Background

Tuberculosis (TB) is the commonest opportunistic infection and cause of death among human immunodeficiency virus (HIV)-infected patients in resource-limited countries [1]. In 2015, there were an estimated 10.4 million new (incident) TB cases worldwide, of which 5.9 million (56%) were among men, 3.5 million (34%) among women, and 1.0 million (10%) among children. In addition, people living with HIV accounted for 1.2 million (11%) of all the new TB cases [2]. In 2014, the World Health Organization (WHO) reported 83% incident TB cases worldwide out of which one third of these new TB cases originated from the African continent with high burden countries (HBCs) [3]. A similar trend was documented in the 2016 WHO global TB report which showed that the proportion of TB cases living with HIV was highest in the WHO African Region (31%) and exceeded 50% in parts of southern Africa [2].

The extent of the combined TB and HIV epidemics in South Africa has created enormous operational challenges for healthcare delivery [4]. Prior to the implementation of TB-HIV integration, the South African healthcare system provided separate vertical programmes for TB and HIV services delivered by different healthcare staff, often located in separate clinics [4,5,6]. The vertical model of delivering TB-HIV care to co-infected patients relied upon referral and linkage to care programmes (between TB and HIV programmes and vice versa). This proved problematic and inefficient as referral between programmes depended chiefly on patients’ resources and health-seeking behavior which was unmonitored [4, 6].

Integration according to Uyei et al. (2014) is operationalized by three domains: functional, organizational, and clinical integration [7]. An integrated model of TB and HIV healthcare service delivery is an efficient use of health system’s resources that would address the two very important co-epidemics [8]. A number of studies provide evidence of the relationships in the integration framework that applies to TB and HIV healthcare delivery [7, 9,10,11]. However, the optimum model for integrated TB-HIV services in a clinical setting is unknown.

South African guidelines on TB-HIV integration

The South African Department of Health (SA DoH) has developed guidelines and policies supportive for the integration of TB-HIV services. The key focus areas for TB-HIV integration as standard of care in the most recently updated SA DoH guidelines [8, 12,13,14] is outlined in Table 1.

Table 1 Focused areas for TB-HIV integration
Full size table

Clinical benefit of known TB-HIV integration interventions

Randomized controlled trials have demonstrated that early initiation of ART during TB therapy improved survival of TB-HIV co-infected patients by 56% [15, 16]. Additionally, initiating ART early during TB treatment (within 2–4 weeks) increased AIDS-free survival by 34–68% among patients with advanced HIV disease [15, 17, 18]. In spite of TB-HIV integration being incorporated into international and South African guidelines, mortality rates in 2015 for TB-HIV co-infected patients in South Africa was 133 per 100,000 population, which is more than three times higher than mortality in HIV-negative TB patients, who had mortality rates of 46 per 100,000 population [2].

Initiation of cotrimoxazole preventive therapy (CPT) before or with ART, irrespective of CD4 count, in co-infected patients has been shown to reduce severe bacterial infections in an observational study [19] and mortality by 27% in a multi-site randomized clinical trial (RCT) conducted in Africa [20]. The clinical benefits associated with the use of CPT was adopted by WHO for use as an adjunctive therapy for improved outcomes in the management of TB and HIV co-infected patients in the 2014 treatment guidelines [20,21,22]. In addition, findings from a RCT conducted in South Africa showed that 12 months of isoniazid preventive therapy (IPT) conferred a 37% reduction in risk of active TB in ART-naïve patients [23]. The benefit of 6 months of IPT and early ART irrespective of baseline CD4 count was also recently confirmed in the TEMPRANO study that showed 44% lower risk of severe HIV-related illness and a 35% lower risk of death from any cause [24].

Rationale of the study

Despite the inclusion of the evidence above in standard TB and HIV treatment guidelines, implementation of these interventions as part of TB-HIV integration in South Africa remains poor [23, 25, 26]. Premature mortality from TB and HIV in women and men account for 53.7 and 24.2% of deaths in the 15–24 year age group and for 44.4 and 47.2% of deaths in the 25–64 age group, respectively [27]. Cost-effective and sustainable strategies to strengthen integration of known effective TB-HIV interventions in primary health care (PHC) clinics, the main service delivery point for millions of patients, will abrogate the high mortality associated with TB-HIV co-infection [28]. A systematic review by Uyei and colleagues in 2011 of observational studies using secondary data on TB-HIV integration in sub-Saharan Africa found several benefits from integration and identified the need for additional research to identify barriers to integration as well as strategies to improve TB-HIV integration [26].

Theoretical framework

Models of TB-HIV integration have ranged from TB clinics referring patients to HIV clinics and vice versa to full integration where both services are available at a single facility, on the same day, by the same healthcare worker [5]. A group of South African researchers developed a model to illustrate the critical health system mechanisms that are essential to operationalize TB and HIV service integration [7]. Implementing integration requires Functional integration which is the extent to which integration is supported at the policy and budget level; Clinical integration which is the extent to which TB and HIV care, treatment, diagnostic testing, and health education activities are taking place concurrently; and Organizational integration which is the extent to which facility level resources (e.g., staff, infrastructure, space, patient files, and data systems) and processes (e.g., patient flow) are integrated [7]. While the model efficiently explains what is needed to integrate TB and HIV services, it does not illustrate how TB and HIV service integration should be integrated. Quality improvement as a method to improve organizational and clinical integration is on the cusp of widespread roll out in South African Department of Health facilities and there is enough support and political will for creating a culture of quality improvement (QI) in facilities [8, 14, 29].

We propose a theoretical framework which is an adaptation of the Uyei et al. (2014) model that demonstrates the central role of QI as a catalyst to operationalizing integration. The SUTHI study intervention has targeted PHC supervisors and frontline clinic staff as the recipients of the QI intervention. Figure 1 illustrates that through continuous QI activities, we theorize that organizational and clinical integration can be improved and strengthened and lead to improved patient and organizational outcomes. Using the collaborative breakthrough series approach, developed by the Institute of Healthcare Improvement (IHI) [30,31,32,33], we propose that a series of timed collaborative learning sessions that brings intervention clinics together coupled with mentorship visits at the facility level will impact clinical and organizational integration activities. Our framework proposes that QI alone may not be sufficient to bring about improvements in integrated service delivery hence evaluating, measuring, and monitoring environmental and contextual factors (confounders) as well as delivering a good quality QI intervention is key to improving TB, HIV, and integrated TB-HIV services.

Fig. 1
figure 1

SUTHI study theoretical framework. IPT isoniazid preventive therapy, CPT cotrimoxazole preventive therapy, PLWHA people living with HIV/AIDS, ICF intensified case finding, HCT HIV counseling and testing

Full size image

Methods

Aims and objectives

This study aims to reduce mortality due to TB, HIV, and TB-HIV co-infection through a QI strategy for scaling up of TB and HIV treatment integration in rural PHC clinics in South Africa. The hypothesis is that survival rates will be lower in TB, HIV, and TB-HIV co-infected patients accessing health care at clinics implementing the study intervention to deliver integrated TB-HIV care, compared to survival in patients accessing health care at clinics that provide only the standard of care for people with TB and or HIV.

Specific objectives include:

  1. 1.

    To determine the impact of a QI-mediated TB-HIV care integration on patient mortality.

  2. 2.

    To determine the effectiveness of peer-led QI approach to enhance integration of TB-HIV healthcare delivery.

  3. 3.

    To identify clinic-level factors that impact on the implementation of integrated TB-HIV services.

  4. 4.

    To determine the cost-effectiveness of implementing TB-HIV integration using the QI approach.

  5. 5.

    To develop an intervention, comprising QI-based change ideas, tools, and approaches for the scale up, implementation, and sustainability of integrated TB-HIV services across South Africa and in other resource-constrained settings.

Study setting and design

This is an open-label cluster randomized controlled trial where the cluster is defined as the group of clinic(s) under the same PHC clinic supervisor, where each of the 16 PHC clinic supervisors may oversee between 3 and 5 PHC clinics. Cluster randomization was chosen for practical reasons because the study will be carried out in pragmatic settings involving 40 PHC clinics within the King Cetshwayo district (formerly called uThungulu) [34] and Ugu districts in KwaZulu-Natal, South Africa. A total of 11.1 million people (19.9% of the South African population) live in KwaZulu-Natal [35]. The province has the highest TB-HIV disease burden in South Africa with an estimated TB-HIV co-infection rate of 70% [12, 36]. The two districts were selected because they are rural with high burden of TB, HIV, and TB-HIV co-infection, despite TB and HIV treatment services in accordance with current guidelines being available [13, 14, 37]. In 2016, 82.3% of TB-HIV co-infected patients were initiated on ART in KwaZulu-Natal, slightly lower than the national ART initiation rate of 84.5% [38]. King Cetshwayo district, located on the northern coast of KwaZulu-Natal, has a population of 937,793, with approximately 80% living in rural settings [39]. Similarly, Ugu district, situated in southern KwaZulu-Natal has 86% of its population of 750,214 living in rural areas [38, 40]. The in-hospital case fatality rate due to TB-HIV co-infection in both King Cetshwayo district (38.4%) and Ugu district (38.8%) remains high despite ART availability [27]. In 2015, over a quarter of all deaths in KwaZulu-Natal, irrespective of age or gender, were caused by TB (15.5%) or HIV (12.2%). PHC clinics offer community level frontline services and chronic care for all health ailments, including TB and HIV. Patient level data for these diseases are routinely captured electronically [41].

Study population

Anonymized clinical and programme outcome data for all TB, HIV, and TB-HIV co-infected patients accessing services in the 40 clinics will be included in the study analysis. According to estimation from the clinic headcount data, an average of 4500 patients are seen in the two districts each month. In addition, the healthcare workers (HCWs) including PHC supervisors will be interviewed to collect clinic level information after obtaining informed consent for participation in interviews. These interviews will be collated and analyzed together with patient level data.

Randomization and study groups

Randomization will occur at the level of the PHC supervisor. This is being done to avoid contamination of intervention effects into control clinics since each supervisor routinely supports multiple clinics. Each PHC clinic supervisor will be randomly assigned in a 1:1 ratio to the intervention or control arm. The study investigators would provide a list of facilities and supervisors to be randomized using a computer-generated randomization list generated by study statistician. In South Africa, a PHC clinic supervisor fulfills a management role at PHC clinics [42], and are mandated to ensure smooth clinic operations through oversight of clinic performance against set targets including patient clinical outcomes and programme performance targets, and oversees the implementation of clinical guidelines. Each PHC clinic supervisor may oversee between three and five PHC clinics within one district. Therefore, through randomization of 14–16 PHC clinic supervisors working in both the Ugu district and King Cetshwayo district, 40 clinics (20 in each arm) will be enrolled into this study as each supervisor oversees on average three to five medium-sized clinics. Not all clinics under the purview of the PHC clinic supervisors randomized will be enrolled into the study. Clinics will be excluded if ART services are not offered on site, services are run by a single nurse, and if they are mobile clinics or clinics located within hospitals. All eligible clinics enrolled in the control arm of this trial will receive the prevailing standard-of-care for TB-HIV integration (Fig. 2) in accordance with the DoH guidelines. While the study team will not interfere with any additional benefit that control clinics may receive (for example, support from other NGOs), changes to the background standard of care will be recorded systematically and accounted for in the final analysis.

Fig. 2
figure 2

SUTHI study enrolment, randomization, follow-up, and outcome following TB-HIV integration intervention

Full size image

The quality improvement intervention

Quality improvement is described as a set of continuous actions through coordinated effort of HCW teams aimed at accelerating improvements to patient care and health outcomes [43]. This is done through iterative changes and peer-to-peer learning about successful changes and has been demonstrated to significantly improve HCW performance and lead to sustainable delivery of quality care with improved health outcomes [30, 43, 44]. The Institute for Healthcare Improvement (IHI) Breakthrough Series (BTS) is an example of quality improvement through learning collaborative, demonstrated to successfully impact health outcomes in developing-country settings [30, 45]. The learning collaborative brings together clinic teams at regular intervals to learn QI methods and exchange challenges and successes in efforts to implement the TB-HIV care pathway (learning sessions). Each clinic will form a QI team that will work with the PHC supervisor and QI mentors between these learning sessions to develop local ideas for implementing a specific area of care, regularly measure performance using agreed-upon indicators, and bring back those results and emerging best practices to other clinic teams through periodic learning sessions. We aim to use this structured implementation approach to improve reliable delivery of published evidence-based interventions known to decrease morbidity and mortality in patients with TB and HIV. The main measure of the collaborative’s success will be measurable improvement in the magnitude, maintenance, and speed of specific steps of TB-HIV integration, ascertained through time-series analyses. These analyses will be collected and acted upon in real-time and will include process and outcome indicators. The magnitude of TB-HIV integration implementation will be measured by the extent to which interventions provide measurable improvements in PHC process indicators from baseline (e.g., coverage of HCT and IPT services among eligible patients, improvements in quality of services, proportion of HIV-infected patients offered IPT following a negative TB screening). We will also track the speed at which each of the eight TB-HIV integration interventions are implemented (e.g., the number of months taken to reach 90% implementation of individual integration interventions) and factors affecting speed of intervention uptake in the clinics.

Systematic testing of change ideas will be evaluated through a rapid sequence of steps called the Plan-Do-Study-Act (PDSA) cycle. The PDSA cycle is a sequential framework for examining problems, deriving solutions, measuring progress, and embedding changes leading to positive outcomes [46, 47]. Selected HCWs from the study intervention clinics will be trained on QI methodology. This training will enable establishment of QI teams in intervention clinics to work test and implement change ideas to advance implementation of the TB-HIV integration interventions in rapid cycle (Fig. 1) and exchange successful ideas for change with other intervention sites through the learning collaborative. Furthermore, measures will be taken to prevent inadvertent cross-contamination of change ideas (e.g., by avoiding any convening of intervention and control clinics or personnel).

Training, coaching, and mentorship for HCWs implementing QI in clinics randomized to the intervention arm

The QI approach promotes front-line staff engagement in the identification of problems affecting performance and catalyzing rapid cycle testing of possible solutions in each of the eight identified TB-HIV integration indicators (Table 1). The QI team at each of the intervention clinics will include the PHC supervisor from the District Health Management team, the QI clinic champion, the clinic’s operation manager, selected clinic staff members, and a QI coach/mentor represented by a member of the study team. The QI champion, usually a clinic’s most senior nursing staff, will be trained to lead and support QI teams in their facilities using QI methods including the PDSA model, systems thinking, and the use of data for improvement.

The QI mentor and champion will be capacitated by the research team to provide peer-leadership and mentorship for implementation of QI methodology. In addition, they will also be capacitated to monitor the weekly performance of their clinics in achieving key successes on the indicators being targeted for improvement. The QI teams will be established through training of selected health workers to become fully fledged QI leaders and implementers within their facilities. Only HCWs experienced in both TB-HIV management and implementation of national TB and HIV guidelines would be eligible for QI training.

Implementation of the intervention

Three collaborative learning sessions and additional QI support (mentoring and coaching) visits at specific time points during the study period (Fig. 1) will be pre-arranged. Collaborative learning session one will coach teams in the use of a range of QI methods and tools (process mapping, fishbone system analysis, PDSA cycles, use of line graphs, and other data for improvement). Areas requiring improvement will be identified and prioritized, and aim statements encompassing specific aims and targets will be developed. Brainstorming by the QI team and process mapping will be done to define a strategy to effect positive change (“a change idea”). The PDSA cycles tracking a set of predefined indicators will be reviewed bi-weekly by the QI team to test if the recently adopted change idea resulted in performance improvement, throughout the study period. This will enable the generation of new ideas for improvement, iterative testing of these ideas, and monitoring of progress attained through use of run charts and graphs. Learning session two, scheduled 6 months later will review learnings for all successful change ideas and challenges against overall clinic performance, with the goal of scaling back on frequency of QI support meetings for clinics that are undertaking reliable improvement work. Learning session three, scheduled at month 12, will be focused on review and scalability of the final successful change package. It is anticipated that after learning session three, the clinic’s QI teams would sustain successful change ideas. Lessons learned will also be shared and documented across the intervention teams. At the successful conclusion of the projects, the intention is to scale up a reliable “change package” of successful, tested changes across all clinics in the two districts, if not across the entire district.

Standard of care in the control clinics: description of TB, HIV, and integration services

Since the 2009 endorsement of integrated TB-HIV services by the South African National AIDS Council (SANAC) [48], and adoption of TB-HIV service integration into policy and practice by the National Department of Health (NDoH) [29], South African guidelines recommend integration of TB and HIV services by a single service provider [8, 49].

Data collection

Data on TB-HIV integration indicators will be obtained from patient files and existing standard customized electronic data management systems supported by SA DoH [4, 41]. Routine clinical information is recorded in paper-based registers and patient clinic folders housed in clinics and subsequently captured onto the relevant TB or HIV electronic system supported by SA DoH. Baseline patient level data will be collected retrospectively for the 12 months prior to study start and will continue prospectively throughout the study period (Tables 2 and 3). In addition, HCW interviews will be conducted, using case report forms, at specific time points during study period from consenting health workers in both the intervention and control clinics. Table 2 summarizes the data to be collected, data source, and outcome measures while Table 3 presents the study’s schedule of activities.

Table 2 Data collection tools, sources, and outcome measures
Full size table
Table 3 Study schedule of activities
Full size table

Study outcome measures

The primary outcome of this study will be all-cause mortality rate among patients newly diagnosed with TB and/or HIV. Secondary outcomes will include time to ART initiation, retention in care, IPT and CPT uptake as per the current SA DoH guidelines, TB cure rates, viral load testing rates, and viral load suppression rates.

Sample size estimation

We anticipate that about 6000 HIV-positive patients will be diagnosed with active TB in the study population during a 12-month period. This is based on the assumption that between 100 and 200 new TB-HIV co-infected patients are seen in each of the 40 clinics per year. This translates to an average of 350 patients per cluster (assume unequal cluster sizes). Assuming a case fatality rate of 15% in the control arm, coefficient of variation between 0.25 and 0.35, and a type I error rate of 5%, we will have 80% power to detect a 30% reduction in mortality. The chances of detecting other levels of effectiveness are shown in Table 4.

Table 4 Power to detect different levels of effectiveness (keeping number of events constant)
Full size table

Data management

Anonymized TB, HIV, and TB-HIV co-infected patient data obtained from electronic software systems supported by the DoH and clinic-based registers will be collated for quality control and for evaluation of impact made by the intervention in data quality. Also, iDataFax version 2014.1.1 which is clinical data management software will be used for the design of case reporting forms for the HCW interviews and its data entry into the study’s database. Both intervention and control clinics will have data mentoring by the study data management team to ensure that quality of data obtained are improved, standardized, reliable, and valid. The study database files will be password-protected, and access to the files will be limited to authorized study staff members. Quality control measure will be carried out periodically throughout the study period prior to the data analysis.

Statistical analysis

The primary outcome of mortality and secondary outcomes will be analyzed using cluster-summary methods. The primary outcome will be analyzed among TB-HIV co-infected patients only while the secondary outcomes will be analyzed among TB-HIV co-infected, HIV-only, and TB-only patients. Mortality rate per arm will be calculated as geometric mean of cluster-level summaries and will be compared using unpaired t test. The same technique will be applied to secondary outcomes. The t test applied in cluster-level summaries is one of the robust methods of analyzing unmatched trials especially when there are small number of clusters per arm [50]. Since we have 16 clusters and 40 sub-clusters, multi-level regression will be used as secondary analyses. Proportional hazard regression with random effects (frailty models) will be used for analyses of time-to-event outcomes. Generalized estimating equations and mixed effects linear models will be used for binary and quantitative outcomes, respectively. These models will take the clustering by PHC supervisor and clinic into account through the random effects. They will also allow us to adjust for baseline variables, especially those with imbalance between arms, as this is likely when clusters are fewer. The HCW interviews will be summarized using descriptive statistics such as means and frequencies. Adjusted baseline descriptive statistics will be calculated as the means of the cluster-level summaries and characteristics of the two arms will be compared using t test or rank sum tests. Individual-level summaries at baseline will be compared using t test or rank sum tests and Fisher’s exact test. Data will be analyzed with SAS version 9.4 (or higher) (SAS Institute INC., Cary N.C, USA).

Discussion

Findings from this trial are expected to provide information on a scalable strategy (a “change package”) to address shortcomings in the implementation of integrated TB-HIV treatment and services. If successful, the strategy could make a contribution to reducing TB-HIV-associated mortality and morbidity in South Africa and other regions of the world where co-infection with TB and HIV is a concern. In 2013, WHO performed a joint review of HIV, TB, and prevention of mother to child transmission (PMTCT) programmes in South Africa, which recommended the need for context-specific mechanisms for the delivery of integrated TB-HIV services at PHC and community level, with particular focus on improving access to TB and HIV services for children, adolescents, and key populations. Some studies undertaken in the South African PHC clinic setting have shown that the QI strategy can be effective as an intervention for PMTCT care [51,52,53,54]. To date, there have been no studies that have explored the use of the QI model as an intervention to improve integration of TB and HIV treatment. An important component of the QI model is the PDSA cycle, comprising learning cycles to test and revise theory-based predictions as recommended by Taylor et al. (2014) in their systematic review on the application of the PDSA method to improve quality in healthcare [55]. We anticipate that the findings from this trial will offer an affordable and sustainable strategy through use of the QI model to effectively improve the integration of TB-HIV programmes. The study results will be communicated to stakeholders through dissemination meetings, conferences, and publication in peer-reviewed journals. Recommendations would be made based on the study findings for appropriate actions to be considered and taken by the department of health and relevant authorities in areas with high burden of TB and HIV.

Trial status

Data collection is currently on-going.