results Amongst the 231 who died during the XTEND trial, caregivers of 137 decedents were interviewed as part of the study on causes of death. Of the 137 deceased, 76 (55.4%) were males, median age was 41 years (interquartile range [IQR] 33-50 years). PCVA assigned 70 (51.1%) deaths to TB; 21(15.3%) to HIV and 46 (33.5%) to other CoD. InterVA 4.3 software assigned 48 (35.0%) deaths to TB, 49 (35.7%) to HIV and 40 (29.1%) to other CoD. Delays in accessing health-care, treatment delay and clinic organisation were perceived to challenge the health-seeking journey of decedents. Amongst 562 participants enrolled in the case manager study, 307(54.6%) were females with a median age 36 years (IQR 29-44). A total of 189/562 (33.6%) participants needed linkage to care: 132 needed only HIV care, 35 needed TB treatment initiation only, 22 needed both HIV care and TB treatment initiation. Of the 57 participants who needed TB treatment, 53 (93%) were initiated on treatment. By the end of three months follow-up, 40 participants (29/132 [22%] needing HIV care, 4/35 [11.4%] needing TB treatment and 7/ 22 [31.8%] needing HIV linkage to care and TB treatment initiation) had not linked to care. Patient (fear and competing priorities) and health system (poor implementation of updated TB and HIV management guidelines as well as the way facilities booked patients for ART initiation)

related factors were highlighted as barriers to HIV linkage. Patients’ ill-health and clinic experience were the main facilitators of initiating TB treatment. In the mHealth study, 457 patients were recorded in the register during the pre implementation period (195 [42.7%] males, median age 34 years [IQR 28-40]). During implementation period, 319 patients were recorded in paper register and mHealth application (131 [41.1%] males, median age 32 years [IQR 27-38]). Implementation of the mHealth intervention in PHC facilities showed no difference in proportion with complete personal details: (mHealth 95% versus paper register 94.0%, [p=0.54]) and proportion with documented results: (mHealth 97.4% versus paper 97.8%, [p=0.8]). The intervention improved the proportion of patients with results available within 48 hours from sample collection, (pre-implementation 68.6% versus implementation period 96.8%; [p>0.001]) but did not improve time to TB treatment initiation (pre-implementation, median = 4 days interquartile range [IQR] 2-6; implementation median = 3 days IQR [2-5]; p=0.5). The delivery of results via text message did not infringe on patients’ privacy. Conclusion TB and HIV are leading CoDs amongst adults being investigated for TB in primary health care clinics in South Africa, as determined by VA. This points to the importance of integrated TB and HIV care and linkage into care for both diseases in this group. InterVA underestimates TB associated deaths and needs to be refined to be able to include extrapulmonary TB as a CoD. Support interventions using lay workers as case managers to link adults undergoing TB investigation into HIV and TB care might help improve TB treatment initiation. However, this kind of support is not adequate to assist HIV positive patients’ to link into HIV care, showing the complexity of the HIV linkage to care process. Future support interventions to be piloted should address both patient and health system factors. Finally, mHealth applications could take the place of current paper-based TB care programme case-finding tools, and could substantially improve results turn-around time and management of TB within the programme. These applications may empower patients with direct receipt of TB results. Implementation of mHealth interventions to support and improve TB care has the potential to improve patient-level outcomes, and large-scale evaluations of these technologies are urgently required. Additionally, future interventions to improve TB treatment initiation

need to focus on how to improve patients’ clinic experience and empower patients to seek care.