Developing a Learning Health System for Primary Care in Thailand

Aims and objectives The overall aim of this study is to conduct a stratified cluster randomized controlled trial to determine whether a Learning Health System approach improves delivery of care and clinical outcomes for people with hypertension and related non-communicable diseases compared to routine care in primary care settings in Thailand. The trial will have three phases. In phase 1, the investigators will trial an intervention for the management of people with hypertension. In phase 2, the investigators will target the management of people with Type 2 Diabetes. In phase 3, the investigators will target the management of people with CKD. The key objectives for each phase of the intervention will be to: 1. Engage with healthcare providers, decision makers, researchers, and the public to co-design a LHS for the management of hypertension, diabetes, and CKD, and develop computerised data systems, decision support tools, and training materials in 16 randomly selected primary care units across two provinces in Thailand. 2. Train multidisciplinary healthcare teams, data scientists, and researchers in the design, delivery, analysis, and evaluation of a LHS and establish a cohort of clinical and community champions to support the intervention. 3. Pilot the delivery of a LHS for improving management and clinical outcomes of hypertension and related NCDs and generate robust new evidence around the acceptability and effectiveness of LHS approaches and delivery of clinical care. 4. Compare outcomes between intervention and control practices and conduct a health economic evaluation of the costs of implementing the LHS versus routine care. 5. Conduct a mixed-methods evaluations to identify provider and patient barriers and facilitators to implementing a LHS to inform subsequent approaches to embedding quality improvement for NCD care in Thai primary care settings and more widely across the region. The investigators' hypotheses are that: 1. Practices implementing the LHS approach will experience improved delivery of care and outcomes for people with hypertension, type 2 diabetes, and CKD vs. routine care 2. Training a multidisciplinary workforce and developing leadership capacity will embed continuous quality improvement approaches into routine clinical practice beyond the lifespan of this project Research plan / methods Study Design: The investigators will conduct a three-phase open-label stratified cluster randomized controlled trial across two provinces in Thailand. Stratified randomisation will be used to allocate 16 primary care units from 419 total eligible primary care units to receive the intervention (273 units in Chiang Mai and 146 units in Lampang). The investigators will firstly stratify all primary care units in the study area according to patient population size (large vs. small practices) and patient case-mix (i.e. proportion of patients with any of hypertension, diabetes, and CKD), to ensure that primary care units within each stratum are balanced according to key characteristics. This will generate four strata (the maximum number recommended for stratified cluster randomized trials of this size29). Four primary care units in each stratum will be randomly selected to receive the intervention, and the rest will receive routine care (approximately 100 primary care units per stratum) The intervention will be an open-label trial as those receiving the intervention will be aware of their allocation. Phased approach of the intervention: The LHS intervention will have three distinct phases. In phase 1, the investigators will design and implement a LHS for improving management of people with hypertension. In phase 2, the investigators will design and implement a LHS for improving management of people with type 2 diabetes. In phase 3, the investigators will design and implement a LHS for improving the management of people with CKD. The implementation of each phase will last 12 months. Workstream 1: Co-designing the Learning Health System Intervention LHSs are complex and must be co-designed with local stakeholders. In addition to the project team, stakeholders will include: the healthcare teams (physicians, nurses, pharmacists, volunteer health workers); patients and members of the public; the Thai Ministry of Public Health (at national, provincial, and district level); the Thai World Health Organization; and the Thai Primary Care Support Office. Together the stakeholders will form a learning community who will drive the work of the LHS. The learning community will come together three times to co-design the LHS for each of the three phases of the intervention. In collaboration with the stakeholder learning community, the investigators will co-create computerised tools to support continuous monitoring, evidence generation, and quality improvement informed by electronic health record data. Tool development and data analysis will be led by the data science team at CMU. Tools will be tested and iterated upon by the learning community and healthcare teams throughout the life of the project. In order to build capacity in embedding a culture of quality improvement into routine clinical care and data analytics, stakeholders will co-develop training modules to train healthcare teams on the design and implementation of LHSs. Workstream 2: Training the multidisciplinary healthcare and research workforce and establishing clinical and community champions Sourcing from each of the intervention practices, the investigators will train a network of clinical and community champions who will support the intervention practices in delivering the LHS. In partnership with the clinical and community champions, the investigatorsat CMU will deliver training to multidisciplinary healthcare and research teams in how to use quality improvement approaches for managing hypertension and related NCDs, use of computerised tools for routine monitoring, and translation of evidence into clinical action. Workstream 3: Trialling implementation of the Learning Health System Healthcare teams will implement the LHS approach agreed during the stakeholder consultation. They will be encouraged to meet regularly to discuss barriers and facilitators to implementing the LHS approach, to freely discuss problems they are experiencing, and to work collaboratively to diagnose and solve problems. Healthcare teams will receive monthly data on their performance and patient outcome data from the CMU team. The data produced will be fed-back to practices via regular network-based peer discussions in a visually accessible format and facilitated by the clinical and community champions. Here, the practices will come together to review the data and share experiences of activities undertaken to improve care. Primary care units not randomized to the LHS intervention will deliver routine care for hypertension, type 2 diabetes, and CKD. While we will collect monthly data on practice performance and key outcome metrics, they will not receive feedback or facilitation. Workstream 4: Conducting quantitative comparisons between intervention and control primary care units and Health Economic Evaluation All primary care units in the two participating Thai provinces will be included in the primary analysis according to their group assignment and regardless of their degree of participation in the study (intention-to-treat analysis). The investigators will first conduct baseline comparisons of the intervention and control primary care units. While outcome data will be collected monthly, the investigators will only conduct the formal evaluation for each phase of the intervention after the entire 12-month implementation period has elapsed. For phase 1, the investigators will test the difference in the proportions of patients with hypertension meeting BP control targets between the intervention and control groups using a two-level generalized linear mixed-effects model which will account for the clustering of individuals within primary care units. We will repeat this approach for phase 2 (difference in the proportions of patients with type 2 diabetes meeting HbA1c control targets) and phase 3 (difference in the proportions of patients with CKD1-4 tested for albuminuria). Health Economic Evaluation: We will develop an evaluation framework informed by Thai population data to support the evaluation of medium and long-term effects of the interventions implemented in the LHS. The project will make use of several unique large administrative databases including individual participant health data from Thai primary care and hospital systems. There will be three key patient populations of interest: people with hypertension, people with diabetes, and people with CKD. Therefore, the investigators will develop an interactive Markov type decision-analytic model management of hypertension, diabetes and CKD using practice level factors, patient socio-demographic factors, clinical characteristics (e.g., systolic, and diastolic blood pressure, HbA1c, eGFR) and other risk factors (comorbidities). The investigators will report reductions in disease risks, quality-adjusted life years gained and additional cost per quality-adjusted life year gained with interventions over medium term (5-10 years) and over lifetime. Workstream 5: Conducting Realist Evaluation of the Learning Health System The investigators will undertake a realist evaluation of the programme alongside the trial to understand what works in promoting change, for whom, in what circumstances, how and why. The evaluation will include both quantitative and qualitative data collection and analyses to generate an overarching programme theory and identify the underlying causal mechanisms. The investigators will start by generating an initial programme theory through an online interactive workshop drawing on learning and evidence gained from Workstream 1 and 2 activities. All research members and key stakeholders from the UK and Thailand will join the workshop. This initial programme theory will hypothesise how learning health systems in specific contexts (C), will trigger mechanisms (M) to generate outcomes (O). The investigators will collate the monthly quantitative data from the 16 intervention sites and all control sites within the same strata gathered by CMU throughout the intervention. Qualitative data collection will include gathering relevant documents for analysis and conducting focus groups. The investigators will collate local documentation, such as board reports and minutes of meetings to understand the written outputs from QI activities. The investigators will also collect qualitative data in a series of focus groups. The investigators will undertake a document analysis of minutes from meetings and QI project reports. This process will lead to the development of an overall programme theory exploring if learning health systems work, in what circumstances, how and why.