Integrated community-based versus facility-based care for people with HIV,

# Integrated community-based versus facility-based care for people with HIV, diabetes, and hypertension in sub-Saharan Africa (INTE-COMM): an open-label, multicountry, cluster-randomised trial *The Lancet 2026* Articles Integrated community-based versus facility-based care for people with HIV, diabetes, and hypertension in sub-Saharan Africa (INTE-COMM): an open-label, multicountry, cluster-randomised trial Francis X Kasujja*, Faith Aikaeli*, Anupam Garrib*, Erik van Widenfelt, Ivan Namakoola, Sokoine Kivuyo, James A Prior, Josephine Birungi, Faith Moyo, Duolao Wang, Stavia Turyahabwe, Gerald Mutungi, Mina Nakawuka Ssali, Omary Said Ubuguyu, Stephen Watiti, Said Aboud, Marie Claire Van Hout, Geoff Gill, Nelson K Sewankambo, Peter G Smith, Sayoki Mfinanga, Kaushik Ramaiya†, Moffat J Nyirenda†, Shabbar Jaffar†, on behalf of the RESPOND-AFRICA Group‡ Summary Lancet 2026; 407: 1084–94 Background In sub-Saharan Africa, the burden of diabetes and hypertension is high, alongside a high prevalence of See Comment page 1033 HIV. Whether these conditions can be managed in an integrated way in the community is unknown. We aim to compare integrated community-based care with integrated facility-based care for people with HIV, diabetes, and *Joint first authors hypertension in Tanzania and Uganda. †Contributed equally ‡Members listed at the end of the Article Methods This open-label, multicountry, cluster-randomised trial was conducted in 14 primary care facilities across Medical Research Council/ Tanzania and Uganda. Adults aged 18 years or older with a diagnosis of HIV, type 2 diabetes, or hypertension (or a Uganda Virus Research combination); receiving regular care at the health facility for at least 6 months; considered clinically stable; living Institute, London School of within the catchment area and planning to stay for at least 6 months; and willing to receive care in the community Hygiene & Tropical Medicine, were enrolled. In each facility, patients were grouped into clusters of 8–14. Each group was randomly assigned (1:1) Uganda Research Unit, Entebbe, Uganda using an online data management system, to integrated facility care or community care. In facility care, participants (F X Kasujja PhD, shared the same registration and waiting areas, were managed by the same physicians and health-care workers, and I Namakoola MSc, J Birungi MSc, used the same pharmacy and laboratory services. In community care, a nurse and a trained lay worker supported the Prof M J Nyirenda PhD); Global groups at focal points in the community with groups meeting once per month. Follow-up was 12 months. The first Health and Migration Unit, Department of Women’s and coprimary endpoint was a composite of blood pressure or fasting glucose control (defined as blood pressure Children’s Health, International <140/90 mm Hg in participants with hypertension alone, fasting glucose <7·0 mmol/L in those with diabetes alone, Maternal, and Child Health, or both indicators controlled in those with both conditions) and the second was plasma viral load suppression for Uppsala University, Uppsala, participants with HIV alone (defined as <1000 copies per mL or undetectable viral load). Both endpoints were assessed Sweden (F X Kasujja); National Institutes for Medical Research, in the intention-to-treat population. Generalised estimating equation models accounted for clustering. This trial was Dar es Salaam, Tanzania registered with the ISRCTN registry, ISRCTN15319595 (completed). (F Aikaeli MSc, S Kivuyo MPhil, Prof S Aboud PhD, Findings Between Jan 30 and Oct 6, 2023, 2940 patients with HIV, diabetes, or hypertension (or a combination of Prof S Mfinanga PhD); Department of Public Health, these conditions) who lived close enough together to be placed into a group were identified as having appointments Kilimanjaro Christian Medical to attend at the participating facilities. 765 (26·0%) patients were not screened and 2175 (74·0%) were screened for University, Kilimanjaro, eligibility. 203 (9·3%) patients were ineligible, four (0·2%) did not consent, and 104 (4·8%) could not be grouped Tanzania (F Aikaeli); Institute into viable clusters. 1864 (63·4%) patients were assigned into 124 groups, and groups were randomised (62 to for Global Health, University College London, London, UK community care and 62 to facility care). There were more females than males (1302 [76·6%] of 1700 vs 398 [23·4%]). (A Garrib MSc, Among those with diabetes or hypertension (or both), 38 (6·3%) of 602 in the community care group versus 43 (7·1%) E van Widenfelt BSc, of 609 in the facility care group were excluded, with nine (3·7%) of 242 versus ten (4·0%) of 247 excluded among Prof S Mfinanga, participants with HIV. The composite of blood pressure or fasting glucose control did not significantly differ Prof S Jaffar PhD); Department of Clinical Sciences, Liverpool between the two groups in participants with hypertension or diabetes (or both; 317 [55·2%] of 574 in the community School of Tropical Medicine, care group vs 304 [53·2%] of 571 in the facility care group; adjusted risk difference 1·80 [95% CI –4·52 to 8·12]; Liverpool, UK (A Garrib, p=0·58), whereas most participants with HIV alone reached viral suppression (227 [99·1%] of 229 Prof D Wang PhD, vs 229 (98·7%) of 232; adjusted risk difference 0·44 [–1·12 to 1·99]; p <0·0001). There were seven deaths in Prof G Gill PhD); Barcelona non-inferiority Institute for Global Health each study group. Hospital Clinic, University of Barcelona, Barcelona, Spain Interpretation In sub-Saharan Africa, integrated community care could reach a high standard of care for people with (S Kivuyo); School of Medicine, Keele University, Keele, UK diabetes or hypertension without adversely affecting outcomes for people with HIV. (J A Prior PhD); Midlands Partnership University NHS Funding National Institute for Health and Care Research. Foundation Trust, Stafford, UK (J A Prior); AIDS Support Organisation, Mulago Hospital Copyright © 2026 The Author(s). Published by Elsevier Ltd. This is an Open Access article under the CC BY 4.0 Complex, Kampala, Uganda license. 1084 Articles (J Birungi); School of Research in context Psychology and Public Health, Evidence before this study designs and quality of implementation were varied. We La Trobe University, We previously conducted a cluster-randomised trial comparing searched PubMed using the terms (“HIV” OR “diabetes” OR Melbourne, VIC, Australia (J Birungi); Bradford Institute integrated facility care with standard vertical facility care “hypertension”) AND “community care” AND “systematic for Health Research, Bradford (ie, organised separately) for patients with HIV, diabetes, and review” OR “trial” AND “Africa”. We found 12 studies showing Teaching Hospitals NHS hypertension in Tanzania and Uganda. We showed that that care provided by non-clinical community health-care Foundation Trust, Bradford, UK integrated facility care was as effective as standard facility care workers improved rates of HIV viral suppression, increased (F Moyo PhD); Communicable Disease Prevention and in terms of clinical indicators (HIV viral suppression, blood engagement with HIV health-care delivery, and contributed Control, Ministry of Health, pressure, and fasting blood glucose control among patients effectively to improved HIV care delivery compared with facility Kampala, Uganda living with HIV, hypertension, and diabetes, respectively), and care. In 2023, another scoping review showed that there were (S Turyahabwe MPH); Non- retention in care and was a cost-saving approach for patients insufficient data on diabetes models of care in sub-Saharan Communicable Diseases Control Programme, Ministry and health services by reducing service duplication. This finding Africa. There was more evidence on community care for of Health, Kampala, Uganda has led to changes in policies for health care of chronic patients with hypertension, although this evidence was (G Mutungi MPH); AIDS Control conditions in Uganda with changes also expected in Tanzania in considered weak in a 2024 systematic review. There were no Programme, Ministry of 2026. The prevalence of diabetes is around 5% and studies or reviews assessing integrated community care. Health, Kampala, Uganda (M Nakawuka Ssali MPH); hypertension is around 25% in most urban and peri-urban Added value of this study Non-Communicable Disease settings in sub-Saharan Africa. Few patients with these Control Programme, Ministry This study is the first to assess integrated community care for conditions are in care and for those who are receiving care, their of Health, Dodoma, Tanzania blood pressure and glucose control are both inadequate. patients with multiple chronic conditions (specifically HIV, (O Said Ubuguyu MMed); diabetes, and hypertension) and the largest trial within National Forum of People Strategies are needed for increasing service reach and community care for any health condition. The study provides Living with HIV Networks in improving accessibility for people with chronic conditions who Uganda, Kampala, Uganda rigorous evidence of both community and integrated care. require repeated care. With this issue in mind, policy makers (S Watiti); South East asked the research team for the evaluation of integrated Implications of all the available evidence Technological University, Waterford, Ireland community care for patients with HIV, diabetes, and The study findings can be used as robust evidence for policy (Prof M C Van Hout PhD); hypertension including services delivered at community points, makers to consider scale-up of integrated care for patients with Makerere University, College of such as places of worship and schools. Before starting this study HIV, diabetes, and hypertension at the community level in sub- Health Sciences, Kampala, Uganda we conducted a scoping review, but could not identify any Saharan Africa. As the burden of non-communicable diseases (Prof N K Sewankambo MMed); studies assessing the management of multiple chronic rises, integrated community management is likely to be a International Statistics and conditions using integrated community care. For managing crucial and cost-effective approach that improves access and Epidemiology Group, London single conditions, research found that patient outcomes, such helps to decongest health facilities, particularly for people with School of Hygiene & Tropical Medicine, London, UK as HIV viral suppression rates, did not differ significantly common multiple chronic conditions. (Prof P G Smith DSc); Tanzania between community and facility care, although the study NCD Alliance, Dar es Salaam, Tanzania (Prof K Ramaiya MMed); Research and Training Unit, Introduction programmes. These included adherence counselling, Shree Hindu Mandal Hospital, In sub-Saharan Africa, around 2 million premature consistent medical records, patient tracing systems, and Dar es Salaam, Tanzania deaths each year are attributed to diabetes or improved medicine supply chains being extended to (Prof K Ramaiya) hypertension, or both, and fewer than 20% of individuals diabetes and hypertension care. Correspondence to: Prof Shabbar Jaffar, Institute for with these conditions are estimated to be in regular care.1 This model was evaluated in a large single-arm cohort Global Health, University College Most people with HIV are receiving treatment and reach study4 and then in a cluster-randomised trial.5 We showed London, London WC1E 6BT, UK effective viral control with a life expectancy similar to that integrated management was associated with high s.jaffar@ucl.ac.uk those without HIV;2 however, health-care services for retention in care for people with diabetes or hypertension HIV operate in parallel to other chronic disease services, (or both), without compromising outcomes for those with separate systems for financing, procurement, with HIV. Integrated care was also cost saving for patient support, monitoring, and delivery of care.3 This patients and health services when compared with vertical approach was crucial for scaling up HIV services, standard vertical care and was popular with patients, but has led to duplicated efforts and low cross-programme possibly by reducing HIV-related stigma.5–7 learning. However, although we identified a more efficient and To address this issue, we partnered with health services effective approach for managing chronic conditions at and civil society in Dar es Salaam in Tanzania and the health facilities, the growing burden of diabetes and Kampala region in Uganda, to combine the care of HIV, hypertension alongside the ongoing HIV epidemic diabetes, and hypertension within primary health-care continue to overwhelm facility-based services. Community facilities. As part of this model, patients with any of these HIV care delivered by nurses and trained workers has conditions (whether occurring individually or in expanded service reach,8–10 but the impact shown on HIV combination) were managed in a single clinic that outcomes has been inconsistent in larger trials.11,12 incorporated best practices adapted from HIV Whether community-based services could help to manage Articles people with multiple chronic conditions and varying All patients were managed by public health-care staff clinical profiles and levels of stigma remains unclear. and invited back to health facilities in their groups, Here, we aim to compare integrated community-based usually 1–2 weeks after enrolment. We held question- care with integrated facility-based care for people with and-answer sessions to assess patient understanding and HIV, diabetes, and hypertension in Tanzania and to improve their knowledge of the study procedures, Uganda.13 including randomisation. All patients provided written informed consent. See Online for appendix Methods The protocol is available in the appendix (pp 12–41) and Study design and participants was approved by ethics committees at the National This open-label, multicountry, cluster-randomised Institute for Medical Research (Dar es Salaam, Tanzania; (INTE-COMM) trial was conducted in 14 medium-to- number NIMR/HQ/R·8a/VolIX/3977), Uganda Virus large sized primary care facilities across urban and Research Institute (Entebbe, Uganda; number peri-urban catchment areas in Dar es Salaam and coastal GC/127/872), London School of Hygiene & Tropical regions of Tanzania and in Kampala, Wakiso, and Mpigi Medicine (London, UK; number 28122), and University districts of Uganda. College London (London, UK; number 2382/01). This As integrated management of these chronic conditions trial was registered with the International Standard was not standard, we established facility integrated care Randomised Controlled Trial Number registry, for people with HIV, diabetes, and hypertension at ISRCTN15319595 (completed). participating sites, with participants managed by government health-care staff. Six of the facilities (in Randomisation and masking Uganda) had participated in earlier integration studies In each health facility, eligible patients were grouped into and maintained the model after the trial.6,14 Procedures clusters of around 8–14 (2:1 ratio of those with diabetes or were reviewed and refresher training provided in these hypertension [or both] to those with HIV). Group size facilities and integrated services were newly established was decided in consultation with patient groups, health- in the other eight facilities. care workers, and policy makers, considering the time The eligibility criteria were age 18 years or older; living participants would need to attend and ease of with HIV, type 2 diabetes, or hypertension (or a management. This ratio considered statistical power for combination); receiving regular care at the health facility both primary endpoints. The clusters were formed based for at least 6 months; considered by the clinical team as on geographical proximity of participants living in the clinically stable (defined as no or well managed same ward or parish. We formed one or two groups at complications or co-infections and self-reported each facility each day and decided the ward or parish to adherence to treatment in the previous 6 months); living target based on appointment schedules. When we within the catchment area of the health facility and suspected insufficient numbers might attend to form a planning to stay for at least 6 months; and willing to group, those scheduled to attend were contacted by receive care in the community. Exclusion criteria were telephone in advance to confirm attendance. blood pressure greater than 160/100 mm Hg at the Once the group size and composition criteria were screening visit or greater than 180/110 mm Hg at any met, each group was randomly assigned (1:1) to integrated For more on the online system time in the previous 6 months; fasting blood glucose facility care or community care using an online data see https://github.com/ greater than 13 mmol/L at any time in the previous management system. The randomisation list was intecomm-trial/ 6 months; any clinical condition that required facility generated by an independent statistician using PROC management; and pregnancy. Women who became PLAN (SAS procedure) and preloaded into the system. pregnant during the study were withdrawn and referred Randomisation was done at the health facilities, with a to antenatal care services. On the advice of health-care representative of each group pressing the return key on a providers and policy makers, we decided to enrol laptop to generate the randomisation group as the final clinically stable patients. step. Trial staff and participants were masked to the Formation of groups was coordinated by experienced randomisation until this step. nurses familiar with each catchment area and the patient population. We used Google maps and landmarks as Procedures points of reference to pinpoint the addresses precisely of Patients randomly assigned to facility care received care patients who did not have a physical address. Patients at dedicated integrated clinics within the facilities. They were invited for enrolment consecutively during were not required to meet in their groups but retained scheduled clinic visits until we reached the target number usual appointment schedules, usually once per month and were given information about the study and for those with diabetes or hypertension and once every procedures in one-to-one consultations; written 3 months for those with HIV. Participants shared the information sheets taken from consent forms; and same registration and waiting areas, were managed by monetary compensation (approximately £4) for their the same physicians and health-care workers, used the time. same pharmacy and laboratory services,4,5,15 and received 1086 Articles care from the same health-care team irrespective of the incapacity. Those meeting the criteria received condition. Participants also received basic health medication at no cost. Some participants had health education and had adherence and psychosocial support, insurance, which covered the costs. and those who missed appointments were contacted by The research team provided a buffer supply of medicines telephone and advised to return to the clinic (ie, track and (including glucometers and blood pressure machines) to trace). Integrated care clinics operated once per week the health facilities in Uganda and Tanzania at the alongside standalone vertical clinics. beginning and middle of the study to ensure a minimum The community care group was designed to deliver supply of medicines, which was administered and made services closer to patients’ homes. The first group available to all patients, not just trial participants. Plasma meeting at the health facility occurred immediately after viral load tests were conducted by the government using the group was formed and after randomisation. A nurse routine testing. Medical data collected by clinical staff at and trained lay worker (community health worker in health facilities were extracted by the research team and Tanzania and village health worker in Uganda) supported entered into the trial database with consistency checks. the group to choose a group representative and identify a Blood pressure was measured using the Omron M2 suitable venue in the community where they would (OMRON Healthcare, Den Bosch, Netherlands). The receive care, such as places of worship or schools. mean was calculated from two measurements on the Locations were evaluated for space, privacy, accessibility, patient’s left arm taken 5 min apart. For patients with and comfort and a venue was chosen collectively, then hypertension, controlled blood pressure was defined as the nurse, trained lay worker, group leader, and systolic blood pressure of less than 140 mm Hg and researchers visited the venue. Community leaders and diastolic blood pressure of less than 90 mm Hg. Fasting other stakeholders at the venue were identified and an glucose was measured using the Contour Plus Blood engagement meeting held to inform stakeholders about Glucose Monitoring System (Ascensia Diabetes Care, the study before service delivery commenced. Basel, Switzerland). Controlled fasting glucose was Groups met once per month at the agreed community defined as a fasting glucose of less than 7 mmol/L. locations. Coordination was managed by the group leader Plasma viral load testing was measured using the with support from the nurse and lay worker. On the day Cobas 8800 system (Roche Diagnostics, Mannheim, before each meeting, patient files were retrieved from the Germany). Viral load suppression was defined as less health facility, prescriptions were prepared by the than 1000 copies per mL. Weight was measured using pharmacist, and medications were prepacked into Seca 813 digital weighing scales and height using identical individually labelled brown envelopes. The Seca 213 portable stadiometers (Seca, Hamburg, nurse and trained lay worker transported the files, Germany). Weight was measured to the nearest 0·1 kg medication, weighing scales, blood pressure machines, and height to the nearest 0·1 cm. BMI was calculated as and glucometers to the community site using taxis. the quotient of weight and the square of height. At each meeting, a nurse led the health education Follow-up was continued for 12 months. The final sessions and conducted clinical assessments, including meeting of groups in the community care group was blood pressure and glucose measurements. Adherence, conducted at the health facility to facilitate data collection. behavioural information, and support were provided by Loss to follow-up was defined as not attending an the trained lay worker with supervision by the nurse. appointment within the previous 6 months (chosen Referrals were made to the health facility when indicated because some participants with HIV were given clinically (eg, the participant had high blood pressure, appointments 3–6 months apart). The final fasting high blood glucose, or reported being unwell, such as glucose, blood pressure, and viral load measurements blurred vision, dizziness, or chest pain). Participants who were done at the 12-month visit or at the visit after missed the community group meeting were contacted by 6 months of follow-up. telephone and advised to collect their prescribed drugs from the integrated clinic at the health facility and Outcomes received the same services as those in the facility care The trial had two coprimary endpoints. The first coprimary group. endpoint was a composite of blood pressure or fasting We took a number of actions to mitigate against glucose control (defined as blood pressure <140/90 mm Hg insufficient supply of medication. In Uganda, we in participants with hypertension alone, fasting glucose encouraged patients receiving care at the health facility to <7·0 mmol/L in those with diabetes alone, and blood set up or join a patient-led medication fund to bulk buy pressure <140/90 mm Hg and fasting glucose medicines at a lower cost when supplies at health <7·0 mmol per L in those with diabetes and hypertension). facilities were low. In Tanzania, public health facilities The second coprimary endpoint was plasma viral load typically provide medication for free to those who cannot suppression for participants with HIV alone (defined as afford medication. Tanzanian facilities assessed <1000 copies per mL or undetectable viral load). participants using designated social welfare offices based Secondary endpoints from baseline to 12-month on predefined eligibility criteria to determine financial follow-up were retention in care; blood pressure control Articles 2940 patients assessed for eligibility 765 not screened 747 missed clinic or screening appointment or did not attend required health talks 17 unwilling to be screened 1 not clinically stable 2175 screened 203 ineligible 82 with history of uncontrolled blood pressure or glucose 61 unwilling to stay in catchment area 28 unsuitable for study at clinician discretion 18 required acute care 8 pregnant 5 in care for less than 6 months 1 with unknown blood pressure 1972 eligible 4 did not consent 1968 provided consent 104 did not meet group size or ratio criteria 77 with diabetes or hypertension (or both) 14 with HIV alone 13 with HIV and diabetes or hypertension (or both) 1864 grouped and randomly assigned (124 groups) 935 assigned to community care (62 groups) 929 assigned to facility care (62 groups) 3 did not attend the baseline assessment 3 did not attend the baseline assessment 88 with HIV and diabetes or hypertension (or both) 70 with HIV and diabetes or hypertension (or both) 242 with HIV alone 602 with diabetes or hypertension 247 with HIV alone 609 with diabetes or hypertension (or both) (or both) 73 with diabetes 72 with diabetes 347 with hypertension 373 with hypertension 182 with diabetes and hypertension 164 with diabetes and hypertension 5 excluded 28 excluded 5 excluded 35 excluded 2 transferred to another health 8 lost to follow-up 3 transferred to another health 13 transferred to another health facility 7 transferred to another health facility facility 1 clinical withdrawal facility 2 lost to follow-up 11 lost to follow-up 1 withdrew consent 6 withdrew consent 6 died 1 lost to follow-up 6 died 5 withdrew consent 1 clinical withdrawal 237 at 6-month follow-up 574 at 6-month follow-up 242 at 6-month follow-up 574 at 6-month follow-up 4 excluded 10 excluded 5 excluded 8 excluded 2 transferred to another health 5 transferred to another health 3 lost to follow-up 5 transferred to another health facility facility 1 transferred to another health facility 1 consent withdrawal 3 lost to follow-up facility 2 lost to follow-up 1 died 1 consent withdrawal 1 pregnant 1 died 1 clinical withdrawal 233 completed 12-month follow-up 564 completed 12-month follow-up 237 completed 12-month 566 completed 12-month follow-up 67 with diabetes follow-up 67 with diabetes 325 with hypertension 344 with hypertension 172 with diabetes and hypertension 155 with diabetes and hypertension Figure: Trial profile 1088 Articles and change in participants with hypertension; fasting into a group were identified from health-care records as blood glucose control and change in participants with having appointments to attend at the participating diabetes; deaths; and process indicators (ie, missed facilities (appendix p 3). Of those, 765 (26·0%) patients appointments and unscheduled facility visits). were not screened and 2175 (74·0%) were screened for eligibility (figure). 203 (9·3%) patients were ineligible, Statistical analysis mostly because they had a history of uncontrolled blood After forming 116 groups of eight people with diabetes or pressure or blood glucose (n=82). An additional hypertension, the trial was estimated to have 80% power four (0·2%) patients did not consent to be part of the to detect an absolute difference of 10% in the composite endpoint between the facility care and community care Participants with diabetes or Participants with HIV alone groups (at 5% two-sided significance level, assuming an hypertension (or both)* intraclass coefficient of 0·02). By including Community care Facility care Community care Facility care four participants with HIV (in each of the 116 groups), (n=602) (n=609) (n=242) (n=247) the trial had more than 90% power to show non- Country of study inferiority in plasma viral load suppression using a Tanzania 273 (45·3%) 300 (49·3%) 112 (46·3%) 121 (49·0%) non-inferiority margin of –10% (at 5% one-sided Uganda 329 (54·7%) 309 (50·7%) 130 (53·7%) 126 (51·0%) significance level, assuming 90% of participants would have viral suppression in the facility care group). Time since diagnosis, years Coprimary and secondary endpoints were assessed in Diabetes† 6·4 (5·6)† 7·1 (6·2)† NA NA the intention-to-treat population (defined as all randomly Hypertension† 6·4 (5·8)† 6·5 (6·3)† NA NA assigned participants). No adjustment for multiple HIV alone NA NA 8·5 (5·2) 9·2 (5·0) comparisons was made as the primary endpoints were Age, years assessed in different populations (ie, those with diabetes Mean 58·6 (11·1) 57·9 (11·8) 44·7 (10·5) 45·7 (10·3) or hypertension [or both] and those with HIV alone as <35 13 (2·2%) 16 (2·6%) 44 (18·2%) 35 (14·2%) per the statistical analysis plan). 35–49 117 (19·4%) 131 (21·5%) 112 (46·3%) 123 (49·8%) Generalised estimating equation (GEE) models were ≥50 472 (78·4%) 462 (75·9%) 86 (35·5%) 89 (36·0%) used for coprimary endpoint analysis to account for data Sex clustering within individuals and groups. The GEE Female 474 (78·7%) 479 (78·7%) 164 (67·8%) 185 (74·9%) model had a binomial distribution and an identity link Male 128 (21·3%) 130 (21·3%) 78 (32·2%) 62 (25·1%) function with treatment allocation as a predictor, and Education patient group as cluster effect was used to calculate the No formal education 101 (16·8%) 92 (15·1%)‡ 16 (6·7%)§ 25 (10·1%) crude risk difference (95% CIs). Adjusted risk difference Primary education 348 (57·8%) 367 (60·4%)‡ 154 (64·2%)§ 168 (68·0%) was also calculated by adding age and sex into the model. Secondary or tertiary 153 (25·4%) 149 (24·5%)‡ 70 (29·2%)§ 54 (21·9%) Sensitivity analyses of coprimary endpoints were education prespecified in the statistical analysis plan and done in Marital status the per-protocol population (according to the group that Single 37 (6·1%) 37 (6·1%)‡ 38 (15·8%)§ 38 (15·4%) participants remained in for at least 6 months), with Married 343 (57·0%) 341 (56·1%)‡ 124 (51·7%)§ 113 (45·7%) imputation of missing primary endpoint data including Divorced 72 (12·0%) 70 (11·5%)‡ 44 (18·3%)§ 50 (20·2%) scenarios of the worst, best, and worst–best possible Widowed 150 (24·9%) 160 (26·3%)‡ 34 (14·2%)§ 46 (18·6%) outcome and with multiple imputation. Secondary Employment binary endpoints were analysed in a similar way to Manual work 292 (48·5%) 284 (46·7%)‡ 168 (70·0%)§ 174 (70·4%) coprimary endpoints. Secondary continuous endpoints Professional 62 (10·3%) 64 (10·5%)‡ 39 (16·3%)§ 34 (13·8%) were analysed with similar GEE models as for analysis of Homemaker 95 (15·8%) 108 (17·8%) 17 (7·1%) 15 (6·1%) binary endpoints but with a normal distribution and Unemployed 27 (4·5%) 27 (4·4%) 8 (3·3%) 11 (4·5%) baseline measurements as an additional covariate, from Unemployed (not seeking 86 (14·3%) 88 (14·5%) 6 (2·5%) 7 (2·8%) which crude and adjusted mean difference (95% CIs) a job) were derived. Retired 32 (5·3%) 32 (5·3%) ·· ·· Other 8 (1·3%) 5 (0·8%) 2 (0·8%) 6 (2·4%) Role of the funding source Smoking status The funder of the study had no role in study design, data Smoker 4 (0·7%) 12 (2·0%)‡ 8 (3·3%)§ 11 (4·5%) collection, data analysis, data interpretation, or writing of Ex-smoker 37 (6·1%) 28 (4·6%)‡ 25 (10·4%)§ 9 (3·6%) the report. Non-smoker 561 (93·2%) 568 (93·4%)‡ 207 (86·3%)§ 227 (91·9%) Alcohol consumption Results Never 558 (92·7%) 552 (90·8%)‡ 168 (70·0%)§ 176 (71·3%) Between Jan 30 and Oct 6, 2023, 2940 patients with HIV, Occasionally or regularly 44 (7·3%) 56 (9·2%)‡ 73 (30·4%)§ 71 (28·7%) diabetes, or hypertension (or a combination of these (Table 1 continues on next page) conditions) who lived close enough together to be placed Articles 158 participants with HIV and concurrent diabetes or Participants with diabetes or Participants with HIV alone hypertension were excluded from the primary analyses hypertension (or both)* and six provided no information. 1700 participants were Community care Facility care Community care Facility care included: 489 (28·8%) with HIV alone, 720 (42·4%) with (n=602) (n=609) (n=242) (n=247) hypertension alone, 145 (8·5%) with diabetes alone, and (Continued from previous page) 346 (20·4%) with diabetes and hypertension. We provide BMI, kg/m² results of the whole cohort including those with HIV and Mean 28·4 (6·1)‡ 28·4 (6·0)‡ 25·2 (5·2)‡ 24·9 (4·6) diabetes or hypertension in the appendix (pp 4–11). <25·0 183 (30·4%)‡ 184 (30·3%)‡ 124 (51·5%)‡ 135 (54·7%) The median number of participants per group 25·0–29·9 208 (34·6%)‡ 210 (34·5%)‡ 81 (33·6%)‡ 81 (32·8%) was 14 (range 9–19) in the community care group ≥30 210 (34·9%)‡ 214 (35·2%)‡ 36 (14·9%)‡ 31 (12·6%) and 14 (10–18) in the facility care group. The median Data are n (%) or mean (SD). Some totals might not equal 100% owing to rounding. NA=not applicable. number per group of patients with HIV alone was 4 (1–7) *73 participants in community care versus 72 in facility care with diabetes alone, 347 versus 373 with hypertension in the community care group versus 4 (0–7) in the facility alone, and 182 versus 164 with diabetes and hypertension. †For participants living with both, the calculation for care group and with diabetes or hypertension (or both) diabetes is based on self-reported diagnosis of diabetes; similarly, the calculation for hypertension is based on when they were diagnosed with hypertension; thus, the number of participants with diabetes was 255 in community care was 10 (7–15) versus 10 (6–13). and 236 in facility care, whereas for hypertension, these numbers were 529 and 537, respectively. ‡Missing for one Baseline social and demographic characteristics were participant. §Missing for two participants. well balanced between the two groups (table 1). Across Table 1: Baseline social and demographic characteristics of participants both study groups, the mean number of years with the condition was 6·4 (SD 6·1; n=1066) for hypertension, 6·7 (5·9; n=491) for diabetes, and 8·9 (5·1; n=489) for HIV. The mean age of participants with HIV alone was Community care Facility care 45·2 years (SD 10·4) and was 58·2 years (11·4) for those (n=602)* (n=609)* with diabetes or hypertension (or both). There were Blood pressure 319/528 (60·4%)† 327/536 (61·0%)† three times more female than male participants enrolled <140/90 mm Hg in those with hypertension or in the study (1302 [76·6%] of 1700 vs 398 [23·4%]). diabetes (or both) Among participants with diabetes or hypertension (or Mean systolic blood 132·0 (16·2)† 131·4 (15·5)† both), the proportion with a BMI of at least 30 kg/m² was pressure, mm Hg 390 (41·0%) of 952 female participants and Mean diastolic blood 80·9 (10·2)† 80·6 (9·8)† 34 (13·2%) of 257 male participants whereas among pressure, mm Hg participants with HIV alone, these proportions were Fasting blood glucose 76/229 (33·2%)‡ 71/208 (34·1%)§ 60 (17·2%) of 348 and seven (5·0%) of 140. <7·0 mmol/L in those with diabetes or hypertension Table 2 summarises the baseline clinical characteristics (or both) by health condition, which were also well balanced Mean fasting blood 8·5 (2·6) 8·4 (2·6) between the groups. At baseline, 268 (46·6%) of 575 in glucose, mmol/L the community care group versus 286 (49·3%) of 580 in Composite of blood pressure 268/575 (46·6%) 286/580 (49·3%) the facility care group had adequate composite of blood or fasting glucose control¶ pressure or fasting glucose control. HIV viral load <1000 copies 236/242 (97·5%) 240/247 (97·2%) per mL in those with HIV Among participants with hypertension alone or with alone diabetes, 319 (60·4%) of 528 in the community care group HIV viral load <400 copies 231/242 (95·5%) 233/247 (94·3%) versus 327 (61·0%) of 536 in the facility care group had a per mL in those with HIV blood pressure of less than 140/90 mm Hg. Among those alone with diabetes alone or with hypertension, 76 (33·2%) of Data are n/N (%) or mean (SD). Some totals might not equal 100% owing to 229 in the community care group versus 71 (34·1%) rounding. *Number of participants with diabetes or hypertension, or both. of 208 in the facility care group had a fasting blood glucose 73 participants in community care versus 72 in facility care with diabetes alone, 347 versus 373 with hypertension alone, and 182 versus 164 with diabetes and of less than 7·0 mmol/L. Among participants with HIV, hypertension. †Missing for one participant. ‡Missing for 26 participants. §Missing 236 (97·5%) of 242 in the community care group versus for 28 participants. ¶Defined as blood pressure of less than 140/90 mm Hg in 240 (97·2%) of 247 in the facility care group reached viral those with hypertension alone, fasting plasma glucose of less than 7·0 mmol in those with diabetes alone, or both indicators controlled in those with both suppression with less than 1000 copies per mL. conditions. Among participants with diabetes or hypertension (or both), 38 (6·3%) of 602 in the community care group Table 2: Baseline clinical characteristics by health conditions versus 43 (7·1%) of 609 in the facility care group could not be evaluated at 12-month follow-up, whereas these study after receiving more information and proportions were nine (3·7%) of 242 versus ten (4·0%) 104 (4·8%) did not meet group size or ratio criteria. of 247 among participants with HIV (figure). 1864 (63·4%) patients formed 124 groups. The groups The composite of blood pressure or fasting glucose were randomised, with 62 groups in community care control coprimary endpoint did not significantly differ and 62 in facility care. between the two groups in participants with hypertension 1090 Articles or diabetes (or both; 317 [55·2%] of 574 in the community vs 566 [92·9%] attended the 12-month appointment; care group vs 304 [53·2%] of 571 in the facility care group; table 3). Among those with hypertension, blood pressure adjusted risk difference 1·80 [95% CI –4·52 to 8·12]; indicators did not differ significantly between the p=0·58; table 3). Sensitivity analyses were consistent two groups at 12-month follow-up. The proportion of with that of the primary analysis (appendix pp 4–5). participants with diabetes reaching good control of Subgroup analyses also show no statistical difference fasting blood glucose (<7·0 mmol/L) also did not differ between the study groups as in the primary analysis significantly between the two groups. (appendix p 7). Coprimary outcome data in participants 126 (38·2%) of 330 participants with diabetes and with HIV and diabetes or hypertension (or both) at hypertension had a blood pressure of less than 12-month follow-up are shown in the appendix (p 11). 140/90 mm Hg and fasting glucose of at least Retention in care was similar between the two study 7·0 mmol/L. 30 (9·1%) had fasting glucose of less than groups (574 [95·3%] of 602 participants in the community 7·0 mmol/L and blood pressure of at least care group vs 573 [94·1%] of 609 in the facility care group 140/90 mm Hg. Among those with hypertension or attended the 6-month appointment and 564 [93·7%] diabetes (or both), there were six (1·0%) deaths in the Community care Facility care Crude risk difference p value Adjusted risk p value (n=602) (n=609) (95% CI)* difference (95% CI)† Coprimary outcome Composite of blood pressure or fasting 317/574 (55·2%) 304/571 (53·2%) 1·76 (–4·58 to 8·10) 0·59 1·80 (–4·52 to 8·12) 0·58 glucose control‡ Secondary outcomes (retention in care) Attended 6-month appointment 574/602 (95·3%) 573/609 (94·1%) 1·25 (–1·36 to 3·87) 0·35 1·36 (–1·25 to 3·97) 0·31 Attended 12-month appointment 564/602 (93·7%) 566/609 (92·9%) 0·75 (–2·16 to 3·66) 0·62 0·85 (–2·04 to 3·73) 0·56 Blood pressure in participants with hypertension Blood pressure <140/90 mm Hg 359/506 (70·9%) 340/504 (67·5%) 3·44 (–2·37 to 9·25) 0·25 3·38 (–2·43 to 9·19) 0·25 Blood pressure >180/110 mm Hg 5/506 (1·0%) 4/504 (0·8%) 0·20 (–1·06 to 1·46) 0·75 0·21 (–1·05 to 1·47) 0·74 Mean systolic blood pressure, mm Hg 128·61 (15·56) 130·11 (16·47) –1·78 (–3·61 to 0·06) 0·058 –1·78 (–3·62 to 0·05) 0·057 Mean diastolic blood pressure, mm Hg 79·45 (10·16) 79·89 (11·20) –0·51 (–1·73 to 0·71) 0·41 –0·50 (–1·72 to 0·72) 0·42 Blood glucose in participants with diabetes Fasting blood glucose <7·0 mmol/L 92/243 (37·9%) 74/222 (33·3%) 4·17 (–5·81 to 14·16) 0·41 4·19 (–5·82 to 14·19) 0·41 Mean fasting blood glucose, mmol/L 8·71 (3·41) 8·55 (2·90) 0·08 (–0·48 to 0·63) 0·79 0·08 (–0·48 to 0·64) 0·78 Deaths in participants with hypertension 6/602 (1·0%) 7/609 (1·1%) –0·15 (–1·25 to 0·95) 0·79 –0·20 (–1·31 to 0·92) 0·73 or diabetes (or both, over the 12-month follow-up) Data are n/N (%) or mean (SD), unless stated otherwise. *Binary outcomes adjusted for clustering and continuous outcomes adjusted for baseline measurements. †Binary outcomes adjusted for clustering, age, and sex, and continuous outcomes adjusted further for baseline measurements. ‡Defined as blood pressure of less than 140/90 mm Hg in those with hypertension alone, fasting plasma glucose of less than 7·0 mmol in those with diabetes alone, or both indicators controlled in those with both conditions. Table 3: Outcomes in participants with hypertension or diabetes (or both) Community care Facility care Crude risk difference p value Adjusted risk p value (n=242) (n=247) (95% CI)* difference (95% CI)† Coprimary outcome Plasma viral load 227/229 (99·1%) 229/232 (98·7%) 0·42 (–1·14 to 1·97)‡ p<0·0001‡ 0·44 (–1·12 to 1·99)‡ p<0·0001‡ <1000 copies per mL Secondary outcomes Plasma viral load <400 copies 224/229 (97·8%) 222/232 (95·7%) 2·12 (–1·23 to 5·47) 0·22 2·20 (–1·15 to 5·55) 0·20 per mL Attended 6-month 237/242 (97·9%) 241/247 (97·6%) 0·36 (–2·16 to 2·88) 0·78 0·43 (–2·09 to 2·96) 0·74 appointment Attended 12-month 233/242 (96·3%) 237/247 (96·0%) 0·36 (–2·99 to 3·70) 0·83 0·38 (–2·96 to 3·73) 0·82 appointment Deaths over 12-month 1/242 (0·4%) 0/247 0·41 (–0·39 to 1·22) 0·32 0·42 (–0·40 to 1·25) 0·32 follow-up Data are n/N (%), unless stated otherwise. *Adjusted for clustering. †Adjusted for clustering, age, and sex. ‡One-sided 95% CI and p . non-inferiority Table 4: Outcomes in participants with HIV alone Articles Participants with hypertension or diabetes (or both) Participants with HIV alone Community care Facility care Risk difference; Community care Facility care Risk difference; (n=602) (n=609) p value (n=242) (n=247) p value Number of scheduled 1099/7745 (14·2%); 1387/6643 (20·9%); –0·45% 1198/3121 (38·4%); 212/1401 (15·1%); 4·09 appointments missed 1·83 (2·43) 2·28 (2·54) (–0·73 to –0·17); 4·95 (2·89) 0·86 (1·81) (3·67 to 4·52); p=0·0016 p<0·0001 Number of participants 354/602 (58·8%) 424/609 (69·6%) –10·70% 224/242 (92·6%) 73/247 (30·0%) 62·44% who missed one or more (–18·31 to –3·09); (53·48 to 71·39); appointments p=0·0059 p<0·0001 Number of participants 19/602 (3·2%)* 13/609 (2·1%)* 0·82% 0 0 .. who visited the facility (–0·14 to 3·01); unscheduled by referral p=0·46 or self-referral Data are n/N (%); mean (SD) or n/N (%), unless stated otherwise. *22 visits in community care and 13 visits in facility care. Table 5: Appointments and referrals measured over 12-month follow-up community care group and seven (1·1%) in the facility To our knowledge, this trial is the first to test an group. integrated community care model for multiple conditions In terms of the coprimary endpoint in participants in Africa. Previous studies have been largely on single with HIV alone, 227 (99·1%) of 229 in the community conditions and have suggested that community care for care group and 229 (98·7%) of 232 in the facility care chronic conditions could be effective when services are group had plasma viral load of less than 1000 copies per delivered vertically, although the evidence for the mL and non-inferiority in viral suppression was shown management of patients with diabetes and hypertension (adjusted risk difference 0·44 [95% CI –1·12 to 1·99]; varies.12,16–18 Supply of medicines for diabetes and p <0·0001; table 4). There was one (0·4%) death hypertension remains a challenge across Africa.19,20 non-inferiority in the community care group. The proportion of Additionally, a large number of people in Africa live with participants with HIV who attended the 12-month multiple chronic conditions21,22 and delivering vertical appointment was high and similar in both groups community care for single conditions might have low (table 4). Sensitivity analyses showed similar results impact. (appendix pp 8–9). Subgroup analyses show non- Patients enrolled in our trial had good control of blood inferiority within the age (<55 years vs ≥55 years) and pressure, blood glucose, and HIV viral load at baseline, sex categories between the study groups (appendix reflecting the inclusion of individuals who were clinically p 10). stable on treatment. By the end of the study at 12 months Among participants with diabetes or hypertension (or of follow-up, control of blood pressure and glucose had both), the number of scheduled appointments missed improved in both community and facility care groups was lower in the community care group than the facility and around 70% of patients with hypertension reached care group (1099 [14·2%] of 7745 vs 1387 (20·9%] of 6643); good blood pressure control. Engagement with the study and 19 (3·2%) of 602 participants versus 13 (2·1%) of was exceptional and the proportion retained in care who 609 visited the facility unscheduled by referral or self- attended the 12-month appointment exceeded 90% in referral (risk difference 0·82% [–0·14 to 3·01]; p=0·46; both groups; whereas in other clinical trials table 5). Among participants with HIV, the number of in low-income and middle-income settings, retention in missed appointments in community care was double care rarely exceeds 50%.23 that of facility care (1198 [38·4%] of 3121 Fewer patients with HIV declined to join or withdrew vs 212 [15·1%] of 1401), but there were no referrals or from the study than those with diabetes or hypertension self-referrals. (or both). The rates of HIV viral suppression remained In a post-hoc analysis, blood pressure control in high and non-inferior in the community care group, participants with HIV and hypertension was significantly despite patients shifting from physician-led clinics with better in community care than facility care groups (20·34 infrequent visits to once per month group community [1·89–38·78]; p=0·031; appendix p 11); but the number of visits. These findings support our earlier observations participants enrolled with HIV and diabetes was small. that integration of services for patients with HIV who have other conditions might reduce HIV associated Discussion stigma.7,24 This multicountry, cluster-randomised trial showed that Our findings have two clear implications. First, that integrated community care is as effective as integrated clinically stable patients in regular care can safely facility care for the management of patients with HIV, transition to be managed in the community, reducing the diabetes, and hypertension. burden on overstretched health facilities and improving 1092 Articles the convenience and accessibility of care. Second, simple Community care should be evaluated further, but we support mechanisms (such as structured follow-up, assume wide scale availability due to the need to be reminder calls, and patient tracing) embedded in HIV organised in more locations and on more days in a month care could substantially improve outcomes for patients to meet demand. Harmonising visit schedules by with diabetes and hypertension. extending diabetes and hypertension appointments to Notably, the proportion of patients with diabetes occur every 3–6 months, which is common in HIV care, reaching glycaemic control was lower than the proportion could improve efficiency and acceptability but will also of those with hypertension reaching blood pressure need evaluation. As health services adapt community care control. Around a third of patients with diabetes reached scale-up, a temptation might be to add community care fasting glucose of less than 7 mmol/L, echoing findings for other conditions (in addition to HIV, diabetes, and from other studies in Africa.5,25 Furthermore, those with hypertension) or offer integrated community care for both diabetes and hypertension had worse glucose clinically unstable patients; these approaches should be control than their blood pressure control. studied further. The consistent gap between diabetes and hypertension A weakness of our study is the open-label design with outcomes points to limitations in diabetes management 12 months of follow-up. We chose endpoints that were strategies rather than health system factors. First, type 2 objective to mitigate this limitation. A further weakness diabetes is a progressive disease with declining β-cell is that the number of patients enrolled with HIV and function and rising insulin resistance over time requiring diabetes or hypertension (or both) was small and we were continued escalation of treatment.26 Second, diabetes in unable to draw conclusions for these groups. Finally, Africa often presents with distinct phenotypes (frequently although participants were managed by government in young and lean individuals) with less insulin resistance health-care staff and according to guidelines, the research and greater insulin deficiency, but the optimal team were observing procedures and recording data; management strategies remain undefined.27 Third, the thus, the outcomes recorded might differ when compared low repertoire of diabetes medications, especially beyond with real-life settings. metformin and sulfonylureas, constrains the effective In summary, this study shows that integrated control of diabetes.28 Together, these factors underscore management of patients with HIV, diabetes, and the urgent need to adapt treatment protocols and expand hypertension from community focal points by nurses therapeutic options for diabetes in Africa. and lay workers is as effective as facility care. By shifting When designing the trial, we assumed that community clinically stable patients from facilities to community care would be challenging for people with HIV since these care, this approach offers a promising pathway to patients were accustomed to dedicated physician-led decongest health services, extend reach, and make clinics—often located separately to preserve anonymity— chronic care more people-centred in sub-Saharan Africa. with conveniently spaced visits every 3–6 months, but the RESPOND-AFRICA Group rates of HIV viral suppression observed in our study were Gerard J Abou Jaoude, Mathias Akugizibwe, Max Bachmann, excellent. Our community care model required attendance Mtumwa Bakari, Ousman Bah, Carlos S Grijalva-Eternod, Peiyun Hu, Caroline Jeffery, Calvin Kaaya, Anne Katahoire, Jerome Kabakyenga, once per month at set times when a nurse and lay worker Aisha Kigongo, Isaac S Kintu, Salome Kitinusa, Zenais Kiwale, visited focal points (such as schools or places of worship), Judith J Lunyungu, Aneth, Mtui, Francis X Namugera, alongside neighbours living with diabetes and hyper- Chaka M Namulundu, Joseph Okebe, Stella Sarwatt, David Serunjogi, tension and in more visible public settings, where they Meshack Shimwela, and Sophia M Tabarani. could be worried about preserving their anonymity. Contributors Despite these concerns, HIV outcomes showed that MJN, SJ, SM, and KR wrote the protocol, secured the funding, and oversaw the programme. MJN, SJ, SM, KR, ST, GM, MNS, OSU, SA, integrated community care was equally as effective as SW, and MCVH designed the study. FXK, FA, IN, JB, SK, AG, FM, and integrated facility care. JAP implemented the study. EvW managed the data. DW and EvW We focused our analysis on two primary populations: analysed the data. FXK, SM, and KR led the engagement with people with diabetes or hypertension (or both) and those communities. MJN and KR led the engagement with the Ministries of Health. NKS, GG, and PGS provided independent input and oversight, with HIV alone. The numbers with HIV and hypertension with PGS chairing the International Steering Committee. FXK, FA, and or diabetes (or both) were small (as testing for diabetes or AG wrote the first draft of the manuscript and subsequent drafts were hypertension [or both] is uncommon), but some blood checked by SJ and MJN, with all other authors contributing. EvW, DW, pressure indicators among these groups were superior in and SJ accessed and verified the data. All authors had full access to all the data in the study and had final responsibility for the decision to the community care group compared with the facility submit for publication. care group. Community care is likely to benefit those Declaration of interests with multiple comorbidities because they have the We declare no competing interests. greatest challenge of accessing facility care, but our study Data sharing was not designed to address this question. Published and de-identified data with the data dictionaries, study Although our study findings show that the concept of protocol, statistical analysis plan, and a sample of the informed consent integrated community care is effective, we do not know form are available upon request to the corresponding author. how effective this approach would be when scaled up. Articles Acknowledgments 14 Barnabas RV, Szpiro AA, van Rooyen H, et al. Community-based This study is funded by the National Institute for Health and Care antiretroviral therapy versus standard clinic-based services for HIV Research (NIHR; Global Health Policy and Systems Research in South Africa and Uganda (DO ART): a randomised trial. Programme grant NIHR131273). We thank patients who participated in Lancet Glob Health 2020; 8: e1305–15. this study, health-care providers, managers of health facilities, and local 15 Mfinanga SG, Nyirenda MJ, Mutungi G, et al. Integrating HIV, and national policy makers for their time and support. We also thank the diabetes and hypertension services in Africa: study protocol for a cluster randomised trial in Tanzania and Uganda. BMJ Open 2021; independent data safety monitoring committee: Pablo Perel, 11: e047979. David Guwatudde, and Sile Molloy. We thank our professional services 16 Firima E, Gonzalez L, Ursprung F, et al. Community-based models teams for administrative and managerial support, in particular Dimitra of care for management of type 2 diabetes mellitus among non- Stamogiannou, whose brilliance allows us to spend concentrated time pregnant adults in sub-Saharan Africa: a scoping review. PLoS One on our research. 2023; 18: e0278353. References 17 Mengesha EW, Tesfaye TD, Boltena MT, et al. Effectiveness of 1 GBD 2021 US Burden of Disease Collaborators. The burden of community-based interventions for prevention and control of diseases, injuries, and risk factors by state in the USA, 1990–2021: hypertension in sub-Saharan Africa: a systematic review. a systematic analysis for the Global Burden of Disease Study 2021. PLOS Glob Public Health 2024; 4: e0003459. Lancet 2024; 404: 2314–40. 18 Fernández LG, Firima E, Robinson E, et al. Community-based care 2 Trickey A, Sabin CA, Burkholder G, et al. Life expectancy after 2015 models for arterial hypertension management in non-pregnant of adults with HIV on long-term antiretroviral therapy in Europe adults in sub-Saharan Africa: a literature scoping review and and North America: a collaborative analysis of cohort studies. framework for designing chronic services. BMC Public Health 2022; Lancet HIV 2023; 10: e295–307. 22: 1126. 3 Jaffar S, Ramaiya K, Karekezi C, et al. Controlling diabetes and 19 Mills KT, Bundy JD, Kelly TN, et al. Global disparities of hypertension in sub-Saharan Africa: lessons from HIV hypertension prevalence and control: a systematic analysis of programmes. Lancet 2021; 398: 1111–13. population-based studies from 90 countries. Circulation 2016; 4 Birungi J, Kivuyo S, Garrib A, et al. Integrating health services for 134: 441–50. HIV infection, diabetes and hypertension in sub-Saharan Africa: 20 Shayo E, Van Hout MC, Birungi J, et al. Ethical issues in a cohort study. BMJ Open 2021; 11: e053412. intervention studies on the prevention and management of diabetes 5 Kivuyo S, Birungi J, Okebe J, et al. Integrated management of HIV, and hypertension in sub-Saharan Africa. BMJ Glob Health 2020; diabetes, and hypertension in sub-Saharan Africa (INTE-AFRICA): 5: e002193. a pragmatic cluster-randomised, controlled trial. Lancet 2023; 21 Chowdhury SR, Chandra Das D, Sunna TC, Beyene J, Hossain A. 402: 1241–50. Global and regional prevalence of multimorbidity in the adult 6 Shiri T, Birungi J, Garrib AV, et al. Patient and health provider costs population in community settings: a systematic review and meta- of integrated HIV, diabetes and hypertension ambulatory health analysis. EClinicalMedicine 2023; 57: 101860. services in low-income settings—an empirical socio-economic 22 Price AJ, Jobe M, Sekitoleko I, et al. Epidemiology of cohort study in Tanzania and Uganda. BMC Med 2021; 19: 230. multimorbidity in low-income countries of sub-Saharan Africa: 7 Akugizibwe M, Zalwango F, Namulundu CM, et al. “After all, findings from four population cohorts. PLOS Glob Public Health we are all sick”: multi-stakeholder understanding of stigma 2023; 3: e0002677. associated with integrated management of HIV, diabetes and 23 Garrib A, Njim T, Adeyemi O, et al. Retention in care for type 2 hypertension at selected government clinics in Uganda. diabetes management in sub-Saharan Africa: a systematic review. BMC Health Serv Res 2023; 23: 20. Trop Med Int Health 2023; 28: 248–61. 8 Lippman SA, Pettifor A, Dufour MK, et al. A community 24 Bukenya D, Van Hout MC, Shayo EH, et al. Integrated healthcare mobilisation intervention to improve engagement in HIV testing, services for HIV, diabetes mellitus and hypertension in selected linkage to care, and retention in care in South Africa: a cluster- health facilities in Kampala and Wakiso districts, Uganda: randomised controlled trial. Lancet HIV 2022; 9: e617–26. a qualitative methods study. PLOS Glob Public Health 2022; 9 Mwai GW, Mburu G, Torpey K, Frost P, Ford N, Seeley J. Role and 2: e0000084. outcomes of community health workers in HIV care in 25 Fina Lubaki JP, Omole OB, Francis JM. Glycaemic control among sub-Saharan Africa: a systematic review. J Int AIDS Soc 2013; type 2 diabetes patients in sub-Saharan Africa from 2012 to 2022: 16: 18586. a systematic review and meta-analysis. Diabetol Metab Syndr 2022; 10 Decroo T, Rasschaert F, Telfer B, Remartinez D, Laga M, Ford N. 14: 134. Community-based antiretroviral therapy programs can overcome 26 Fonseca VA. Defining and characterizing the progression of type 2 barriers to retention of patients and decongest health services in diabetes. Diabetes Care 2009; 32: S151–56. sub-Saharan Africa: a systematic review. Int Health 2013; 5: 169–79. 27 Katte JC, Squires S, Dehayem MY, et al. Non-autoimmune, insulin- 11 Hanrahan CF, Schwartz SR, Mudavanhu M, et al. The impact of deficient diabetes in children and young adults in Africa: evidence community- versus clinic-based adherence clubs on loss from care from the young-onset diabetes in sub-Saharan Africa (YODA) cross- and viral suppression for antiretroviral therapy patients: findings sectional study. Lancet Diabetes Endocrinol 2025; 13: 745–53. from a pragmatic randomized controlled trial in South Africa. 28 Babar ZU, Ramzan S, El-Dahiyat F, Tachmazidis I, Adebisi A, PLoS Med 2019; 16: e1002808. Hasan SS. The availability, pricing, and affordability of essential 12 Jaffar S, Amuron B, Foster S, et al. Rates of virological failure in diabetes medicines in 17 low-, middle-, and high-income countries. patients treated in a home-based versus a facility-based HIV-care Front Pharmacol 2019; 10: 1375. model in Jinja, southeast Uganda: a cluster-randomised equivalence trial. Lancet 2009; 374: 2080–89. 13 Aikaeli F, Kasujja FX, Moyo F, et al. Integrated community-based management of HIV, diabetes, and hypertension in Tanzania and Uganda: protocol for a cluster-randomized trial. 2024. https:// openresearch.nihr.ac.uk/articles/4-37/pdf (accessed Feb 14, 2026). 1094 --- [PDF原文](https://sci-net.xyz/storage/7932541/41712da18220e958ff07153c79ab0f0590d59b13a7bb575afcc84532f4cee10e/Integrated-community-based-versus-facility-based-care-for-people-with-HIV-diabetes-and-hypertension.pdf) DOI: 10.1016/S0140-6736(25)02641-8