Optimizing Antihypertensive Care for Tanzanians Living with HIV: Effectiveness Outcomes from the COACH Pilot Trial

Abstract

Background

In Tanzania, many people living with HIV (PLWH) have uncontrolled hypertension, yet integration of hypertension management into HIV care remains limited. The Community Health Worker Optimization of Antihypertensive Care in HIV (COACH) intervention was developed to integrate standardized hypertension management into HIV clinic workflows.

Setting

A single-arm pilot trial was conducted from December 2024 to September 2025 at two public HIV clinics in Moshi, Tanzania.

Methods

COACH included six monthly community health worker (CHW)-delivered hypertension education sessions within HIV clinic visits, monthly blood pressure (BP) checks, care coordination, subsidized antihypertensive medications, a standardized treatment algorithm, and provider training. The primary effectiveness outcome was BP control at six months (<140/<90 mmHg). Secondary effectiveness outcomes included systolic and diastolic BP (SBP/DBP), hypertension knowledge (HK-LS score), medication adherence, BMI, and five-year cardiovascular risk. Paired t-tests and McNemar’s tests compared baseline and follow-up values.

Results

Of 100 participants, 96 completed follow-up. Six-month BP control was achieved in 73 (76%). Mean SBP decreased from 159.6 mmHg at baseline to 130.8 mmHg at follow-up (−28.9 mmHg, p<0.001); DBP decreased from 100.8 to 85.5 mmHg (−15.4 mmHg, p<0.001). HK-LS scores increased from 15.0 to 20.4 (p<0.001), and self-reported adherence improved from 14.0% to 95.0% (p<0.001). Mean BMI decreased from 27.0 to 26.6 kg/m² (p=0.019), and the proportion with ≥20% five-year cardiovascular risk declined from 33% to 7% (p<0.001).

Conclusions

COACH resulted in substantial improvements in BP control, hypertension knowledge, medication adherence, and cardiovascular risk among PLWH in Tanzania.

Introduction

People living with HIV (PLWH) face a 1.5- to 2-fold higher risk of cardiovascular disease (CVD) compared to the general population.¹ Over the past two decades, the global burden of CVD among PLWH has tripled, now accounting for 2.6 million disability-adjusted life-years annually.² The greatest attributable fraction of this burden is found in sub-Saharan Africa (SSA), the epicenter of the global HIV/AIDS epidemic where 25.6 million PLWH reside.³ One risk factor accelerating HIV-related CVD in SSA is hypertension - the leading risk factor for CVD globally. Estimates indicate that more than one in five PLWH in SSA has hypertension,⁴ attributed to factors like aging, inflammation, and cardiometabolic side effects of antiretroviral therapy (ART).⁵ However, hypertension remains underdiagnosed and undertreated among PLWH in SSA largely because of low patient awareness and siloed HIV and non-communicable disease (NCD) care systems.^6–9 In Tanzania, for example, our team found that only one-third of PLWH with hypertension were aware of their diagnosis and none had achieved blood pressure control.¹⁰

To address the growing burden of poorly-controlled hypertension among PLWH, we developed the Community Health Worker (CHW) Optimization of Antihypertensive Care in HIV (COACH) intervention in Tanzania.¹¹ The COACH intervention, originally adapted from a community-based hypertension intervention from southeast Asia,^12–16 was refined using intervention mapping and insights from our prior research in Tanzania.^{6–8,10,17–19} COACH comprises monthly hypertension education sessions and blood pressure monitoring delivered by CHWs within HIV clinics, coordinated follow-up care, and access to subsidized antihypertensive medications guided by an evidence-based, standardized treatment algorithm.²⁰ In this study, we report clinical effectiveness outcomes from a single-arm pilot trial of COACH implemented in two HIV clinics in northern Tanzania. We expect trial findings to provide evidence on the potential effectiveness and scalability of the COACH intervention across Tanzania and in similar resource-limited settings globally.

Methods

The COACH Intervention

Development and components of the COACH intervention have been published in detail elsewhere.^11,20 Briefly, an international team of healthcare providers, researchers, CHWs, and patients leveraged the ADAPT-ITT framework and a participatory co-design approach to adapt and integrate the COBRA¹² and CHAMP¹⁷ interventions into the COACH intervention. The COACH intervention includes five key components: (1) six monthly HIV clinic-based, one-on-one hypertension counseling sessions with CHWs, supported by hypertension educational handouts and follow-up calls; (2) blood pressure monitoring by CHWs at each HIV clinic visit, with referrals to HIV providers as needed for hypertension treatment; (3) specialized hypertension management training and a simplified treatment algorithm for HIV care providers; (4) hypertension care coordination by CHWs, including patient registries, clinic attendance tracking, and outreach to those missing visits; and (5) free on-site antihypertensive medications for participants and supplementary stipends for CHWs. The COACH hypertensive treatment algorithm was based on World Health Organization (WHO) guidelines,²¹ and included amlodipine, losartan, and hydrochlorothiazide (see Figure, Supplemental Digital Content 1 for full algorithm).²² The CHWs who delivered the COACH intervention were employed at HIV clinics to provide HIV-related counseling, and they received additional training in blood pressure measurement and in hypertension screening, diagnosis, complications and care to conduct the COACH counseling sessions. During hypertension counseling sessions with participants, CHWs provided education on the definition and symptoms of hypertension, emphasized the importance of daily medication adherence, addressed common misconceptions, and promoted lifestyle modifications including tobacco cessation, alcohol reduction, exercise, and weight loss.

Study Setting & Recruitment

The six-month pilot trial of COACH was conducted in the Kilimanjaro region of Tanzania, where our preliminary studies took place.^{6–8,10,17,18,23,24} The intervention was implemented at two government-funded HIV clinics, Majengo Care and Treatment Center and Pasua Care and Treatment Center, which provide HIV care services free of charge to a combined total of over 2300 PLWH. Routine hypertension care is not provided at these HIV clinics and is typically accessed at separate healthcare facilities. Like most HIV clinics in Tanzania, the COACH study sites are staffed by CHWs who provide HIV-related educational counseling to patients as part of routine HIV care.

From December 2024 to March 2025, trained research assistants screened patients (aged ≥18) attending HIV care appointments at the HIV clinic study sites. Those meeting eligibility criteria were given detailed study information, offered enrollment, and provided written informed consent. Recruitment continued until 50 participants were enrolled from each site, for a total of 100 participants.

Inclusion & Exclusion Criteria

Eligibility criteria included age ≥ 18 years old, living with HIV, and enrolled in care at one of the HIV clinic study sites. All patients were screened for hypertension during the recruitment period at these two sites. Hypertension eligibility was defined as follows: (1) systolic blood pressure (SBP) ≥ 160 mmHg and/or diastolic blood pressure (DBP) ≥ 100 mmHg; (2) SBP 140 to 159 mmHg and/or DBP 90 to 99 mmHg with previously documented elevated blood pressure (SBP ≥ 140 mmHg and/or DBP ≥ 90 mmHg); or (3) SBP 140 to 159 mmHg and/or DBP 90 to 99 mmHg without prior documentation of elevated blood pressure, in which case patients were asked to return for reassessment in 1–2 weeks. If blood pressure was still ≥ 140/90 mmHg at this return visit, they were eligible for enrollment.

In addition to meeting hypertension criteria, participants were required to be receiving care at one of the study sites for the subsequent six months and to have the ability to attend monthly clinic visits during this period. Pregnant individuals were excluded from study participation because they receive HIV care at separate, antenatal clinics in Tanzania, where antihypertensive regimens unsafe in pregnancy can be avoided and care can be more appropriately tailored. Individuals failing to provide written, informed consent were also excluded from this study.

Study procedures

During the initial study visit, research staff obtained baseline survey data on demographics, medical history, antihypertensive and HIV ART medication use and adherence, as well as participant hypertension knowledge via the Hypertension Knowledge Level Scale.²⁵ Research staff also measured patients’ height, weight, and waist circumference. Information on CD4 count and HIV viral load was gathered directly from participants’ medical records. Point-of-care glucose was also measured using a glucometer (GlucoPlus Blood Glucose Monitoring System, GlucoPlus, Montreal, Canada), with fasting and random blood glucose values determined based on participant self-report. At enrollment, trained research assistants obtained two baseline blood pressure measurements, five minutes apart, using the Omron M2 automated blood pressure cuff (Omron Healthcare, Kyoto, Japan). Blood pressure was measured with participants seated with arm at heart level and feet resting on the floor. Research assistants ensured that patients had rested and had not consumed caffeine for at least 30 minutes prior to blood pressure measurement. Enrolled participants then underwent the first component of the COACH intervention, including a CHW-led counseling session and referral to the HIV clinician for consideration of antihypertensive therapy. Importantly, COACH was implemented entirely by clinic staff and integrated into routine HIV care. The research team was not involved in the delivery of COACH; their role was limited to training CHWs and conducting fidelity assessments.

After the initial visit, participants were followed for six months, attending monthly clinic visits in the HIV clinic. Participants received a 5,000 Tanzanian shilling (approximately 2 USD) travel stipend per visit to offset transportation costs. At each monthly appointment, CHWs provided individualized counseling sessions following the COACH-specific hypertension curriculum. Blood pressure was measured by the CHW at every session, and all participants were referred to their HIV providers for potential antihypertensive treatment adjustments according to the COACH treatment algorithm. CHWs also tracked appointment attendance, contacted participants who missed visits to reschedule, and maintained detailed registries of participants, clinic attendance, referrals, and outreach activities. The research team used these registries to monitor study retention, identify participants lost to follow-up, and document reasons for withdrawal when applicable. Additionally, the research assistants collected blood pressure values and antihypertensive prescription data directly from participants’ medical records, which were recorded by HIV clinicians and CHWs during their monthly visits.

At the final six-month follow-up visit, research assistants administered a structured follow-up questionnaire to assess antihypertensive medication adherence, hypertension knowledge, and lifestyle behaviors. The surveys also included questions on what influenced their ability to adhere to their medication regimen. At the final follow-up visit, research assistants again obtained two blood pressure measurements and performed point-of-care glucose testing, using the same procedure as at baseline.

Study Definitions and Clinical Effectiveness Outcomes

Baseline and follow-up blood pressures were defined as the mean of the two blood pressure readings at baseline and six-month follow-up, respectively. We chose to define blood pressure by the mean of two readings in accordance with WHO and American Heart Association guidelines.^26,27 Moreover, two blood pressure measurements were used to reflect typical clinical practice in the Tanzanian HIV clinic setting. Baseline and follow-up body mass index (BMI) was calculated directly from measured weight and height. Baseline and follow-up cardiovascular risk was calculated via the NHANES risk score, an evidence-based tool incorporating demographic, clinical, and behavioral variables to estimate 5-year cardiovascular risk and has demonstrated strong predictive performance for cardiovascular events.^28,29 As a supplemental assessment, the 2019 WHO CVD Risk Charts for East Africa were also used to estimate baseline and follow-up 5-year cardiovascular risk.^30,31

Both effectiveness and implementation outcomes were measured in this pilot trial;¹¹ implementation outcomes will be reported separately. Effectiveness outcomes included the proportion of participants achieving a controlled blood pressure < 140/90 mmHg at six months, systolic and diastolic blood pressure, hypertension knowledge, self-reported antihypertensive medication adherence, five-year NHANES cardiovascular risk category, and BMI. Exploratory outcomes to inform future larger trials included waist circumference, tobacco use, alcohol consumption, and physical activity.³²

Physical activity was assessed based on participants’ self-reported daily minutes of exercise. In accordance with WHO recommendations that adults engage in at least 150 minutes of moderate-intensity physical activity per week, we derived two standardized exercise variables for analysis: currently exercising (yes/no) and meeting the ≥150 minutes per week threshold (equivalent to ≥21.4 minutes per day).³³

Medication adherence was measured using a shortened version of the Voils instrument,³⁴ a self-reported questionnaire assessing adherence to antihypertensive medications. We used the extent item to capture the number of missed antihypertensive doses over the past 7 days. For participants reporting any missed doses, we provided a context-adapted list of reasons for nonadherence such as cost, time constraints, and social support, excluding items that were not relevant in this setting. The Voils instrument has been used in multiple countries but has not been validated in Tanzania; there are currently no validated tools for assessing adherence to anti-hypertensives in Tanzania.

Hypertension knowledge was measured using the Hypertension Knowledge Level Scale (HK-LS), a 22-item survey encompassing domains such as causes, risk factors, complications, and lifestyle modifications related to hypertension.²⁵ Although there are currently no validated instruments for measuring hypertension knowledge in Tanzania, the HK-LS was selected for use in our study because it has been previously used elsewhere in SSA.^35–37 Each item is scored on as a binary 0 or 1, with 0 indicating incorrect or ‘I don’t know’ and 1 indicating a correct response. The total possible score ranges from 0 to 22, with higher scores indicating greater knowledge about hypertension. Informed by our prior work describing common misconceptions about hypertension in Tanzania,^7,8,18,22 four additional supplemental hypertension knowledge questions were added. These items asked about hypertension symptoms, whether worrying was a major cause of hypertension, and whether garlic or water were effective treatments for hypertension. In our prior qualitative work informed by a phenomenology theoretical framework, these four misconceptions were identified as major barriers to hypertension medication adherence among Tanzanians.^7,8,18 These supplemental knowledge questions were not included in the HK-LS score. Although there is some evidence that garlic may have modest effects on blood pressure,³⁸ a recent Cochrane review found that insufficient evidence to recommend garlic as treatment for hypertension,³⁹ and international guidelines do not recommend garlic as a substitute for conventional anti-hypertensive medications.⁴⁰ Our prior formative qualitative work found that many Tanzanians do not take anti-hypertensive medications because they believe that cooking with garlic is an acceptable substitute for evidence-based therapies. Thus, participants were asked about their perceptions of garlic as treatment for hypertension at baseline and at six-month follow-up.

Data Analysis

All statistical analyses were conducted using the R Suite. Baseline participant characteristics were described using summary statistics: continuous variables were reported with means and standard deviations, and categorical variables were reported as counts and percentages. To compare differences in primary and secondary effectiveness outcomes at baseline and follow-up, paired t-tests were used for continuous variables and McNemar’s test was used for dichotomous categorical variables. The Wilcoxon signed-rank test was used for all ordinal variables, including cardiovascular risk category and alcohol use frequency, to assess paired changes between baseline and six-month follow-up. Patients lost to follow-up were excluded from these analyses. As a supplemental exploratory analysis, multivariable logistic regression was used to identify baseline factors associated with blood pressure control at six-month follow-up, with results presented as odds ratios and 95% confidence intervals. Patients lost to follow-up (n=4) were excluded from these analyses. All remaining participants (n=96) had complete data for included variables.

Ethics & Data Availability

This study was reviewed and approved by institutional review boards (IRBs) at Duke Health (PRO00090902 approved on July 22^nd, 2024), NIMR (NIMR/HQ/R.8a/Volpi/4615, approved on May 21^st, 2024), and KCMC (Proposal 1454, approved on February 12^th, 2024). All participants provided written, informed consent prior to their enrollment. Data from this study are available from the corresponding author on reasonable request.

Results

During study recruitment at two HIV clinics in Moshi, Tanzania, 958 adult patients were screened, of whom 102 (10.6%) had uncontrolled hypertension and met eligibility criteria for participation (see Figure, Supplemental Digital Content 2, which presents the participant flow from screening through follow-up). Two eligible patients (2%) declined participation, and 100 (98%) eligible patients consented to participation and were enrolled in the COACH study. Among those enrolled, 96 (96%) completed 6-month follow-up, while two participants (2%) withdrew and two (2%) were lost to follow-up.

Among the 100 participants enrolled in the COACH study, the mean (SD) age was 53.4 (9.2) years, and 77 (77.0%) were female (Table 1). One-fifth of participants had greater than primary school education and most lacked health insurance (91.0%). The baseline mean (SD) BMI was 26.9 (±5.4) kg/m² and mean (SD) waist circumference 95.0 (±13.7) cm. Participants reported living with HIV for a median of 9.5 years (IQR 5–16). Immunologic status was reflected by most recently available CD4 measurement, with median value of 406 cells/mm³ (IQR 224.5–584.8). Most (88%) had undetectable viral load. Despite uncontrolled hypertension at enrollment, only 72% of participants reported a previous diagnosis of hypertension and only 30% reported taking antihypertensive medication at baseline. Additional self-reported comorbidities included prior stroke (10%), diabetes (6%), and hyperlipidemia (3%). Most engaged in some physical activity (88%), and few used tobacco (2%).

Table 1.

Baseline characteristics of hypertensive participants attending 2 HIV treatment centers and enrolled in the COACH study, northern Tanzania, 2024–2025

Participant Characteristics	n = 100
Demographics
Age (years), mean (SD)	53.4 (9.2)
Sex (female), n (%)	77 (77%)
Education
Primary school or less, n (%)	80 (80%)
Secondary school or higher, n (%)	20 (20%)
Household income per month (USD), median (IQR)	40.8 (20.4–81.6)
Health Insurance
National Health Insurance Fund (NHIF), n (%)	9 (9%)
None, n (%)	91 (91%)
Anthropometrics
BMI (kg/m2), mean (SD)	26.9 (5.4)
Waist circumference (cm), mean (SD)
Female	95.6 (13.7)
Male	92.9 (13.6)
HIV-related characteristics
HIV duration (years), median (IQR)	9.5 (5–16)
ART duration (years), median (IQR)	8 (5–13)
Current ART
TDF, 3TC, DTG, n (%)	96 (96%)
Other, n (%)	4 (4%)
CD4 current (cells/mm3), median (IQR)	406 (225–585)
HIV RNA level (copies/mL)
Undetectable (≤40), n (%)	88 (88%)
40 – 200, n (%)	11 (11%)
>200, n (%)	1 (1%)
Hypertension & risk factors
Systolic BP average (mmHg), mean (SD)	159.8 (17.9)
Diastolic BP average (mmHg), mean (SD)	100.9 (9.7)
Known history of hypertension (self-reported), n (%)	72 (72%)
Currently taking blood pressure medication, n (%)	30 (30%)
Current BP medication
Amlodipine, n (%)	18 (18%)
Losartan, n (%)	11 (11%)
Nifedipine, n (%)	10 (10%)
Thiazide diuretic, n (%)	4 (4%)
Other, n (%)	5 (5%)
Alcohol use
Never, n (%)	45 (45%)
1–4 per month, n (%)	21 (21%)
2–3 drinks per week, n (%)	14 (14%)
4+ drinks per week, n (%)	20 (20%)
Smoking or tobacco use (current), n (%)	2 (2%)
Currently exercises (yes/no), n (%)	77 (77%)
Exercises ≥150 minutes per week, n (%)	64 (64%)
Comorbidities (self-reported)
Stroke, n (%)	10 (10%)
Diabetes, n (%)	6 (6%)
Hyperlipidemia, n (%)	3 (3%)
Heart disease, n (%)	2 (2%)
Kidney disease, n (%)	1 (1%)
5-Year Cardiovascular risk
Very Low (<5%), n (%)	15 (15%)
Low (5–10%), n (%)	18 (18%)
Moderate (>10–20%), n (%)	33 (33%)
High (>20–30%), n (%)	20 (20%)
Very High (>30%), n (%)	14 (14%)

HIV: Human immunodeficiency virus; ART: Antiretroviral therapy; TDF: tenofovir disoproxil fumarate; 3TC: lamivudine; DTG: dolutegravir; BP: Blood pressure

Table 2 summarizes participants’ baseline hypertension knowledge and medication adherence. The mean HK-LS score was 15.0 (SD 4.9) out of 22. Detailed item-level responses are provided in Supplemental Digital Content 3. Only one-third (33%) correctly recognized that elevated diastolic pressure indicates hypertension, and about half answered correctly regarding the need for daily medication use (52%) and lifelong treatment (51%). Among the supplemental questions added by the research team, 55% believed that everyone with hypertension has symptoms, 77% attributed the most common cause of hypertension to be due to “thinking too much,” and many endorsed home remedies such as garlic (54%) and drinking large amounts of water (75%) as effective treatments. Among the 30 participants taking antihypertensive medication at baseline, the median number of missed days per week was 1.5 [IQR 0–4]. Over half (57%) reported missing at least one dose in the past week, with cost (37%) and running out of medication (30%) being the most common reasons for missed doses.

Table 2.

Baseline hypertension knowledge and medication adherence of hypertensive participants attending 2 HIV treatment centers and enrolled in the COACH study, northern Tanzania, 2024–2025

Hypertension Knowledge-Level Scale (HK-LS) (n = 100)
HK-LS total score, mean (SD)	15.0 (4.9)
Supplemental Extra Knowledge Questions (n = 100)
Question	Yes, n (%)	No, n (%)	Don’t know, n (%)	Correct Answer	Correct, n (%)
Everyone with high blood pressure has symptoms of high blood pressure.	55 (55)	25 (25)	22 (22)	No	25 (25)
The most common cause of high blood pressure is having many thoughts or worries.	77 (77)	5 (5)	18 (18)	No	5 (5)
Garlic is a very excellent treatment for high blood pressure.	54 (54)	12 (12)	34 (34)	No	12 (12)
Drinking lots of water is an effective treatment for high blood pressure.	75 (75)	7 (7)	18 (18)	No	7 (7)
Baseline Medication Adherence			n = 30a
Self-reported days missed taking BP medication in a typical week, median [IQR]			1.5 [0–4]
Self-reported days missed taking ART medication in a typical week, median [IQR]			0 [0–0]
Participant agreement with “I missed or skipped at least one dose of my blood pressure medication in the past 7 days”
Strongly agree, n (%)			16 (53%)
Agree, n (%)			1 (3%)
Neutral, n (%)			0 (0%)
Disagree, n (%)			0 (0%)
Strongly disagree, n (%)			13 (43%)
Reasons contributing to missed BP dose(s), past 7 days
Cost of medication			11 (37%)
Ran out of medication			9 (30%)
Side Effects			2 (7%)
Fear of dependence on medication			2 (7%)
No symptoms of high BP			1 (3%)
Fear BP medication would harm body			1 (3%)
Forgot to take medication			1 (3%)
Feeling too ill to take medication			1 (3%)
Fear BP medication would interact with other medication			1 (3%)
Other			3 (10%)
Main reason contributing to missed BP dose(s), past 7 days
Cost of medication			5 (17%)
Ran out of medication			4 (13%)
Side Effects			3 (10%)
Fear of dependence on medication			1 (3%)
No symptoms of high BP			1 (3%)
Feeling too ill to take medication			1 (3%)
Other			2 (7%)

^aParticipants taking anti-hypertensive medication at baseline

Over the course of the six-month COACH intervention, four participants (4%) did not complete the study, including 2 who withdrew consent and 2 who were lost to follow-up. Ninety-six (96%) of enrolled participants completed the six-month intervention. Overall, blood pressure control (SBP<140 mmHg and DBP<90 mmHg), was achieved in 73 of 96 participants (76%, 95% CI 66–84%) at six-month follow-up. Figure 1 presents the SBPs and DBPs of the 96 participants who completed the COACH intervention at each of their monthly clinic visits. Supplemental Digital Content 4 presents the proportion of participants achieving blood pressure control at each monthly visit. To account for potential attrition bias, we conducted a sensitivity analysis treating participants missing follow-up data as not achieving BP control (missing=failure). Results of this sensitivity analysis are shown in Supplemental Digital Content 5.

Figure 1.

Systolic (A) and diastolic (B) blood pressures of COACH participants across the six-month intervention, northern Tanzania (N=96).

Table 3 summarizes clinical and behavioral outcomes at baseline and six-month follow-up, among the 96 participants who completed the six-month follow-up. Mean (SD) SBP decreased from 159.6 (17.4) mmHg to 130.8 (14.5) mmHg, with a mean reduction of 28.9 (95% CI −33.0, −24.8) mmHg (p < 0.001). Similarly, mean (SD) DBP decreased from 100.8 (9.9) mmHg to 85.5 (8.7) mmHg, with a mean reduction of 15.4 (95% CI −17.6, −13.2) mmHg (p < 0.001). Mean (SD) BMI decreased from 27.0 (5.2) kg/m² to 26.6 (5.1) kg/m² (p = 0.019). When stratified by gender, waist circumference did not significantly change in both female and male participants.

Table 3.

Clinical and behavioral outcomes at baseline and six-month follow-up among participants in the COACH intervention, northern Tanzania, 2024–2025

Outcome Variable (n=96)	Baseline, mean (SD) or n (%)	Six-month follow-up, mean (SD) or n (%)	Mean change (95 % CI) or Δ % points	p-value
Systolic BP (mmHg)	159.6 (17.4)	130.8 (14.5)	−28.9 (−33.0, −24.8)	<0.001*
Diastolic BP (mmHg)	100.8 (9.9)	85.5 (8.7)	−15.4 (−17.6, −13.2)	<0.001*
HK-LS total score (0–22)a	15.0 (4.9)	20.4 (2.0)	+5.4 (4.4, 6.5)	<0.001*
BMI (kg/m2)	27.0 (5.2)	26.6 (5.1)	−0.4 (−0.7, −0.1)	0.019*
Waist circumference (cm) by gender
Female	95.9 (13.4)	95.8 (12.4)	−0.2 (−2.3, 2.0)	0.872
Male	92.9 (13.6)	92.4 (10.1)	−0.5 (−3.4, 2.3)	0.722
Hypertension risk factors
Alcohol consumption
Never, n (%)	43 (45%)	55 (57%)
<1 occasions per month, n (%)	8 (8%)	9 (9%)
2–4 occasions per month, n (%)	11 (11%)	9 (9%)
2–3 occasions per week, n (%)	14 (15%)	17 (18%)
4+ occasions per week, n (%)	20 (21%)	6 (6%)
Overall shiftb				<0.001*
Smoking or tobacco use (current), n (%)	2 (2%)	1 (1%)	−1% (−3%, 1%)	0.999
Currently exercises (yes/no), n (%)	75 (78%)	91 (95%)	+17% (12%, 19%)	<0.001*
Exercises ≥150 minutes per week, n (%)	62 (65%)	66 (69%)	+4% (−3%, 10%)	0.626
# with >20% five-year cardiovascular riskc	32 (33%)	7 (7%)	−26% (−35%, −17%)	<0.001*
5-year cardiovascular risk category
Very Low (<5%)	12 (12%)	15 (16%)
Low (5–10%)	19 (20%)	25 (26%)
Moderate (>10–20%)	33 (34%)	49 (51%)
High (>20–30%)	19 (20%)	7 (7%)
Very High (>30%)	13 (14%)	0 (0%)
Overall shiftb				<0.001*
Supplemental knowledge questions
# identifying not everyone with hypertension has symptoms	24 (25%)	66 (69%)	+44% (31%, 56%)	<0.001*
# identifying worrying is not the main cause of hypertension	5 (5%)	83 (86%)	+81% (73%, 89%)	<0.001*
# identifying garlic as not an excellent treatment for hypertension	11(11%)	89 (93%)	+81% (73%, 89%)	<0.001*
# identifying water not effective for treating hypertension	7 (7%)	65 (68%)	+60% (51%, 70%)	<0.001*
Medication adherence d
# Agreeing or Strongly Agreeing that they took all BP medication doses in the past 7 days	13 (14%)	91 (95%)	+81% (73%, 89%)	<0.001*

BP: Blood pressure; HK-LS: Hypertension Knowledge-Level Scale; BMI: Body Mass Index

^aHigher scores indicate greater hypertension knowledge.

^bp-value derived from the Wilcoxon signed-rank test comparing paired baseline and follow-up risk categories.

^cCalculated using the NHANES risk score.

^dAdherence defined as participants who disagreed or strongly disagreed with the statement “I missed or skipped at least one dose of my blood pressure medication in the past 7 days.” Participants not taking antihypertensive medication at baseline were classified as non-adherent for this measure.

^*p < 0.05 considered statistically significant

Mean (SD) HK-LS scores increased from 15.0 (4.9) to 20.4 (2.0) (p < 0.001). Improvements were also observed across all supplemental knowledge items: the proportion correctly identifying that not everyone with hypertension has symptoms increased from 25% to 69%, that worrying is not a main cause from 5% to 86%, and that neither garlic (11% to 93%) nor drinking large amounts of water (7% to 68%) are effective treatments (p < 0.001 for all). Medication adherence improved substantially: the proportion of participants agreeing or strongly agreeing that they took all their antihypertensive doses in the past seven days increased from 14% at baseline to 95% at follow-up (p < 0.001).

These gains in hypertension knowledge were accompanied by favorable shifts in behavioral risk factors over the six-month period. Alcohol consumption shifted toward lower-risk categories (p<0.001), with the proportion reporting ≥4 drinking occasions per week decreasing from 21% to 6% and those reporting no alcohol use increasing from 45% to 57%. Tobacco use remained infrequent and unchanged (2% to 1%, p=0.999). The proportion of participants reporting any daily exercise increased from 78% to 95% (p<0.001), while the proportion meeting guideline-recommended ≥150 minutes per week remained similar (65% vs 69%; p=0.626).

The distribution of five-year cardiovascular risk shifted markedly toward lower categories over six months. The proportion of participants with a risk greater than 20% declined from 32 (33%) at baseline to 7 (7%) at follow-up, reflecting a 26% reduction (p < 0.001). No participants remained in the very high (>30%) risk category at follow-up. Most were classified as moderate risk (51%), followed by low (26%), very low (16%), and high (7%) risk. The overall distribution shifted significantly toward lower risk (Wilcoxon signed-rank test, p < 0.001). The proportions of participants classified as very high (>30%) and high (20–30%) risk decreased from 14% and 20% at baseline to 0% and 7% at follow-up, respectively, while those in the moderate (>10–20%) and low (5–10%) risk categories increased from 34% to 51% and from 17% to 26%. The transition matrix (see Supplemental Digital Content 6) indicates that nearly all participants in the highest baseline categories moved to lower risk categories, including nine from >30% to 10–20% and four to 20–30%. Two participants (2.1%) experienced upward shifts in cardiovascular risk category, while the remainder demonstrated improvement (41.7%) or no change (56.2%). Both participants with increased cardiovascular risk had lower BMI and blood pressure at follow-up but were reclassified upward due to new diabetes diagnoses from point-of-care glucose testing at their six-month follow-up visit. Supplemental assessment using the 2019 WHO CVD Risk Charts (see Supplemental Digital Content 7) yielded similar baseline and follow-up risk profiles and likewise demonstrated a significant shift toward lower risk from baseline to follow-up (Wilcoxon signed-rank test, p < 0.001).

Table 4 summarizes follow-up medication adherence outcomes. Self-reported medication adherence was excellent, with a median of 0 [IQR 0–0] missed antihypertensive doses in a typical week. On the adherence item “I missed or skipped at least one dose of my blood pressure medication in the past seven days,” 93% strongly disagreed, 2% disagreed, and 5% strongly agreed. Reported reasons for missed doses were infrequent and included running out of medication (2%), travel (2%), or being in a public place (1%).

Table 4.

Medication adherence among hypertensive participants attending 2 HIV treatment centers following the COACH intervention, northern Tanzania, 2024–2025

Follow-Up Medication Adherence
Self-reported days missed taking BP medication in a typical week, median [IQR]	0 [0–0]
Participant agreement with “I missed or skipped at least one dose of my blood pressure medication in the past 7 days”
Strongly agree, n (%)	5 (5)
Agree, n (%)	0 (0)
Neutral, n (%)	0 (0)
Disagree, n (%)	2 (2)
Strongly disagree, n (%)	89 (93)
Reasons contributing to missed BP dose(s), past 7 days, n (%)
Ran out of medication	2 (2)
Was traveling	2 (2)
Was in a public place	1 (1)
Other	1 (1)
Main reason contributing to missed BP dose(s), past 7 days, n (%)
Ran out of medication	2 (2)
Was traveling	2 (2)
Was in a public place	1 (1)
Other	1 (1)

Supplemental Digital Content 8 summarizes baseline factors associated with blood pressure control at six-month follow-up. Older age was independently associated with greater odds of achieving control (OR 1.09, 95% CI 1.02–1.18, p = 0.019), corresponding to roughly a 2.3-fold increase in odds for every 10-year increase in age. Higher baseline systolic blood pressure was associated with lower odds of control (OR 0.97, 95% CI 0.94–1.00, p = 0.040), equivalent to about a 26% decrease in odds for every 10 mmHg higher baseline SBP. Gender, baseline BMI, hypertension knowledge score, and diabetes status were not significantly associated with blood pressure control.

Discussion

To our knowledge, this is among the first multicomponent, clinic-integrated, task sharing hypertension interventions for PLWH in SSA to demonstrate significant improvements across multiple domains of hypertension care. In this single-arm pilot conducted in two public HIV clinics in northern Tanzania, mean SBP and DBP decreased by approximately 29 and 15 mmHg, respectively, over six months. These reductions were accompanied by improvements in hypertension knowledge, self-reported medication adherence, and overall cardiovascular-risk profile. Retention was high, with 96% of participants completing follow-up. Together, these findings suggest that a multicomponent, task-sharing approach combining CHW counseling, point-of-care blood pressure monitoring, simplified treatment algorithms, and consistent medication access all within the HIV clinic can strengthen hypertension control for PLWH receiving routine HIV care.

The COACH intervention was adapted from COBRA, a multicomponent hypertension program in Southeast Asia that combined task sharing, protocolized treatment, and patient education.^12,14 In the cluster-randomized COBRA-BPS trial conducted in Bangladesh, Pakistan, and Sri Lanka, the intervention reduced SBP and DBP by 9 mmHg and 3 mmHg, respectively, over 24 months compared with usual care. In contrast, the present COACH pilot study, which used a single-arm pre-post design, observed substantially larger reductions in SBP and DBP over six months. Differences in study design and context likely explain these findings: COACH was implemented within two HIV clinics among participants already engaged in longitudinal care and medication adherence programs, whereas COBRA was delivered in community settings to individuals with more variable health-system contact. Importantly, the controlled design of COBRA provides a more conservative—and externally valid—estimate of intervention effect, while the uncontrolled, pilot nature of COACH may magnify apparent changes due to regression to the mean, higher baseline blood pressures (mean SBP/DBP approximately 160/101 mmHg), and closer clinical follow-up.

Similarly, a CHW-led trial in Nepal tested a hypertension program involving home visits and provider follow-up and demonstrated more modest effects than COACH.⁴¹ After twelve months, SBP and DBP decreased by 4 and 3 mmHg, respectively, with small gains in hypertension knowledge and no measurable change in BMI or medication adherence.⁴¹ These findings suggest that CHW counseling delivered outside of structured, clinic-based management may have limited effect. In a high-income setting, a multicomponent intervention in Singapore combined physician training, subsidized antihypertensive therapy, nurse-delivered motivational counseling, and telephone follow-up, achieving a 13 mmHg reduction in SBP and a 2.3% decline in 10-year cardiovascular risk over 24 months.⁴² Despite these resource-intensive supports, nearly half of participants remained at high cardiovascular risk, highlighting the magnitude of improvement achieved by COACH under routine-care conditions.⁴²

Prior studies of CHW-led hypertension management have used both community-based models, such as COBRA in Southeast Asia,^12,14 and clinic-based models, such as the multicomponent intervention implemented in primary health clinics in Singapore.⁴² We prioritized a clinic-based approach in COACH because our target population was PLWH, and in Tanzania, HIV care services provide a robust infrastructure, including appointment-tracking systems, medication supply chains, and CHWs trained in routine HIV services.⁴³ Integrating hypertension screening, counseling, and treatment directly into HIV clinic workflows therefore allowed us to capitalize on this existing platform, strengthen linkage between HIV and NCD care, and enhance the feasibility, cost-effectiveness, and potential scalability of the intervention in Tanzania and elsewhere in SSA.^44,45 Moreover, our prior formative work in Tanzania found that both patients and providers preferred clinic-based hypertension care.^7,8,17

In SSA, several integrated HIV-hypertension programs using multicomponent, task-sharing approaches have similarly improved blood pressure control and retention in care. Programs most comparable to COACH, single-arm or early-implementation pilots embedded within HIV clinics, include the InterCARE pilot in Botswana, which increased hypertension control from approximately 47% to 71% over twelve months,⁴⁶ and the Lighthouse HIV clinic study in Malawi, which reported that 38% of participants with mild hypertension and 30% with moderate hypertension achieved blood pressure control at six months.⁴⁷ A facility-based HEARTS-aligned program in Uganda similarly reported large gains, increasing control from roughly 9% to 74% at six months, although participants received intensified follow-up that may not reflect routine practice.⁴⁸

In our pilot, COACH achieved mean reductions of 29 and 15 mmHg in SBP and DBP, respectively, with 76% achieving control over six months. However, differences in study design, baseline blood pressure, follow-up intensity, and patient populations limit direct comparison of relative effectiveness across interventions. We note that while several prior SSA programs incorporated counseling or patient-education components, few were embedded within existing clinic workflows or assessed hypertension knowledge or medication adherence. COACH extends this evidence by embedding CHW-led counseling, structured education, and subsidized antihypertensive medicines within existing HIV care, demonstrating improvement in blood pressure control together with concurrent gains in knowledge, adherence, BMI, and cardiovascular-risk profile under real-world conditions. Within this context, the improvements observed in COACH are encouraging but underscore the need for larger, controlled evaluations to rigorously assess effectiveness and generalizability.

Several features of COACH likely contributed to these effectiveness outcomes. Integration within HIV clinics leveraged established infrastructure and frequent patient contact, minimizing loss to follow-up between HIV and NCD services. In addition, the CHW-led counseling curriculum, adapted from prior formative research,^7,8,11,18,19 targeted locally prevalent misconceptions about hypertension symptoms, causation, and home remedies—misconceptions previously identified as key barriers to awareness and adherence. Following COACH, we observed large reductions in endorsement of these misconceptions, as reflected in the supplemental knowledge questions, indicating that the curriculum effectively addressed key cognitive barriers. These sessions also covered the importance of lifestyle modification, including tobacco cessation, alcohol reduction, increased physical activity, and weight management. Reinforcing these messages during monthly HIV visits likely supported gradual behavior change and may help explain the substantial decline in higher-frequency alcohol use and the increased proportion of participants reporting regular exercise at follow-up. This combination of clarified beliefs and behavior-focused counseling likely supported more consistent engagement in heart-healthy behaviors alongside improvements in hypertension knowledge and medication adherence. Importantly, provision of free, on-site antihypertensive medication guided by a simplified, WHO-supported algorithm²¹ reduced cost and access barriers described by patients and providers. In light of our prior formative work demonstrating gaps in HIV clinician training in evidence-based antihypertensive management,^7,8,11,18,19 the combination of a standardized treatment algorithm and targeted provider training likely contributed to the large BP reduction observed.

COACH was implemented entirely by HIV-clinic staff within routine workflows, suggesting feasibility for integration into Tanzania’s public-sector HIV program. Its components align closely with the WHO HEARTS technical package²¹ and the Tanzanian National NCD Strategic Plan,⁴⁹ both of which emphasize standardized treatment, task sharing, and reliable medicine supply. Integrating hypertension management into HIV platforms could substantially expand NCD care in SSA without major new resource requirements, although additional cost-effectiveness analyses would be needed to inform this work. COACH thus represents a pragmatic, potentially scalable model for strengthening chronic disease management within existing health systems. Indeed, if the reductions in blood pressure and cardiovascular risk observed in this pilot were achieved at a national scale, COACH could prevent a large number of cardiovascular events among PLWH in Tanzania.

Given the high cardiovascular risk among PLWH and the strong association between excess body weight and incident hypertension,¹ COACH’s structured counseling on lifestyle modification and weight management within routine HIV visits may yield additional preventive benefit. Indeed, we observed a small but statistically significant reduction in BMI following the COACH intervention. If this finding is confirmed in a larger controlled trial, COACH would be one of the first interventions in SSA to improve both blood pressure control and weight among PLWH.

Older age was associated with higher odds of achieving blood pressure control at the six-month follow-up. Although this finding should be interpreted cautiously as this pilot was not powered to examine age-stratified outcomes in detail, it does suggest important age-related trends in hypertension management among PLWH in SSA. While older age is an independent risk factor for hypertension in both PLWH and the general population,^50,51 prior work from SSA has shown that older PLWH often have better retention in HIV care and higher ART adherence compared to younger adults.^52,53 Meanwhile, younger PLWH may face greater challenges in accessing and engaging with care for both HIV and hypertension, including economic constraints, competing work and family responsibilities, and lower health literacy.^54,55 These factors may help explain why older participants in COACH were more likely to achieve blood pressure control, but more research is needed to fully elucidate age differences in hypertension management among PLWH.

This study had several strengths. The intervention was fully integrated within routine HIV care and implemented entirely by clinic-employed CHWs and HIV clinicians, demonstrating feasibility in real-world conditions and strengthening external validity. Retention and data completeness were high, and blood pressure was measured using standardized protocols and calibrated automated devices to reduce observer bias. Effectiveness outcomes captured patient knowledge, medication adherence, physiologic measures, and a validated cardiovascular risk score, allowing assessment across multiple domains of hypertension care.

Several limitations should also be noted in this study. First, as a single-arm, pilot study with a small sample size, this analysis was not powered to detect modest effects or subgroup differences, limiting the precision of effect estimates. Second, the study was conducted in two urban HIV clinics, which may reduce generalizability to other regions of Tanzania or SSA, particularly more rural settings. Third, as with any longitudinal pre-post study without a control group, unmeasured temporal or contextual factors may have contributed to the observed improvements. Although no major secular changes in hypertension management were identified during the study period, background trends independent of the COACH intervention, including regression to the mean, cannot be excluded. Hence, the pre-post changes observed in blood pressure may overstate the true effect of the COACH intervention relative to what would be seen in a randomized trial. Fourth, the six-month follow-up limits conclusions about the durability of blood pressure control, adherence, and cardiovascular-risk reduction. Fifth, adherence was assessed by self-report and may have been influenced by recall or social-desirability bias, though the magnitude and consistency of improvement suggest a true behavioral change. Sixth, Tanzanian standard treatment guidelines recommend that hypertension be defined by three readings;³¹ our study used the mean of two readings, which may have led to slightly less accurate blood pressure determinations. Seventh, only 10.6% of adults screened at the two HIV clinics met eligibility for uncontrolled hypertension, a proportion lower than prior estimates from the same region.^6,10 Differences in defining uncontrolled hypertension, exclusion of patients with controlled hypertension, and prior hypertension-focused activities at these sites may partly explain this discrepancy and may indicate possible selection bias. Eighth, the HK-LS instrument was initially developed in a Turkish population and has not been validated in Tanzania; therefore its validity in our setting is unknown. Unfortunately, there are currently no validated instruments for measuring hypertension knowledge in Tanzania. Similarly, the Voils instrument, which was used to explore reasons for non-adherence, has not been validated in Tanzania, and its psychometric properties in our setting is unknown. Finally, the NHANES cardiovascular-risk equation has not been calibrated for Tanzanian PLWH, so absolute risk estimates should be interpreted cautiously even though relative changes remain informative.

A larger, multi-site pragmatic trial will be needed to confirm the effectiveness of COACH and assess its scalability across clinical settings. Cluster randomization would allow evaluation of both patient- and system-level effects, and longer follow-up would clarify the durability of blood pressure control. Economic evaluation will also be important to assess cost-effectiveness. Qualitative inquiry with HIV clinicians, CHWs, and facility administrators will be essential to identify barriers and facilitators to integrated cardiovascular-risk management and to refine supervision, workflow, and counseling content prior to scale-up. Finally, long-term follow-up assessing cardiovascular outcomes and program adoption across facility types will help guide national strategies for integrating NCD management into HIV platforms.

Supplementary Material

Supplemental Digital Content 3

Item-level baseline responses to the Hypertension Knowledge-Level Scale (HK-LS) among participants in the COACH study, northern Tanzania, 2024–2025.

Supplemental Digital Content 2

Flow of COACH study participants from screening through 6-month follow-up.

Supplemental Digital Content 1

COACH hypertension management protocol for HIV providers.

Supplemental Digital Content 4

Percentage of COACH participants with controlled blood pressure across the six-month intervention, northern Tanzania (N = 96).

Supplemental Digital Content 5

Percentage of COACH participants with controlled blood pressure across the six-month intervention, assuming participants lost to follow-up did not achieve blood pressure control (N = 100).

Supplemental Digital Content 6

Transition matrix comparing five-year cardiovascular risk among participants with HIV and hypertension at baseline and at six-month follow-up.

Supplemental Digital Content 7

5-year WHO CVD risk at baseline and six-month follow-up among participants in the COACH intervention, northern Tanzania, 2024–2025 (N=96).

Supplemental Digital Content 8

Baseline factors associated with blood pressure control at six-month follow-up among COACH participants, northern Tanzania, 2024–2025 (N = 96).