Integrated community-based management of HIV, diabetes, and hypertension in Tanzania and Uganda: protocol for a cluster-randomized trial

Patient and Public Involvement

Our group, RESPOND-AFRICA, has established a strong relationship with health services in Tanzania and Uganda through previous trials on integrated facility-based care. We have an active Patient and Public Involvement (PPI) group that engages with researchers throughout the entire study cycle. Together with the researchers, the PPI group conceptualized and defined the research questions and designed the INTECOMM study.

The PPI group consists of key community stakeholders, including policymakers, patients’ representatives, health service providers, such as health facility managers and front-line healthcare workers, NGOs providing care for HIV and NCDs, civil society organizations, such as the National NCDs Alliance, Network of People Living with HIV, the Tanzania Diabetic Association, and community advisory boards (CAB). Our aim is to leverage patient and public involvement to ensure that our research priorities, questions, and design are aligned with the insights, needs, and experiences of individuals living with chronic conditions.

In the INTECOMM trial, we seek to answer key questions about whether integrated care for HIV and NCDs could be decentralized to community settings as a response to key needs observed in facility settings. During stakeholder engagement meetings, issues such as clinic congestion, scarcity of human resources for health, and poor control of glycemia and blood pressure were discussed. As a result, stakeholders proposed and advocated for the conceptualization of decentralized care and discussed how it could be designed and evaluated. This trial is grounded in the practical realities of chronic disease management in low-resource settings with high disease burden in Africa.

The PPI group contributed significantly to the choice of primary and secondary outcome measures for this study. Their insights were instrumental in identifying key themes for exploration alongside traditional clinical measures, broadening the study’s scope to better evaluate the intervention. Patient representatives, healthcare workers, policymakers, and other stakeholders were involved in discussions about which patients would benefit most from community-integrated care (inclusion and exclusion criteria) and how they should be recruited into community care from their health facilities. The PPI group's input remains pivotal during the recruitment phase as they will receive updates and provide feedback during CAB and steering committee meetings.

Throughout the study, an iterative dialogue will be maintained with the PPI group, incorporating their feedback into the ongoing research cycle. Findings from the study will be shared with this group, and they will provide guidance on how best to communicate the main messages from the study to participants and the community. Through ongoing workshops and meetings, they will continually guide on contextually appropriate avenues for public dissemination.

Development of the integrated community care model

The integrated community model was designed using lessons learnt from HIV community care. Healthcare systems, patient diagnosis and management tools, and implementation strategies developed to provide continuity of care for HIV in SSA can efficiently and effectively support patients with other chronic conditions such as diabetes and hypertension¹¹.

To develop a community care model to be tested in this trial, a scoping review was conducted to identify existing or potential interventions for managing HIV, diabetes, and hypertension through integrated care (Supplemental material 1)¹⁷. The scoping review was based on a framework developed by Arksey & O’Malley and refined by Levac et al., which was further validated through a broad interprofessional team experience^18–20. The framework consisted of five steps that included; identifying the research question, identifying relevant studies, selecting the studies, charting the data, and collating, summarising, and reporting the results.

Based on the findings of the scoping review, we categorised community-based care into six general design groups. These included improving access to medications (specifically ART), providing treatment support programmes, multifaceted support programmes, home visits from community health workers, support through SMS text messages, and differentiated care. Research comparing patient outcomes between community and alternative care found that outcomes improved or at least no worse than the standard of care, although the study design and the quality of implementation varied between different studies. This body of research suggests that a suite of interventions may be more effective than a single measure.

From a range of government policies from Tanzania & Uganda, we found four community-based interventions recommended for the treatment of patients with HIV. These interventions are accompanied by clear and detailed methods for their implementation in practice. Several tentative models of integrated community care were developed using the HIV community care models as a template. These were presented to policy makers, healthcare providers, and patient representatives in Uganda and Tanzania.

In these workshops, stakeholders, together with researchers, addressed important questions such as; "What are the challenges associated with the different proposed models of community integrated care and how can they be mitigated from the design stage?", "Where in the community should services be provided?", "Which patients should receive care in the community?", "How often should routine blood pressure and glycemia monitoring be performed?", "How often should patients visit the facility to see a doctor?" and "What kind of community model would be attractive to policymakers and what type of data would they need for policy considerations?". Through these deliberations, a model of integrated community care was developed for HIV, diabetes, and hypertension, which will be tested in this trial.

Study design and setting

Community care is an innovative approach that will include major changes to the way health care is currently organized. For example, it will involve medicines being dispensed by non-pharmacists in a nonclinical setting, as well as monitoring the control of chronic conditions in the community. It is crucial that this approach is rigorously evaluated in a trial to inform policy in Tanzania and Uganda, as well as to facilitate generalization to other countries in Africa. Due to the need for rigorous evidence, we planned a pragmatic cluster randomized trial comparing community-based integrated care for HIV, diabetes, and hypertension with integrated services offered at health facilities in Tanzania and Uganda.

The trial will involve the recruitment of residents of the catchment areas of 10-16 health facilities in Tanzania and Uganda, which are mainly urban and peri-urban. In each facility, patients who meet the inclusion criteria will be formed into groups of approximately 8–14 people based on their address and their disease/condition. Grouping of study participants will be done collaboratively between the research team, the clinical team of the health facility, and lay workers who know the area.

Patients with HIV, diabetes, and/or hypertension who live close together in the same ward or parish and have met the eligibility criteria (Table 1) and provided informed consent will be placed in a trial group. Each group will have a 2:1 ratio of participants living with diabetes and/or hypertension to those with HIV. Patients will be recruited consecutively as they come in for their scheduled clinic visit and by reviewing patient records at the facility. A group is finalized when the target number and ratios are met. The group will then be randomized to either of the study arms (Figure 1 – The trial schema).

Figure 1. INTECOMM study schema.

Abbreviation: MOCCA – Management of Chronic Conditions in Africa, INTE-AFRICA – Integrated management of HIV, diabetes and hypertension in sub-Saharan Africa.

Randomization will be computer-generated. To ensure allocation concealment, the randomization procedures will be built into the study’s electronic database. The group arm allocation will be visible only after randomisation. It will be stratified by health facility, based on its infrastructure. This allows the results of the findings to be generalizable. Equipoise is robust, as we are truly uncertain which of the trial arms is most likely to benefit the patient²¹. This trial protocol is reported using the SPIRIT Checklist for reporting protocols of randomised controlled trials²².

The intervention

The Integrated Community care (intervention) arm. Groups of 8–14 patients randomized to community-based integrated care will hold their first group meeting at the health facility. In this meeting, a nurse and a trained lay worker (community health worker in Tanzania and village health worker in Uganda) will lead the group in choosing a group leader and suggest venues in the community where they would like to receive care. The nurse, the trained lay worker, and the group leader will then visit these venues and agree on the most suitable in terms of accessibility and comfort for a group meeting. They will then map the community stakeholders around the proposed venue and hold an engagement meeting to inform them about the study prior to the first meeting at the community venue.

The nurse and a trained lay worker will prepare patient files one day before the community group meeting, and a health facility clinician will write prescriptions for patients at community sites. The facility pharmacist will then dispense these medications ready for pickup by a nurse.

Group meetings will occur monthly. Patients will receive health education, monitoring of their blood pressure and blood sugar, and the nurse and lay worker will deliver drugs in the community. The nurse will also conduct a clinical review to assess the progress of the patients and refer those who have developed danger signs to the facility for further management by a physician. Adherence and behavioural information and support will be provided by a trained lay worker supervised by the nurse.

The baseline and end of the study (12^th month) visit will occur at the health facility and monthly follow-up meetings will occur in the community. The trained lay worker will identify and trace patients who miss their community visit. They will be instructed to visit the health facility to pick their drugs and be informed of their next community meeting.

The integrated facility care (control) arm. Participants randomised to the control arm will receive integrated care in health facilities led by a clinician. They are effectively in artificial groups of 8–14 patients to allow for comparison.This study will be conducted in facilities in Tanzania and Uganda that were involved in previous studies on integrated facility-based care. In Uganda, some facilities continued the integrated clinics after the trial ended. However, facilities in Tanzania did not continue providing integrated care, as the decision to do so required evidence of effectiveness for implementation. Therefore, integrated clinics will be re-established in these facilities for the current trial.

Participants living with HIV, diabetes, and/or hypertension will receive care from a single clinic. They will have one registration and waiting area and will be managed by the same clinicians, nurses, counsellors, and other staff. There is one pharmacy where drug dispensing is integrated. Patient records will be integrated, and laboratory samples will be managed and tested in the same laboratory where possible.

Patients will receive health education and clinical monitoring, collect their drugs from the health facility pharmacy, and receive adherence and behavioural information and support as standard provided by the health facility to all patients.

The standard of care for participants who declined to join the study or withdrew. The current standard of care for HIV, diabetes, and hypertension varies depending on the level of the health facility. Although patients with diabetes or hypertension are typically seen in outpatient clinics in smaller hospitals, larger facilities have separate specialised clinics for these conditions.

HIV care features separate waiting areas, consultation rooms, and a dedicated pharmacy. Medical records and laboratory samples are managed separately from those of diabetes and hypertension. Patients with HIV have different schedules depending on their level of stability. They usually attend care three or six monthly for routine appointments when stable, while patients with diabetes and hypertension attend monthly or even more frequently. Patients with multimorbid chronic conditions must attend different clinics on different days since the clinics are run separately.

Trial setting

This trial will be run in close to real-life conditions, and participants will be managed by government staff in public facilities and routine staff in private or non-profit facilities. In both arms of the trial, we will train clinical staff on how to manage HIV, diabetes, and hypertension. They will also receive training on the trial protocol and standard operating procedures. The staff members will then undergo on-the-job training regularly. Clinical staff experts in one disease will support staff who are learning to manage the other two conditions in integrated care clinics. For example, doctors who have historically treated patients with diabetes and hypertension would periodically observe those in HIV clinics and provide constructive feedback and support.

Patients randomised to the control arm will receive care in the integrated clinic of the health facility and will be seen by routine staff at the facility. Those receiving care in the community will also have their community clinical review notes and prescriptions recorded in their hospital files by facility staff, while the facility pharmacist will dispense and package the drugs to be delivered to the community. This approach is economical, allows facility personnel to learn transferable clinical skills, and is sustainable beyond the life of the trial if the facility decides to carry out this mode of service delivery, as happened in Uganda²³.

Primary endpoints

The trial has two coprimary endpoints, measured in 2 different populations:

Previous studies of integration of services at the facility level showed that integration of services did not adversely affect gains from vertical HIV programming, as viral suppression and retention in care were comparable in both arms. With the decentralization of integrated care to the community, it is important to evaluate how integrated community care will affect HIV viral suppression in the community compared to standard integrated care in the health facility. A plasma viral load of less than 1000 copies/ml is the WHO marker for viral suppression adopted by the country guidelines in Tanzania and Uganda²⁴.

Control of blood pressure and blood glucose is poor even when retention is high⁶. Good control of chronic conditions is important at the patient and health service level, as this will prevent further progression of the disease and the development of complications. The management of complications is costly, reduces quality of life, and leads to poor outcomes for patients¹¹.

Community care offers a more personalized approach to patient management by taking patients out of clinical settings and placing them in smaller groups. Although some may perceive it as an inferior service, this approach allows health care staff to provide better services without the time pressure caused by the high volume of patients in the health facility. It is crucial to assess the impact of this change on the management of hypertension and diabetes, specifically the control of blood pressure and blood sugar, in patients with these conditions.

Secondary endpoints will include the equivalence of glycemia and blood pressure control, retention in care, costs, and cost-effectiveness.

Retention in care is fundamental to disease control. In our previous studies on integrated care in health facilities, patients maintained high retention. With the scepticism and controversy surrounding community care, it is important to assess whether participants will remain in care and how this might differ between participants with different conditions.

Economic evaluation

A within-trial cost-effectiveness analysis guided by the Gates reference case on cost-effectiveness will be conducted²⁵. Cost analyses will estimate the total societal cost of delivery, including provider and patient or household costs. Provider cost data will be collected prospectively at both sites through project accounts, a facility-based cost capture tool, project records, and interviews with project staff. Data on patient costs will be collected using an endline questionnaire that asks about medication and other medical costs, travel costs, and time spent travelling and receiving care for a given visit. Any additional patient care-seeking in the month before the visit will also be captured. Costs will be captured in Ugandan and Tanzanian shillings. All costs will be adjusted for inflation using the Consumer Price Index (CPI) for Uganda and Tanzania and will be converted to 2023 International Dollars (INT$) using the 2023 Purchasing Power Parity (PPP) conversion factor for both countries²⁶.

Results will be presented in terms of incremental cost-effectiveness ratios (ICERs), calculated as the arithmetic mean difference in cost divided by the arithmetic mean difference in effect between the community and facility arms. ICERs will be calculated for the co-primary outcome measures, i.e., plasma viral load suppression and a composite of glycemia and blood pressure control. ICERs will also be calculated based on the self-reported Luganda and Swahili versions of the EQ-5D-3L²⁷, to generate an incremental cost per quality-adjusted life year gained. This will improve the comparability of the results with other studies and health interventions²⁵. Sensitivity analyses will be conducted to assess the impact on cost-effectiveness results, of changes in parameters with the greatest uncertainty, or with the greatest impact on the total costs. Costs and outcomes will be converted to present values using an annual discount rate of 3% in the base-case, and annual rates of 0% and 6% in sensitivity analyses²⁵. Reporting of economic evaluation results will be guided by the Consolidated Health Economic Evaluation Reporting Standards (CHEERS) 2022²⁸. A separate economic evaluation protocol will describe planned analyses in more detail.

Process evaluation

A process evaluation will be conducted alongside the intervention²⁹. The aim of the process evaluation of this community-based integrated care trial is to explore implementation from diverse perspectives (patients, community health workers and volunteers, healthcare providers, policymakers, NGOs, community members and leaders) to gain insight into the influence of socio-structural and contextual factors on service integration at the community level.

Sample size consideration

Glycaemia and blood pressure control. We hypothesize that community care will be superior to facility care because health services and support will be more accessible as it is provided close to the patient’s home and more personalised due to the small size of the groups run by a nurse alongside a community/village health worker and peer leaders. In an earlier pilot study on integrated care (MOCCA study)⁶, which was conducted in 10 facilities with more than 2,000 participants followed for 6–12 months, 54% of hypertension participants had good control of blood pressure (BP<140/90 mmHg) and 39% of diabetes had achieved fasting glucose <7 mmol/L⁶.

We assume that 50% of participants in the control arm with diabetes or hypertension will achieve a good level of control after 12 months, that is, blood pressure <140/90 mmHg and fasting plasma glucose <7 mmol/L. This is plausible since we will be recruiting stable patients who are in regular care and should therefore start with a better level of control, and patients in both arms will be monitored and supported as per guidelines.

The calculation assumes 80% power and a 5% two-sided significance level. Table 2 shows the number of participants needed to compare integrated community care with integrated facility-based care for the control of blood pressure and/or glycemia.

Therefore, with 116 groups of 8 people each with diabetes or hypertension, the trial will have 80% power to detect a difference in the absolute rise of 10% (i.e., 50% versus 60% achieving good control in the 2 arms would be statistically significant at the 5% two-sided significance level if the intra-class coefficient is 0.02 or lower) (Table 2). In the MOCCA study (a pilot study on integrated facility based care in Tanzania and Uganda), the risk differences between baseline and the end of the study among participants with hypertension were 10.5% and among those with diabetes 7.4%, after excluding those in care for less than 6 months. Therefore, we will use an absolute increase of 10%, as power will be very high for differences larger than this.

HIV viral load suppression. For the HIV viral suppression endpoint, the MOCCA study showed viral suppression close to 90%, the UNAIDS target, in this population⁶. We hypothesize that patients in the community arm will be able to achieve and retain viral suppression similar to those attending care at the facility-integrated clinic. This is because they will have the same access to medication, monitoring, adherence support, and counselling. Thus, the primary aim is to show non-inferiority in the community care arm.

The non-inferiority margin in trials is often set at 10% (i.e., that the upper one-sided 95% CI of the difference between the control and intervention arm in terms of viral suppression will be within 10%). If we form 116 groups in total, each with 4 HIV participants, then the trial will have more than 80% power to show non-inferiority at delta= 8.5%, 7.5%, and 5.5% assuming viral suppression is 85%, 90% and 95%, respectively (assuming an intraclass coefficient of 0.02).

Therefore, we need a sample size of 116 groups, each comprising 12 persons, of which 8 should have diabetes or hypertension and 4 should have HIV. This equates to 1,392 evaluable participants in total. We propose enrolling 124 groups to allow for just over a 5% loss to follow-up in the number of groups. In each of these 124 groups, we will enrol 14 persons to allow for a loss to follow-up of just over 10% in the number of participants. Thus, our target sample size is 1,736 participants. Approximately half of the enrolment will be in Tanzania and half in Uganda. Thus, each country will enrol a total of 62 groups and 868 participants.

Data management

All members of the study team will be trained in good clinical practices (GCP), the study protocol, and standard operating procedures. Study group meetings will be conducted regularly to review the study's progress, address difficulties, and provide feedback on the implementation of the study. The trial will also be regularly monitored following the study monitoring plan. An electronic database custom-designed for the trial will be used that has a built-in data type and logic checks for real-time electronic data validation. Any data modifications are tracked on a comprehensive electronic audit trail and any changes to the source code are tracked and versioned. Data can be created, viewed, modified, deleted, or exported by delegated personnel according to their access roles associated with their personal accounts. Data security is ensured by using authentication and encryption to make subject identity and personal health information unreadable and undecipherable to unauthorised individuals. The study data are not stored on any device in the field.

An independent Data Safety Monitoring Board (DSMB) will be responsible for evaluating outcomes and receive and review reports of serious adverse events.

Statistical analysis

The statistical analysis will employ Generalised Estimating Equations to account for clustering based on the intention-to-treat principle. The primary analysis will compare the proportions of patients who achieve viral suppression, blood pressure, and glycaemic control.

Cost data analysis will calculate the costs of healthcare delivery and the societal costs incurred by patients to access healthcare. These data will then be linked to effectiveness estimates to determine potential cost-effectiveness.

Qualitative data analysis will be conducted thematically using NVivo, an electronic data management package. This method will facilitate the identification of key themes and patterns in the data.

Ethics and dissemination

This study will be conducted in accordance with the Revised Declaration of Helsinki (Ethical Principles for Medical Research Involving Human Subjects) and with the laws and regulations of Tanzania and Uganda, as well as the Good Clinical Practices Guidelines.

The study protocol, information sheets and consent forms for patients were submitted for approval to the local National Ethics Committees for Health Research, and to the sponsor institution’s Research Ethics Committee at UCL.

Written approval from all Ethics Committees has been obtained; the National Institute of Medical Research in Tanzania (NIMR/HQ/R.8a/Vol. IX/3977 (28^th April 2022)), the Uganda Virus Research Institute(GC/127/872 (23^rd March 2022)), the Uganda National Council for Science and Technology (HS2278TS (13^th July 2022)), the London School of Hygiene and Tropical Medicine Ethics Committee (28122 (31^st August 2022)) and the University College of London (2382/001 (25^th October 2022)), which is also the trial sponsor.

All research staff have undertaken Good Clinical Practice Training, and tailored, study-specific research training addressing ethical concerns and considerations for engaging participants in research study in each setting. As required, any changes to the protocol will be submitted to the approving ethics committee as an amendment.

Before data collection, the delegated research staff will read and explain the information sheet to the participants in Swahili or Luganda and answer any questions that may arise. The participant will be informed about the purpose of the study, the study procedures, and the risks and benefits of participation. The information sheet will also emphasise that participation in the study is entirely voluntary and that participants are free to withdraw at any time without compromising their standard care received at the study site.

If the individual agrees to participate, the participant and the research staff will write their full name, and date, and sign the consent form. In case of illiteracy, oral consent in the presence of a witness who is not part of the health personnel at the trial site is acceptable. In this case, the signature of the witness will be collected. The information sheet and a copy of the signed informed consent form will be given to the participant. The dates of the information and the signature of the consent form will be recorded in the electronic CRF database with the name of the research staff who obtained the informed consent.

The findings of the study will be disseminated through stakeholder meetings, local and international conferences, policy briefs, peer-reviewed journal articles, and publications.