Integration of a multicomponent intervention for hypertension into primary healthcare services in Singapore-A cluster randomized controlled trial.

BACKGROUND: Despite availability of clinical practice guidelines for hypertension management, blood pressure (BP) control remains sub-optimal (<30%) even in high-income countries. This study aims to assess the effectiveness of a potentially scalable multicomponent intervention integrated into primary care system compared to usual care on BP control. METHODS AND
FINDINGS: A cluster-randomized controlled trial was conducted in 8 government clinics in Singapore. The trial enrolled 916 patients aged ≥40 years with uncontrolled hypertension (systolic BP (SBP) ≥140 mmHg or diastolic BP (DBP) ≥90 mmHg). Multicomponent intervention consisted of physician training in risk-based treatment of hypertension, subsidized losartan-HCTZ single-pill combination (SPC) medications, nurse training in motivational conversations (MCs), and telephone follow-ups. Usual care (controls) comprised of routine care in the clinics, no MC or telephone follow-ups, and no subsidy on SPCs. The primary outcome was mean SBP at 24 months' post-baseline. Four clinics (447 patients) were randomized to intervention and 4 (469) to usual care. Patient enrolment commenced in January 2017, and follow-up was during December 2018 to September 2020. Analysis used intention-to-treat principles. The primary outcome was SBP at 24 months. BP at baseline, 12 and 24 months was modeled at the patient level in a likelihood-based, linear mixed model repeated measures analysis with treatment group, follow-up, treatment group × follow-up interaction as fixed effects, and random cluster (clinic) effects. A total of 766 (83.6%) patients completed 2-year follow-up. A total of 63 (14.1%) and 87 (18.6%) patients in intervention and in usual care, respectively, were lost to follow-up. At 24 months, the adjusted mean SBP was significantly lower in the intervention group compared to usual care (-3.3 mmHg; 95% CI: -6.34, -0.32; p = 0.03). The intervention led to higher BP control (odds ratio 1.51; 95% CI: 1.10, 2.09; p = 0.01), lower odds of high (>20%) 10-year cardiovascular risk score (OR 0.67; 95% CI: 0.47, 0.97; p = 0.03), and lower mean log albuminuria (-0.22; 95% CI: -0.41, -0.02; p = 0.03). Mean DBP, mortality rates, and serious adverse events including hospitalizations were not different between groups. The main limitation was no masking in the trial.
CONCLUSIONS: A multicomponent intervention consisting of physicians trained in risk-based treatment, subsidized SPC medications, nurse-delivered motivational conversation, and telephone follow-ups improved BP control and lowered cardiovascular risk. Wide-scale implementation of a multicomponent intervention such as the one in our trial is likely to reduce hypertension-related morbidity and mortality globally. TRIAL REGISTRATION: Trial Registration: Clinicaltrials.gov NCT02972619.

Introduction

High blood pressure (BP) is the leading risk factor for mortality globally, accounting for 19% (10.8 million) of all deaths [1]. Despite numerous clinical practice guidelines for hypertension management, implementation remains inadequate with BP control less than 30% even in high-income countries [2,3].

Several barriers to hypertension care have been identified at multiple levels worldwide [4,5]. The significant patient-level challenges include health illiteracy about hypertension and related cardiovascular disease (CVD), lack of motivation to adopt a healthy lifestyle, poor medication adherence, concern about side effects from treatment, complex dosing regimens, and high cost [6,7]. The physician-level issues are clinical inertia to initiate or intensify treatment, reluctance to prescribe single-pill combination (SPC) antihypertensive drugs, which are tolerated better than individual drugs and improve adherence, lack of training in health communication [8,9]. The health systems-level barriers are inadequate funding, lack of performance standards, and time constraints in busy clinical practices [6,7,10].

One strategy to improve hypertension management involves adopting multicomponent intervention to address many of the above barriers yet are scalable in the public sector [11]. However, the vast majority of studies testing the effectiveness of multicomponent interventions are of short duration (6 months or less), and did not address the barriers at all levels [11]. The few studies with longer-term follow-up from did not show benefit persisting beyond 1 year [12,13].

Thus, a scarcity of information exists on effective, scalable, and potentially sustainable primary care interventions for managing hypertension in high-income countries.

By leveraging the public sector primary care clinic infrastructure, we conducted a pragmatic, cluster-randomized controlled trial (RCT) over 2 years in Singapore to evaluate the effectiveness of a potentially scalable, multicomponent intervention designed specifically to manage hypertension.

We adapted the intervention strategy from our pilot feasibility study to fit the primary care clinic workflow, receiving favorable feedback from stakeholders including the healthcare providers and patients with hypertension [14,15]. The intervention addressed several of the barriers identified above and included physician training on risk-based management, motivational conversations with patients by trained nurses, telephone follow-ups of all patients with hypertension by trained nurses, and subsidy on SPC antihypertensive medication.

The comparator was usual care in the primary care clinics. We hypothesized that a multicomponent intervention tailored to the existing primary care infrastructure would be more effective than usual care in lowering systolic BP (SBP) among adults with uncontrolled hypertension.

Methods

Design and oversight

The study was a pragmatic cluster RCT in 8 government primary care clinics in Singapore conducted according to the framework of the United Kingdom Medical Research Council (MRC) for implementing complex intervention trials [16,17].

Because intervention delivery was via the primary care system, a cluster RCT design was chosen to minimize intervention “spill-over”—participants in the usual care arm receiving any component of the multicomponent intervention.

Randomization and masking

We used computer-generated codes to randomly assign 4 clinics to the multicomponent intervention and 4 to usual care. Following randomization, 916 individuals with uncontrolled hypertension were recruited from these clinics. Patient enrollment started in January 2017 and ended in May 2018, and follow-up was conducted from December 2018 to September 2020. There was no masking in the trial. The trial protocol (S1 Protocol in S1 Appendix) and statistical analysis plan (SAP) (S1 Statistical Analysis Plan in S1 Appendix) were published previously [15,18].

The Ethics Review Committees at SingHealth institutional review board and National University of Singapore approved the study. Informed consent was signed by all participants before screening. An independent Data Safety and Monitoring Committee reviewed the trial conduct. Funders had no role in the design, conduct, analysis, interpretation, or reporting of results.

Study participants

Public sector clinics in Singapore provide subsidized primary care services to a multiethnic patient population from diverse socioeconomic backgrounds [19]. These clinics are staffed by physicians and nurses, and have onsite pharmacies and laboratories. Patients are assigned to a pool or team of physicians at every visit for a nominal co-payment. Patients may also choose to be seen by a family physician in the polyclinic albeit at a significantly higher (2 to 3 times) cost. Each provider attends to about 50 to 70 patients daily, with an average encounter time of about 7 to 10 minutes per patient [20]. Medications for chronic conditions are dispensed on prescriptions only. All major classes of antihypertensive medications (thiazide-like diuretics, calcium channel blockers, renin-angiotensin-aldosterone-system blockers, and beta blockers) are available at the pharmacies, and their generic formulation are subsidized. Generic and branded SPC antihypertensive medications are also available, albeit neither are subsidized as part of usual care.

We enrolled patients aged ≥40 years with previously diagnosed hypertension (SBP ≥140 mmHg or diastolic BP (DBP) ≥90 mmHg) based on mean of last 2 of 3 measurements who had visited an enrolled clinic at least twice during the prior 12 months and were Singaporean citizens or permanent residents.

Procedures

The multicomponent intervention is described in detail in the protocol paper [15]. Briefly, it incorporated the following components:

Physician training on risk-based management: Physicians were trained in a risk-based approach treatment algorithm requiring cardiovascular risk stratification of patients based on the 10-year Framingham risk score (FRS) for CVDs modified for the local population that is already embedded in the electronic health record system at the clinics [21]. SPC recommended for patients at high-risk, i.e., those with high (≥20%) 10-year CVD risk based on FRS, or those with preexisting CVD, or chronic kidney disease (CKD). Statins were also initiated if low-density lipoproteins (LDL) cholesterol was 2.6 mmol/L (100 mg/dl) or higher, as per the algorithm [22]. The least expensive, generic, SPC angiotensin II receptor blocker (ARB) and a thiazide diuretic (losartan-HCTZ) available in the polyclinic pharmacies was preferred. For patients at medium or low risk, the first-line antihypertensive medication was calcium channel blockers, followed by angiotensin-converting enzyme inhibitors (ACEIs) (preferably in those aged <55 years) or thiazide diuretics (preferably in those ≥55 years) initiated at half-standard dose and up-titrated as necessary. Use of beta blockers as first-line antihypertensive agents without a compelling indication was discouraged. The general target BP was <140/90 mmHg. However, for hypertensive individuals with proteinuria or preexisting CVD, the target BP was <130/80 mmHg. Physicians completed a standardized management checklist. The initial training session was over 2 hours, with re-training at 3 months and then 12 and 24 months. Details of the checklists and algorithm are described in Sections S2 to S5 in S1 Appendix.

Motivational conversations: A motivational conversation curriculum for hypertensive patients was developed in a feasibility study [23]. The counseling approach is intended to empower patients with hypertension to manage risk factors, set self-care goals for a healthy lifestyle (diet, physical activity, alcohol intake moderation, and smoking cessation), use a home BP monitor, and adhere to medication regimens [24]. A consultant specialist psychologist trained the clinic nurses in motivational conversation skills for 4 hours using the curriculum [23,24]. Refresher training was provided at months 12 and 24. Trained nurses delivered one 15- to 20-minute face-to-face motivational conversation session to patients identified at high risk of CVD.

Telephone-based follow-up by trained nurses: The nurses trained in motivational conversations also followed all hypertensive patients by telephone for adherence reinforcement to healthy lifestyles and treatment, using a standardized checklist. The calls were scheduled at monthly intervals for the first 3 months and then 3-month intervals for the 2-year duration of the project for all patients. Regular meetings were scheduled at 6- to 8-week intervals among physicians and nurses to review patients not achieving BP targets to provide advice or action, as needed.

Subsidy on SPC medications: A 50% discount on losartan-HCTZ SPC medication was offered to patients at high risk of CVD in the intervention clinics throughout the trial. Discounts were channeled through clinic pharmacies. The savings amounted to about S$7 per month per patient.

Usual care consisted of existing clinic services with routine care by physicians and nurses. Physicians were free to prescribe medications of their choice. Although SPC medications (including losartan-HCTZ costing S$12 to 15, per 1-month supply) were available in the clinic pharmacies, there was no trial-based subsidy for usual care participants. However, all participants in usual care received subsidies on single agent generic antihypertensive medications as per routine clinical practice.

Baseline screening

An on-site trained clinical research coordinator (CRC) at the participating clinics approached potentially eligible patients with a prior diagnosis of hypertension and invited them for prescreening. CRCs measured BP 3 times with 3-minute intervals between readings, with the patient in a sitting position, arm rested, using an upper arm calibrated automated Omron device (HEM-7130). Individuals with SBP ≥140 mmHg or DBP ≥90 mmHg (mean of last 2 of 3 BP measurements) were considered to have uncontrolled BP. Thus, the patients knew the group to which their practice was assigned at enrollment. Informed consent specific for intervention and usual care participants was obtained for further screening. Patients with active systemic illness (fever, known liver disease), clinically unstable heart failure, advanced CKD (estimated glomerular filtration rate (eGFR) <40 ml/minute/1.73m2), proteinuria ≥3 g/d, hospitalization in the prior 4 months, deemed mentally unfit to give informed consent, and pregnant women, were excluded.

This trial is reported according to the Consolidated Standards of Reporting Trials (S1 CONSORT Checklist).

The baseline, 12- and 24-month questionnaire was administered to all eligible patients and information was collected on sociodemographics (baseline only), comorbidities, diet, lifestyle and tobacco use, direct and indirect healthcare cost, and quality of life (EuroQol-5 Dimension-5 Level questionnaire (EQ-5D-5L)).

Anthropometric measurements (weight, height, waist circumference) were obtained. Fasting blood samples for serum electrolytes, glucose, and lipids were collected as part of the hypertension panel test. A random urine sample was collected for albumin, creatinine, and sodium.

Outcomes assessments

All participants were assessed in the clinic by trained CRC independent to treatment at 1-year and 2-year post-baseline.

Standardized measurement of BP was obtained using the same procedure as in baseline. CRCs also performed telephone follow-up assessments every 4 months over the 2-year study period.

Information on adverse events (including falls, hypotension, cough, hyperkalemia, musculoskeletal pain, peripheral edema, coronary heart disease, stroke, and heart failure) were reported. Hospitalizations and deaths were tracked (details are provided in the protocol). Deaths due to myocardial infarction, heart failure, or stroke (per ICD-10 codes) were categorized as cardiovascular deaths. Fasting blood and random urine samples were collected at 24 months. Patients were reimbursed for cost of travel to the clinic for their final follow-up visit.

For use in a future cost-effectiveness analysis, costs related to the training and time of nurses, physicians’ training, telephone follow-ups, and SPC subsidy were recorded.

Reporting in this publication is consistent with the CONSORT statement.

Outcomes

The primary endpoint was SBP based on a mean of last 2 of 3 measurements at the 2-year post-baseline final follow-up.

Prespecified key BP-related secondary outcomes included percentage of participants with BP control (SBP <140 mmHg and DBP <90 mmHg), DBP, number of antihypertensive medications, high (>20%) 10-year CVD risk based on FRS [21], and albuminuria measured as urine albumin to creatinine ratio (ACR) at 2-year post-baseline. Other secondary outcomes also at 2-year post-baseline included various thresholds of BP control, lifestyle measures (physical activity, body mass index (BMI), waist circumference, smoking status, intake of fruits and vegetables, dietary sodium intake (urinary sodium excretion)), clinical measures and outcomes (lipid levels, eGFR, dose of antihypertensive medications) [25], integrated 10-year CVD risk score (validated for a Chinese population) [26], fasting blood glucose (FBG), new-onset diabetes, and participant-reported health status. The latter was measured according to the mean score on the visual analogue scale (VAS) of the EQ-5D-5L (range, 0 to 100, with higher scores indicating better health) and mean score on the EQ-5D-5L utility index calculated using the Indonesian value set (range, −0.865 to 1, with higher scores indicating better health) [27]. In addition, the safety events included mortality, any serious adverse event (SAE), SAEs of special interest (cardiovascular death, hospitalization for CVDs due to coronary heart disease or acute coronary syndrome or elective revascularization or myocardial infarction, stroke, heart failure, peripheral vascular disease, and peripheral edema), adverse events (AEs), and hospitalizations potentially related to the intervention.

Statistical analysis

An a priori sample size of n = 125 patients was recruited at each of the 8 study clinics (4 intervention, 4 usual care) to detect a (cohen) effect size of 0.28 between the intervention and usual care at the 2-year follow up with 80% power at α = 0.05 assuming a within-clinic intra-class correlation of 0.01, based on assumptions from previous trials, including our own [28,29].

An SAP was published previously [18]. The original analysis for the primary outcome for the trial was patient-level change from baseline in SBP at 24-month post randomization. However, at the publication of the SAP, the analysis was modified to SBP at 24 months in a clinic-level comparison. At that time, an unexpectedly higher potential for patient dropout was anticipated, as the Coronavirus Disease 2019 (COVID-19) pandemic led to delays in clinic attendance. The SAP reviewer underscored that analysis of change from baseline can be inefficient, leads to unnecessary exclusion of subjects with missing data, and that a superior method would be to analyze a multivariate (repeated) outcome that includes the baseline measure of the response [30,31]. Therefore, we elected to use a repeated measures analysis of SBP inclusive of baseline SBP as one of the outcomes vector in the model [18]. Notwithstanding the modification, the analysis on the original primary outcome of patient-level change from baseline in SBP is reported in a sensitivity analysis.

Briefly, all analyses were performed based on the intention-to-treat (ITT) principle. BP at baseline, 12 and 24 months was modeled at the participant level in a likelihood-based, linear mixed model repeated measures (MMRMs) analysis with cluster random effects for clinic assuming participant and cluster (clinic) Gaussian error distributions, identity link function, using Satterthwaite approximation denominator degrees of freedom (DDF). Treatment arm and follow-up time and treatment × follow-up time interaction were modeled as fixed effects and random effects for clusters (clinics) and participants. All other secondary outcomes were analyzed using the same modeling approach but incorporating generalized techniques appropriate for the type of outcome and using between-within DDF.

Sensitivity analyses: (1) ITT for outcome of change from baseline in SBP and key BP-related secondary outcomes at 24 months with baseline BP (or baseline level of the respective secondary outcome) as a covariate performed at the patient level using between-within DDF, per ITT (2) per protocol analysis of the primary outcome using MMRM. Participants who were high CVD risk as per the checklist and did not receive single pill combination (SPC) in the multicomponent intervention (n = 73) as per protocol were excluded from this analysis. (3) Primary outcome restricted to patients completing the 24-month follow-up, (4) ITT for primary outcome using MMRM with adjustment for clinically important variables at baseline (including age, gender, waist circumference, diabetes, and 10-year FRS CVD risk score), (5) ITT after multiple imputations, in which missing values of 1- (15.0%) and 2-year (16.4%) SBP were replaced through a process of multiple imputation (details in Section S6 in S1 Appendix) [32]. After imputation, each of the 20 imputed datasets was analyzed using a MMRM model similar to the approach used for the non-imputed dataset. The parameter estimates obtained from each dataset were pooled for inference and (6) primary outcome using MMRM analysis restricted to participants followed up prior to 12 March 2020 (onset of COVID-19 pandemic). Details regarding sensitivity analyses are provided in the Supporting information (Section S6 in S1 Appendix).

Cost: The incremental cost of intervention delivery was computed. These included the cost of administration and oversight, physicians’ and nurses’ training, implementation of motivational conversation, subsidy for SPC antihypertensive medication, cost of other antihypertensive and lipid lowering medications, and cost of related laboratory (safety) tests. Details are reported in the Section S7 in S1 Appendix.

A detailed incremental cost-effectiveness analysis taking a lifetime perspective will be the subject of another paper.

Results

A total of 3,836 individuals aged ≥40 years were prescreened at the clinics, and 916 (23.9%) with uncontrolled hypertension who met the eligibility criteria were enrolled (447 and 469 participants in intervention and usual care, respectively) from January 2017 to May 2018 (Fig 1).

Fig 1

Participant flow (CONSORT) chart.

The 24-month follow-up ended in September 2020 with 85.9% and 81.4% retention in the intervention and usual care arms, respectively (Fig 1).

Mean (SD) age of trial participants was 64.5 (9.8) years; 49.6% were women; 73.5% were Chinese, 13.3% Malay, 9.7% Indian, and 3.5% other ethnicities; 73.3% were overweight or obese (Asian cutoff of 23.5 kg/m2 or more) [33]; 34.3% had diabetes; and 59.8% had at least 1 comorbid chronic disease (diabetes, CKD, self-reported heart disease, or stroke) (Table 1 and Table A of Section S8 in S1 Appendix).

Table 1

Baseline characteristics.

		All clinics (N = 916)
Characteristicsa	Total(N = 916, 8 clinics)	Multicomponent intervention(N = 447, 4 clinics)	Usual care(N = 469, 4 clinics)	P value
Age, mean (SD), years	64.5 (9.8)	63.0 (9.7)	65.9 (9.7)	0.41
Female, n (%)	454 (49.6)	203 (45.4)	251 (53.5)	0.13
Ethnicity, n (%)				0.88
Chinese	673 (73.5)	334 (74.7)	339 (72.3)
Malay	122 (13.3)	57 (12.8)	65 (13.9)
Indian	89 (9.7)	36 (8.1)	53 (11.3)
Other	32 (3.5)	20 (4.5)	12 (2.6)
Secondary or higher education, n (%)	670 (73.1)	351 (78.5)	319 (68.0)	0.22
Currently employed, n (%)	445 (48.6)	240 (53.7)	205 (43.7)	0.33
Overweight or obese, n (%)b	671 (73.3)	324 (72.5)	347 (74.0)	0.80
High waist circumference, n (%)c	658 (71.8)	308 (68.9)	350 (74.6)	0.39
Chronic diseases
Self-reported heart disease, n (%)	92 (10.0)	33 (7.4)	59 (12.6)	0.24
Self-reported stroke, n (%)	37 (4.0)	20 (4.5)	17 (3.6)	0.62
Diabetes, n (%)d	314 (34.3)	138 (30.9)	176 (37.5)	0.45
CKD, n (%)e	351 (38.3)	165 (36.9)	186 (39.7)	0.56
Any of above chronic diseases, n (%)f	548 (59.8)	245 (54.8)	303 (64.6)	0.10
BP, mmHg
Systolic, mean (SD)	149.3 (13.0)	148.4 (11.8)	150.1 (14.0)	0.30
Diastolic, mean (SD)	88.3 (9.8)	89.4 (9.2)	87.3 (10.2)	0.26
Current smoker, n (%)	78 (8.5)	40 (8.9)	38 (8.1)	0.88
Physical activity, mean (SD), log MET-minute/weekg	6.1 (2.4)	6.3 (2.4)	6.0 (2.3)	0.28
10-year calculated CVD FRS %, mean (SD)	22.2 (7.6)	21.7 (7.6)	22.7 (7.5)	0.34
Ln urine albumin to creatinine (ACR), mean (SD)	1.4 (1.0)	1.3 (0.9)	1.4 (1.0)	0.31
EQ-5D-5L VAS, mean (SD)h	75.4 (13.4)	77.0 (12.3)	74.0 (14.2)	0.50
Currently on antihypertensive medications, n (%)i				0.29
0	26 (2.8)	9 (2.0)	17 (3.6)
1	408 (44.5)	208 (46.5)	200 (42.6)
2	315 (34.4)	158 (35.3)	157 (33.5)
3 or more	154 (16.8)	70 (15.7)	84 (17.9)

aBaseline characteristics are represented as means (SD) or proportions (%) as appropriate.

bDefined as BMI ≥23.5 kg/m2.

cHigh waist circumference was defined as waist circumference of ≥90 cm in males and ≥80 cm in females.

dDiabetes is defined as physician diagnosed or FBG ≥7 mmol/L or glycated hemoglobin (HbA1c) ≥6.5%.

eCKD is defined as urine ACR ≥3 mg/mmol or estimated glomerular filtration rate (eGFR) <60 ml/minute/1.73m2.

fAt least 1 comorbid chronic diseases of diabetes, CKD, self-reported heart disease, and self-reported stroke.

gPhysical activity was assessed using the IPAQ score. The estimates are log MET-minute/week, higher score indicates higher activity level.The IPAQ Group, 2005 Guidelines.

hHealth status was reported on a scale of 0 (worst imaginable health) to 100 (best imaginable health) using EQ-5D-5L VAS, where higher values indicate better health.

iA total of 13 participants had missing pharmacy data for antihypertensive medications.

ACR, albumin to creatinine ratio; BMI, body mass index; BP, blood pressure; CKD, chronic kidney disease; CVD, cardiovascular diseases; FBG, fasting blood glucose; FRS, Framingham risk score; IPAQ, International Physical Activity Questionnaire; Ln, natural logarithm; SD, standard deviation.

By design, all participants had uncontrolled BP at baseline. The mean SBP (SD) and DBP (SD) were 149.3 (13.0) mmHg and 88.3 (9.8) mmHg, respectively. Baseline characteristics were generally balanced between the intervention and usual care arms (Table 1 and Table A of Section S8 in S1 Appendix).

The intervention had high implementation fidelity: 100% of the invited physicians in the intervention clinics were trained and completed the participant management checklist; 99.6% of participants received at least 1 telephone-based follow-up, and a mean (95% CI) of 7.2 (7.0, 7.4) calls per participant were delivered during 2 years; 94.5% of high-risk participants received 1 motivational conversation session; 63.7% received prescribed SPC, and 96.9% of those prescribed SPC received the subsidy (Table B of Section S8 in S1 Appendix).

No participant assigned to a usual care clinic sought treatment at an intervention clinic during the course of the study and vice versa.

Primary outcome

At 24 months, the adjusted mean SBP (95% CI) in the intervention and usual care arms was 135.4 (133.1, 137.7) mmHg and 138.7 (136.4, 141.1) mmHg, respectively. The primary outcome, adjusted mean (95% CI) SBP, was significantly lower by −3.33 (−6.34, −0.32) mmHg (p = 0.03) in the intervention arm versus usual care (Fig 2A and Table 2).

Fig 2

Mean SBP and proportion BP controlled by treatment group over time.

Panel A: Mean SBP was estimated with a generalized linear MMRM for SBP, with cluster random effects for clinic, and random effects for participants. The I bars indicate 95% CIs. *The p-value for the difference in the mean SBP between 2 treatment groups at 24 months was 0.03. Panel B: Proportions BP controlled were estimated at 12 months and 24 months using a generalized linear mixed-effects model for BP control (SBP <140 mmHg and DBP <90 mmHg) as repeated measures, with fixed effect of baseline SBP, cluster random effects for clinic, and random effects for participants. The I bars indicate 95% CIs. †The p-value for the difference in the proportion BP controlled between 2 treatment groups at 24 months was 0.01. BP, blood pressure; CI, confidence interval; DBP, diastolic BP; MMRM, mixed model repeated measure; SBP, systolic BP.

Table 2

Primary and BP-related key secondary outcomes by treatment group.

	n	Multicomponent interventionMean (SE)(95% CI)(n = 447, 4 clinics)	n	Usual careMean (SE)(95% CI)(N = 469, 4 clinics)	Adjusted mean differenceb/OR (95% CI)c	P value
Primary outcome
SBP, mean (95% CI), mmHg
24 months	384	135.4 (0.98)(133.1, 137.7)	382	138.7 (1.01)(136.4, 141.1)	−3.33(−6.34, −0.32)	0.03
Secondary outcomes
DBP, mean (95% CI), mmHg
24 months	384	81.9 (0.86)(79.5, 84.3)	382	81.4 (1.31)(77.5, 85.4)	0.45(−3.39, 4.29)	0.78
BP controlled to conventional goal (SBP <140 mmHg and DBP <90 mmHg), % (95% CI)
24 months	384	61.3 (11.0)(55.2, 67.1)	382	51.2 (10.6)(45.9, 56.5)	1.51(1.10, 2.09)c	0.01
High FRS 10-year CVD risk, % (95% CI)d
24 months	363	47.3 (14.5)(40.3, 54.4)	365	57.2 11.8(51.4, 62.7)	0.67(0.47, 0.97)	0.03
Ln Urine ACR, mean (95% CI)
24 months	366	1.32 (0.07)(1.16, 1.48)	355	1.53 (0.06)(1.39, 1.68)	−0.22(−0.41, −0.02)	0.03
Number of antihypertensive medications per day, mean (95% CI)
24 months	393	1.9 (0.06)(1.8, 2.0)	382	1.7 (0.07)(1.6, 1.9)	0.18(0.01, 0.36)	0.04

aAll analyses based on the ITT principle.

bAdjusted means differences between multicomponent intervention and usual care using repeated measures analysis.

cOR (95% CI).

dHigh 10-year FRS CVD risk score was defined as >20% risk of CVD at 10 years.

ACR, albumin to creatinine ratio; BP, blood pressure; CI, confidence interval; CVD, cardiovascular disease; DBP, diastolic BP; FRS, Framingham risk score; ITT, intention-to-treat; Ln, natural logarithm; OR, odds ratio; SBP, systolic BP.

At 24 months, the adjusted decline in SBP (95% CI) was −13.3 (−14.8, −11.8) in the intervention and −10.7 (−12.1, −9.2) in usual care. The adjusted decline (95% CI) in SBP from baseline was greater by −2.65 (−4.73, −0.57) mmHg (p = 0.01) in the intervention versus usual care (Table 3).

Table 3

Sensitivity analyses of intervention effect on primary and key secondary outcomes.

	n	Multicomponent interventionMean (SE)(95% CI)(N = 447, 4 clinics)	n	Usual careMean (SE)(95% CI)(N = 469, 4 clinics)	Adjusted mean differencea(95% CI)	P value
Change from Baseline in SBP and Key Secondary Outcomes b
SBP, mmHg
24 months	384	−13.3 (0.7)(−14.8, −11.8)	382	−10.7 (0.7)(−12.1, −9.2)	−2.65(−4.73, −0.57)	0.01
DBP, mmHg
24 months	384	−7.0 (0.4)(−7.9, −6.2)	382	−6.2 (0.7)(−7.1, −5.4)	−0.81(−1.99, 0.37)	>0.05
FRS 10-year CVD risk scorec
24 months	359	−2.3 (0.3)(−2.9, −1.8)	357	−1.3 (0.3)(−1.9, −0.8)	−1.00(−1.77, −0.22)	0.01
Ln urine ACR
24 months	363	0.0 (0.0)(−0.1, 0)	353	0.1 (0.1)(0.0, 0.2)	−0.17(−0.31, −0.02)	0.02
Number of antihypertensive medications per day
24 months	393	0.2 (0.1) (0.1, 0.4)	382	0 (0) (0, 0.1)	0.20(0.07, 0.34)	0.01
SBP using MMRM d
Per protocol analysise
24 months	325	134.8 (1.0)(132.6, 137.0)	382	138.7 (1.0)(136.3, 141.2)	−3.94(−6.93, −0.94)	0.01
Restricted to patients completing 24-month follow-upf
24 months	384	135.3 (0.9)(133.1, 137.5)	382	138.7 (0.8)(136.9, 140.6)	−3.43(−6.07, −0.80)	0.01
Final analysis adjusting for clinically important variables at baselineg
24 months	384	135.7 (0.8)(133.8, 137.6)	382	138.2 (1.0)(136, 140.4)	−2.53(−5.00, −0.09)	0.04
After multiple imputationh
24 months	447	136.3 (1.0)(134.2, 138.4)	469	139.8 (1.0)(137.9, 141.7)	−3.39(−6.32, −0.65)	0.02
After restricting to participants followed before COVID-19i
24 months	322	134.9 (1.1)(132.3, 137.4)	372	138.9 (1.0)(136.6, 141.2)	−4.00(−7.14, −0.86)	0.02

aAdjusted means differences between multicomponent intervention and usual care using repeated measures analysis.

bChange from baseline in each outcome (at year-1 and year-2) was analyzed at the participant level after adjusting for baseline values of respective outcome as a fixed covariate. The analyses were performed based on the ITT principle.

cTen-year FRS CVD risk score.

dPrimary outcome of SBP at baseline, 12 and 24 months was modeled at the patient level in a likelihood-based, linear MMRM analysis with cluster random effects for clinic.

eParticipants who are high CVD risk as per the checklist and did not receive SPC in the multicomponent intervention (n = 73) as per protocol were excluded from this analysis.

fParticipants who did not complete the year-2 outcomes assessments (n = 150) were excluded from this analysis.

gPrimary outcome analysis at the participant level after adjusting for important baseline characteristics as covariates. The characteristics included as covariates in adjustments were age, gender, waist circumference, diabetes, and 10-year FRS CVD risk score.

hPrimary outcome analysis after multiple imputations to replace missing values of SBP at 1 year: 137 (15.0%) and 2 years: 150 (16.4%).

iSBP readings of participants followed at year-2 outcomes assessments on or after 12 March 2020 (n = 72, intervention– 62, and usual care– 10) were excluded from the analysis to assess the effect of multicomponent intervention before COVID-19 was announced as pandemic by WHO.

ACR, albumin to creatinine ratio; BP, blood pressure; CI, confidence interval; COVID-19, Coronavirus Disease 2019; CVD, cardiovascular disease; DBP, diastolic BP; FRS, Framingham risk score; ITT, intention-to-treat; Ln, natural logarithm; MMRM, mixed-effects model repeated measures; OR, odds ratio; SE, standard error; SBP, systolic BP; SPC, single pill combination; WHO, World Health Organization.

Key secondary outcomes

Most of the key secondary outcomes related to BP, vascular risk, and antihypertensive medication use showed improvements consistent with the primary outcome (Table 2). A greater percentage of participants achieved BP control at 24 months in the intervention (61.3%) arm versus usual care (51.2%) with associated odds ratio (95% CI) 1.51 (1.10, 2.09) (p = 0.01) (Fig 2B and Table 2). At 24 months, the percentage with high (>20%) 10-year CVD risk based on FRS was lower in the intervention arm with associated odds ratio (95% CI) 0.67 (0.47, 0.97) (p = 0.03) compared to usual care. Adjusted mean (95% CI) log albuminuria levels were lower by −0.22 (−0.41, −0.02) (log ACR) (p = 0.03) for intervention compared to usual care (Table 2). The number of antihypertensive medications used was higher by a mean (95% CI) of 0.18 (0.01, 0.36) (p = 0.04) per participant for intervention compared to usual care (Table 2) [25]. However, the mean DBP did not differ significantly between study arms (Table 2).

The findings for sensitivity analysis for SBP and BP-related key secondary outcomes were also consistent when analyzed using change from the baseline level (Table 3). Consistent results were obtained across all supplementary sensitivity analyses: per-protocol analysis, analysis restricted to participants completing the 24-month follow-up, analysis after accounting for clinically important baseline characteristics, multiple imputations analysis, and pre-COVID lockdown analysis (Table 3).

Other secondary outcomes

Secondary outcomes related to health behaviors and quality of life also showed incremental improvements in response to the intervention (Table C of Section S8 in S1 Appendix). Self-reported overall health status was better in the intervention arm. The mean score (95% CI) on the EQ-5D-5L VAS in the intervention arm was higher than usual care by 6.28 (0.56, 11.99) points (p = 0.03). Secondary outcomes of BMI, waist circumference, current smokers, cholesterol, and estimated GFR were not significantly different between study arms (p > 0.20 for all) (Supporting information Table C).

The proportion of participants with incident SAEs did not differ significantly between study arms. Total mortality was not different between intervention (0.5%) compared to usual care (1.3%) (p = 0.29). Fewer participants reported injuries or falls with intervention compared to usual care (0.5 versus 2.8%, respectively) (p = 0.01) (Table D of Section S8 in S1 Appendix).

Effects of intervention on the primary outcome of SBP were consistent for all predefined subgroups, with no evidence of effect heterogeneity (Fig 3).

Fig 3

Subgroup analyses for SBP at 24 months, according to participant characteristics at baselinea.

aSubgroups defined based on the status of the participants at the time of baseline interview. bNormal weight is defined as BMI <23.5 kg/m2 per Asian cutoffs. cHigh waist circumference was defined as waist circumference of ≥90 cm in males and ≥80 cm in females. dPhysical activity was assessed using the IPAQ score. The IPAQ Group, 2005. eDefined as SBP ≥160 mmHg and DBP ≥100 mmHg. fDefined using the standardized physician management checklist developed for the study. The participant is categorized as high CVD risk if any of the following criteria is met: high CVD risk by FRS adapted for Singapore ≥20, diabetes, ACR >34 mg/mmol, eGFR <60 ml/minute/1.73 m2, physician diagnosed heart disease, and self-reported stroke. gDiabetes is defined as physician diagnosed or FBS >7 mmol/L or glycated hemoglobin (HbA1c) >6.5%. hCKD is defined as eGFR <60 ml/minute/1.73 m2 or urine ACR ≥3 mg/mmol. ACR, albumin to creatinine ratio; BMI, body mass index; BP, blood pressure; CVD, cardiovascular disease; DBP, diastolic BP; eGFR, estimated glomerular filtration rate; FBS, fasting blood sugar; FRS, Framingham risk score; IPAQ, International Physical Activity Questionnaire; SBP, systolic BP.

The intraclass cluster coefficient was 0.009 for SBP and 0.02 for change in SBP, respectively (Table E of Section S8 in S1 Appendix). The raw means of primary and key secondary outcomes were consistent with the adjusted means. (Table F of Section S8 in S1 Appendix).

Cost of intervention

The incremental cost of intervention delivery per eligible participant was SGD $231 (USD $170 annually, exchange rate 1 SGD $ = 0.74 USD $ as of January 2022) (Table G of Section S8 in S1 Appendix).

Discussion

In this cluster-randomized trial in adults with uncontrolled hypertension visiting primary care clinics in Singapore, a multicomponent intervention incorporating risk-based treatment by trained physicians, subsidized SPC medications, nurse-led motivational conversation, and telephone follow-ups, significantly lowered SBP and improved BP control at 24 months, compared to usual care. The intervention also increased use of antihypertensive medications, lowered cardiovascular risk and albuminuria, improved some aspects of health-related quality of life, and there were no safety concerns. The potential ease of adaptation into the existing primary care system and low cost (USD $14 per participant per month) indicate rapid scalability and sustainability of the intervention in Singapore and other countries with similar infrastructure.

The vast majority of previous multicomponent intervention trials demonstrating BP lowering effectiveness in high-income countries were of short duration (median, 6 months) [11]. A strategy of ACEI (Lisinopril)-hydrochlorothiazide SPC medication-based therapy was shown effective in California adults [34]. Trials of interventions in the United States that included telephone follow-ups did not show persistent intervention benefits beyond 12 months [12,13]. However, those trials did not include risk-based management with SPC medications or any related subsidies. Trials longer than 1 year were mostly from low- and middle-income countries (LMICs) with primary care infrastructures differing from Singapore and other high-income countries [35,36].

Our intervention strategy departs from previous trials in its composition as a combination of risk-based management and subsidized SPC medications coupled with proactive motivational counseling delivered via the routine primary care infrastructure.

An additional feature of our trial was a prespecified assessment of cardiovascular risk and albuminuria outcomes—both robust prognostic indicators of CVD [37]. Our trial was not designed to unravel how much of the benefit of the multicomponent intervention was attributable to each individual component. It is likely that the intervention benefit on BP observed in our trial is a reflection of positive interaction among its components, which addressed the barriers to hypertension care at the patient, physician, and health systems levels. First, trained intervention-arm physicians are more likely to prescribe antihypertensive medications, intensify antihypertensive treatments, and prescribe SPC medications [38]. SPC medications are better tolerated than higher doses of individual medications, which improves adherence [8,9]. Second, the motivational conversation and telephone follow-ups likely enhanced uptake of healthy lifestyles (potentially physical activity) and medications. Third, the subsidy component of the intervention likely served as an additional stimulus to SPC medication utilization by reducing the copayment by the patients at high risk of CVD [39].

We found that although BP declined in both groups at 12 and 24 months, the benefit of intervention on lowering BP was apparent only at the latter time point. The longevity of the benefit of intervention at 24 months is unique for high-income countries. These findings are consistent with our previous work in LMICs and possibly reflect synergies among the intervention components over time [35]. For example, messages on adherence to lifestyle and antihypertensive medications were reinforced during repeated telephone-based follow-ups by trained nurses. In addition, the patients identified with elevated BP were flagged for the clinic visit when trained physicians escalated antihypertensive drugs per the study algorithm, including SPC antihypertensive medications for high-risk patients who were incentivized by the subsidy [29,40]. Nurse and physician competencies may have enhanced over time with the annual retraining.

Hypertension is an immense public health challenge, implicated in over 50% of heart disease, stroke, and heart failure cases [41]. BP control remain suboptimal globally even in high income countries [1]. Our findings regarding benefits of a multicomponent intervention leveraging the existing healthcare infrastructure have implications for clinical practice and policy globally. Discussions with the health departments and advisory committees are ongoing for scaling-up the intervention across primary care clinics in Singapore.

Our intervention is a potentially viable option for many Asian countries such as Japan, Korea, Taiwan, and urban China with equally high-risk populations and hypertension-related CVD burden, similar public health infrastructure with wide availability of generic SPC antihypertensive medications [42]. Looking beyond Asia, the potential impact of the intervention could be even more significant in the US, Canada, UK, and many European countries where recent trends indicate a plateauing or rise in uncontrolled hypertension rates [43,2,44-46]. Our intervention addressed barriers to hypertension care at multiple levels, yet it was relatively straightforward. In addition, the implementation fidelity indicated excellent uptake of strategies (patients, physicians, nurses, and clinic) levels demonstrating successful leveraging of the existing healthcare infrastructure. Therefore, it would be logistically possible to scale up the intervention rapidly with marginal costs. Moreover, the physician’s and nurses’ staffing ratios relative to the population (2.4 and 6.2 per 1,000 population, respectively) and the electronic health record system in Singapore’s public sector are comparable to healthcare systems in high- and upper-middle-income countries [47]. Generic formulations of antihypertensive medications, including SPC drugs, are available [10]. Various forms of vouchers, which have been used to incentivize care in other diseases [48,49], could substitute for our use of a subsidy for the SPC.

We suggest that suitably modified forms of a multicomponent intervention such as ours be evaluated in other settings and their effects on direct measures of vascular disease. The major strengths of our trial are the cluster-randomized design, intervention delivered with high fidelity using the existing public health infrastructure that minimized cost and enhances sustainability and generalizability, and 2-year follow-up. Consistent results of ITT, per-protocol, and sensitivity analyses at both the clinic and patient level for BP, CVD risk, and albuminuria indicate the robustness of our findings.

Our trial has limitations. First, since this was a pragmatic trial, it was not possible to mask the clinical practice teams to the random group assignment. Moreover, the patients were aware of the group to which their practice was assigned at enrollment. However, the key outcome (BP) was measured by using an automated device, which minimized the risk for biased outcome assessment. Likewise, the assessment of several vascular risk factors including fasting glucose, lipids, and albuminuria were objectives measures obtained directly from the laboratory and medication data from pharmacy records. Thus, the potential of reporting bias for the primary and key secondary outcomes is minimal. Second, almost 16% of the trial cohort was lost to follow-up during 2 years. However, retention was considered very good, given the COVID-19 pandemic restrictions imposed during the last 3 months of follow-up. Moreover, sensitivity analysis restricted to participants before COVID-19-related lockdown, as well as multiple imputation accounting for missing values yielded consistent results. Third, our trial did not have sufficient power to detect the intervention effect on several secondary outcomes especially DBP and lifestyle measures. However, several large-scale epidemiological studies and systematic reviews of trials have shown that SBP—our primary outcome—is a much stronger predictor of CVD morbidity and mortality than DBP [50,51]. Moreover, the intervention improved BP control, accounting for both SBP and DBP target levels. It is also possible that physicians were not aggressive with titrating therapy to the DBP target in some patients, perhaps due to concerns with excessive lowering of DBP levels and paradoxical increase in CVD [52]. A post-intervention acceptability study is planned to explore these factors. Fourth, the trial duration was insufficient to assess cardiovascular events. However, even a 2-mmHg decline in SBP has been associated with a 7% to 10% decrease in deaths from coronary disease and stroke [51,53-55]. Moreover, improvement in the secondary outcomes of CV risk and albuminuria—albeit multiplicity unadjusted—indicate potential in the downstream benefit of the strategy on vascular risk reduction. Fifth, the benefit of the intervention on albuminuria reduction could be a chance association, pending further studies for confirmation. However, blockers of the renin-angiotensin-aldosterone system (RAAS) such as losartan are known to have antiproteinuric effects that are in addition to their BP-lowering effect, especially when used together with diuretics [56,57]. RAAS blockers and diuretics were part of the SPC medications used more frequently in patients in the intervention group. Sixth, the existence of some elements of the intervention among control clinics, or contamination, remains a possibility with implications of a diluted intervention effect. Moreover, usual care participants also experienced a decline in BP, which was most likely due to behavior modification in response to BP screening by the CRCs. These factors would contribute to underestimation of the full benefit of the intervention. Moreover, regression to the mean could account for some reduction in BP, although it would apply equally to both arms [56,58]. Finally, the number of clusters was relatively small, which increases the possibility of randomization imbalance at baseline. However, we noted no statistically or clinically meaningful differences in any baseline characteristics between the randomized groups. Furthermore, we employed a contemporary analytic approach with no assumptions for baseline [30,31]. Moreover, sensitivity analysis adjusting for baseline characteristics for the primary outcome yielded consistent results. Thus, we believe our results are robust.

Conclusions

In conclusion, a multicomponent intervention, including physician training on risk-based subsidized SPC antihypertensive medications, nurse-led motivational conversation, and proactive telephone follow-ups, led to reduced SBP and improved BP control over 24 months among patients with uncontrolled hypertension in Singapore’s primary care clinics. Given the potential ease of adaptation and rapid scalability into many existing health systems at marginal costs, implementation of our strategy would represent a step forward in the high-priority global effort to reduce hypertension-related morbidity and death.

CONSORT 2010 checklist of information to include when reporting a cluster randomized trial.

(DOCX)

Click here for additional data file.

Supplementary Appendix: S1 Protocol S1 Statistical Analysis Plan.

Section S1: Research Contributions. Section S2: Nurse telephone follow-up checklist. Section S3: Physician Management Checklist. Section S4: Summary of intervention training. Section S5: Antihypertensive treatment algorithm. Section S6: Statistical methods for sensitivity for systolic blood pressure (BP). Section S7: Program delivery cost estimation—methods, assumptions, and data sources. Section S8: Tables A to G. Section S9. References.

(DOCX)

Click here for additional data file.

16 Feb 2022

Dear Dr Jafar,

Thank you for submitting your manuscript entitled "Integration of a Multicomponent Intervention for Hypertension into Primary Healthcare Services - A Cluster Randomised Controlled Trial" for consideration by PLOS Medicine.

Your manuscript has now been evaluated by the PLOS Medicine editorial staff and I am writing to let you know that we would like to send your submission out for external peer review.

However, before we can send your manuscript to reviewers, we need you to complete your submission by providing the metadata that is required for full assessment. To this end, please login to Editorial Manager where you will find the paper in the 'Submissions Needing Revisions' folder on your homepage. Please click 'Revise Submission' from the Action Links and complete all additional questions in the submission questionnaire.

Please re-submit your manuscript within two working days, i.e. by Feb 18 2022 11:59PM.

Login to Editorial Manager here: https://www.editorialmanager.com/pmedicine

Once your full submission is complete, your paper will undergo a series of checks in preparation for peer review. Once your manuscript has passed all checks it will be sent out for review.

Feel free to email us at plosmedicine@plos.org if you have any queries relating to your submission.

Kind regards,

Callam Davidson

Associate Editor

PLOS Medicine

4 Apr 2022

Dear Dr. Jafar,

Thank you very much for submitting your manuscript "Integration of a Multicomponent Intervention for Hypertension into Primary Healthcare Services - A Cluster Randomised Controlled Trial" (PMEDICINE-D-22-00512R1) for consideration at PLOS Medicine.

Your paper was evaluated by an associate editor and discussed among all the editors here. It was also discussed with an academic editor with relevant expertise, and sent to independent reviewers, including a statistical reviewer. The reviews are appended at the bottom of this email and any accompanying reviewer attachments can be seen via the link below:

[LINK]

In light of these reviews, I am afraid that we will not be able to accept the manuscript for publication in the journal in its current form, but we would like to consider a revised version that addresses the reviewers' and editors' comments. Obviously we cannot make any decision about publication until we have seen the revised manuscript and your response, and we plan to seek re-review by one or more of the reviewers.

In revising the manuscript for further consideration, your revisions should address the specific points made by each reviewer and the editors. Please also check the guidelines for revised papers at http://journals.plos.org/plosmedicine/s/revising-your-manuscript for any that apply to your paper. In your rebuttal letter you should indicate your response to the reviewers' and editors' comments, the changes you have made in the manuscript, and include either an excerpt of the revised text or the location (eg: page and line number) where each change can be found. Please submit a clean version of the paper as the main article file; a version with changes marked should be uploaded as a marked up manuscript.

In addition, we request that you upload any figures associated with your paper as individual TIF or EPS files with 300dpi resolution at resubmission; please read our figure guidelines for more information on our requirements: http://journals.plos.org/plosmedicine/s/figures. While revising your submission, please upload your figure files to the PACE digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at PLOSMedicine@plos.org.

We hope to receive your revised manuscript by Apr 25 2022 11:59PM. Please email us (plosmedicine@plos.org) if you have any questions or concerns.

***Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.***

We ask every co-author listed on the manuscript to fill in a contributing author statement, making sure to declare all competing interests. If any of the co-authors have not filled in the statement, we will remind them to do so when the paper is revised. If all statements are not completed in a timely fashion this could hold up the re-review process. If new competing interests are declared later in the revision process, this may also hold up the submission. Should there be a problem getting one of your co-authors to fill in a statement we will be in contact. YOU MUST NOT ADD OR REMOVE AUTHORS UNLESS YOU HAVE ALERTED THE EDITOR HANDLING THE MANUSCRIPT TO THE CHANGE AND THEY SPECIFICALLY HAVE AGREED TO IT. You can see our competing interests policy here: http://journals.plos.org/plosmedicine/s/competing-interests.

Please use the following link to submit the revised manuscript:

https://www.editorialmanager.com/pmedicine/

Your article can be found in the "Submissions Needing Revision" folder.

To enhance the reproducibility of your results, we recommend that you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. Additionally, PLOS ONE offers an option to publish peer-reviewed clinical study protocols. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols

Please ensure that the paper adheres to the PLOS Data Availability Policy (see http://journals.plos.org/plosmedicine/s/data-availability), which requires that all data underlying the study's findings be provided in a repository or as Supporting Information. For data residing with a third party, authors are required to provide instructions with contact information for obtaining the data. PLOS journals do not allow statements supported by "data not shown" or "unpublished results." For such statements, authors must provide supporting data or cite public sources that include it.

We look forward to receiving your revised manuscript.

Sincerely,

Callam Davidson,

Associate Editor

PLOS Medicine

plosmedicine.org

-----------------------------------------------------------

Comments from the Academic Editor:

This is a well conducted, pragmatic study of a multifaceted intervention. I think the effect size is in keeping with pragmatic health care interventions of these nature and so I am not that concerned by reviewer 4 comments on that particular issue. The longer follow-up is certainly a value add as the authors point out most studies are of much shorter duration.

I do agree with reviewer 2 that the major gap in the paper is more information on implementation fidelity. There is absolutely no information on numbers of people reached by each of the intervention components including medication use and adherence rates and whether fidelity changed over time which could help explain why effects emerged in year 2 (assuming much of the year 1 reduction was regression to the mean). It really is a bit of a black box trying to work out the causal path to effectiveness and this does not help readers with generating new knowledge or replicating the study in other settings (as recommended by the authors).

The cost analysis is cursory. It only assesses costs of the intervention and does not include costs of care utilisation, medications, diagnostic tests etc. The EQ5D-5L VA scores only are reported which shows a difference but the mean scores are not reported. It is difficult to see how this intervention might generate a quality of life gain in such a short period of time with just a small reduction in BP. That said a cost per QALY estimate could be relatively easily generated from the data available to determine an incremental cost effectiveness ratio and I suggest this be included in the paper to help us understand this in more detail.

A few other issue of note:

1. I also was quite perplexed by the large rise in albuminuria in the usual care group. This is quite a spurious finding and not at all in keeping with the effect sizes observed in the other outcome variables. This warrants greater attention.

2. I was also surprised that other cardimetabolic parameters such as lipids and blood glucose were not included. This information was collected and lipids are stated as a secondary outcome measure on p.11 but it is not reported. I have not checked the protocol and SAP but worth checking if all outcomes measures are reported as per protocol.

3. The authors state total mortality was lower for intervention (0.5%) compared to usual 437 care (1.3%) (p=0.29) - this should be amended as there was no statistically significant difference in mortality

Requests from the editors:

Please structure your abstract using the PLOS Medicine headings (Background, Methods and Findings, Conclusions).

Please report your abstract according to CONSORT for abstracts, following the PLOS Medicine abstract structure (Background, Methods and Findings, Conclusions) http://www.consort-statement.org/extensions?ContentWidgetId=562

Abstract Background: Provide the context of why the study is important. The final sentence should clearly state the study question.

Abstract Methods and Findings:

Please state that there was no masking in the trial.

Please define both the intervention and control states.

Please provide the number of participants lost to follow up in each group.

Please indicates the dates during which study enrollment and follow up occurred.

Please include the important dependent variables that are adjusted for in the analyses.

Please include a summary of adverse events.

Please ensure relevant p values are reported.

In the last sentence of the Abstract Methods and Findings section, please describe the main limitation(s) of the study's methodology.

The Data Availability Statement (DAS) requires revision. If the data are not freely available, please describe briefly the ethical, legal, or contractual restriction that prevents you from sharing it. Please also include an appropriate contact (web or email address) for inquiries (this cannot be a study author).

Thank you for including an Author Summary. Please update as follows:

* Relocate such that Author Summary is placed after Abstract and before Introduction.

* Combine bullets one and two in a single-sentence bullet.

* In the final bullet under ‘Why was this study done?’, please replace ‘were’ with ‘are’ and relocate the comma after ‘mixed’ to come after ‘(majority <6 months)’.

* Bullets under the ‘What did the researchers do and find?’ can be slightly more succinct (aim for 3 single sentence bullet points, if possible).

* Please quantify the main findings (i.e. lowering of blood pressure and improvement in BP control relative to usual care) and consider stating ‘was not associated with any safety concerns’ rather than ‘was safe’ (same comment applies at Line 471).

Please remove funding details from the first page (this information is captured in your Financial Disclosure statement in the submission form).

Citations should be in square brackets and preceding punctuation.

Thanks for completing the CONSORT checklist. Please cite your checklist in the Methods (‘This trial is reported according to the Consolidated Standards of Reporting Trials (S1 Checklist)’ at line 224.

Please update the checklist to use section and paragraph numbers, rather than page numbers.

Please define what is meant by dropouts/loss to follow up.

Please present the safety data for the study including numbers of specific events and whether or not adverse events are thought to be related to treatment.

Please indicates the dates during which study enrollment and follow up occurred.

Line 217: Please cite your protocol as an item within the Supporting Information (see our guidance here https://journals.plos.org/plosmedicine/s/supporting-information).

Line 210: Outcomes could be described more clearly (they are easier to determine in the protocol/on clinicaltrials.gov).

Line 271: Please cite the SAP you have included in your Supporting Information.

Line 281: Should Sensitivity Analyses have their own subsection?

Given that randomisation has been performed, please remove p values from Table 1 and Table S1.

Please clarify the difference between Table 1 and Table S1.

Please ensure consistency when reporting p values in terms of decimal places (e.g. Table 3).

Line 433: Should this not be P>0.05 for all given threshold for significance?

Line 437: Why is this reported as a difference when P=0.29?

Supplementary sections S1-4 are not cited in the text as far as I can see?

Please remove funding details from the Acknowledgements.

The Data Availability Statement can be removed from the end of the main text – please ensure all relevant information is captured in the relevant question on the Submission Form.

Comments from the reviewers:

Reviewer #1: Prof. Tazeen H Jafar and his colleague conducted the trail to assess the effectiveness of a multicomponent intervention on blood pressure control compared to usual care, and the findings showed a significant lowering of SBP level and a higher control rate of BP in management group than that in usual group. Several comments:

1. How about the comparability of the Clinics between two groups? What`s the proportion of education degree of the physicians, how many years did they work, how about the social economic level, etc.

2. Although four trainings (beginning, 3m, 12m, 24m) for doctors, three times (beginning, 12m, 24m) for nurses, were done in intervention group, except the last one, only 3 times for doctors, 2 times for nurses, if there were the other effect beside the "intervention". Subsidies on drug?

3. According to the design, there was no power to test the effect on the secondary outcomes, why added many items?

Reviewer #2: This was a cluster-based randomized program which compared multicomponent intervention vs usual care for control of BP in Singapore setting. In a nutshell, 8 clinics participated, 4 were intervention clinics and 4 were usual care clinics. The intervention included training physicians to stratify risk, 50% subsidy for fixed dose combination (FDC) of ARB and thiazide, motivational interview by nurses and regular telephone calls. A clinical research coordinator also performed evaluation at month 0, 12 and 24 in both groups. The intervention group was supposed to be seen by doctors/nurses at 3-4 monthly intervals. Over 400 patients were recruited in each group and the endpoint was SBP reduction at 24 months along with Framingham 10-year risk score and ACR. There was also mention of EQ5D evaluation and cost effectiveness analysis. About 80% of patients returned at month 24 for evaluation. The study was influenced by the COVID-19 outbreak.

The analysis was very technical using multiple statistical methods to compare clinic-based rather than individual patient-level data with adjustment for clusters and time trends. Analyses were also performed using imputed data with multiple modelling as well as using data of patients who returned at month 24 in a per protocol analysis. The conclusion was the intervention group achieved lower SBP by 3 mmHg and had an 1.5 odds ratio of achieving a BP<140/90 with 40% risk reduction of 10-year CVD risk score and reduction in ACR compared with usual care.

There is now consensus that the challenges in control of NCD lie in implementation which needs support from healthcare planners and payors in addition to providers and patients. Multicomponent interventions are also well accepted although its implementation depends on context which needs to take into consideration the usual care model for scalability and sustainability. It is here that the paper lacks granularity as to what the usual care model is. Do people in Singapore have a regular family doctor or do they shop around for doctors? For WHO defined essential lab tests (e.g. lipid, A1c, RFT) and medications (statins, ACEi, metformin, SU etc), do they have to co-pay or are they fully covered? Is there a ceiling for government coverage and do patients have to pay every time out of pocket to see their doctor or a nurse? Do patients get medications from the clinic or pharmacy and can they self purchase medications without a prescription? Are there huge differences amongst these clinics in terms of staff to patient ratio, care pathways and payment structure? All these factors may influence the quality of care in the intervention and usual care groups.

In its current format, it is challenging to visualize the implementation procedures and patient/provider adherence as well as pre and end of study clinical profiles in both groups. Apart from BP and some demographic data, there was no information on eGFR, lipids and glycaemic indexes, some of which were needed to compute the 10-year CVD risk score. There was mention of EQ5D but not reported in the paper. The cost-effectiveness analysis was very minimal and it is not clear what was the average number of man-hour involved (doctors/nurses), number of face to face visits and telephone consultations per patient during this 2-year program in the intervention group. What was the no of clinic visits in the usual care group? These information should have been captured by the project coordinator, indeed, what information was collected during these important evaluation visits at 0, 12 and 24 month.

The authors proposed that the model is applicable to many countries. If this is their message, then they have to be very specific on the implementation details. How many nurses and doctors were involved in each of these participating clinics? What is the provider:patient ratio? What were the drop out and adherence rates and why did people drop out? Why did patients do not join the program?

From a clinical perspective, were there any changes in other medications for lipids, BP and glucose? What were the changes in these risk factors that contribute to the 40% reduction in 10-year CVD risk score? Was there any difference in new cases of diabetes between the 2 groups? Since patients have to pay out of pocket for 50% of the FDC, how many patients agreed to the switch and persisted with the new medications? On this point, did patients have access to a list of free medications for BP control? At baseline, the patients were already on multiple drugs and what was the pattern of drug usage between the 2 groups at baseline and end of 24 months?

In the intervention group, were the patients informed of their BP goals and were they taught to self monitor BP? Indeed having a coordinator to relay these 2 messages every 6 months or so might go a long way while allowing the doctors and patients make their own choices in term of FU visits, drug regimens and lab tests. Indeed, this may be the reason why both groups had dropped their BP significantly with the yearly review by the project coordinator.

The authors should list the names of all 8 participating clinics and their PIs in the acknowledgement.

Reviewer #3: Alex McConnachie, Statistical Review

Jafar et al present the results of a cluster randomised trial of a multicomponent intervention in primary care to improve blood pressure control. This review considers the statistical aspects of the paper.

Overall, this is an excellent study, very well reported. The statistics are generally very good, and my comments are therefore relatively minor.

One difficulty with cluster randomised trials is always the timing of randomisation at the cluster level and recruitment of individual patients. My impression from the paper is that practices were randomised first, and then the patients were recruited. What is not clear is whether the patients were aware of the allocation of their practice at the time of giving consent to be included in the study, or when they provided their baseline data. Also, how long was it between randomisation of the practices and recruitment/baseline? What was the timing of the intervention - had any training taken place before the patients had completed their baseline assessments? Were the follow-ups conducted 12- and 24-months after each patient's baseline assessment, or relative to the randomisation of the practice? Line 228 refers to the 2-year follow-up post-randomisation, but was it actually 2 years post-baseline?

Line 231 refers to blood pressure <140/90 as an outcome, though there was an earlier reference (line 159) to <130/80 as a target for some patients. Was this lower target used for relevant patients in defining this outcome?

The sample size calculation appears correct, but it is not possible to replicate exactly without knowing what loss to follow-up was assumed.

The repeated measures approach to the analysis is good, and is described very well. The sensitivity analyses are all very sensible, though the definition of "per protocol" is not clear - I did find it in the footnote to one of the tables, but it could perhaps be included in the main text. In sensitivity analysis 4, I assume the covariates were baseline measures?

Some people would complain about reporting p-values for baseline differences, but I think it is a good thing with cluster randomised trials, and it is good that these comparisons take account of the clustering.

In Tables 2 and 3, we are given the mean and 95% CI, or percentage with 95% CI, in each randomised arm. This is not wrong, but for me, it does not show the distribution of the data in each group, since the CI depends on the sample size. Personally, I would rather see a summary of the data, i.e. mean and SD, or number and percentage, within each group.

Reporting ICC statistics (line 377) is good, though it would be good if these could be reported for all outcomes, perhaps within the supplement somewhere.

Finally, in Figure 1, the variance in cluster sizes in the two arms are reported as 21 and 13 "per cluster". I'm not sure that "per cluster" makes sense, since the units of variance is the square of the units of measurement. Would it be easier to report the SD of cluster sizes?

Reviewer #4: MS: PMEDICINE-D-22-00512R1

Summary:

This study investigates the effects of multicomponent intervention for managing hypertension. In this cluster randomized controlled trial (RCT) within a primary care setting in Singapore, an intervention that integrates CVD risk assessment, subsidised single pill combination drugs, motivational conversations and telephone follow-up was compared to standard care in lowering blood pressure (BP). After a follow-up of up to two years, the investigators report a reduction of 3 mmHg systolic BP and better control of hypertension in the active arm compared to standard care.

Comments:

This trial is indeed interesting in that factors beyond that of the drug therapy itself is being considered and reflect effectiveness of an intervention that are relevant to practice settings. The inclusion of CVD risk assessment is an important strength, as treatment benefit tends to be higher in people with a higher background risk, at least in absolute terms. However, an important limitation is that it recruited participants with baseline BP levels of 140/90 mmHg or higher. I would have thought that CVD risk assessment would have benefited those at increased risk but their baseline BP may be slightly below this BP thresholds (e.g., hypertension treatment may be initiated in patients with diabetes whose BP are below this threshold as some clinical guidelines would suggest).

The achieved BP difference (of just over 3 mmHg systolic) seems very modest - clinically important at a group level, but likely less so at individual level. Because the aim of the intervention is to control hypertension or bring BP levels below the 140/90 mmHg threshold, which is the same for both treatment arms, it seems that the added benefit of all the components only amounted to 3 mmHg of effectiveness - a huge undertaking for a modest impact. Rather than focusing on the various components, would it have been better to focus on a component of the intervention that is most likely to achieved a greater reduction, perhaps by ensuring reaching an optimal target that is well below 140/90 mmHg? Very few (<6%) were at 'low-risk' of CVD at baseline, and the vast majority were either at medium or high risk. What would have been the impact if everyone was simply offered single pill combination drugs - it would have simplified further the intervention, and even if the costs might be slightly higher, the benefit in reducing BP perhaps might be freater?

The inclusion criteria included BP levels of ≥140/≥90 mmHg based on mean of last 2 of 3 measurements. Since a good proportion of participants were on antihypertensive medications at baseline, does it mean that either these patients had poor BP control and/or had higher baseline BP if they were not on medications?

Making the most of BP measurements in reporting BP differences is indeed more appropriate than simply comparing BP changes from baseline. However, it is expected that BP measured a few months after randomisation tend not to reflect the full effects of the intervention since several adjustments/titration take place during this period, as well as regression to the mean/random variability in BP. I suggest that the authors consider a sensitivity analysis that exclude BP measured, say, in the first 3 or 6 months - it might be possible to demonstrate a bigger treatment effect than the 3 mmHg systolic BP being reported.

The difference in the albumin:creatinine ratio seems very substantive which is not in keeping with differences in other parameters between treatment arms. Was this an expected finding? Is this a plausible result given the small difference in achieved BP reduction between the comparison groups?

Any attachments provided with reviews can be seen via the following link:

[LINK]

22 Apr 2022

Submitted filename: Reviewer comments_PlosMed.docx

Click here for additional data file.

10 May 2022

Dear Dr. Jafar,

Thank you very much for re-submitting your manuscript "Integration of a Multicomponent Intervention for Hypertension into Primary Healthcare Services - A Cluster Randomised Controlled Trial" (PMEDICINE-D-22-00512R2) for review by PLOS Medicine.

I have discussed the paper with my colleagues and the academic editor and it was also seen again by four reviewers. I am pleased to say that provided the remaining editorial and production issues are dealt with we are planning to accept the paper for publication in the journal.

The remaining issues that need to be addressed are listed at the end of this email. Any accompanying reviewer attachments can be seen via the link below. Please take these into account before resubmitting your manuscript:

[LINK]

***Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.***

In revising the manuscript for further consideration here, please ensure you address the specific points made by each reviewer and the editors. In your rebuttal letter you should indicate your response to the reviewers' and editors' comments and the changes you have made in the manuscript. Please submit a clean version of the paper as the main article file. A version with changes marked must also be uploaded as a marked up manuscript file.

Please also check the guidelines for revised papers at http://journals.plos.org/plosmedicine/s/revising-your-manuscript for any that apply to your paper. If you haven't already, we ask that you provide a short, non-technical Author Summary of your research to make findings accessible to a wide audience that includes both scientists and non-scientists. The Author Summary should immediately follow the Abstract in your revised manuscript. This text is subject to editorial change and should be distinct from the scientific abstract.

We hope to receive your revised manuscript within 1 week. Please email us (plosmedicine@plos.org) if you have any questions or concerns.

We ask every co-author listed on the manuscript to fill in a contributing author statement. If any of the co-authors have not filled in the statement, we will remind them to do so when the paper is revised. If all statements are not completed in a timely fashion this could hold up the re-review process. Should there be a problem getting one of your co-authors to fill in a statement we will be in contact. YOU MUST NOT ADD OR REMOVE AUTHORS UNLESS YOU HAVE ALERTED THE EDITOR HANDLING THE MANUSCRIPT TO THE CHANGE AND THEY SPECIFICALLY HAVE AGREED TO IT.

Please ensure that the paper adheres to the PLOS Data Availability Policy (see http://journals.plos.org/plosmedicine/s/data-availability), which requires that all data underlying the study's findings be provided in a repository or as Supporting Information. For data residing with a third party, authors are required to provide instructions with contact information for obtaining the data. PLOS journals do not allow statements supported by "data not shown" or "unpublished results." For such statements, authors must provide supporting data or cite public sources that include it.

To enhance the reproducibility of your results, we recommend that you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. Additionally, PLOS ONE offers an option to publish peer-reviewed clinical study protocols. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols

Please review your reference list to ensure that it is complete and correct. If you have cited papers that have been retracted, please include the rationale for doing so in the manuscript text, or remove these references and replace them with relevant current references. Any changes to the reference list should be mentioned in the rebuttal letter that accompanies your revised manuscript.

Please note, when your manuscript is accepted, an uncorrected proof of your manuscript will be published online ahead of the final version, unless you've already opted out via the online submission form. If, for any reason, you do not want an earlier version of your manuscript published online or are unsure if you have already indicated as such, please let the journal staff know immediately at plosmedicine@plos.org.

If you have any questions in the meantime, please contact me or the journal staff on plosmedicine@plos.org.

We look forward to receiving the revised manuscript by May 17 2022 11:59PM.

Sincerely,

Callam Davidson,

Associate Editor

PLOS Medicine

plosmedicine.org

------------------------------------------------------------

Requests from Editors:

Please update your title to ‘Integration of a Multicomponent Intervention for Hypertension into Primary Healthcare Services in Singapore - A Cluster Randomised Controlled Trial’.

The Data Availability Statement (DAS) requires revision. If the data are not freely available, please describe briefly the ethical restriction that prevents you from sharing it.

Please note that line numbers in the comments below pertain to the tracked changes word document you provided.

Line 169: Please cite your protocol using our Supporting Information guidelines (e.g., S1 protocol, or similar).

Guidance can be found here: https://journals.plos.org/plosmedicine/s/supporting-information

Line 285: This sentence can now be removed as you already state this on line 256.

Table 1 and Line 409: With apologies for the conflicting advice I gave on the previous version, as noted by the statistical reviewer, in this particular case it is OK to present p-values for a comparison of baseline data (see comments from Reviewer 3 for more information).

Lines 688-690: Please remove funding information from the Acknowledgements.

Reference 8 contains COI information that can be removed.

Supporting Information: Please include references to sections S1-4 in the main text.

Figure 3 and Table 3: The terms gender and sex are not interchangeable (as discussed in http://www.who.int/gender/whatisgender/en/ ); please use the appropriate term.

Comments from Reviewers:

Reviewer #1: no other comments.

Reviewer #2: The revised paper is considerably improved and the authors have addressed all the concerns satisfactorily. A few points need clarification:

1. Line 172-173: Generic and branded SPC antihypertensive medications are also available, albeit they are not subsidized - are generic medications also NOT subsidized?

2. Ref 1 and 2 need correction

3. Ref 8 is mixed with declaration on conflicts of interest

Reviewer #3: Alex McConnachie, Statistical Review

Revision of "Integration of a Multicomponent Intervention for Hypertension into Primary Healthcare Services - A Cluster Randomised Controlled Trial", Jafar et al.

I thank the authors for their consideration of my original points.

Whilst I appreciate that the paper is clear that the study was not blinded, I think it is worth stating explicitly that the patients were aware of the group that their practice had been assigned to at the point of enrollment. This should also be recognised as a limitation.

This has implications for the presentation of the baseline data. As the editors say, for randomised trials it is not usually correct to present p-values for a comparison of baseline data. However, the fact that patients were aware of their allocated group at baseline data collection means that there could be systematic differences between the patients recruited in the two arms. We need reassurance in the form of statistical comparisons that this is not the case.

In Tables 2 and 3, I don't think adding the standard errors improves things (this is essentially the same information as given by the CIs). I am aware that the mean values reported are model predictions, which was my point; I would rather see a summary (mean and SD, or N and %) of the actual data in each group at each time point. Reporting model-predicted values and CIs is fine in the figures, and of course, we need the estimated between-group differences and ORs, but I feel it is important to present some summaries of the data that was actually collected.

Reviewer #4: I have no further comments.

Any attachments provided with reviews can be seen via the following link:

[LINK]

20 May 2022

Dear Dr Jafar,

On behalf of my colleagues and the Academic Editor, Dr David Peiris, I am pleased to inform you that we have agreed to publish your manuscript "Integration of a Multicomponent Intervention for Hypertension into Primary Healthcare Services in Singapore - A Cluster Randomised Controlled Trial" (PMEDICINE-D-22-00512R3) in PLOS Medicine.

Before your manuscript can be formally accepted you will need to complete some formatting changes, which you will receive in a follow up email. Please be aware that it may take several days for you to receive this email; during this time no action is required by you. Once you have received these formatting requests, please note that your manuscript will not be scheduled for publication until you have made the required changes.

In the meantime, please log into Editorial Manager at http://www.editorialmanager.com/pmedicine/, click the "Update My Information" link at the top of the page, and update your user information to ensure an efficient production process.

PRESS

We frequently collaborate with press offices. If your institution or institutions have a press office, please notify them about your upcoming paper at this point, to enable them to help maximise its impact. If the press office is planning to promote your findings, we would be grateful if they could coordinate with medicinepress@plos.org. If you have not yet opted out of the early version process, we ask that you notify us immediately of any press plans so that we may do so on your behalf.

We also ask that you take this opportunity to read our Embargo Policy regarding the discussion, promotion and media coverage of work that is yet to be published by PLOS. As your manuscript is not yet published, it is bound by the conditions of our Embargo Policy. Please be aware that this policy is in place both to ensure that any press coverage of your article is fully substantiated and to provide a direct link between such coverage and the published work. For full details of our Embargo Policy, please visit http://www.plos.org/about/media-inquiries/embargo-policy/.

To enhance the reproducibility of your results, we recommend that you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. Additionally, PLOS ONE offers an option to publish peer-reviewed clinical study protocols. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols

Thank you again for submitting to PLOS Medicine. We look forward to publishing your paper.

Sincerely,

Callam Davidson

Associate Editor

PLOS Medicine