Calcium Blockers

The n e w e ng l a n d j o u r na l of m e dic i n e
Original Article
A Community-Based Intervention for

Managing Hypertension in Rural South Asia
Tazeen H. Jafar, M.D., M.P.H., Mihir Gandhi, Ph.D.,
H. Asita de Silva, D.Phil., F.R.C.P., Imtiaz Jehan, F.C.P.S.,
Aliya Naheed, M.B., B.S., Ph.D., Eric A. Finkelstein, Ph.D.,
Elizabeth L. Turner, Ph.D., Donald Morisky, Sc.D.,
Anuradhani Kasturiratne, M.B., B.S., M.D., Aamir H. Khan, F.C.P.S.,
John D. Clemens, Ph.D., Shah Ebrahim, D.M., Pryseley N. Assam, Ph.D.,
and Liang Feng, Ph.D., for the COBRA-BPS Study Group*
A BS T R AC T
BACKGROUND
The burden of hypertension is escalating, and control rates are poor in low- and From the Program in Health Services and
middle-income countries. Cardiovascular mortality is high in rural areas. Systems Research (T.H.J., E.A.F., L.F.) and
the Center for Quantitative Medicine (M.G.),
Duke–NUS Medical School, the Department
METHODS of Renal Medicine, Singapore General Hos-
We conducted a cluster-randomized, controlled trial in rural districts in Bangla- pital (T.H.J.), and the Department of Biosta-
desh, Pakistan, and Sri Lanka. A total of 30 communities were randomly assigned tistics, Singapore Clinical Research Institute
(M.G., P.N.A.) — all in Singapore; the Duke
to either a multicomponent intervention (intervention group) or usual care (control Global Health Institute (T.H.J., E.A.F., E.L.T.)
group). The intervention involved home visits by trained government community and the Department of Biostatistics and Bio-
health workers for blood-pressure monitoring and counseling, training of physi- informatics, Duke University (E.L.T.) — both
in Durham, NC; the Center for Child Health
cians, and care coordination in the public sector. A total of 2645 adults with hyper- Research, Tampere University, Tampere, Fin-
tension were enrolled. The primary outcome was reduction in systolic blood pres- land (M.G.); the Clinical Trials Unit, Depart-
sure at 24 months. Follow-up at 24 months was completed for more than 90% of ment of Pharmacology (H.A.S.), and the De-
partment of Public Health (A.K.), Faculty of
the participants. Medicine, University of Kelaniya, Ragama,
Sri Lanka; the Department of Community
RESULTS Health Sciences (I.J.) and the Section of Car-
At baseline, the mean systolic blood pressure was 146.7 mm Hg in the intervention diology, Department of Medicine (A.H.K.),
group and 144.7 mm Hg in the control group. At 24 months, the mean systolic Aga Khan University, Karachi, Pakistan; the
International Center for Diarrheal Disease
blood pressure fell by 9.0 mm Hg in the intervention group and by 3.9 mm Hg in Research, Bangladesh, Dhaka, Bangladesh
the control group; the mean reduction was 5.2 mm Hg greater with the interven- (A.N., J.D.C.); the UCLA Fielding School of
tion (95% confidence interval [CI], 3.2 to 7.1; P<0.001). The mean reduction in Public Health, Department of Community
Health Sciences, Los Angeles (D.M.); and
diastolic blood pressure was 2.8 mm Hg greater in the intervention group than in the London School of Hygiene and Tropical
the control group (95% CI, 1.7 to 3.9). Blood-pressure control (<140/90 mm Hg) Medicine, London (S.E.). Address reprint re-
was achieved in 53.2% of the participants in the intervention group, as compared quests to Dr. Jafar at Duke–NUS Medical
School, 8 College Rd., Singapore 169857, Sin-
with 43.7% of those in the control group (relative risk, 1.22; 95% CI, 1.10 to 1.35). gapore, or at tazeen.jafar@duke-nus.edu.sg.
All-cause mortality was 2.9% in the intervention group and 4.3% in the control
*A list of the members of the COBRA-
group. BPS Study Group is provided in the
Supplementary Appendix, available at
CONCLUSIONS NEJM.org.
In rural communities in Bangladesh, Pakistan, and Sri Lanka, a multicomponent Drs. Gandhi, de Silva, Jehan, and Na-
intervention that was centered on proactive home visits by trained government heed contributed equally to this article.
community health workers who were linked with existing public health care infra- N Engl J Med 2020;382:717-26.
structure led to a greater reduction in blood pressure than usual care among adults DOI: 10.1056/NEJMoa1911965
with hypertension. (Funded by the Joint Global Health Trials scheme; COBRA-BPS Copyright © 2020 Massachusetts Medical Society.
ClinicalTrials.gov number, NCT02657746.)
n engl j med 382;8 nejm.org February 20, 2020 717

The New England Journal of Medicine
Downloaded from nejm.org on July 1, 2024. For personal use only.
No other uses without permission. Copyright © 2020 Massachusetts Medical Society. All rights reserved.
U
ncontrolled high blood pressure ties) of the districts Tangail and Munshiganj in
is the leading attributable risk factor for Bangladesh, Thatta in Pakistan, and Puttalam in
death globally.1 Treatment of hyperten- Sri Lanka. Because the intervention was delivered
sion reduces risk, but less than one third of through the health systems in the rural areas of
persons with hypertension have controlled blood these countries, a cluster-randomized, controlled
pressure.2-4 Asians have enhanced susceptibility to trial design was chosen to minimize contamina-
vascular disease.5-7 Uncontrolled blood pressure tion (i.e., to prevent participants in the control
is particularly prevalent in rural areas in low- group from actively or passively receiving some
and middle-income countries where health litera- or all of the multicomponent intervention). The
cy and health systems are weakest and case fatal- trial protocol and statistical analysis plan were
ity rates for cardiovascular disease are highest.8,9 published previously14,16 and are available with
Our previous trial in urban Pakistan sug- the full text of this article at NEJM.org. The au-
gested that a combined intervention of home thors vouch for the completeness and accuracy
health education delivered by community health of the data and for the fidelity of the trial to the
workers, coupled with training of physicians, protocol. The ethics review committee at each
lowered blood pressure and was cost-effective.10,11 participating institution approved the trial. All
However, the trial intervention used a privately the participants provided written informed con-
contracted health care workforce, which was not sent before screening. The conduct of the trial
integrated into the existing community infra- was independently reviewed by the trial steering
structure, and would not be sustainable or scal- committee and the data and safety monitoring
able. More than 30 trials on hypertension man- committee. The funders had no role in the de-
agement in low- and middle-income countries sign, conduct, analysis, or reporting of the trial.
have similar limitations.12,13
We conducted a cluster-randomized, con- Participants
trolled trial (Control of Blood Pressure and Risk The main eligibility criteria were an age of 40
Attenuation–Bangladesh, Pakistan, and Sri Lanka years or older and hypertension, defined as cur-
[COBRA-BPS]) in rural communities in three rent use of antihypertensive medications or per-
South Asian countries over a period of 2 years to sistently elevated blood pressure (systolic blood
evaluate the effectiveness of a scalable, multi- pressure ≥140 mm Hg or diastolic blood pres-
component intervention designed specifically for sure ≥90 mm Hg) based on each set of the last
hypertension management in rural areas.14 The two of three measurements from 2 separate days.
intervention was conceptually based on our pre- Pregnant women and persons with advanced ill-
vious intervention in urban Pakistan and was ness (e.g., those receiving dialysis or with liver
modified for delivery in rural settings in the failure), terminal illness, or an inability to travel
three South Asian countries.10,15 Additional com- to the clinic were excluded. (Additional details
ponents were added (blood-pressure monitoring are provided in the protocol.)
by government community health workers, check-
lists, care coordinators, and compensation for Randomization
additional services) in response to the results ofThe unit of randomization was a cluster of 250
a feasibility study.10,14,15 We hypothesized that a
to 300 households served by one or two com-
low-cost, multicomponent intervention integrated munity health workers and one government
into the existing public health system would be clinic.14,16 A total of 30 clusters were randomly
more effective than usual care in lowering blood selected from designated districts in the three
pressure among adults with hypertension in rural countries (10 per country). Randomization was
communities. stratified according to country and distance from
the government clinic (near [≤2 km] or far [>2 km]),
and clusters were assigned in a 1:1 ratio to either
Me thods
the multicomponent intervention (intervention
Trial Design group) or usual care (control group) with the use
The trial was a multicountry, cluster-randomized, of a computer-generated program (Table S1 in the
controlled trial in 30 rural villages (communi- Supplementary Appendix, available at NEJM.org).
718 n engl j med 382;8 nejm.org February 20, 2020

Managing Hypertension in Rur al South Asia
Trial Groups The fourth component was a designated hyper-

Multicomponent Intervention tension triage reception desk and hypertension
The details of the intervention components are care coordinator at the government clinics. A
described in the trial protocol; a condensed de- hypertension triage reception desk to reduce wait-
scription is provided here. The first component ing time was established at the intervention clin-
was blood-pressure monitoring and the use of ics. A hypertension care coordinator was appointed
checklists to guide monitoring and referral to to track participants with very poorly controlled
physicians. Government community health work- blood pressure.
ers were trained in measuring blood pressure The fifth component was compensation for
with the use of a digital blood-pressure monitor. additional health services and targeted subsidies.
They monitored participants’ blood pressures at Compensation was paid to the community health
home visits every 3 months. On the basis of a workers at the discretion of the local district
checklist (see the Supplementary Appendix), par- health office. The cost of medications and diag-
ticipants with very poorly controlled blood pres- nostics was borne primarily by the patients in
sure (systolic blood pressure ≥160 mm Hg or Bangladesh and Pakistan and by publicly funded
diastolic blood pressure ≥100 mm Hg) or those clinics in Sri Lanka, in accordance with the local
at high risk for cardiovascular disease were re- norms.
ferred to a physician at the government primary
care facility. Usual Care
The second component was home health edu- Usual care consisted of existing services in the
cation by government community health workers. community, with routine home visits by com-
Community health workers were trained in a munity health workers for maternal and child
curriculum regarding home health education and care only. The clinics did not have designated
in strategies regarding behavior-change commu- triage reception desks or care coordinators for
nication over a period of 5 days (see the Supple- hypertension.
mentary Appendix), followed by retraining in
2 months and then annually. Details of the train- Trial Assessments
ing curricula are provided in the protocol. Home Trained research staff who were unaware of
health education was delivered to all the partici- randomization status visited all households and
pants and their family members at home visits invited adults 40 years of age or older to partici-
every 3 months. All the participants were encour- pate. Written informed consent was obtained be-
aged to adhere to antihypertensive medications fore assessment for trial eligibility. Blood pres-
and to follow up with their physicians. A check- sure was measured with an Omron HEM-7300
list was completed by the community health automatic digital monitor (Omron Healthcare)
workers and submitted to their supervisors. with the person in a sitting position according
The third component was training of physi- to the standard protocol.19 Three blood-pressure
cians in blood-pressure monitoring, management readings were taken consecutively 3 minutes
of hypertension, and use of the checklist. A treat- apart with the use of a cuff of the appropriate
ment algorithm was based on the Joint National size. Persons with consistently elevated blood
Committee and 2013 European Society of Cardi- pressure (systolic blood pressure ≥140 mm Hg
ology guidelines.17 Generic antihypertensive med- or diastolic blood pressure ≥90 mm Hg based on
ications (thiazide-like diuretics, angiotensin-con- the last two of three readings) were visited again
verting–enzyme inhibitors or angiotensin-receptor after 2 weeks for remeasurement. Those with
blockers, and calcium-channel blockers) and persistently elevated blood pressure at the sec-
statins (for patients at high risk for cardiovascular ond screening visit were invited for enrollment.
disease) were used as indicated.17,18 The target blood Persons who were using antihypertensive medi-
pressure was a systolic blood pressure of less cations were also invited to enroll.
than 140 mm Hg and a diastolic blood pressure Information on sociodemographic character-
of less than 90 mm Hg (see the treatment algo- istics, health-seeking behavior, and associated
rithm in the Supplementary Appendix). Physicians costs was collected. Adherence to antihyperten-
were retrained in 2 months and annually thereafter. sive medications and statins was assessed by

means of the Morisky Medication Adherence value set (range, −0.865 to 1, with higher scores
Scale (MMAS-8; scores range from 0 to 8, with indicating better health).23
higher scores indicating better adherence).20-22 Other prespecified secondary outcomes were
Body-mass index (BMI) and waist circumference BMI, waist circumference, physical activity, smok-
were measured. ing status, intake of fruits and vegetables, dietary
All persons in both the intervention group sodium intake (urinary sodium excretion), labo-
and the control group with uncontrolled hyper- ratory measures (plasma glucose level, lipid
tension (systolic blood pressure ≥140 mm Hg or levels, estimated glomerular filtration rate, and
diastolic blood pressure ≥90 mm Hg) were asked urinary albumin-to-creatinine ratio), adverse
by the research staff to consult their local physi- events, new-onset diabetes, death from any cause,
cians. Persons with very high blood pressure and hospitalizations for cardiovascular disease.
(systolic blood pressure ≥180 mm Hg or dia- Because information on secondhand smoking
stolic blood pressure ≥120 mm Hg) or those was missing at baseline, the prespecified mea-
with acute associated symptoms (e.g., chest pain sure of the INTERHEART score for the risk of
or breathlessness) were referred urgently to the cardiovascular disease was replaced with the
district hospital. Framingham score for the 10-year risk of cardio-
Follow-up assessments of blood pressure were vascular disease.24 Although not prespecified,
conducted at home visits every 6 months in both the daily dose of antihypertensive medications
the intervention group and the control group. and the causes of death were evaluated. A pre-
Adverse events including falls, hypotension, cor- specified cost-effectiveness analysis14 is being
onary heart disease, stroke, and heart failure conducted, but costs of the intervention are re-
were reported. Hospitalizations and deaths were ported below.
tracked extensively (details are provided in the
protocol). Fasting blood and random urine sam- Statistical Analysis
ples were collected at baseline and 24-month The estimated sample size was 2550 partici-
visits. pants, under the assumptions of 85 participants
per cluster, 10 clusters per country, an intraclass
Trial Outcomes correlation coefficient of 0.02,10,15 80% retention,
The prespecified primary outcome was the mean and a two-sided type I error rate of 5%. The
change in systolic blood pressure from baseline trial had more than 99% power to detect a dif-
to 24 months. The mean of the second and third ference of 5 mm Hg in the change in systolic
blood-pressure readings was used for all analy- blood pressure between the two groups at 24
ses, and the first was discarded. months.16
Prespecified secondary outcomes included All analyses were performed with the use of
diastolic blood pressure, the percentage of par- the intention-to-treat principle. As prespecified
ticipants with blood-pressure control (systolic in the statistical analysis plan, for the primary
blood pressure <140 mm Hg and diastolic blood outcome analysis, the changes from baseline
pressure <90 mm Hg), blood-pressure response measurements were modeled with the use of a
(blood-pressure control or decline in systolic generalized linear mixed-effects model for re-
blood pressure by 5 mm Hg), very poorly con- peated measures based on a participant-level
trolled blood pressure (systolic blood pressure analysis.25 The primary outcome model included
≥160 mm Hg or diastolic blood pressure fixed effects for baseline systolic blood pressure,
≥100 mm Hg), use of and mean MMAS-8 scores country, distance of the cluster from the clinic,
for adherence to antihypertensive medications age, sex, trial group, time, and the interaction of
and statins, and participant-reported health sta- trial group with time. No imputation technique
tus according to the mean score on the visual- was used because the analysis model accounts
analogue scale of the EuroQol 5-Dimension for missing data and is valid under the missing-
5-Level questionnaire (EQ-5D-5L; range, 0 to at-random assumption. Similar analyses were
100, with higher scores indicating better health) used for secondary outcomes.25 We also con-
and the mean score on the EQ-5D-5L utility in- ducted post hoc sensitivity and exploratory analy-
dex calculated with the use of the Indonesian ses, as explained in the Supplementary Appendix.

The incremental cost of intervention delivery Table 1. Baseline Characteristics of the Participants.*
was prospectively estimated for each country with
the use of an activity-based costing approach Intervention Control
Characteristic (N = 1330) (N = 1315)
that quantified all nonrecurring labor, rental
space, materials, supplies, and services required Age — yr 58.5±11.2 59.0±11.8
to deliver the intervention. Further details are Female sex — no. (%) 877 (65.9) 824 (62.7)
presented in the Supplementary Appendix. Formally educated — no. (%) 834 (62.7) 725 (55.1)
Overweight or obese — no. (%)† 814 (61.2) 683 (51.9)
R e sult s Participant-reported heart disease — no. (%) 177 (13.3) 167 (12.7)
Participants Participant-reported stroke — no. (%) 165 (12.4) 159 (12.1)

Of 11,510 persons who were screened for the Diabetes — no. (%)‡ 374 (28.1) 308 (23.4)
trial, 2645 (23.0%) were enrolled from April 2016 Chronic kidney disease — no. (%)§ 558 (42.0) 549 (41.7)
through February 2017, with follow-up ending in Blood pressure — mm Hg
March 2019. The 24-month follow-up ended with Systolic 146.7±22.4 144.7±21.0
retention of 92.1% of the participants in the inter-
Diastolic 89.1±14.7 87.8±13.8
vention group and 89.3% of those in the control
Uncontrolled blood pressure — no. (%)¶ 934 (70.2) 907 (69.0)
group (Figs. S1 through S4). The baseline char-
acteristics were generally balanced between the Very poorly controlled blood pressure — no. (%)‖ 416 (31.3) 366 (27.8)
intervention group and the control group. The Current smoker — no. (%) 138 (10.4) 132 (10.0)
mean (±SD) age of the participants was 58.8±11.5 No. of antihypertensive medications — no. (%)
years, 64.3% were women, 25.8% had diabetes, 0 404 (30.4) 422 (32.1)
and 41.9% had chronic kidney disease. Blood 1 607 (45.6) 552 (42.0)
pressure was uncontrolled in 69.6% of the par-
2 257 (19.3) 258 (19.6)
ticipants and very poorly controlled in 29.6%
(Table 1 and Table S2). ≥3 62 (4.7) 83 (6.3)
* Plus–minus values are means ±SD. A total of 30 rural communities in Bangla

Implementation and Intervention Adherence desh, Pakistan, and Sri Lanka were randomly assigned to either a multicom-
In the intervention clusters during 2 years, 92.5% ponent intervention (intervention group) or usual care (control group).
† Overweight or obesity was defined as a body-mass index (the weight in kilo-
of the planned home health education checklists grams divided by the square of the height in meters) of 23.5 or more.
were completed; 91.5% of the participants re- ‡ Diabetes was defined as a fasting plasma glucose level of 126 mg or more per
ceived at least 80% of planned home visits (up to deciliter, the use of antidiabetes medications, or a previous diagnosis of diabetes.
§ Chronic kidney disease was defined as an estimated glomerular filtration rate
eight visits every 3 months over a period of 24 (calculated with the Chronic Kidney Disease Epidemiology Collaboration equa-
months) by community health workers for blood- tion on the basis of Pakistan data) of less than 60 ml per minute per 1.73 m2
pressure monitoring and home health education; of body-surface area or a urinary albumin-to-creatinine ratio of 30 or more
(with albumin measured in milligrams and creatinine in grams).
and 76.8% of physician management checklists ¶ Uncontrolled blood pressure was defined as a systolic blood pressure of
were completed for participants referred to clin- 140 mm Hg or more or a diastolic blood pressure of 90 mm Hg or more.
ics (Table S3). ‖ Very poorly controlled blood pressure was defined as a systolic blood pressure
of 160 mm Hg or more or a diastolic blood pressure of 100 mm Hg or more.
Blood-Pressure Outcomes
At baseline, the mean systolic blood pressure
was 146.7±22.4 mm Hg in the intervention group tolic blood pressure increased over time (Fig. 1A
and 144.7±21.0 mm Hg in the control group. At and Table 2).
24 months, the mean systolic blood pressure fell At baseline, the mean diastolic blood pressure
by 9.0 mm Hg (95% confidence interval [CI], 7.7 to was 89.1±14.7 mm Hg in the intervention group
10.4) in the intervention group and by 3.9 mm Hg and 87.8±13.8 mm Hg in the control group.
(95% CI, 2.5 to 5.3) in the control group. The From baseline to 24 months, the mean reduction
mean reduction in systolic blood pressure was in diastolic blood pressure was 2.8 mm Hg
5.2 mm Hg greater in the intervention group greater in the intervention group than in the con-
than in the control group (95% CI, 3.2 to 7.1; trol group (95% CI, 1.7 to 3.9) (Table 2). Con-
P<0.001). The between-group differences in sys- trolled blood pressure was achieved in 53.2% of

ence to antihypertensive medications and other

A Systolic Blood Pressure
2
secondary outcomes are provided in Table S4.)
Mean Change from Baseline
0
Other Secondary Outcomes
−2 −3.87 (95% CI, −5.27 to −2.47)
Control Participants in the intervention group reported
(mm Hg)
−4 better overall health status than those in the con-

−6 trol group: the mean increase in the score on the
Intervention
−8 EQ-5D-5L visual-analogue scale was greater by
−10
2.41 with the intervention (95% CI, 0.15 to 4.66).
−9.04 (95% CI, −10.42 to −7.65) Similar results were observed for the EQ-5D-5L
−12
Baseline 6 12 18 24 utility index. The mean MMAS-8 score for ad-
Month herence to statins increased more in the inter-
vention group than in the control group (mean
B Diastolic Blood Pressure difference, 0.42; 95% CI, 0.15 to 0.68).
2
Mean Change from Baseline
Safety and Mortality

0
Control −3.24 (95% CI, −4.03 to −2.45)
There was no intervention-related serious adverse
event in either group. All-cause mortality was
(mm Hg)
−2
2.9% (39 deaths) in the intervention group and
−4
Intervention 4.3% (56 deaths) in the control group (P = 0.06).
−6 The number of deaths from cardiovascular events
was lower in the intervention group (8 deaths,
−6.07 (95% CI, −6.85 to −5.29)
−8 0.6%) than in the control group (23 deaths,
Baseline 6 12 18 24
1.7%), (P = 0.006) (Table S5).
Month
No. of Participants Sensitivity and Subgroup Analyses
Control 1315 1225 1201 1187 1164
Intervention 1330 1278 The results with respect to the intervention ef-
1250 1243 1211
fect were consistent in sensitivity analyses that
Figure 1. Mean Change in Systolic Blood Pressure and Diastolic Blood used models for each time point separately (Ta-
Pressure over Time. ble S6). The cluster-level analysis, which takes a
Mean changes from baseline were estimated with a generalized linear mixed- summary measure for each cluster (as opposed
effects model for repeated measures for change in systolic blood pressure
(Panel A) or diastolic blood pressure (Panel B), with fixed effects for base-
to the primary analysis, which is at patient level
line systolic or diastolic pressure, country, distance of the cluster from the but accounts for clustering), also showed consis-
clinic, age, sex, time, and interaction of time with trial group and with ran- tent results (Table S7). The results with respect
dom effects for clusters. The I bars indicate 95% confidence intervals. to the intervention effect were also consistent in
the prespecified subgroups (Fig. 2 and Table S8),
post hoc subgroups (Fig. S5), and country-spe-
the participants in the intervention group, as cific analyses (Tables S9 through S11).
compared with 43.7% in the control group (rela-
tive risk, 1.22; 95% CI, 1.10 to 1.35) (Table 2). Cost of Intervention
The estimated cost of scale-up per eligible patient
Use of Antihypertensive Medication with hypertension in rural areas in Bangladesh,
At 24 months, the mean number of antihyperten- Pakistan, and Sri Lanka was $10.70, $10.50, and
sive medications per participant increased more in $4.70 (U.S. dollars), respectively (Table S12).
the intervention group than in the control group
(mean difference, 0.11) (Table 2), and the mean Discussion
increase in the daily dose was greater by 6.3 mg
(95% CI, 2.7 to 9.8). The mean MMAS-8 score In a cluster-randomized trial involving adults
for adherence to antihypertensive medications with hypertension in villages in Bangladesh,
increased more in the intervention group than in Pakistan, and Sri Lanka, blood-pressure control
the control group (mean difference, 0.60; 95% CI, was improved by a multicomponent interven-
0.24 to 0.96). (Detailed results regarding adher- tion, which included community health workers,

Table 2. Intervention Effect on Blood-Pressure Outcomes and Use of Antihypertensive Medications.*
Difference or Relative
Variable Intervention Control Risk (95% CI)†
No. of Mean or Percentage No. of Mean or Percentage

Participants (95% CI) Participants (95% CI)
Change in mean systolic blood pressure (mm Hg) from baseline‡
At mo 6 1278 −6.38 (−7.71 to −5.05) 1225 −5.08 (−6.43 to −3.74) −1.30 (−3.19 to 0.59)
At mo 12 1250 −4.11 (−5.47 to −2.74) 1201 −2.61 (−3.99 to −1.24) −1.50 (−3.43 to 0.44)
At mo 18 1243 −7.66 (−9.03 to −6.29) 1187 −4.54 (−5.93 to −3.16) −3.12 (−5.07 to −1.17)
At mo 24 1211 −9.04 (−10.42 to −7.65) 1164 −3.87 (−5.27 to −2.47) −5.17 (−7.13 to −3.20)§
Change in mean diastolic blood pressure (mm Hg) from baseline
At mo 6 1278 −4.10 (−4.85 to −3.35) 1225 −3.45 (−4.21 to −2.69) −0.65 (−1.72 to 0.42)
At mo 12 1250 −2.49 (−3.26 to −1.73) 1201 −1.63 (−2.40 to −0.85) −0.87 (−1.96 to 0.23)
n engl j med 382;8

At mo 18 1243 −5.29 (−6.07 to −4.50) 1187 −2.56 (−3.37 to −1.76) −2.72 (−3.85 to −1.60)
At mo 24 1211 −6.07 (−6.85 to −5.29) 1164 −3.24 (−4.03 to −2.45) −2.83 (−3.94 to −1.72)
nejm.org
Percentage of participants with controlled blood pressure¶
At mo 6 1278 48.4 (45.1 to 52.0) 1225 45.7 (42.4 to 49.2) 1.06 (0.96 to 1.17)
At mo 12 1250 42.2 (39.1 to 45.5) 1201 42.1 (39.0 to 45.5) 1.00 (0.90 to 1.12)
At mo 18 1243 48.7 (45.4 to 52.2) 1187 44.3 (41.0 to 47.8) 1.10 (0.99 to 1.22)
At mo 24 1211 53.2 (49.7 to 56.9) 1164 43.7 (40.4 to 47.3) 1.22 (1.10 to 1.35)

February 20, 2020
Change in mean no. of antihypertensive medications from baseline
At mo 6 1281 −0.01 (−0.05 to 0.04) 1233 0.00 (−0.05 to 0.04) 0.00 (−0.07 to 0.06)
At mo 12 1268 0.08 (0.03 to 0.12) 1212 0.02 (−0.03 to 0.07) 0.06 (−0.01 to 0.13)

At mo 18 1251 0.15 (0.10 to 0.20) 1194 0.05 (0.00 to 0.10) 0.10 (0.03 to 0.17)
At mo 24 1224 0.15 (0.10 to 0.20) 1174 0.05 (−0.01 to 0.10) 0.11 (0.04 to 0.18)
* Mean changes from baseline and percentages were estimated from a generalized linear mixed-model for repeated measures for the outcome variable, with fixed effects for baseline value
of the outcome (if a quantitative outcome), country, distance of the cluster from the clinic, age, sex, time, and interaction of time with trial group and with random effects for clusters.
† Difference (intervention − control) is shown for mean changes from baseline. Relative risk (intervention:control) is shown for percentages.
‡ The mean intraclass correlation coefficient for systolic blood pressure was 0.0115 (range, 0.0111 to 0.0123).
§ P<0.001.
¶ Controlled blood pressure was defined as a systolic blood pressure of less than 140 mm Hg and a diastolic blood pressure of less than 90 mm Hg.
723
Subgroup Intervention Control Intervention Control Difference (95% CI)

no. of participants mean change
in systolic blood pressure
(mm Hg)
All participants 1211 1164 −9.04 −3.87 −5.17 (−7.13 to −3.20)
Country
Bangladesh 404 389 −7.40 −3.00 −4.40 (−7.87 to −0.94)
Pakistan 410 402 −8.93 −4.04 −4.89 (−8.34 to −1.45)
Sri Lanka 397 373 −10.03 −3.79 −6.24 (−9.73 to −2.74)
Distance of cluster from clinic
Far 476 466 −8.50 −3.41 −5.09 (−8.30 to −1.89)
Near 735 698 −9.71 −4.31 −5.40 (−8.01 to −2.79)
Sex
Male 400 429 −8.70 −2.81 −5.88 (−8.74 to −3.03)
Female 811 735 −9.49 −4.61 −4.88 (−7.21 to −2.54)
Socioeconomic status
Low to middle 753 833 −8.11 −3.71 −4.39 (−6.63 to −2.16)
High 455 327 −9.92 −3.42 −6.50 (−9.41 to −3.59)
Very poorly controlled blood pressure
Yes 376 308 −12.16 −3.84 −8.32 (−11.31 to −5.34)
No 835 856 −7.26 −3.54 −3.72 (−5.87 to −1.57)
Receiving antihypertensive medication
Yes 842 786 −8.78 −3.73 −5.04 (−7.26 to −2.82)
No 369 378 −8.79 −3.34 −5.46 (−8.41 to −2.50)
−12 −10 −8 −6 −4 −2 0 2
Intervention Better Control Better
Figure 2. Subgroup Analyses for Change in Systolic Blood Pressure at 24 Months, According to Participant Characteristics at Baseline.
Mean changes and differences (intervention − control) were estimated with a generalized linear mixed-effects model for repeated mea-
surements for change in systolic blood pressure, with fixed effects for baseline systolic pressure, country, distance of the cluster from
the clinic, age, sex, time, and interaction among time, trial group, and subgroup and with random effects for clusters. Socioeconomic
status was defined as low to middle and high on the basis of an International Wealth Index range of the 67th percentile or lower and higher
than the 67th percentile, respectively, for each country sample separately. Participants with a systolic blood pressure of 160 mm Hg or
more or a diastolic blood pressure of 100 mm Hg or more were considered to have very poorly controlled blood pressure.
was tailored to the rural setting, and was deliv- each component of the intervention on the effect,
ered through the existing public health care infra- our previous work indicated synergies among the
structure. The intervention also increased adher- components10,11; the current trial suggests that
ence to antihypertensive medication and improved appropriate use of medications may have played
some aspects of participant-reported health at a substantial role. For example, participants with
an annual cost of less than $11 per patient. The elevated blood pressure were referred to clinics
major strengths of our trial are a cluster design; in which trained physicians prescribed a greater
rural settings in three countries, with stratifica- number and a higher dose of antihypertensive
tion according to the distance from the clinic; medications than in the control group, coordina-
the inclusion of all adults with hypertension tors facilitated tracking, and community health
(uncontrolled and controlled); excellent recruit- workers monitored blood pressure and reinforced
ment and retention rates; and a prespecified and messages about adherence during repeated home
prepublished statistical analysis plan.16 visits. The annual retraining of community
During this 24-month trial, the benefit of the health workers and physicians may have en-
intervention with respect to blood-pressure low- hanced their competencies over time.
ering increased with a longer duration of follow- Many systematic reviews largely from high-
up, which suggests potential longevity of the income countries have shown the benefit of
intervention effect. Although our trial was not multicomponent strategies on hypertension con-
designed to dissect the relative contributions of trol.26 Our trial is different from previous studies

that have either focused primarily on urban strategies for blood-pressure control are lacking.
populations12,13,27 or used technology to deliver Our low-cost intervention (<$11 per patient an-
interventions in rural environments with con- nually), if scaled up, might translate into sub-
flicting results on blood-pressure reduction.27-30 stantial reductions in premature deaths and dis-
Community health workers are an integral ability, as well as social and economic returns.38,39
part of the primary care infrastructure for the Discussions with provincial health departments
successful delivery of maternal and child health and national advisory committees are ongoing
care in South Asia,31 as well as in China, Mexico, to facilitate the scale-up of the intervention in
and Africa.32-34 Our findings show that commu- the three countries, with the same fidelity as
nity health workers who are employed in the implemented in the trial.
public sector can have an important role in man- We found that a multicomponent intervention
aging hypertension. for hypertension care, which centered on proac-
Our trial has limitations. First, the interven- tive home visits by trained government commu-
tion effect could have been underestimated be- nity health workers who were linked with exist-
cause participants in the usual-care group may ing public health care infrastructure, led to a
have modified their behavior in response to clinically meaningful reduction in blood pres-
blood-pressure measurements performed by re- sure in rural communities in Bangladesh, Paki-
searchers to assess outcomes. Second, the trial stan, and Sri Lanka.
was underpowered to detect changes in many
A data sharing statement provided by the authors is available
secondary outcomes. Third, the short duration with the full text of this article at NEJM.org.
meant that there was insufficient power to as- Supported by a grant (MR/N006178/1) from the Joint Global
sess cardiovascular events. However, a reduction Health Trials scheme, which includes the Medical Research
Council, the U.K. Department for International Development, the
of 2 mm Hg in systolic blood pressure has been National Institute for Health Research, and the Wellcome Trust.
associated with a reduction of 7 to 10% in the Disclosure forms provided by the authors are available with
risk of coronary heart disease, stroke, and re- the full text of this article at NEJM.org.
We thank all the investigators, coordinators, and staff of the
lated death.35-37 trial at the respective institutions, including at the International
Our findings have public health implications. Center for Diarrheal Disease Research, Bangladesh (Dhaka); Aga
Cardiovascular mortality continues to rise in low- Khan University (Karachi, Pakistan); the Faculty of Medicine,
University of Kelaniya (Ragama, Sri Lanka); the London School
and middle-income countries, especially in rural of Hygiene and Tropical Medicine (London); and Duke–NUS
areas with a high burden of poverty and frag- Medical School (Singapore, Singapore). We also thank all the
mented health systems.9 There is ample evidence members of the trial steering committee and the data and safety
monitoring committee (listed in the Supplementary Appendix)
of the benefit of blood-pressure reduction on and all the trial participants, without whose cooperation the
cardiovascular mortality; however, affordable trial would not have been possible.
References
1. GBD 2016 Risk Factors Collaborators. Global disparities of hypertension preva- HA, et al. Patients’ experiences on access-
Global, regional, and national compara- lence and control: a systematic analysis of ing health care services for management
tive risk assessment of 84 behavioural, population-based studies from 90 coun- of hypertension in rural Bangladesh, Pak-
environmental and occupational, and tries. Circulation 2016;134:441-50. istan and Sri Lanka: a qualitative study.
metabolic risks or clusters of risks, 1990- 5. Jafar TH, Islam M, Poulter N, et al. PLoS One 2019;14(1):e0211100.
2016: a systematic analysis for the Global Children in South Asia have higher body 9. Yusuf S, Rangarajan S, Teo K, et al.
Burden of Disease Study 2016. Lancet mass-adjusted blood pressure levels than Cardiovascular risk and events in 17 low-,
2017;390:1345-422. white children in the United States: a middle-, and high-income countries. N Engl
2. Chow CK, Teo KK, Rangarajan S, et al. comparative study. Circulation 2005;111: J Med 2014;371:818-27.
Prevalence, awareness, treatment, and con- 1291-7. 10. Jafar TH, Hatcher J, Poulter N, et al.
trol of hypertension in rural and urban 6. Shah AD, Kandula NR, Lin F, et al. Community-based interventions to pro-
communities in high-, middle-, and low- Less favorable body composition and adi- mote blood pressure control in a develop-
income countries. JAMA 2013; 310:959- pokines in South Asians compared with ing country: a cluster randomized trial.
68. other US ethnic groups: results from the Ann Intern Med 2009;151:593-601.
3. Geldsetzer P, Manne-Goehler J, Mar- MASALA and MESA studies. Int J Obes 11. Jafar TH, Islam M, Bux R, et al. Cost-
cus ME, et al. The state of hypertension (Lond) 2016;40:639-45. effectiveness of community-based strate-
care in 44 low-income and middle-income 7. Misra A, Tandon N, Ebrahim S, et al. gies for blood pressure control in a low-
countries: a cross-sectional study of na- Diabetes, cardiovascular disease, and income developing country: findings from
tionally representative individual-level data chronic kidney disease in South Asia: cur- a cluster-randomized, factorial-controlled
from 1·1 million adults. Lancet 2019;394: rent status and future directions. BMJ trial. Circulation 2011;124:1615-25.
652-62. 2017;357:j1420. 12. He J, Irazola V, Mills KT, et al. Effect
4. Mills KT, Bundy JD, Kelly TN, et al. 8. Legido-Quigley H, Naheed A, de Silva of a community health worker-led multi-

component intervention on blood pres- women and men. Blood Press Monit 2009; health intervention for people assessed at
sure control in low-income patients in 14:87-90. high cardiovascular disease risk in rural
Argentina: a randomized clinical trial. 20. Morisky DE, Ang A, Krousel-Wood M, India. PLoS One 2019;14(3):e0213708.
JAMA 2017;318:1016-25. Ward HJ. Predictive validity of a medica- 30. Jafar TH, Samad Z, Bloomfield GS.
13. Anand TN, Joseph LM, Geetha AV, tion adherence measure in an outpatient Parallel community solutions for cardio-
Prabhakaran D, Jeemon P. Task sharing setting. J Clin Hypertens (Greenwich) vascular risk reduction. Lancet 2019;394:
with non-physician health-care workers 2008;10:348-54. 1207-8.
for management of blood pressure in 21. Bress AP, Bellows BK, King JB, et al. 31. Bhutta ZA, Soofi S, Cousens S, et al.
low-income and middle-income countries: Cost-effectiveness of intensive versus Improvement of perinatal and newborn
a systematic review and meta-analysis. standard blood-pressure control. N Engl J care in rural Pakistan through communi-
Lancet Glob Health 2019;7(6):e761-e771. Med 2017;377:745-55. ty-based strategies: a cluster-randomised
14. Jafar TH, Jehan I, de Silva HA, et al. 22. Berlowitz DR, Foy CG, Kazis LE, et al. effectiveness trial. Lancet 2011;377:403-12.
Multicomponent intervention versus usual Effect of intensive blood-pressure treat- 32. Huang W, Long H, Li J, et al. Delivery
care for management of hypertension in ment on patient-reported outcomes. N Engl of public health services by community
rural Bangladesh, Pakistan and Sri Lanka: J Med 2017;377:733-44. health workers (CHWs) in primary health
study protocol for a cluster randomized 23. Purba FD, Hunfeld JAM, Iskandarsyah care settings in China: a systematic re-
controlled trial. Trials 2017;18:272. A, et al. Quality of life of the Indonesian view (1996-2016). Glob Health Res Policy
15. Jafar TH, Silva Ad, Naheed A, et al. general population: test-retest reliability 2018;3:18.
Control of blood pressure and risk attenu- and population norms of the EQ-5D-5L 33. Carvajal SC, Huang S, Bell ML, et al.
ation: a public health intervention in rural and WHOQOL-BREF. PLoS One 2018; Behavioral and subjective health changes
Bangladesh, Pakistan, and Sri Lanka: fea- 13(5):e0197098. in US and Mexico border residing partici-
sibility trial results. J Hypertens 2016;34: 24. D’Agostino RB Sr, Vasan RS, Pencina pants in two promotora-led chronic dis-
1872-81. MJ, et al. General cardiovascular risk pro- ease preventive interventions. Health Educ
16. Gandhi M, Assam PN, Turner EL, file for use in primary care: the Framing- Res 2018;33:522-34.
Morisky DE, Chan E, Jafar TH. Statistical ham Heart Study. Circulation 2008;117: 34. Singh P, Sachs JD. 1 Million commu-
analysis plan for the Control of Blood 743-53. nity health workers in sub-Saharan Africa
Pressure and Risk Attenuation–Rural 25. Tango T. Repeated measures design by 2015. Lancet 2013;382:363-5.
Bangladesh, Pakistan, Sri Lanka (COBRA- with generalized linear mixed models for 35. Psaty BM, Lumley T, Furberg CD, et al.
BPS) trial: a cluster randomized trial for a randomized controlled trials. Boca Raton, Health outcomes associated with various
multicomponent intervention versus usual FL:CRC Press, 2017. antihypertensive therapies used as first-
care in hypertensive patients. Trials 2018; 26. Mills KT, Obst KM, Shen W, et al. line agents: a network meta-analysis. JAMA
19:658. Comparative effectiveness of implemen- 2003;289:2534-44.
17. Mancia G, Fagard R, Narkiewicz K, tation strategies for blood pressure con- 36. Ettehad D, Emdin CA, Kiran A, et al.
et al. 2013 ESH/ESC guidelines for the trol in hypertensive patients: a systematic Blood pressure lowering for prevention of
management of arterial hypertension: the review and meta-analysis. Ann Intern Med cardiovascular disease and death: a sys-
Task Force for the management of arterial 2018;168:110-20. tematic review and meta-analysis. Lancet
hypertension of the European Society of 27. Schwalm JD, McCready T, Lopez-Jara- 2016;387:957-67.
Hypertension (ESH) and of the European millo P, et al. A community-based com- 37. Ford ES, Ajani UA, Croft JB, et al. Ex-
Society of Cardiology (ESC). J Hypertens prehensive intervention to reduce cardio- plaining the decrease in U.S. deaths from
2013;31:1281-357. vascular risk in hypertension (HOPE 4): coronary disease, 1980–2000. N Engl J Med
18. James PA, Oparil S, Carter BL, et al. a cluster-randomised controlled trial. Lan- 2007;356:2388-98.
2014 evidence-based guideline for the cet 2019;394:1231-42. 38. Bertram MY, Sweeny K, Lauer JA, et al.
management of high blood pressure in 28. Joshi R, Thrift AG, Smith C, et al. Investing in non-communicable diseases:
adults: report from the panel members Task-shifting for cardiovascular risk fac- an estimation of the return on investment
appointed to the Eighth Joint National tor management: lessons from the Global for prevention and treatment services.
Committee (JNC 8). JAMA 2014;311:507- Alliance for Chronic Diseases. BMJ Glob Lancet 2018;391:2071-8.
20. Health 2018;3:Suppl 3:e001092. 39. Jafar TH, Haaland BA, Rahman A, et al.
19. Grim CE, Grim CM. The Omron Elite 29. Peiris D, Praveen D, Mogulluru K, et al. Non-communicable diseases and injuries
7300W home blood pressure monitor SMARThealth India: a stepped-wedge, clus- in Pakistan: strategic priorities. Lancet
passes the European Society of Hyperten- ter randomised controlled trial of a com- 2013;381:2281-90.
sion International Validation Protocol for munity health worker managed mobile Copyright © 2020 Massachusetts Medical Society.


Calcium Blockers

Uploaded by

Copyright:

Available Formats

Calcium Blockers

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Calcium Blockers

Uploaded by

Copyright:

Available Formats

The n e w e ng l a n d j o u r na l of m e dic i n e

A Community-Based Intervention for

ClinicalTrials.gov number, NCT02657746.)

n engl j med 382;8 nejm.org February 20, 2020 717

718 n engl j med 382;8 nejm.org February 20, 2020

The New England Journal of Medicine

Trial Groups The fourth component was a designated hyper-

n engl j med 382;8 nejm.org February 20, 2020 719

720 n engl j med 382;8 nejm.org February 20, 2020

The New England Journal of Medicine

Participants Participant-reported stroke — no. (%) 165 (12.4) 159 (12.1)

* Plus–minus values are means ±SD. A total of 30 rural communities in Bangla­

n engl j med 382;8 nejm.org February 20, 2020 721

ence to antihypertensive medications and other

−4 better overall health status than those in the con-

Safety and Mortality

722 n engl j med 382;8 nejm.org February 20, 2020

The New England Journal of Medicine

No. of Mean or Percentage No. of Mean or Percentage

n engl j med 382;8

The New England Journal of Medicine

Downloaded from nejm.org on July 1, 2024. For personal use only.

Subgroup Intervention Control Intervention Control Difference (95% CI)

Intervention Better Control Better

724 n engl j med 382;8 nejm.org February 20, 2020

The New England Journal of Medicine

n engl j med 382;8 nejm.org February 20, 2020 725

726 n engl j med 382;8 nejm.org February 20, 2020

The New England Journal of Medicine

You might also like

* Plus–minus values are means ±SD. A total of 30 rural communities in Bangla