Hypertension 2020 76 2 333-41
Hypertension 2020 76 2 333-41
Hypertension 2020 76 2 333-41
Abstract—Elevated blood pressure remains the single biggest risk factor contributing to the global burden of disease and
mortality. May Measurement Month is an annual global screening campaign aiming to improve awareness of blood
pressure at the individual and population level. Adults (≥18 years) recruited through opportunistic sampling were screened
at sites in 92 countries during May 2019. Ideally, 3 blood pressure readings were measured for each participant, and data
on lifestyle factors and comorbidities were collected. Hypertension was defined as a systolic blood pressure ≥140 mm Hg,
or a diastolic blood pressure ≥90 mm Hg (mean of the second and third readings) or taking antihypertensive medication.
When necessary, multiple imputation was used to estimate participants’ mean blood pressure. Mixed-effects models were
used to evaluate associations between blood pressure and participant characteristics. Of 1 508 130 screenees 482 273
(32.0%) had never had a blood pressure measurement before and 513 337 (34.0%) had hypertension, of whom 58.7%
were aware, and 54.7% were on antihypertensive medication. Of those on medication, 57.8% were controlled to <140/90
mm Hg, and 28.9% to <130/80 mm Hg. Of all those with hypertension, 31.7% were controlled to <140/90 mm Hg,
and 350 825 (23.3%) participants had untreated or inadequately treated hypertension. Of those taking antihypertensive
medication, half were taking only a single drug, and 25% reported using aspirin inappropriately. This survey is the
largest ever synchronized and standardized contemporary compilation of global blood pressure data. This campaign is
Received February 11, 2020; first decision February 14, 2020; revision accepted May 14, 2020.
From the Imperial Clinical Trials Unit (T.B., S.C., N.R.P.) and Department of Primary Care and Public Health, Imperial College London, United Kingdom
(T.B.); Faculty of Medicine, University of New South Wales, George Institute for Global Health, Sydney, Australia (A.E.S.); South Africa Medical Research
Council, North-West University, Potchefstroom (A.E.S.); School of Medicine, Hypertension Center STRIDE-7 National and Kapodistrian University of
Athens Third Department of Medicine, Sotiria Hospital, Greece (G.S.S.); Department of Medical and Surgical Sciences, Universita di Bologna, Italy (C.B.);
Kidney Research Centre, Ottawa Hospital Research Institute, University of Ottawa, Canada (D.B.); Federation University Australia, Ballarat, VIC (F.C.);
Neurology Section, University of Santo Tomas Hospital, Philippines (A. Diaz); Faculty of Medicine, Eduardo Mondlane University, Maputo, Mozambique
(A. Damasceno, D.P.); Internal Medicine, Hospital San Martin de la Plata, Argentina (W.E.); Centre for Chronic Conditions and Injuries, Public Health
Foundation for India, Gurugram, Haryana (A.P.J.); Faculty of Medicine, University of British Colombia, Vancouver, Canada (N.K.); Department of
Preventive Cardiology, National Cerebral and Cardiovascular Centre, Suita, Japan (Y.K.); Department of General Medicine, Babu Banarasi Das University,
Lucknow, Uttar Pradesh, India (A. Maheshwari); Hospital Italiano de San Justo, Departamento Clínica Médica, Sección Hipertensión Arterial, San Justo,
Argentina (M.J.M.); Institute of Hypertension, Rural Health Progress Trust, Maharashtra, India (A. More); Nepal Development Society, Chitwan, Nepal
(D.N.); Welch Center for Prevention, Epidemiology, and Clinical Research, Johns Hopkins University (D.N.); Department of Clinical Sciences, Lund
University, Skane University Hospital, Malmo, Sweden (P.N.); Indian Association of Parenteral and Enteral Nutrition, India (M.P.); Hospital Universitario
Fundación Favaloro, Buenos Aires, Argentina (A.R.); Hypertension Clinic, Instituto Cardiovascular, Argentinian Society of Hypertension, Buenos Aires,
Argentina (P.R.); Dobney Hypertension Centre, School of Medicine, Royal Perth Hospital Unit, University of Western Australia, Perth, Australia (M.S.);
Department of Cardiovascular and Renal Research, Institute of Molecular Medicine, University of Southern Denmark (U.M.S.); Division of Cardiovascular
Sciences, University of Manchester, United Kingdom (M.T.); CARIM School for Cardiovascular Diseases, Maastricht University, the Netherlands (T.U.);
Department of Pharmacology and Experimental Therapeutics and the Whitaker Cardiovascular Institute, Boston University School of Medicine (R.W.);
The Shanghai Institute of Hypertension, Rujin Hospital, Shanghai Jiaotong University School of Medicine, China (J.W.); and Institute of Cardiovascular
Sciences, University College London, United Kingdom (B.W.).
*A list of all MMM Investigators is given in the Data Supplement.
This paper was sent to Morris J. Brown, Guest Editor, for review by expert referees, editorial decision, and final disposition
The Data Supplement is available with this article at https://www.ahajournals.org/doi/suppl/10.1161/HYPERTENSIONAHA.120.14874.
Correspondence to Neil R. Poulter, Imperial Clinical Trials Unit, Imperial College London, 68 Wood Ln, London, W12 7RH. Email n.poulter@imperial.
ac.uk or Thomas Beaney, Imperial Clinical Trials Unit, Imperial College London, W12 7RH, United Kingdom. Email thomas.beaney@imperial.ac.uk
© 2020 The Authors. Hypertension is published on behalf of the American Heart Association, Inc., by Wolters Kluwer Health, Inc. This is an open access
article under the terms of the Creative Commons Attribution Non-Commercial License, which permits use, distribution, and reproduction in any medium,
provided that the original work is properly cited and is not used for commercial purposes.
Hypertension is available at https://www.ahajournals.org/journal/hyp DOI: 10.1161/HYPERTENSIONAHA.120.14874
333
334 Hypertension August 2020
needed as a temporary substitute for systematic blood pressure screening in many countries worldwide. (Hypertension.
2020;76:333-341. DOI: 10.1161/HYPERTENSIONAHA.120.14874.) Data Supplement •
Key Words: adults ◼ awareness ◼ blood pressure ◼ hypertension ◼ risk factor ◼ screening ◼ treatment
Organization–derived single-age world-standard population taken before. There was moderate negative correlation be-
was used, according to the Surveillance, Epidemiology, and tween the proportion of participants never having had a BP
End Results (SEER) group, and assuming an equal ratio of measured within each country and gross national income
females to males.13 (r=−0.32, P=0.002).
For those participants missing either the second or third The majority of screening took place in hospitals or clinics
BP measurement (or both), multiple imputation using chained (36.0%), with 25.7% in outdoor public areas, 9.1% in indoor
equations was used to estimate the missing mean reading, public areas, 8.2% in the workplace, and 3.4% in pharma-
to provide better comparison across all participants. This cies (with 17.6% unrecorded or recorded as other). Of all
assumed missingness was Missing-at-Random and therefore screenees, 304 101 (20.2%) had participated in either MMM
dependent on the observed data only. Two separate imputa- 2017 or MMM 2018 (or both).
tion models were run: the first, complete imputation model, There was a wide geographic spread of participants,
imputed the missing systolic, and diastolic BP value for only with the majority screened in South Asia (31.3%), followed
those individuals with fully recorded data on age, sex, eth- by East Asia (18.6%) and the Americas (17.4%; Table 1).
nicity, use of antihypertensive medication, and where sex Across 7 geographic regions, the distribution of age and sex
was not recorded as other. Also included were all variables of screenees varied significantly. The highest mean age (51.2
included in the subsequent analyses, following guidance from years) was found in Europe and the lowest mean age (40.8
White et al.14 Variables which were used to compute the vari- years) in Sub-Saharan Africa. In Europe, 61.3% of those
ables within the analysis models (such as the individual BP screened were women, while in South Asia, 43.0% were
readings used to calculate the mean reading) were also in- women. Significant differences were also seen in the pro-
cluded, following the just another variable approach.15 portions of participants on antihypertensive medication, with
To handle cases missing the second or third BP measure- 30.0% on treatment in South-East Asia and Australasia and
ment (or both) and ≥1 of age, sex, ethnicity, or use of antihy- only 9.6% on treatment in Sub-Saharan Africa (Table 1 and
pertensive medication, a second, reduced imputation model Table S2 in the Data Supplement).
was run, imputing the missing systolic and diastolic BP values Globally, 280 958 (18.6%) participants were taking anti-
based only on the available BP readings. Imputed results from hypertensive medication, and of the 203 719 with a recorded
the 2 models were combined, with the imputations from the number of medication classes, 53.1% were taking a single
reduced imputation model used only where the values could medication, and 33.3%, 9.5%, and 3.2% were taking 2, 3,
not be imputed by the complete imputation model. A total of and 4, respectively. Only 0.9% were taking ≥5 antihyper-
25 imputations were created, corresponding to the percentage tensive drugs.
of missing data in the mean BP readings and aiming for a The characteristics for all 1 508 130 participants globally
Monte Carlo error of the estimates at under 10%.15 A full de- are given in Table S3. Of all participants, 88.6% had a doc-
scription of the imputation models and sensitivity analyses umented ethnicity of whom the majority reported their eth-
can be found in the Data Supplement. nicity as South Asian (33.7%) or East Asian (18.9%). Of all
Analysis of measures of association used only those indi- screenees, 116 369 (7.7%) reported having diabetes mellitus
viduals with complete data on age, sex, ethnicity, and use (either type I or type II), 55 189 (3.7%) a history of myocardial
of antihypertensive medication, and imputations performed infarction, 36 667 (2.4%) a history of stroke, 184 225 (12.2%)
under the complete imputation model. Linear mixed-effects were current smokers, and 82 726 (5.5%) reported drinking
models were run separately for systolic and diastolic BP, alcohol at least once per week. Among women, 17 762 (2.3%)
assuming a random intercept model to account for country- were pregnant at the time of screening, whereas 19 120 (2.5%)
level clustering effects. It was decided a priori to adjust for reported a history of hypertension in a previous pregnancy.
age, sex, and antihypertensive medication, along with an in- The mean BMI was 25.0 kg/m2 (SD 5.3) in women and 25.1
teraction between age and sex, given the known strong effects kg/m2 (SD 4.7) in men. 197 021 (25.4%) women and 214 395
of these variables on BP. Age, as a continuous variable, was (29.5%) men were overweight, and 104 690 (13.5%) women
incorporated into models as a restricted cubic spline with 5 and 89 170 (12.3%) men were obese.
knots to allow for flexibility in its relationship with BP.
BP Readings
Results Based on the inclusion criteria, all participants had at least
one BP reading and 1 133 008 (75.1%) had all 3 BP read-
Study Participants ings recorded with a further 119 669 (13.2%) having at least
Data on 1 521 974 participants were submitted from 92 coun- two readings. Analysis of only those with all three readings
tries during MMM 2019. After data cleaning, and excluding showed that BP fell, on average, by 3.1/1.8 mmHg, from a
participants who did not have at least one valid BP reading mean of 126.0/78.9 mmHg for the first reading to a mean of
(see appendix), 1 508 130 participants were included in the 122.9/77.1 mmHg for the third reading, whereas the corre-
study. Of these, 15.8% were submitted via the mobile appli- sponding proportion with hypertension fell from 37.6% to
cation. Recording of key demographic factors was improved 33.6%. The mean of the second and third readings identi-
from the previous year, with 1.0% of participants missing data fied the lowest proportion of participants with hypertension
on age and 0.4% missing data on sex. (33.5%), compared with any single or combination of meas-
Of all screenees, 482 273 (32.0%: 28.4% of women and ures despite a higher average BP (123.8/77.7 mmHg) com-
35.9% of men) reported never having had a BP measurement pared to the mean of the third reading (see Table S4).
336 Hypertension August 2020
Table 1. Total Participants and Distribution of Age, Sex, and Use of Antihypertensive Medication, Worldwide and by Region
Female Male
On Antihypertensive
Region Total Participants Total Mean Age, y Total Mean Age, y Medication
South Asia 471 302 (31.3%) 202 379 (43.0%) 42.2 267 590 (56.9%) 43.2 59 514 (12.6%)
East Asia 280 863 (18.6%) 148 843 (53.3%) 49.2 130 618 (46.7%) 50.0 47 060 (16.8%)
Americas 261 676 (17.4%) 156 615 (60.0%) 49.2 104 368 (40.0%) 50.0 75 056 (28.7%)
Sub-Saharan Africa 177 692 (11.8%) 92 921 (52.5%) 40.8 84 100 (47.5%) 40.8 17 114 (9.6%)
Southeast Asia and Australasia 121 767 (8.1%) 66 886 (55.1%) 45.1 54 493 (44.9%) 45.3 36 546 (30.0%)
Europe 107 608 (7.1%) 65 008 (61.3%) 50.5 41 079 (38.7%) 52.2 30 171 (28.0%)
Northern Africa and Middle East 87 222 (5.8%) 42 787 (49.2%) 41.8 44 103 (50.7%) 43.5 15 497 (17.8%)
Worldwide 1 508 130 775 439 (51.6%) 45.7 726 351 (48.4%) 45.8 280 958 (18.6%)
Four hundred and ten participants with sex recorded as other and 5930 participants with sex unknown not shown in the table.
Further analyses make use of the mean of the second and mm Hg and 28.9% controlled to <130/80 mm Hg. Of all hy-
third BP reading for each participant, as the most conserva- pertensive participants, 31.7% were controlled to <140/90
tive estimate. Where either, or both of the second and third mm Hg and 15.8% to <130/80 mm Hg.
BP readings were missing, multiple imputation using chained Of those participants not taking antihypertensive medi-
equations was performed to estimate the missing mean reading cation, 232 379 (18.9%) were found to have hypertension. In
based on observed data. Measurements for a total of 372 120 total, 350 825 (23.3%) participants were found to have un-
participants were imputed—imputations for 201 810 partici- treated or inadequately treated hypertension. Of these, 47.9%
pants from the complete imputation model and imputations had a BP in the range 140/90 to 149/94 mm Hg, and 25.5%
for 170 310 participants from the reduced imputation model. had a BP in the range 150/95 to 159/99 mm Hg (Table S5).
Worldwide, the mean BP (based on 1 136 010 individu- 6.6% of participants with hypertension, 6.6% had a BP over
als with the second and third BP reading available) was 180/110 mm Hg.
123.7/77.7 mm Hg before imputation and, following imputa- Using the lower threshold of systolic BP ≥130 mm Hg
tion, of all 1 508 130 participants, was 124.1/77.7 mm Hg. The or diastolic BP ≥ 80 mm Hg (or in those on antihypertensive
mean systolic and diastolic BPs, worldwide and by region, are medication) to diagnose hypertension, 775 068 (51.4%) of
displayed in Table S8, before and after standardization for age screenees were identified as hypertensive.
and sex. After imputation, and standardizing for age and sex, Sensitivity analyses were performed comparing results
in those not taking antihypertensive medication, the mean BP from the complete case analysis to the analyses using imputa-
was 121.6/76.7 mm Hg and in those taking antihypertensive tions from the reduced imputation model, complete imputa-
medication was 130.8/81.7 mm Hg. tion model, and the combined imputation model, which are
shown in Tables S6 and S7. The estimates from each model
Participants With Hypertension were similar, with the global proportion with hypertension,
Following imputation, of all 1 508 130 participants, 513 337 and raised BP stratified by medication use within a 1.0% ab-
(34.0%) had hypertension (Table 2). Of those with hyperten- solute range. In the complete case analysis, the proportion of
sion, 58.7% were aware of their diagnosis, and 54.7% were on participants with hypertension was 33.5% out of a total of
antihypertensive medication. Of the 280 958 participants on 1 136 010. The corresponding proportions from the reduced
medication, 162 512 (57.8%) had a BP controlled to <140/90 imputation and complete imputation models were 34.1% and
Table 2. Participant Numbers and Proportions With Hypertension, Proportions Aware, Treated, and Controlled, Worldwide and by Region
33.1%, respectively, compared with 34.0% in the combined to females until the age of about 80 years, after which the
imputation model. mean systolic was higher in females. Similarly, diastolic BP
After standardization for age and sex, the proportion with was higher in males until 80 years, after which there were no
hypertension worldwide reduced slightly to 32.5%, with ab- significant differences between the sexes. Increasing heart rate
solute reductions in the proportion with hypertension of >5% showed a strong linear association with increasing diastolic
in East Asia, the Americas, and Europe and an increase in the BP, but a weaker, less clear relationship was apparent with
proportion with hypertension in Sub-Saharan Africa, South systolic BP (Figure S2 and Table S10).
Asia, and Northern Africa, and the Middle East. The propor- Of all risk factors analyzed, reported use of antihyper-
tions of hypertensive awareness, and proportions with raised tensive medication and a previous diagnosis of hypertension
BP stratified by antihypertensive medication use, following were the strongest predictors of higher levels of systolic and
standardization, are given in Table 3. diastolic BP. After adjusting for age and sex, participants tak-
ing antihypertensive medication had a higher mean systolic
Medication Use BP (8.8 mm Hg higher, P<0.001) and higher diastolic BP
In participants taking a single antihypertensive medication, (3.7 mm Hg higher, P<0.001) compared with those not taking
39.3% were uncontrolled, and in those taking 2 drug classes, medication (Figure 1 and Table S11). After adjusting for age,
44.8% were uncontrolled. Proportions with uncontrolled BP sex, and antihypertensive medication use, those with known
were similar in those on 3 (47.9%), 4 (48.0%) or ≥5 (44.5%) hypertension had a significantly higher mean systolic (8.0
medications (Table S9). In total, 17 532 participants were de- mm Hg higher, P<0.001) and diastolic BP (4.5 mm Hg higher,
fined as treatment-resistant, which is 8.6% of the hypertensive P<0.001) compared with those without known hypertension.
population included in the study for whom data on the number Women who reported a history of previous hypertension in
of medications were available. pregnancy had significantly higher systolic (3.6 mm Hg, P<0.001)
Of the 209 048 participants taking antihypertensive med- and diastolic (2.6 mm Hg, P<0.001) BPs compared with women
ication for whom concomitant use of a statin was recorded, with no previous history of hypertension in pregnancy (Figure 2
76 480 (36.5%) were on a statin, of whom 14 013 (18.3%) had and Table S12). Adjusting for BMI in addition to age, sex and
a previous myocardial infarction or stroke. Of the 207 220 antihypertensive medication use had no impact on the association
participants taking antihypertensive medication for whom the between BP and previous hypertension in pregnancy.
concomitant use of aspirin was documented, 67 149 (32.4%) BMI was also strongly linked to both systolic and dias-
were taking aspirin, of whom only 14 871 (22.1%) reported a tolic BP, with a linear increase in both with increasing BMI
history of myocardial infarction or stroke. Of the 52 278 hy- category. The difference in mean systolic and diastolic BP in
pertensive patients not taking aspirin for secondary preven- those participants with a BMI in the obese range, compared
tion, 18 131 (34.7%) had a BP ≥ 150/90 mm Hg. with those of healthy weight was 4.6 mm Hg and 3.1 mm Hg,
respectively (Figure 3 and Appendix Table S13).
Measures of Association Several smaller but significant differences in systolic and
Based on linear mixed models, mean systolic BP displayed diastolic BP were observed in association with several con-
a roughly linear increase with age in both men and women ditions or risk factors. For example, participants with dia-
who were not using antihypertensive medication (Figure S1). betes had significantly higher systolic BPs, but significantly
In contrast, mean diastolic BP showed an inverted U-shaped lower diastolic BPs, whereas those with a history of myocar-
curve, with BP peaking at 50 to 55 years and then gradu- dial infarction or stroke had lower systolic and diastolic BPs
ally decreasing. Systolic BP was higher in males compared (Figure 1 and Table S11).
Table 3. Proportions With Hypertension, of Those Taking/Not Taking Antihypertensive Medication, After Imputation and Standardization for Age and Sex According to
the WHO World-Standard Population
In addition, a significant dose-dependent increase in for measurement) and basing the diagnosis on only one set
both systolic and diastolic BP was seen in alcohol drinkers of 3 readings, which is by no means ideal and not in keeping
compared with nondrinkers, after adjusting for age, sex, and with recommendations for diagnosis at the individual level.8–10
antihypertensive medication use (Figure 2 and Table S12). Although the screening was volunteer-based and oppor-
Participants who currently smoked, and those fasting during tunistic with the propensity to ascertainment bias, it is striking
the period of MMM had a small, but statistically significant that almost one-third of all screenees reported never hav-
increase in both systolic and diastolic BP. Adjusting for BMI ing had their BPs measured previously. This proportion was
in addition had no significant impact on the association be- higher in men (35.9%) than women (28.4%), which may re-
tween BP and smoking. Conversely, women who were preg- flect the routine BP measurement in women associated with
nant had significantly lower systolic and diastolic BPs than oral contraceptive use and pregnancy.
those who were not. In 2019, the number of medication classes taken by those
On average, BPs measured in pharmacies were higher participants on antihypertensive medication was recorded for
than in any other setting, whereas those measured in indoor the first time. Of those taking medication, more than half were
public areas were the lowest (Figure S3 and Table S14). Small only on a single agent, a further third were taking 2 medica-
variations in mean BPs taken on different days of the week tions, and only 13.6% were taking ≥3 medications. Almost 4
were apparent but were of limited, if any, clinical significance in 10 of those on a single drug were uncontrolled, suggesting
(Figure S4 and Table S15). a significant enhanced treatment potential through the use of
additional agents. These findings add support to the increasing
Discussion recommendation to initiate drug treatment with 2 agents.8,10 The
MMM 2019 expanded on the preceding 2 campaigns, including concomitant use of aspirin by one-third of those on antihyper-
over 1.5 million participants from 92 countries. The campaign tensive medication is at odds with current recommendations16
reached significant numbers of new participants, with 1.2 mil- in that many such users did not report established cardiovas-
lion never having participated in a previous campaign and al- cular disease and many also had inadequately controlled BP.
most half a million never having had their BP measured before. Despite not being designed to provide nationally represen-
Over half a million screenees met the criteria for hypertension tative samples, and including different screening sites in dif-
based on a cutoff of 140/90 mm Hg or being on treatment for ferent countries, each year the global findings are remarkably
hypertension, and over half of all those screened were clas- consistent with previous estimates from MMM.6,7 The overall
sified as hypertensive using the lower threshold of 130/80 proportions with hypertension in 2017 and 2018 were, re-
mm Hg. Furthermore, over one-third of a million participants spectively, 34.9% and 33.4%, compared with 34.0% in 2019.
were found to have either untreated or inadequately treated hy- Rates of awareness amongst hypertensives were 58.7% in
pertension. When using the lower BP target of <130/80 mm Hg 2019, compared with 59.5% in 2018, whereas the proportion
as the definition of control, which more accurately reflects cur- treated was 54.7% in 2019 compared with 55.3% in 2018. The
rent guideline recommendations,8,10 the proportion controlled proportion of all hypertensives controlled to <140/90 mm Hg
globally was particularly low at 15.8%. was 31.7% in 2019, marginally lower than in 2018 (33.2%).
The proportion of screenees classified as hypertensive in Although most major guidelines recommend the use of
MMM is susceptible to spurious elevation due to a combina- ambulatory or home BP measurement in the diagnosis of hy-
tion of selection bias (those worried about their BP presenting pertension,8–10 such an approach was not feasible in this study
Beaney et al Global Blood Pressures: May Measurement Month 2019 339
due to the cost and logistics involved, and so diagnosis was antihypertensive medication use. The lower BPs in those with
based on a single set of clinic readings. Our results suggest a history of myocardial infarction may reflect the routine use
that if clinic readings are used, the mean of the second and of cardioprotective agents such as ACE (angiotensin-con-
third of 3 readings results in the most conservative estimate verting enzyme)-inhibitors and beta-blockers in that setting
of hypertension, which again is consistent with the results of irrespective of the presence of raised BP. Alternatively, both
previous campaigns.6,7 in those with a history of myocardial infarction or stroke,
Although these reported estimates of parameters of hy- the lower BPs may reflect stricter BP management in these
pertension management are not population-based, adjusted groups, in turn, reflecting greater interaction with healthcare
measures of association within the MMM cohort are less sub- professionals.
ject to selection bias and remain valid.17 Compared with pre- Strong positive associations were seen between BP and risk
vious years, a similar pattern of the difference in mean BP factors, such as increasing BMI and increasing alcohol intake,
with age and sex was seen. Participants with diabetes mel- and as found in previous years, pregnant women had lower
litus had on average higher systolic and diastolic BPs, but systolic and diastolic BPs. Women with a previous history
those with a history of myocardial infarction or stroke had, of hypertension in pregnancy had significantly higher mean
surprisingly, lower BPs, after adjusting for age, sex, and BPs compared with women without previous hypertension in
pregnancy. This difference was unaffected by adjustment for less-than-ideal but all-too-common situation in which diag-
BMI, implying that raised BMI was not a common explana- nosis and treatment initiation are based on a single clinic visit.
tory etiological mechanism. Higher rates of hypertension and Despite lacking a population-based representative design,
cardiovascular disease are established in women with a his- the estimates at the global level have been remarkably con-
tory of hypertensive pregnancy disorders,18 and hence women sistent across MMM campaigns and are in line with other pub-
who experience pregnancy-associated hypertension may ben- lished estimates of hypertension prevalence. Furthermore, the
efit from more regular BP checks in ensuing years. results of MMM provide real-world estimates of the numbers of
Much of the costs of the MMM campaign are borne by the individuals with hypertension that could be detected through an
generous support from local benefactors and thousands of vol- opportunistic campaign, which may make the estimates more
unteers around the world, but the central coordination remains applicable to the potential impact of screening in settings where
relatively inexpensive and equates to a cost of 0.65 USD per systematic population-based screening is not feasible.
case of untreated or inadequately treated hypertension detected. Due to the cross-sectional design of the study, outcomes
in participants found to have raised BP cannot be evaluated.
Limitations However, those found to have high BP were provided with
As an opportunistic study aimed primarily at raising aware- lifestyle and dietary advice, and advice to seek further medical
ness, the results of MMM should be viewed in the context of assessment based on locally available facilities. This approach
its limitations. Study participants were self-selected through mirrors precisely the intervention used in a recent commu-
convenience sampling, and hence recruitment is unlikely to nity-based BP screening program of older adults in China
generate representative samples of the population due to se- which, using regression discontinuity analysis, resulted in a
lection bias. Hence, prevalence at a global, regional, or na- 6.3 mm Hg reduction in systolic BP after 2 years.19 If similar
tional level should not be inferred. However, the overall study BP-lowering was associated with the MMM campaign inter-
aim was to raise awareness of the importance of BP measure- vention huge benefits in terms of cardiovascular disease pre-
ment and so investigators were not encouraged by design to vention would accrue.20
seek representative samples. Nevertheless, although the aim
was to target those who had not had a BP measured in the last Advantages
12 months, no one was excluded on this basis. The MMM19 campaign includes contemporary data from
Similarly, differences in the estimates between countries over 1.5 million adults from 92 countries that were collated
and regions should be interpreted cautiously, due to differ- in a synchronized survey following a common protocol. Over
ential characteristics of those screened. Standardization of 350 000 individuals were detected with untreated or inade-
estimates according to the World Health Organization world quately treated hypertension and advised on nonpharmacolog-
age-standard population can reduce differential distribu- ical management and further follow-up. Meanwhile utilizing
tions of age and sex, but selection differences will remain. multimedia promotion campaigns the importance of raised BP
Likewise, although standardization improves direct compara- was enhanced at the population level.
bility across regions, it does not ensure that the proportions of Although systematic screening is still a distant prospect
hypertensives estimated are any more representative at the re- for many nations in the world, we think that the MMM cam-
gional level, as the regional distribution of age will not match paign should continue annually to raise awareness at the indi-
the world-standard population. vidual and population level of this treatable condition which
Although the protocol was unified across all screening currently leads to approximately 28 000 deaths per day.1
sites, and efforts were made to train volunteers in the meas-
urement and recording of BPs, due to the scale of the study, Perspectives
inconsistencies in measurement may have arisen within and MMM has included over 4.2 million screenings across the first
between sites. Data were not fully recorded for all individuals, 3 annual campaigns and grown each year in terms of number
and 13 844 (0.9%) of submitted participants were excluded of countries involved and total participants.6,7 The detection of
due to data quality. However, all included participants had at over 900 000 adults with untreated or inadequately treated hy-
least one BP measurement and 99.0% and 99.6% had age and pertension during these 3 years, attests to the benefits of such
sex recorded, respectively. opportunistic screening. As a condition for which lifestyle
The protocol specified that 3 BP measurements be taken for changes can have major benefits, and effective treatments are
each participant, but in around one-quarter of cases, this was relatively cheap and accessible, the finding that fewer than a
not the case, reflecting local logistical challenges or participant third of participants with hypertension were controlled must
preference. Our findings showed significant differences in the motivate efforts to better detect and manage raised BP. MMM
mean BPs between the first, second, and third readings, which continues to supply an inexpensive means of raising BP aware-
had the potential to bias the results in favor of a higher propor- ness at the individual and population level around the world.
tion with hypertension in those with only one or two readings.
However, by using multiple imputation, we were able to pro- Acknowledgments
vide a more reliable estimate of what the mean reading would Our sincere thanks to Omron for the donations of blood pressure (BP)
have been, reducing any bias. Sensitivity analyses showed the devices, World Hypertension League (WHL), and Professor Daniel
T. Lackland for endorsing the extension of World Hypertension Day
results were robust to different imputation models, with only to May Measurement Month (MMM) and to Judith Bunn (MMM
small differences in the proportions with hypertension. These Project Manager) and Ranjit Rayat (Editing Assistant) for their su-
data serve to inform optimal BP screening particularly in the perb dedicated efforts and to all volunteer staff and participants.
Beaney et al Global Blood Pressures: May Measurement Month 2019 341
Sources of Funding May measurement month 2017: an analysis of blood pressure screen-
ing results worldwide. Lancet Glob Health. 2018;6:e736–e743. doi:
May Measurement Month (MMM) is an initiative of the International
10.1016/S2214-109X(18)30259-6
Society of Hypertension. MMM 2019 was generously supported by
7. Beaney T, Burrell LM, Castillo RR, Charchar FJ, Cro S, Damasceno A,
Servier through the Institut la Conference Hippocrate. As a supporter Kruger R, Nilsson PM, Prabhakaran D, et al. May Measurement Month
of the study, Servier had no role in study design, data collection, data 2018: a pragmatic global screening campaign to raise awareness of blood
analysis, data interpretation, or writing of the report. The first and pressure by the International Society of Hypertension. Eur Heart J.
corresponding authors had full access to all the data in the study and 2019;40:2006–2017. doi: 10.1093/eurheartj/ehz300
had final responsibility for the decision to submit for publication. 8. Williams B, Mancia G, Spiering W, Agabiti Rosei E, Azizi M, Burnier M,
Clement DL, Coca A, de Simone G, Dominiczak A, et al. 2018 ESC/ESH
Disclosures guidelines for the management of arterial hypertension. Eur Heart J.
N.R. Poulter has received financial support from several pharmaceutical 2018;39:3021–3104. doi: 10.1093/eurheartj/ehy339
companies which manufacture blood pressure (BP)–lowering agents, for 9. National Institute for Health and Clinical Excellence, Hypertension
consultancy fees (Servier), research projects and staff (Servier, Pfizer) (NG136): clinical management of primary hypertension in adults. https://
www.nice.org.uk/guidance/ng136. Accessed September 28, 2019.
and for arranging and speaking at educational meetings (AstraZeneca,
10. Whelton PK, Carey RM, Aronow WS, Casey DE Jr, Collins KJ, et al.
Lri Therapharma, Napi, Servier, Sanofi, Eva Pharma, and Pfizer). He
ACC/AHA/AAPA/ABC/ACPM/AGS/APhA/ASH/ASPC/NMA/
holds no stocks and shares in any such companies. M.P. Schlaich re-
PCNA guideline for the prevention, detection, evaluation, and manage-
ports nonfinancial support from OMRON, nonfinancial support from ment of high blood pressure in adults. a report of the American College
A&D, during the conduct of the study, personal fees from Medtronic, of Cardiology/American Heart Association Task Force on Clinical
personal fees from Abbott, personal fees from Novartis, grants from Practice Guidelines 2017. Hypertension. 2018;71:1269–1324. doi:
Boehringer ingelheim, outside the submitted work. G.S. Stergiou has 10.1161/HYP.0000000000000066
received consultation fees and travel fees by Omron and consultation 11. UN Statistics Division. Standard country or area codes for statistical use
fees, travel fees, and research support by Servier. J.G. Wang reports hav- (M49) [Internet]. Available at: https://unstats.un.org/unsd/methodology/
ing received research grants from Bayer, MSD and Omron, and lecture m49/. Accessed September 28, 2019.
and consulting fees from AstraZeneca, Bayer, Omron, Salubris, Servier, 12. The World Bank. World Bank Country and Lending Groups – World
and Takeda. The other authors report no conflicts. Bank Data Help Desk [Internet]. Available from: https://datahelpdesk.
worldbank.org/knowledgebase/articles/906519-world-bank-country-and-
References lending-groups. Accessed September 28, 2019.
13. Surveillance Epidemiology and End Results (SEER) Program. Standard
1. GBD 2017 Risk Factor Collaborators. Global, regional, and national
Populations - Single Ages. 2013. [Internet] Available at: https://seer.
comparative risk assessment of 84 behavioural, environmental and oc-
cancer.gov/stdpopulations/stdpop.singleages.html. Accessed September
cupational, and metabolic risks or clusters of risks for 195 countries
and territories, 1990-2017: a systematic analysis for the Global Burden 28, 2019.
of Disease Study 2017. Lancet. 2018;392:1923-1994. doi: 10.1016/ 14. White IR, Royston P, Wood AM. Multiple imputation using chained equa-
S0140-6736(18)32225-6 tions: issues and guidance for practice. Stat Med. 2011;30:377–399. doi:
2. GBD 2015 Risk Factor Collaborators. Global, regional, and national 10.1002/sim.4067
comparative risk assessment of 79 behavioural, environmental and oc- 15. Seaman SR, Bartlett JW, White IR. Multiple imputation of miss-
cupational, and metabolic risks or clusters of risks, 1990–2015: a sys- ing covariates with non-linear effects and interactions: an evalua-
tematic analysis for the Global Burden of Disease Study 2015. Lancet. tion of statistical methods. BMC Med Res Methodol. 2012;12:46. doi:
2016;388:1659–1724. doi: 10.1016/S0140-6736(16)31679-8 10.1186/1471-2288-12-46
3. Chow CK, Teo KK, Rangarajan S, Islam S, Gupta R, Avezum A, 16. Arnett DK, Blumenthal RS, Albert MA, et al. 2019 ACC/AHA guide-
Bahonar A, Chifamba J, Dagenais G, Diaz R, et al. Prevalence, awareness, line on the primary prevention of cardiovascular disease: executive sum-
treatment, and control of hypertension in rural and urban communities in mary: a report of the American College of Cardiology/American Heart
high-, middle-, and low-income countries. JAMA. 2013;310:959–968. doi: Association Task Force on Clinical Practice Guidelines. J Am Coll
10.1001/jama.2013.184182 Cardiol. 2019;74:1376–1414. doi: 10.1016/j.jacc.2019.03.009
4. NCD Risk Factor Collaboration (NCD-RisC). Long-Term and recent 17. Sedgwick P. Convenience sampling. BMJ. 2013;347:f6304.
trends in hypertension awareness, treatment, and control in 12 high- 18. Ying W, Catov JM, Ouyang P. Hypertensive disorders of pregnancy and
income countries: an analysis of 123 nationally representative surveys. future maternal cardiovascular risk. J Am Heart Assoc. 2018;7:e009382.
Lancet. 2019;394:639–651. doi: 10.1016/S0140-6736(19)31145-6 doi: 10.1161/JAHA.118.009382
5. Geldsetzer P, Manne-Goehler J, Marcus ME, Ebert C, Zhumadilov Z, 19. Chen S, Sudharsanan N, Huang F, Liu Y, Geldsetzer P, Bärnighausen T.
Wesseh CS, Tsabedze L, Supiyev A, Sturua L, Bahendeka SK, et al. The Impact of community based screening for hypertension on blood pressure
state of hypertension care in 44 low-income and middle-income coun- after two years: regression discontinuity analysis in a national cohort of
tries: a cross-sectional study of nationally representative individual- older adults in China. BMJ. 2019;366:l4064. doi: 10.1136/bmj.l4064
level data from 1·1 million adults. Lancet. 2019;394:652–662. doi: 20. Ettehad D, Emdin CA, Kiran A, Anderson SG, Callender T, Emberson J,
10.1016/S0140-6736(19)30955-9 Chalmers J, Rodgers A, Rahimi K. Blood pressure lowering for prevention
6. Beaney T, Schutte AE, Tomaszewski M, Ariti C, Burrell LM, of cardiovascular disease and death: a systematic review and meta-analy-
Castillo RR, Charchar FJ, Damasceno A, Kruger R, Lackland DT, et al. sis. Lancet. 2016;387:957–967. doi: 10.1016/S0140-6736(15)01225-8