Confidential: For Review Only · 2019. 11. 11. · Confidential: For Review Only 1 Title The effectiveness of first-line antihypertensive drugs according to age and ethnicity: a cohort

Confidential: For Review OnlyThe effectiveness of first-line antihypertensive drugs

according to age and ethnicity: a cohort study in UK primary care

Journal: BMJ

Manuscript ID BMJ-2019-053556

Article Type: Research

BMJ Journal: BMJ

Date Submitted by the Author: 11-Nov-2019

Complete List of Authors: Sinnott, Sarah-Jo; London School of Hygiene and Tropical Medicine, Non-communicable disease epidemiologyDouglas, Ian; London School of Hygiene and Tropical Medicine, Non-communicable disease epidemiologySmeeth, Liam; London School of Hygiene and Tropical Medicine, Non-communicable disease epidemiologyWilliamson, Elizabeth; London School of Hygiene and Tropical Medicine, Medical StatisticsTomlinson, Laurie; LSHTM, Non-communicable disease epidemiology

Keywords: hypertension, blood pressure, calcium channel blocker, ace-inhibitor, angiotensin receptor blocker, effectiveness, age, ethnicity

https://mc.manuscriptcentral.com/bmj

BMJ

Confidential: For Review Only

1

Title

The effectiveness of first-line antihypertensive drugs according to age and ethnicity: a

cohort study in UK primary care

Author names

Sarah-Jo Sinnott PhD1, Ian J Douglas PhD1, Liam Smeeth PhD1, Elizabeth Williamson PhD2,

Laurie A Tomlinson PhD1

1 Department of Non-Communicable Disease Epidemiology, London School of Hygiene and

Tropical Medicine, Keppel Street, London, WC1E 7HT, United Kingdom2 Department of Medical Statistics, London School of Hygiene and Tropical Medicine, Keppel

Street, London, WC1E 7HT, United Kingdom

Correspondence:

Sarah-Jo Sinnott

[email protected]

Department of Non-Communicable Disease Epidemiology, London School of Hygiene and

Tropical Medicine, Keppel Street, London, WC1E 7HT, United Kingdom

Tables: 2

Figures: 5

Page 1 of 50


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


2

Abstract

Objective: To study whether age-based (</≥55yrs) and ethnicity-based (black/non-black) treatment

recommendations in United Kingdom (UK) clinical guidelines for hypertension translate to better

reductions in blood pressure (BP) in current routine clinical care.

Design: Observational cohort study

Setting: UK Primary Care

Participants: New-users of angiotensin converting enzyme inhibitors/angiotensin receptor blockers

(ACEI/ARB), calcium channel blockers (CCB) and thiazide diuretics.

Main outcome measure: Change in systolic blood pressure in new-users of ACEI/ARB versus CCB,

stratified by age and gender. In secondary analyses we compared new-users of CCB versus diuretics.

We used a negative outcome (Herpes Zoster) to detect residual confounding and a positive outcome

(expected drug effects) to show that the study identified expected associations.

Results: We identified 84,440 new-users of ACEI/ARB, 67,274 of CCB and 22,020 of thiazides with a

median of 4 (IQR 2-6) BP measurements during one year follow-up. For non-black, non-diabetic

people <55 years initiating CCB, systolic BP fell by 1.69mmHg (99%-CI -2.52,0.86) more than

ACEI/ARB users at 12 weeks and by 0.40mmHg (99%-CI -0.98,0.18) in those ≥55 years. Across six

finer age categories, we found that CCB were more effective than ACEI/ARB in reducing systolic BP

only in those ≥75 years. Comparing initiators of CCB versus ACEI/ARB, among black people systolic

BP reduction was 2.15mmHg (99%-CI -6.17,1.87) greater with CCB, while for non-black people the

reduction was 0.98mmHg (99%-CI -1.49,-0.47) greater for CCB.

Conclusions: We found similar patterns of effectiveness for CCB versus ACEI/ARB in those aged <55

and ≥55 years. CCB produced numerically greater BP reductions than ACEI/ARB among black people

compared to non-black people, but the confidence intervals for reductions achieved were

overlapping for the two strata. Our results suggest that the current UK algorithmic approach to first

line antihypertensive treatment does not lead to greater BP reductions. Specific indications could be

considered in treatment recommendations.

Page 2 of 50


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


3

Summary box

What is already known on this topic

Current NICE guidance for first-line antihypertensive choice is based on age and ethnicity,

with the understanding that people of black African or African-Caribbean family ethnicity

(hereafter referred to as ‘black people’ to reflect diversity) and older people have lower

levels of renin and a salt-sensitive hypertension phenotype. Calcium channel blockers (CCB)

are thus recommended in non-diabetic people aged ≥55 years and black people. Angiotensin

converting enzyme inhibitors/angiotensin receptor blockers (ACEI/ARB) are recommended in

people aged <55 years.

Since the current guidance was developed the medical literature has evolved incrementally,

including older populations in trials and with meta-analysis indicating that the effectiveness

of first-line antihypertensive drugs does not differ by age (</>65 years). Additionally, the use

of categorisations of ethnicity to guide clinical decision making has been questioned.

It is not known if current recommended drug choices lead to greater reductions in BP in

routine clinical care.

What this study adds

This propensity score matched cohort study shows that initiation of either CCB or ACEI/ARB

leads to similar blood pressure (BP) reductions in people aged <55 years and those aged ≥55

years.

In analyses using finer age strata, we found that CCB lead to greater reductions in BP only in

those aged ≥75 years.

Our results for ethnicity indicated numerically greater reductions in BP for black people

initiating CCB versus ACEI/ARB than seen in non-black people but the confidence intervals

overlapped between the two groups.

Page 3 of 50


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


4

Introduction

High blood pressure (BP), or hypertension, affects more than one in four adults globally and is a

major modifiable risk factor for morbidity and mortality.1 Guideline based approaches to

pharmacotherapy for hypertension have been adopted internationally to simplify clinical practice

and improve BP control.2-4 Central to the recommendations of these guidelines is the understanding

that the effect of antihypertensive drugs differs among specific populations.

In the United Kingdom (UK), National Institute for Health and Care Excellence (NICE) guidelines

recommend angiotensin-converting enzyme inhibitors and angiotensin receptor blockers (ACEI/ARB)

as first-line treatment for hypertension in people aged <55 years, and calcium channel blockers

(CCBs) for those aged ≥55 years, replacing CCB with thiazide diuretics for those with drug

intolerance.2 The presence of an age-based recommendation is unique amongst major international

guidelines for the treatment of hypertension,3,4 and is based on differences in the activity of the

renin-angiotensin system with age.5-7 Since this threshold was introduced in the first iteration of

NICE hypertension guidance in 2004, the evidence base for hypertension treatment in older age,

including the use and safety of ACEI/ARB drugs in older populations has evolved.8,9

Secondly, in NICE guidelines CCB or thiazides are recommended as first-line treatment for black

people of African or Caribbean family origin (hereafter referred to as ‘black people’ to reflect

diversity), and the same drugs are recommended, after consideration of comorbidities, in

international guidelines.2-4 The pathophysiology of hypertension in this population has been thought

to differ importantly from people of Caucasian heritage; lower levels of renin result in a reduced

antihypertensive response to drugs that block the renin-angiotensin system such as ACEI/ARB.10

However, treatment recommendations based on historical categorisations of ethnicity have recently

been criticised given that ethnicity can be considered a social construct rather than a biological one,

and as the proportion of people with mixed ethnic heritage increases.11

Contemporary routine care is characterised by an increasingly older, more ethnically diverse and

multi-morbid population. For people initiating antihypertensives, it is not known whether current

age- and ethnicity-based treatment recommendations translate to greater BP reductions in these

settings.

The Quality and Outcomes Framework in the UK ensures that BP is regularly measured and recorded

in patients’ electronic health records (EHR) in primary care.12 This, along with complete information

on drugs prescribed, means that these anonymised data are a rich and high quality resource for

examining antihypertensive drug effectiveness. Therefore, framing our questions around the current

Page 4 of 50


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


5

NICE algorithm for drug treatment of hypertension, we sought to determine whether initiation of

CCB compared to ACEI/ARB led to differences in BP reduction across age and ethnicity strata.

Page 5 of 50


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


6

Methods

Data

We conducted a cohort study, using the Clinical Practice Research Database (CPRD-GOLD) linked to

Hospital Episodes Statistics (HES). CPRD-GOLD is a nationally representative repository of

prospectively collected, anonymised EHRs from primary care in the UK, that has been extensively

validated.13 It holds data on demographic information, health related behaviours, test results

including BP measurements, diagnoses, and prescriptions for more than 11 million people.13 The HES

database records all hospital admissions for patients covered by the National Health Service (NHS)

who receive treatment from either English NHS trusts or independent providers.14 Almost 60% of

general practices included in the CPRD-GOLD are linked to HES.13 We used linked data in this study

to improve completeness of ethnicity recording.15

Cohort entry

We identified new-users of antihypertensive drugs (CCB, ACEI/ARB and thiazides), defined as people

who had a prescription for one of these drugs in the study period 1st January 2007 to 31st December

2017, with no prior use of any antihypertensive in the preceding year. People entered the cohort

(index date) on the date of the first prescription. People were eligible for inclusion from the latest of:

study start date (1st January 2007); one year after GP registration (to allow time for recording of

covariates in the GP record); the date the GP practices’ data recording practices were considered of

adequate standard to be included in CPRD-GOLD; or the persons’ 18th birthday. People were

included until the earliest of: end of the study period (31st December 2017), last data collection from

the GP practice, date of leaving the GP practice, or death.

Exclusion

We sought to determine the effects of drugs on BP solely in those treated for hypertension.

Therefore we excluded people who initiated any of the study drugs without recorded BP

measurements in the year prior to cohort entry, and those whose BP was at target or lower

(<140/90mmHg, according to current NICE guidelines).2 We also excluded people who initiated more

than one antihypertensive drug on the index date.

Outcomes

For our main analysis we investigated the change in systolic BP from baseline at 12 weeks, 26 weeks

and 52 weeks. In a secondary analysis, we repeated this for diastolic BP.

To investigate the reliability of our data and methods in this context we included positive control

analyses; outcomes that we expected to differ according to drug prescribed. These outcomes were

Page 6 of 50


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


7

incidence of ankle swelling, gout and angioedema. We also included a negative control outcome;

Herpes Zoster. Our rationale was that none of the antihypertensive drugs studied should be

associated with an altered risk of Herpes Zoster – thus, if an association was found between any

drug group and Herpes Zoster this would suggest unmeasured confounding in our analyses.16

Covariates

Baseline systolic and diastolic BP values were taken from CPRD, using the closest measurement to,

or on, the index date. Using a priori knowledge about factors that could influence treatment choice

or response we defined multiple covariates: age, sex, smoking, alcohol use, body mass index

diabetes, myocardial infarction, stroke, heart failure, arrhythmia, peripheral vascular disease, cancer,

depression and chronic obstructive pulmonary disease. Chronic comorbidities were defined using

records from information recorded in both CPRD and HES using all available data. We determined

whether each person had ever previously been prescribed statins, anti-platelet agents, proton pump

inhibitors, insulin and loop diuretics. To capture polypharmacy, we measured medication usage from

multiple British National Formulary chapters in the year prior to index date. To capture health

service use in the year prior to index date, we determined how often a person visited the GP

surgery. To account for socioeconomic status we used quintiles of person-level deprivation based on

the 2010 English Index Multiple Deprivation scores. We calculated baseline estimated Glomerular

Filtration Rate (eGFR) from the most recent creatinine value recorded in CPRD within one year prior

to index date using the CKD-EPI equation, categorised as follows: no CKD (eGFR≥60); stage 3a (eGFR

45-59); stage 3b (eGFR 30-44); stage 4 (eGFR 15-29); stage 5 (eGFR<15).

To retain power in our main analysis, we imputed missing data under the missing at random

assumption. Data for renal function were approximately 30% missing; our previous work has

demonstrated that people without recorded renal function have similar health outcomes to those

with normal renal function after adjusting for covariates such as age, diabetes and vascular

disease.17 In the imputation model we included all explanatory variables listed above, including the

outcome variable (first systolic and diastolic BP measurement during follow-up). We conducted

diagnostics using the midiagplots function in Stata.18 We combined treatment effects across five

imputed datasets to obtain one overall estimate in each of our analyses.19

Analysis

Reflecting current UK NICE guidance2, we sought to answer two specific research questions in

relation to BP reduction: 1) in a non-diabetic, non-black population does the effectiveness of CCB

versus ACEI/ARB differ in those aged </≥55 years? 2) In a non-diabetic population, does the

effectiveness of CCB versus ACEI/ARB differ by black or non-black ethnicity? We also compared

Page 7 of 50


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


8

thiazides versus CCB to give insight into the comparative effectiveness of drugs that are

recommended as alternative choices.2

Therefore, among new-users of ACEI/ARB or CCB, within each of four strata defined by age (</≥55

years) and ethnicity (black/non-black), we estimated the propensity to be prescribed CCB versus

ACEI/ARB using a logistic regression model including all the covariates listed above, except diastolic

BP. Then, within each stratum we matched one CCB user to one ACEI/ARB user on their propensity

score within a caliper of 1%. We used linear mixed models to model change in systolic BP from

baseline at 12, 26 and 52 weeks with a random intercept for each person and fixed effects for

exposure drug, using splines over time to allow the effects of exposure to vary flexibly over time.

Such models allow for correlations within persons for longitudinal BP measurements and also

accommodate the unbalanced nature of BP recordings in the data.20 People without a BP

measurement during one year follow up were dropped from the BP analysis (n=10,840). The same

analysis was used to compare thiazides versus CCB.

We used Cox proportional hazards models for the control outcomes within each of the propensity

score matched strata defined above.

To reduce the risk of type I error with multiple testing we defined a priori that we would report 99%

confidence intervals (CI).

Subgroup and sensitivity analyses

To look more closely at the effectiveness of CCB versus ACEI/ARB and TZD versus CCB in reducing

systolic BP according to age in a non-black and non-diabetic population we used fine age categories

for stratum specific analyses: <55, 55-<60, 60-<65, 65-<70, 70-<75 and ≥75 years.

We explored the effectiveness of thiazide-like diuretics (indapamide and chlortalidone) versus CCB

and also the effectiveness of older thiazide agents versus CCB.

As our main analysis represents an ‘intention-to-treat’ approach, we also carried out an ‘on-

treatment’ analysis whereby we censored each persons’ follow-up BP data if they switched to or

added another antihypertensive or discontinued their drug, defined as no repeat prescription within

90 days of prior prescription. We carried out a complete case analysis to explore the impact of our

multiple imputation approach. Finally, as a post-hoc addition we repeated our main analysis with

95% CI.

Patient involvement

No patients were involved in setting the research question or the outcome measures, nor were they

involved in developing plans for design or implementation of the study. No patients were asked to

Page 8 of 50


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


9

advise on interpretation or writing up of results. We are not able to disseminate the results of the

research directly to study participants because the data used were anonymised.

Page 9 of 50


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


10

Results

From more than 1 million users of antihypertensive drugs between 2007-2017, we identified 84,440

new users of ACEI/ARB, 67,274 new users of CCB and 22,040 new users of thiazides (Figure 1).

People taking thiazides were more likely to be female and were older than other antihypertensive

users, while people taking ACEI/ARB group had a higher prevalence of myocardial infarction, heart

failure and diabetes. For each drug comparison, we matched people on propensity score and

achieved good balance on all covariates within age and ethnicity strata (Appendix 1 and 2). Within

the matched stratum of people with black ethnicity for each drug comparison, there was a 1mmHg

difference in systolic BP at baseline. There was a median of 4 (IQR 2-6) BP measurements available

during one year follow up for each people.

*Insert Figure 1*

*Insert Table 1*

Change in systolic BP for CCB vs ACEI/ARB by age

In those aged <55 years, systolic BP changed from 162.8mmHg to 140.7mmHg at 12 weeks in new

users of CCB in comparison to a change from 162.7mmHg to 142.2mmHg in new users of ACEI/ARB.

In relative terms, CCB reduced systolic BP by 1.69mmHg (99% CI -2.52, -0.86) more than ACEI/ARB in

those aged <55 years (Table 2). In those aged ≥55 years, systolic BP changed from 163.8mmHg to

143.2mmHg at 12 weeks in new users of CCB in comparison to a change from 164.6mmHg to

144.6mmHg in new users of ACEI/ARB (Figure 2), a relative difference of -0.40mmHg (99% CI -0.98,

0.18) in favour of CCB (Table 2). Over one year follow-up, there was no difference in effectiveness

between CCB and ACEI/ARB for systolic BP in either age group (Table 2 and Figure 2).

*Insert Table 2*

*Insert Figure 2*

Change in systolic BP for CCB vs ACEI/ARB by ethnicity

Page 10 of 50


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


11

Among black people, systolic BP changed from 156.0mmHg to 140.1mmHg at 12 weeks in new users

of CCB in comparison to a change from 158.4mmHg to 144.1mmHg in new users of ACEI/ARB. In

relative terms, CCB reduced systolic BP by 2.15mmHg (99% CI -6.17, 1.87) more than ACEI/ARB in

black people (Table 2). Among non-black people, systolic BP changed from 163.6mmHg to

143.5mmHg at 12 weeks in new users of CCB in comparison to a change from 163.7mmHg to

143.8mmHg in new users of ACEI/ARB (Figure 2), a relative difference of 0.98mmHg (99% CI -1.49, -

0.47) greater reduction among CCB initiators. Over one year follow-up, there was no difference in

effectiveness between CCB and ACEI/ARB for either non-black or black people (Table 2 and Figure

2).

We found results of similar magnitude and direction for diastolic BP (Appendices 3- 4)

Change in systolic BP for thiazides versus CCB

For those aged >55 years, CCB were more effective at lowering systolic BP compared to thiazide

diuretics at 12 and 26 weeks. However, the reductions observed were similar to those observed in

those aged <55 years, indicated by overlapping confidence intervals. Similarly, overlapping

confidence intervals for the reductions in each of the ethnicity strata indicated no evidence for a

difference in effectiveness for thiazides versus CCB in black and non-black people (Table 2 and

Figure 3).

*Insert Figure 3*

We observed similar results for diastolic BP (Appendices 3-4).

Control analyses

The incidence of drug specific side effects was as expected for each drug comparison; there was

increased incidence of ankle swelling for new users of CCB in comparison to new users of ACEI/ARB

(HR 1.77, 99% 1.61, 1.94) and reduced ankle swelling for new users of thiazides in comparison to

new users of CCB (HR 0.69, 99% CI 0.61, 0.78) (Figure 4). There was no evidence that the incidence

of Herpes Zoster differed between drug groups in either drug comparison (Figure 4). Stratum

specific negative and positive control analyses are in Appendix 6.

*Insert Figure 4*

Figure 3: Achieved systolic blood pressures for thiazides and CCB in each stratum

Page 11 of 50


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


12

Sensitivity analyses

Across all age categories, among non-black, non-diabetic people, we found that the CCB were more

effective than ACEI/ARB at reducing systolic BP only in those aged ≥75 years (12 week change -

3.0mmHg, 99% -4.05, -1.93, 26 weeks change -1.90mmHg, 99% CI -3.03, -0.77 and 52 weeks change -

1.86mmHg, 99% CI -3.08, -0.64) (Figure 5). In the thiazide versus CCB comparison, CCB were more

effective than thiazides in all age categories at 12 weeks, but the difference between the drugs

became negligible with increasing time (Figure 5).

*Insert Figure 5*

In the whole propensity score matched cohort, we found that thiazide-like drugs (indapamide and

chlortalidone) were of similar effectiveness to CCB in terms of lowering systolic BP, and that CCB

were more effective than older thiazide agents (Appendix 7).

Our ‘on treatment’ analysis provided results that were broadly similar to the main analysis. The

numerical reduction in BP was greater among black people commencing CCB than ACEI/ARB (-

3.65mmHg, 99% CI -7.99, 0.70) than we observed in non-black people (-0.35mmHg, 99% CI -0.94,

0.24), although the 99% confidence intervals were overlapping for both strata indicating no strong

evidence for a difference, similar to the main analysis (Appendix 8).

The complete case analysis found similar results to the main analysis (Appendix 9).

Page 12 of 50


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


13

Discussion

In this propensity-score matched cohort study of more than 150,000 hypertensive people in routine

care, we found no evidence to support a difference in the effectiveness of CCB versus ACEI/ARB

across those aged above and below 55 years of age. We did, however, find that CCB were more

effective than ACEI/ARB at reducing systolic BP in people older than 75 years. Our results indicated a

numerically greater reduction in systolic BP for CCB versus ACEI/ARB among black people compared

to non-black people but the confidence intervals for the reductions overlapped between the two

strata indicating no evidence of difference.

It is possible that people initiating antihypertensives other than first-line options suggested by

current guidance were prescribed them for other indications. However, we endeavoured to reduce

confounding by indication by: including only people with hypertension; undertaking our analyses

only in non-diabetics; and using propensity-score matching on multiple factors, inclusive of cardiac

failure. This latter approach achieved highly balanced cohorts for each drug comparison within each

stratum. Furthermore, we used multiple strategies to help detect bias in our matched cohorts; our

negative outcome analysis demonstrated no evidence of selection bias or residual confounding by

worse health status given that there was no increased risk of Herpes Zoster in any drug group.16

Additionally, our positive outcome analyses demonstrated that our data and methods were sensitive

to drug effects that we would anticipate to be causally associated with specific antihypertensive

drugs.

Our results related to ethnicity have wide margins of uncertainty which may have occurred for

several reasons. First, we compared changes in BP between black and non-black people, consistent

with current UK guidance.2 Much of the original clinical research that informs this guidance

compared people of African-Caribbean family history with people classified as Caucasian. If people

of non-black, non-Caucasian ethnicity have intermediate responses to the antihypertensive drug

classes, our inclusion of them in the non-black group may have contributed to the overlapping

confidence intervals we observed for non-black and black people. Secondly, we had small numbers

of black people initiating ACEI/ARB as a first line antihypertensive, suggesting that current NICE

guidance is closely adhered to. Third, the uncertainty we observed may be truly reflective of wide

variation in the activity of the renin-angiotensin system across all ethnicities.

Our outcomes relied on clinic measurements of BP, rather than BP measured using ambulatory

devices, which is recommended by NICE for the confirmation of hypertension. However, clinic

measurements of BP are used in primary care to monitor hypertension and to guide treatment

decisions. Therefore, the outcomes we have used in this study reflect the provision of care as it

Page 13 of 50


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


14

occurs routinely. Furthermore, we do not expect any differential misclassification of the measured

blood pressure by variation in measurement depending on the drug prescribed.

NICE guidelines recommend ACEI/ARB as first-line treatment for hypertension for those aged <55

years.2 This age threshold, currently absent in international guidelines, is based on the idea that

hypertension in younger people is more commonly characterised by high renin levels.5,6 The age cut-

off contained in NICE guidelines is supported by studies limited by small sample size (<60 people),5,6

short follow up periods (<6 weeks),5,6 restricted to men7 and geographical location,5 and unpublished

data.21 To support drug recommendations in those aged ≥55 years the ASCOT, VALUE and ALLHAT

trials are cited although none demonstrate clear effect modification by age (usually <60/>60 or

<65/>65 age categorisation) for any included clinical outcome.22-24 Our age-stratified results are in

line with a meta-analysis of 31 trials including more than 190,000 patients that demonstrated no

difference in BP reduction or cardiovascular events for any antihypertensive drug class between

young (<65yrs) or older (>65 years) patients.25 Our result of no difference in effectiveness between

CCB and ACEI/ARB in those aged ≥55years also agrees with the results of a randomised study that

examined effectiveness of first line BP-lowering agents across the <55/≥55years age categorisation.26

Since the first inclusion of the </≥55 years age threshold in NICE guidance (CG18) in 2004, the

literature has evolved to capture older populations with subgroup analyses based on older ages

(</>75years) or subgroups defined by frailty.8,9 In our study, greater effectiveness of CCB over

ACEI/ARB was seen only in those aged ≥75 years. This may reflect a higher prevalence of isolated

systolic hypertension in this subgroup (57% in ≥75 years versus 40% in ≥55 years in our data) for

which CCB may be more effective.27,28

International guidelines for hypertension are similar to NICE guidelines in that ethnicity is a major

consideration in choosing a first-line antihypertensive drug, although do not use the algorithmic

approach of UK guidance. Ethnicity-based recommendations are supported by strong evidence from

the ALLHAT randomized clinical trial which demonstrated thiazides or CCBs are more effective at

lowering BP compared to ACEI/ARB in black people.24 We have suggested above a number of

reasons why our results may differ from established clinical evidence. Notwithstanding, the value of

making clinical decisions based on the colour of an individual’s skin is increasingly questioned given

the growing proportion of people who classify themselves as multi-ethnic or of mixed heritage.11

Furthermore, our understanding of gene-environment interactions in the development of

hypertension is now improving.29 Thus, the value of using categorisations of ethnicity as a proxy

predictor for drug response is uncertain in current society, though is likely to be informed by ongoing

research.30

Page 14 of 50


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


15

In conclusion, we found that CCB, relative to ACEI/ARB, achieved reductions in systolic BP of similar

magnitude in those aged above and below 55 years. However, in those aged ≥75 years, CCB was

more effective than ACEI/ARB in reducing BP. This suggests that people aged between ≥55 and 75

years who are currently offered CCB, but have indications for renin-angiotensin blockade such as

proteinuria, may be missing the therapeutic benefits of ACEI/ARB. We found no evidence that CCB

was more effective than ACEI/ARB among black people compared to other ethnic groups, although

power was limited and there was uncertainty in the estimate. Our results suggest that the algorithm

for choice of pharmacotherapy in UK NICE guidance could be simplified. Moving towards a choice of

any of the three major antihypertensive drug classes with suggested compelling indications for their

use would align the UK with international guidance, in particular with regard to age.

Page 15 of 50


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


16

Author’s contributions

S-JS, LAT, ID, EW and LS conceived and devised the study, S-JS and EW analysed the data. All authors

contributed to the interpretation of the data. S-JS drafted the article and all authors (IJD, EW, LS,

LAT) reviewed and edited the manuscript, and approved the version to be published. The

corresponding author attests that all listed authors meet authorship criteria and that no others

meeting the criteria have been omitted. S-JS is guarantor and accepts full responsibility for the

conduct of the study, had access to the data, and controlled the decision to publish.

Funding

S-JS is funded by a Wellcome Trust Sir Henry Wellcome fellowship (107340/Z/15/Z). LAT is funded by

a Wellcome Trust Intermediate Clinical Fellowship (101143/Z/13/Z). The funders had no role in the

design or conduct of this research.

Competing interests:

IJD is funded by, holds stock in and has consulted for GSK. LS has received grants from GSK and has

received grants from the Wellcome trust, the MRC, the NIHR, the British Heart foundation and

Diabetes UK outside of the submitted work and is a Trustee of the British Heart Foundation. S-JS, LAT

and EW have no relevant conflicts of interest to disclose.

Ethical approval

The study protocol was approved by the London School of Hygiene and Tropical Medicine ethics

committee (No. 16251) and the Independent Scientific Advisory Committee for the Medicines and

Healthcare products Regulatory Agency (No. 18_312R). The protocol was made available to

reviewers as supplementary material.

Transparency declaration

The manuscript’s guarantor (SJS) affirms that the manuscript is an honest, accurate, and transparent

account of the study being reported; that no important aspects of the study have been omitted; and

that any discrepancies from the study as planned have been explained.

Data sharing

No additional data available.

Page 16 of 50


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


17

License

“I Sarah-Jo Sinnott the Corresponding Author of this article contained within the original manuscript which includes any diagrams & photographs within and any related or stand alone film submitted (the Contribution”) has the right to grant on behalf of all authors and does grant on behalf of all authors, a licence to the BMJ Publishing Group Ltd and its licencees, to permit this Contribution (if accepted) to be published in the BMJ and any other BMJ Group products and to exploit all subsidiary rights, as set out in our licence set out at: http://www.bmj.com/about-bmj/resources-authors/forms-policies-and-checklists/copyright-open-access-and-permission-reuse.”

Please tick one or more boxes as appropriate:

. I am one author signing on behalf of all co-owners of the Contribution.

Page 17 of 50


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


18

1. Olsen MH, Angell SY, Asma S, et al. A call to action and a lifecourse strategy to address the global burden of raised blood pressure on current and future generations: the Lancet Commission on hypertension. The Lancet 2016; 388(10060): 2665-712.2. National Institute for Health and Care Excellence. Hypertension in adults: diagnosis and management.[NG136]. Available from https://www.nice.org.uk/guidance/ng136/evidence. 2019.3. Whelton PK, Carey RM, Aronow WS, et al. 2017 ACC/AHA/AAPA/ABC/ACPM/AGS/APhA/ASH/ASPC/NMA/PCNA guideline for the prevention, detection, evaluation, and management of high blood pressure in adults: a report of the American College of Cardiology/American Heart Association Task Force on Clinical Practice Guidelines. Journal of the American College of Cardiology 2018; 71(19): e127-e248.4. Williams B, Mancia G, Spiering W, et al. 2018 ESC/ESH Guidelines for the management of arterial hypertension: The Task Force for the management of arterial hypertension of the European Society of Cardiology and the European Society of HypertensionThe Task Force for the management of arterial hypertension of the European Society of Cardiology and the European Society of Hypertension. Journal of hypertension 2018; 36(10): 1953-2041.5. Dickerson JC, Hingorani AD, Ashby MJ, Palmer CR, Brown MJ. Optimisation of antihypertensive treatment by crossover rotation of four major classes. The Lancet 1999; 353(9169): 2008-13.6. Deary AJ, Schumann AL, Murfet H, Haydock SF, Foo RS-Y, Brown MJ. Double-blind, placebo-controlled crossover comparison of five classes of antihypertensive drugs. Journal of hypertension 2002; 20(4): 771-7.7. Materson BJ, Reda DJ, Cushman WC, et al. Single-Drug Therapy for Hypertension in Men--A Comparison of Six Antihypertensive Agents with Placebo. New England Journal of Medicine 1993; 328(13): 914-21.8. Beckett NS, Peters R, Fletcher AE, et al. Treatment of hypertension in patients 80 years of age or older. New England Journal of Medicine 2008; 358(18): 1887-98.9. SPRINT Research Group. A randomized trial of intensive versus standard blood-pressure control. New England Journal of Medicine 2015; 373(22): 2103-16.10. Sagnella G. Why is plasma renin activity lower in populations of African origin? Journal of human hypertension 2001; 15(1): 17.11. Bonham VL, Green ED, Pérez-Stable EJ. Examining how race, ethnicity, and ancestry data are used in biomedical research. Jama 2018; 320(15): 1533-4.12. Hippisley-Cox J, Vinogradova Y, Coupland C. Time Series Analysis for selected clinical indicators from the Quality and Outcomes Framework 2001-2006. Leeds: The Information Centre for Health and Social Care 2007.13. Herrett E, Gallagher AM, Bhaskaran K, et al. Data Resource Profile: Clinical Practice Research Datalink (CPRD). Int J Epidemiol 2015; 44(3): 827-36.14. Herbert A, Wijlaars L, Zylbersztejn A, Cromwell D, Hardelid P. Data Resource Profile: Hospital Episode Statistics Admitted Patient Care (HES APC). International Journal of Epidemiology 2017; 46(4): 1093-i.15. Mathur R, Bhaskaran K, Chaturvedi N, et al. Completeness and usability of ethnicity data in UK-based primary care and hospital databases. Journal of public health 2013; 36(4): 684-92.16. Lipsitch M, Tchetgen ET, Cohen T. Negative controls: a tool for detecting confounding and bias in observational studies. Epidemiology (Cambridge, Mass) 2010; 21(3): 383.17. Schmidt M, Mansfield KE, Bhaskaran K, et al. Serum creatinine elevation after renin-angiotensin system blockade and long term cardiorenal risks: cohort study. BMJ 2017; 356: j791.18. Eddings W, Marchenko Y. Diagnostics for multiple imputation in Stata. Stata Journal 2012; 12(3): 353.19. Leyrat C, Seaman SR, White IR, et al. Propensity score analysis with partially observed covariates: How should multiple imputation be used? Stat Methods Med Res 2017: 962280217713032.

Page 18 of 50


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

https://www.nice.org.uk/guidance/ng136/evidence


19

20. Fitzmaurice G.M, Laird N.M, J.H. W. Applied Longitudinal Analysis (Second Edition): Chapter 8 Linear Mixed Effects Models.: Wiley; 2011.21. ASCOT. From reference number 55 in Full NICE guideline for Hypertension "ASCOT Steering Committee. Age-Stratified Analysis of Blood Pressure Responses. Personal communication: 06". 2006.22. Dahlöf B, Sever PS, Poulter NR, et al. Prevention of cardiovascular events with an antihypertensive regimen of amlodipine adding perindopril as required versus atenolol adding bendroflumethiazide as required, in the Anglo-Scandinavian Cardiac Outcomes Trial-Blood Pressure Lowering Arm (ASCOT-BPLA): a multicentre randomised controlled trial. The Lancet 2005; 366(9489): 895-906.23. Zanchetti A, Julius S, Kjeldsen S, et al. Outcomes in subgroups of hypertensive patients treated with regimens based on valsartan and amlodipine: An analysis of findings from the VALUE trial. Journal of hypertension 2006; 24(11): 2163-8.24. Wright J, Bakris G, Greene T. Major outcomes in high-risk hypertensive patients randomized to angiotensin-converting enzyme inhibitor or calcium channel blocker vs. diuretic. the Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial (ALLHAT). ACC Current Journal Review 2003: 37.25. Blood Pressure Lowering Treatment Trialist's Collaboration. Effects of different regimens to lower blood pressure on major cardiovascular events in older and younger adults: meta-analysis of randomised trials. BMJ 2008; 336(7653): 1121-3.26. Mahmud A, Feely J. Choice of first antihypertensive: simple as ABCD? American journal of hypertension 2007; 20(8): 923-7.27. Staessen JA, Fagard R, Thijs L, et al. Randomised double-blind comparison of placebo and active treatment for older patients with isolated systolic hypertension. The Systolic Hypertension in Europe (Syst-Eur) Trial Investigators. Lancet 1997; 350(9080): 757-64.28. Hansson L, Lindholm LH, Ekbom T, et al. Randomised trial of old and new antihypertensive drugs in elderly patients: cardiovascular mortality and morbidity the Swedish Trial in Old Patients with Hypertension-2 study. Lancet 1999; 354(9192): 1751-6.29. Jones DW, Hall JE. Racial and ethnic differences in blood pressure: biology and sociology. Am Heart Assoc; 2006.30. Mukhtar O, Cheriyan J, Cockcroft JR, et al. A randomized controlled crossover trial evaluating differential responses to antihypertensive drugs (used as mono-or dual therapy) on the basis of ethnicity: The comparIsoN oF Optimal Hypertension RegiMens; part of the Ancestry Informative Markers in HYpertension program—AIM-HY INFORM trial. American heart journal 2018; 204: 102-8.

Page 19 of 50


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


20

Table 1: Descriptive characteristics for new-users of ACEI/ARB, CCB and thiazides at baseline.ACEI/ARB

87,440n (%)

CCB 67,274n (%)

Thiazides 22,040n (%)

Females 37158 (42.5) 33506 (49.8) 14277 (64.8)

Age (years)<50 29079 (33.3) 9204 (13.7) 2372 (10.8)50-54 15178 (17.4) 5377 ( 8.0) 1341 ( 6.1)55-59 11190 (12.8) 9522 (14.2) 2302 (10.4)60-64 10215 (11.7) 11686 (17.4) 3394 (15.4)65-69 7795 ( 8.9) 10746 (16.0) 3205 (14.5)70-74 5763 ( 6.6) 8421 (12.5) 3139 (14.2)75+ 8220 ( 9.4) 12318 (18.3) 6287 (28.5)

Systolic BP mean (sd) 160.2 (16.6) 165.8 (18.1) 164.3 (17.2)Diastolic BP mean (sd) 94.8 (11.4) 93.8 (12.1) 91.5 (11.2)

EthnicityWhite 73247 (83.8) 55477 (82.5) 19,253 (87.4)South Asian 2949 (3.4) 1,684 (2.5) 372 (1.7)Black 914 (1.1) 3,096 (4.6) 389 (1.8)Other/mixed/unknown 3,214 (3.7) 2394 (3.6) 707 (3.2)Missing 7116 (8.1) 4623 (6.9) 1319 (6.0))

Index Multiple Deprivation1 (least deprived) 20,174 (23.1) 15,540 (23.1) 40,491 (22.9)2 20,481 (23.4) 15,822 (23.5) 41,637 (23.6)3 17,616 (20.2) 13,457 (20.0) 35,634 (20.2)4 16,301 (18.6) 12,795 (19.0) 33,233 (18.8)5 (most deprived) 12,810 (14.7) 9,631 (14.3) 25,655 (14.5)Missing 58 (0.1) 29 (0.04) 17 (0.1)

Body Mass IndexUnderweight <18.5 646 ( 0.7) 863 ( 1.3) 361 ( 1.6)Healthy weight 18.5-24.9 17317 (19.8) 17609 (26.2) 6152 (27.9)Overweight 25-29.9 31198 (35.7) 25329 (37.7) 7755 (35.2)Obesity ≥30 35119 (40.2) 20498 (30.5) 6841 (31.0)Missing 3160 ( 3.6) 2975 ( 4.4) 931 ( 4.2)

SmokingNon-smoker 29956 (34.3) 23302 (34.6) 7984 (36.2)Current smoker 18183 (20.8) 12632 (18.8) 3830 (17.4)Ex-smoker 39206 (44.8) 31168 (46.3) 10171 (46.1)Missing 95 ( 0.1) 172 ( 0.3) 55 ( 0.2)

AlcoholNon-drinker 7929 ( 9.1) 7060 (10.5) 2663 (12.1)Current drinker 67932 (77.7) 50627 (75.3) 16236 (73.7)Ex-drinker 7332 ( 8.4) 6197 ( 9.2) 2093 ( 9.5)Missing 4247 ( 4.9) 3390 ( 5.0) 1048 ( 4.8)

ComorbiditiesMyocardial Infarction 2489 ( 2.8) 449 ( 0.7) 158 ( 0.7)Stroke 2671 ( 3.1) 2199 ( 3.3) 785 ( 3.6)Heart Failure 1122 ( 1.3) 294 ( 0.4) 114 ( 0.5)Peripheral Vascular Disease 2427 ( 2.8) 2872 ( 4.3) 896 ( 4.1)Diabetes 15095 (17.3) 4289 ( 6.4) 975 ( 4.4)Depression 6677 ( 7.6) 4756 ( 7.1) 1547 ( 7.0)

Page 20 of 50


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


21

COPD 3141 ( 3.6) 3748 ( 5.6) 1250 ( 5.7)Cancer 6455 ( 7.4) 7822 (11.6) 2729 (12.4)Herpes Zoster 4872 ( 5.6) 5067 ( 7.5) 1939 ( 8.8)Gout 4365 ( 5.0) 3207 ( 4.8) 645 ( 2.9)Angioedema 61 ( 0.1) 67 ( 0.1) 5 ( 0.0)

CKDNo CKD 57963 (66.3) 37990 (56.5) 11824 (53.6)Stage 3a 6132 ( 7.0) 4726 ( 7.0) 2126 ( 9.6)Stage 3b 1541 ( 1.8) 999 ( 1.5) 393 ( 1.8)Stage 4 218 ( 0.2) 251 ( 0.4) 44 ( 0.2)Stage 5 28 ( 0.0) 96 ( 0.1) ~Missing 21558 (24.7) 23212 (34.5) 7651 (34.7)

DrugsAntiplatelet agents 11868 (13.6) 9590 (14.3) 3699 (16.8)Statins 19769 (22.6) 13740 (20.4) 4232 (19.2)Proton Pump Inhibitors 29775 (34.1) 26374 (39.2) 8139 (36.9)Insulin 2352 ( 2.7) 422 ( 0.6) 84 ( 0.4)Loop diuretics 3109 ( 3.6) 2813 ( 4.2) 1607 ( 7.3)NSAIDs 56108 (64.2) 44292 (65.8) 14663 (66.5)

Primary Care Consultations<5 consultations 22829 (26.1) 17330 (25.8) 5563 (25.2)5-9 consultations 28577 (32.7) 20855 (31.0) 7237 (32.8)10-14 consultations 16106 (18.4) 12243 (18.2) 4121 (18.7)15-19 consultations 8091 ( 9.3) 6591 ( 9.8) 2081 ( 9.4)≥20 consultations 11837 (13.5) 10255 (15.2) 3038 (13.8)Missing 0 ( 0.0) 0 ( 0.0) 0 ( 0.0)

Page 21 of 50


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


22

Table 2: Difference in systolic blood pressure since drug initiation for CCB versus ACEI/ARB and thiazide versus CCB.

Difference between CCB and ACEI/ARB for change in systolic BP(mmHg) since baseline (99% CI)

n 12 weeks 26 weeks 52 weeks

Age <55* 20,964 -1.69 (-2.52, -0.86) -0.48 (-1.47, 0.51) -0.53 (-1.96, 0.91)Age≥55* 58,396 -0.40 (-0.98, 0.18) 0.63 (0.06, 1.21) 0.85 (-0.11, 1.81)

Non-black** 73,726 -0.98 (-1.49, -0.47) 0.11 (-0.42, 0.64) 0.08 (-0.81, 0.96)

Black** 894 -2.15 (-6.17, 1.87) 0.55 (-3.39, 4.49) 2.28 (-5.36, 9.92)

Difference between thiazide and CCB for change in systolic BP(mmHg) since baseline (99% CI)

Age <55* 5,724 1.51 (-0.03, 3.06) 0.07 (-1.70, 1.83) 0.34 (-2.39, 3.07)Age≥55* 32,464 2.16 (1.35, 2.96) 1.43 (0.56, 2.31) 0.17 (-0.95, 1.30)

Non-black** 35,876 2.10 (1.37, 2.82) 1.19 (0.40, 2.0) 0.32 (-0.74, 1.38)

Black** 648 -0.28 (-5.69, 5.14) -1.75 (-6.98, 3.48) 2.01 (-5.46, 9.47)

*In a non-black, non-diabetic population** In a non-diabetic populationAll analyses are 1:1 PS matched within each stratum. Each n consists of half CCB and half ACEI/ARB (or half thiazide and half CCB). For the CCB vs ACEI/ARB comparison a negative result means CCB produced larger reductions in systolic BP and a positive results indicates that ACEI/ARB produced larger reductions in systolic BP. For the thiazide vs CCB comparison a negative result means thiazides produced larger reductions in systolic BP and a positive results indicates that CCB produced larger reductions in systolic BP. Changes in systolic BP with 95% CI are in Appendix 5.

Page 22 of 50


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


All antihypertensive drug users in CPRD eligible

for HES linkage and eligible for study inclusion,

2007 – 2017

N= 1,034,036

Prevalent users

ACE/ARB = 615,740 CCB = 454,499

Thiazide = 310,552

New-users

ACE/ARB = 120,722 CCB = 91,432

Thiazide = 31,473

New-users

ACE/ARB = 120,722 CCB = 91,432

Thiazide = 31,473

ACE/ARB = 116,601 CCB = 87,934

Thiazide = 29,059

Remove people who initiate more than

one agent on index date:

ACEI/ARB + CCB = 2,727 ACEI/ARB + thiazide = 1,667

CCB + thiazide = 1,027 ACE + CCB + thiazide = 0

Single drug new-users

ACE/ARB = 114,463 CCB = 85,346

Thiazide = 29,053

Remove second instances of new-use:

ACEI/ARB= 2,138 CCB = 2,588 Thiazide= 6

Number of new users with BP readings in year prior to cohort entry

ACEI/ARB =106,948

CCB = 77,816 Thiazide = 25,956

Number of new users with high BP readings in

year prior to cohort entry

ACEI/ARB =87,440 CCB = 67,274

Thiazide = 22,040

Page 23 of 50


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


Page 24 of 50


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


Page 25 of 50


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


Page 26 of 50


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


Page 27 of 50


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


1

List of Appendices

Appendix 1: Descriptive characteristics and propensity score distributions before and after

matching, CCB vs ACE-I/ARB .......................................................................................................... 2

Appendix 2: Descriptive characteristics and propensity score distributions before and after

matching, thiazides vs CCB............................................................................................................. 5

Appendix 3: Diastolic blood pressure results .................................................................................. 8

Appendix 4: Diagrams of diastolic blood pressure results ............................................................... 9

Appendix 5: 95% CI for changes in systolic blood pressure ........................................................... 11

Appendix 6: Negative and positive outcomes in age-based and ethnicity-based strata ................. 12

Appendix 7: Comparison of thiazide-like diuretics with calcium channel blockers ........................ 14

Appendix 8: On treatment analysis .............................................................................................. 15

Appendix 9: Complete case analysis ............................................................................................. 16

Appendix 10: RECORD-PE checklist .............................................................................................. 17

Page 28 of 50


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


2

Appendix 1: Descriptive characteristics and propensity score distributions before and after matching, CCB vs ACE-I/ARB

Appendix 1 Table : Descriptive characteristics for initiators of calcium channel blockers (CCB) versus initiators of ACE-inhibitors/angiotensin receptor blockers (ACE-I/ARB) in matched stratum specific cohorts

Aged <55yrs Aged ≥55yrs Non-Black Black

ACE-I/ARB CCB ACE-I/ARB CCB ACE-I/ARB CCB ACE-I/ARB CCB

10482 10482 29198 29198 36863 36863 447 447

Females 4620 (44.1) 4675 (44.6) 13768 (47.2) 13549 (46.4) 17249 (46.8) 17380 (47.1) 251 (56.2) 250 (55.9)

Age (years)

<50 6256 (59.7) 6278 (59.9) 0 ( 0.0) 0 ( 0.0) 5723 (15.5) 5723 (15.5) 258 (57.7) 258 (57.7)

50-54 4226 (40.3) 4204 (40.1) 0 ( 0.0) 0 ( 0.0) 3801 (10.3) 3801 (10.3) 75 (16.8) 75 (16.8)

55-59 0 ( 0.0) 0 ( 0.0) 7403 (25.4) 7403 (25.4) 6669 (18.1) 6669 (18.1) 42 ( 9.4) 42 ( 9.4)

60-64 0 ( 0.0) 0 ( 0.0) 7229 (24.8) 7229 (24.8) 6591 (17.9) 6591 (17.9) 22 ( 4.9) 22 ( 4.9)

65-69 0 ( 0.0) 0 ( 0.0) 5417 (18.6) 5417 (18.6) 5055 (13.7) 5055 (13.7) 20 ( 4.5) 20 ( 4.5)

70-74 0 ( 0.0) 0 ( 0.0) 3818 (13.1) 3818 (13.1) 3646 ( 9.9) 3646 ( 9.9) 19 ( 4.3) 19 ( 4.3)

75+ 0 ( 0.0) 0 ( 0.0) 5331 (18.3) 5331 (18.3) 5378 (14.6) 5378 (14.6) 11 ( 2.5) 11 ( 2.5)

Ethnicity**

White 8627 (82.3) 8247 (78.7) 25797 (88.4) 25593 (87.7) 34686 (94.1) 34172 (92.7) 0 ( 0.0) 0 ( 0.0)

South asian 415 ( 4.0) 588 ( 5.6) 442 ( 1.5) 515 ( 1.8) 827 ( 2.2) 1077 ( 2.9) 0 ( 0.0) 0 ( 0.0)

Black 0 ( 0.0) 0 ( 0.0) 0 ( 0.0) 0 ( 0.0) 0 ( 0.0) 0 ( 0.0) 447 (100.0) 447 (100.0)

Other/mixed/unknown 421 ( 4.0) 673 ( 6.4) 893 ( 3.1) 925 ( 3.2) 1350 ( 3.7) 1614 ( 4.4) 0 ( 0.0) 0 ( 0.0)

Missing 1019 (9.72) 974 (9.29) 2066 (7.08) 2165 (7.41) 0 (0.00) 0 (0.00) 0 (0.00) 0 (0.00)

BMI**

Underweight <18.5 71 ( 0.7) 67 ( 0.6) 348 ( 1.2) 323 ( 1.1) 386 ( 1.0) 378 ( 1.0) 0 ( 0.0) 2 ( 0.4)

Healthy weight 18.5-24.9 1893 (18.1) 1857 (17.7) 7863 (26.9) 7758 (26.6) 8969 (24.3) 8975 (24.3) 77 (17.2) 81 (18.1)

Overweight 25-29.9 3402 (32.5) 3438 (32.8) 11505 (39.4) 11653 (39.9) 13976 (37.9) 13917 (37.8) 178 (39.8) 179 (40.0)

Obesity >=30 4718 (45.0) 4663 (44.5) 8260 (28.3) 8267 (28.3) 12119 (32.9) 12093 (32.8) 178 (39.8) 178 (39.8)

Missing 398 ( 3.8) 457 ( 4.4) 1222 ( 4.2) 1197 ( 4.1) 1413 ( 3.8) 1500 ( 4.1) 14 ( 3.1) 7 ( 1.6)

SMOKING**

Non-smoker 3874 (37.0) 3835 (36.6) 9774 (33.5) 9785 (33.5) 12379 (33.6) 12557 (34.1) 250 (55.9) 246 (55.0)

Current smoker 2740 (26.1) 2756 (26.3) 4931 (16.9) 4823 (16.5) 7122 (19.3) 7087 (19.2) 64 (14.3) 58 (13.0)

Ex-smoker 3859 (36.8) 3878 (37.0) 14458 (49.5) 14511 (49.7) 17328 (47.0) 17136 (46.5) 133 (29.8) 143 (32.0)

Missing 9 ( 0.1) 13 ( 0.1) 35 ( 0.1) 79 ( 0.3) 34 ( 0.1) 83 ( 0.2) 0 ( 0.0) 0 ( 0.0)

ALCOHOL**

Non-drinker 1055 (10.1) 1028 ( 9.8) 2499 ( 8.6) 2475 ( 8.5) 3202 ( 8.7) 3129 ( 8.5) 124 (27.7) 120 (26.8)

Current drinker 8094 (77.2) 8086 (77.1) 22763 (78.0) 23019 (78.8) 28856 (78.3) 28885 (78.4) 264 (59.1) 262 (58.6)

Ex-drinker 822 ( 7.8) 799 ( 7.6) 2391 ( 8.2) 2362 ( 8.1) 2978 ( 8.1) 3099 ( 8.4) 38 ( 8.5) 48 (10.7)

Missing 511 ( 4.9) 569 ( 5.4) 1545 ( 5.3) 1342 ( 4.6) 1827 ( 5.0) 1750 ( 4.7) 21 ( 4.7) 17 ( 3.8)

Comorbidities

Myocardial Infarction 14 ( 0.1) 16 ( 0.2) 326 ( 1.1) 335 ( 1.1) 371 ( 1.0) 347 ( 0.9) 1 ( 0.2) 0 ( 0.0)

Stroke 157 ( 1.5) 159 ( 1.5) 1225 ( 4.2) 1325 ( 4.5) 1438 ( 3.9) 1461 ( 4.0) 6 ( 1.3) 8 ( 1.8)

Heart Failure 11 ( 0.1) 11 ( 0.1) 205 ( 0.7) 215 ( 0.7) 267 ( 0.7) 225 ( 0.6) 2 ( 0.4) 3 ( 0.7)

Peripheral Vascular Disease 119 ( 1.1) 121 ( 1.2) 1159 ( 4.0) 1161 ( 4.0) 1279 ( 3.5) 1208 ( 3.3) 2 ( 0.4) 6 ( 1.3)

Diabetes 0 ( 0.0) 0 ( 0.0) 0 ( 0.0) 0 ( 0.0) 0 ( 0.0) 0 ( 0.0) 0 ( 0.0) 0 ( 0.0)

depression 1022 ( 9.8) 994 ( 9.5) 1746 ( 6.0) 1684 ( 5.8) 2658 ( 7.2) 2674 ( 7.3) 23 ( 5.1) 34 ( 7.6)

COPD 168 ( 1.6) 170 ( 1.6) 1645 ( 5.6) 1594 ( 5.5) 1812 ( 4.9) 1729 ( 4.7) 5 ( 1.1) 8 ( 1.8)

Cancer 380 ( 3.6) 376 ( 3.6) 3581 (12.3) 3578 (12.3) 3889 (10.5) 3819 (10.4) 14 ( 3.1) 21 ( 4.7)

Herpes Zoster 391 ( 3.7) 359 ( 3.4) 2312 ( 7.9) 2345 ( 8.0) 2553 ( 6.9) 2575 ( 7.0) 10 ( 2.2) 10 ( 2.2)

Gout 407 ( 3.9) 409 ( 3.9) 1555 ( 5.3) 1587 ( 5.4) 1864 ( 5.1) 1882 ( 5.1) 11 ( 2.5) 9 ( 2.0)

Angioedema 16 ( 0.2) 17 ( 0.2) 15 ( 0.1) 11 ( 0.0) 25 ( 0.1) 27 ( 0.1) 0 ( 0.0) 1 ( 0.2)

CKD**

no CKD 7041 (67.2) 5619 (53.6) 17340 (59.4) 15650 (53.6) 22626 (61.4) 19675 (53.4) 271 (60.6) 238 (53.2)

stage 3a 156 ( 1.5) 139 ( 1.3) 3556 (12.2) 3166 (10.8) 3550 ( 9.6) 3260 ( 8.8) 30 ( 6.7) 31 ( 6.9)

Page 29 of 50


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


3

stage 3b 32 ( 0.3) 35 ( 0.3) 784 ( 2.7) 681 ( 2.3) 852 ( 2.3) 724 ( 2.0) 4 ( 0.9) 5 ( 1.1)

stage 4 15 ( 0.1) 9 ( 0.1) 121 ( 0.4) 93 ( 0.3) 131 ( 0.4) 112 ( 0.3) 2 ( 0.4) 2 ( 0.4)

stage 5 5 ( 0.0) 14 ( 0.1) 13 ( 0.0) 33 ( 0.1) 13 ( 0.0) 45 ( 0.1) 1 ( 0.2) 1 ( 0.2)

Missing 3233 (30.8) 4666 (44.5) 7384 (25.3) 9575 (32.8) 9691 (26.3) 13047 (35.4) 139 (31.1) 170 (38.0)

Drugs

Antiplatelet agents 516 ( 4.9) 496 ( 4.7) 4625 (15.8) 4731 (16.2) 4996 (13.6) 5084 (13.8) 26 ( 5.8) 29 ( 6.5)

Statins 742 ( 7.1) 751 ( 7.2) 6456 (22.1) 6542 (22.4) 7038 (19.1) 6967 (18.9) 31 ( 6.9) 34 ( 7.6)

Proton Pump Inhibitors 3270 (31.2) 3213 (30.7) 11057 (37.9) 10853 (37.2) 13830 (37.5) 13754 (37.3) 149 (33.3) 156 (34.9)

Insulin 0 ( 0.0) 0 ( 0.0) 0 ( 0.0) 0 ( 0.0) 0 ( 0.0) 0 ( 0.0) 0 ( 0.0) 0 ( 0.0)

Loop diuretics 216 ( 2.1) 194 ( 1.9) 1331 ( 4.6) 1332 ( 4.6) 1481 ( 4.0) 1584 ( 4.3) 9 ( 2.0) 12 ( 2.7)

NSAIDs 6478 (61.8) 6537 (62.4) 19290 (66.1) 19380 (66.4) 24392 (66.2) 24547 (66.6) 286 (64.0) 303 (67.8)

Primary Care Consultations**

<5 consultations 3351 (32.0) 3370 (32.2) 7734 (26.5) 7794 (26.7) 9633 (26.1) 9639 (26.1) 128 (28.6) 130 (29.1)

5-9 consultations 3045 (29.0) 3055 (29.1) 9925 (34.0) 10039 (34.4) 12300 (33.4) 12284 (33.3) 139 (31.1) 117 (26.2)

10-14 consultations 1687 (16.1) 1654 (15.8) 5268 (18.0) 5269 (18.0) 6736 (18.3) 6834 (18.5) 75 (16.8) 87 (19.5)

15-19 consultations 800 ( 7.6) 823 ( 7.9) 2598 ( 8.9) 2521 ( 8.6) 3357 ( 9.1) 3308 ( 9.0) 49 (11.0) 57 (12.8)

>=20 consultations 1599 (15.3) 1580 (15.1) 3673 (12.6) 3575 (12.2) 4837 (13.1) 4798 (13.0) 56 (12.5) 56 (12.5)

Missing 0 ( 0.0) 0 ( 0.0) 0 ( 0.0) 0 ( 0.0) 0 ( 0.0) 0 ( 0.0) 0 ( 0.0) 0 ( 0.0)

Systolic BP mean (sd) 162.7 (17.2) 162.8 (19.2) 164.6 (16.8) 163.8 (16.2) 163.7 (17.0) 163.6 (17.1) 158.4 (15.4) 156.0 (14.5)

Diastolic BP mean (sd) 100.4 (10.0) 101.0 (11.3) 92.4 (11.2) 92.5 (11.2) 94.2 (11.5) 94.5 (11.9) 96.7 (10.2) 96.9 ( 9.9)

Page 30 of 50


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


4

Before Matching After Matching

<55years

≥55years

Non-Black

Black

02

46

8

kde

nsity p

sco

re_im

p

0 .2 .4 .6 .8propensity score

_ACE/ARB _CCB

02

46

8

kde

nsity p

sco

re_im

p


_ACE/ARB _CCB

01

23

4

kde

nsity p

sco

re_im

p

0 .2 .4 .6 .8 1propensity score

_ACE/ARB _CCB

01

23

4

kde

nsity p

sco

re_im

p


_ACE/ARB _CCB

01

23

kde

nsity p

sco

re_im

p


_ACE/ARB _CCB

0.5

11.5

22.5

kde

nsity p

sco

re_im

p


_ACE/ARB _CCB

02

46

8

kde

nsity p

sco

re_im

p


_ACE/ARB _CCB

01

23

45

kde

nsity p

sco

re_im

p

.4 .6 .8 1propensity score

_ACE/ARB _CCB

Page 31 of 50


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


5

Appendix 2: Descriptive characteristics and propensity score distributions before and

after matching, thiazides vs CCB

Appendix 2 Table: Descriptive characteristics for thiazides versus calcium channel blockers (CCB) in matched stratum specific cohorts

Aged <55yrs Aged ≥55yrs Non-Black Black

CCB Thiazides CCB Thiazides CCB Thiazides CCB Thiazides

2862 2862 16232 16232 17938 17938 324 324

Females 1881 (65.7) 1902 (66.5) 10392 (64.0) 10414 (64.2) 11708 (65.3) 11636 (64.9) 222 (68.5) 223 (68.8)

Age (years)

<50 1751 (61.2) 1778 (62.1) 0 ( 0.0) 0 ( 0.0) 1616 ( 9.0) 1616 ( 9.0) 161 (49.7) 161 (49.7)

50-54 1111 (38.8) 1084 (37.9) 0 ( 0.0) 0 ( 0.0) 985 ( 5.5) 985 ( 5.5) 54 (16.7) 54 (16.7)

55-59 0 ( 0.0) 0 ( 0.0) 2048 (12.6) 2048 (12.6) 1861 (10.4) 1861 (10.4) 32 ( 9.9) 32 ( 9.9)

60-64 0 ( 0.0) 0 ( 0.0) 3035 (18.7) 3035 (18.7) 2795 (15.6) 2795 (15.6) 19 ( 5.9) 19 ( 5.9)

65-69 0 ( 0.0) 0 ( 0.0) 2879 (17.7) 2879 (17.7) 2708 (15.1) 2708 (15.1) 26 ( 8.0) 26 ( 8.0)

70-74 0 ( 0.0) 0 ( 0.0) 2819 (17.4) 2819 (17.4) 2676 (14.9) 2676 (14.9) 14 ( 4.3) 14 ( 4.3)

75+ 0 ( 0.0) 0 ( 0.0) 5451 (33.6) 5451 (33.6) 5297 (29.5) 5297 (29.5) 18 ( 5.6) 18 ( 5.6)

Ethnicity**

White 2304 (80.5) 2322 (81.1) 14478 (89.2) 14689 (90.5) 16833 (93.8) 17023 (94.9) 0 ( 0.0) 0 ( 0.0)

South asian 142 ( 5.0) 126 ( 4.4) 280 ( 1.7) 178 ( 1.1) 444 ( 2.5) 303 ( 1.7) 0 ( 0.0) 0 ( 0.0)

Black 0 ( 0.0) 0 ( 0.0) 0 ( 0.0) 0 ( 0.0) 0 ( 0.0) 0 ( 0.0) 324 (100.0) 324 (100.0)

Other/mixed/unknown 187 ( 6.5) 149 ( 5.2) 459 ( 2.8) 463 ( 2.9) 661 ( 3.7) 612 ( 3.4) 0 ( 0.0) 0 ( 0.0)

Missing 229 (8.00) 265 (9.26) 1015 (6.25) 902 (5.56) 0 (0.00) 0 (0.00) 0 (0.00) 0 (0.00)

BMI**

Underweight <18.5 16 ( 0.6) 14 ( 0.5) 319 ( 2.0) 302 ( 1.9) 309 ( 1.7) 308 ( 1.7) ~ ~

Healthy weight 18.5-24.9 454 (15.9) 462 (16.1) 4983 (30.7) 5000 (30.8) 5109 (28.5) 5157 (28.7) 75 (23.1) 54 (16.7)

Overweight 25-29.9 842 (29.4) 856 (29.9) 5936 (36.6) 6034 (37.2) 6472 (36.1) 6473 (36.1) 80 (24.7) 92 (28.4)

Obesity >=30 1431 (50.0) 1449 (50.6) 4251 (26.2) 4219 (26.0) 5267 (29.4) 5300 (29.5) 160 (49.4) 170 (52.5)

Missing 119 ( 4.2) 81 ( 2.8) 743 ( 4.6) 677 ( 4.2) 781 ( 4.4) 700 ( 3.9) 7 ( 2.2) 7 ( 2.2)

SMOKING**

Non-smoker 1074 (37.5) 1086 (37.9) 5642 (34.8) 5755 (35.5) 6255 (34.9) 6354 (35.4) 183 (56.5) 185 (57.1)

Current smoker 700 (24.5) 723 (25.3) 2548 (15.7) 2543 (15.7) 3048 (17.0) 3073 (17.1) 46 (14.2) 50 (15.4)

Ex-smoker 1085 (37.9) 1050 (36.7) 7993 (49.2) 7895 (48.6) 8575 (47.8) 8470 (47.2) 95 (29.3) 89 (27.5)

Missing 3 ( 0.1) 3 ( 0.1) 49 ( 0.3) 39 ( 0.2) 60 ( 0.3) 41 ( 0.2) 0 ( 0.0) 0 ( 0.0)

ALCOHOL**

Non-drinker 268 ( 9.4) 304 (10.6) 1771 (10.9) 1863 (11.5) 2033 (11.3) 2072 (11.6) 111 (34.3) 102 (31.5)

Current drinker 2221 (77.6) 2212 (77.3) 12029 (74.1) 12090 (74.5) 13250 (73.9) 13377 (74.6) 178 (54.9) 177 (54.6)

Ex-drinker 226 ( 7.9) 228 ( 8.0) 1588 ( 9.8) 1551 ( 9.6) 1729 ( 9.6) 1719 ( 9.6) 22 ( 6.8) 32 ( 9.9)

Missing 147 ( 5.1) 118 ( 4.1) 844 ( 5.2) 728 ( 4.5) 926 ( 5.2) 770 ( 4.3) 13 ( 4.0) 13 ( 4.0)

Comorbidities

Myocardial Infarction ~ ~ 121 ( 0.7) 130 ( 0.8) 133 ( 0.7) 130 ( 0.7) ~ ~

Stroke 30 ( 1.0) 27 ( 0.9) 642 ( 4.0) 637 ( 3.9) 646 ( 3.6) 660 ( 3.7) ~ ~

Heart Failure ~ ~ 83 ( 0.5) 90 ( 0.6) 104 ( 0.6) 92 ( 0.5) ~ ~

Peripheral Vascular Disease 26 ( 0.9) 25 ( 0.9) 707 ( 4.4) 711 ( 4.4) 768 ( 4.3) 727 ( 4.1) ~ ~

Diabetes 0 ( 0.0) 0 ( 0.0) 0 ( 0.0) 0 ( 0.0) 0 ( 0.0) 0 ( 0.0) 0 ( 0.0) 0 ( 0.0)

Depression 294 (10.3) 301 (10.5) 1027 ( 6.3) 986 ( 6.1) 1213 ( 6.8) 1251 ( 7.0) 7 ( 2.2) 13 ( 4.0)

COPD 53 ( 1.9) 45 ( 1.6) 1011 ( 6.2) 1013 ( 6.2) 1067 ( 5.9) 1040 ( 5.8) 6 ( 1.9) ~

Cancer 102 ( 3.6) 105 ( 3.7) 2242 (13.8) 2267 (14.0) 2365 (13.2) 2330 (13.0) 9 ( 2.8) 13 ( 4.0)

Herpes Zoster 124 ( 4.3) 115 ( 4.0) 1618 (10.0) 1595 ( 9.8) 1659 ( 9.2) 1623 ( 9.0) 8 ( 2.5) 8 ( 2.5)

Gout 51 ( 1.8) 54 ( 1.9) 495 ( 3.0) 497 ( 3.1) 549 ( 3.1) 529 ( 2.9) 6 ( 1.9) 5 ( 1.5)

Angioedema ~ ~ ~ ~ ~ ~ ~ ~

CKD**

no CKD 1586 (55.4) 1626 (56.8) 8828 (54.4) 8623 (53.1) 9731 (54.2) 9601 (53.5) 170 (52.5) 166 (51.2)

stage 3a 31 ( 1.1) 36 ( 1.3) 1653 (10.2) 1850 (11.4) 1634 ( 9.1) 1811 (10.1) 14 ( 4.3) 17 ( 5.2)

stage 3b ~ ~ 333 ( 2.1) 319 ( 2.0) 311 ( 1.7) 319 ( 1.8) ~ ~

stage 4 0 ( 0.0) 0 ( 0.0) 38 ( 0.2) 33 ( 0.2) 39 ( 0.2) 33 ( 0.2) ~ ~

Page 32 of 50


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


6

stage 5 0 ( 0.0) ~ 7 ( 0.0) ~ 12 ( 0.1) ~ ~ ~

Missing 1244 (43.5) 1198 (41.9) 5373 (33.1) 5406 (33.3) 6211 (34.6) 6172 (34.4) 138 (42.6) 136 (42.0)

Drugs

Antiplatelet agents 140 ( 4.9) 132 ( 4.6) 3001 (18.5) 2944 (18.1) 3002 (16.7) 3026 (16.9) 19 ( 5.9) 21 ( 6.5)

Statins 173 ( 6.0) 168 ( 5.9) 3296 (20.3) 3319 (20.4) 3364 (18.8) 3368 (18.8) 27 ( 8.3) 26 ( 8.0)

Proton Pump Inhibitors 846 (29.6) 845 (29.5) 6183 (38.1) 6190 (38.1) 6932 (38.6) 6838 (38.1) 83 (25.6) 79 (24.4)

Insulin 0 ( 0.0) 0 ( 0.0) 0 ( 0.0) 0 ( 0.0) 0 ( 0.0) 0 ( 0.0) ~ ~

Loop diuretics 132 ( 4.6) 120 ( 4.2) 1167 ( 7.2) 1131 ( 7.0) 1257 ( 7.0) 1234 ( 6.9) 7 ( 2.2) 11 ( 3.4)

NSAIDs 1799 (62.9) 1820 (63.6) 11103 (68.4) 10872 (67.0) 12221 (68.1) 12092 (67.4) 195 (60.2) 212 (65.4)

Primary Care Consultations**

<5 consultations 779 (27.2) 799 (27.9) 4069 (25.1) 4068 (25.1) 4357 (24.3) 4380 (24.4) 113 (34.9) 103 (31.8)

5-9 consultations 1008 (35.2) 1003 (35.0) 5494 (33.8) 5438 (33.5) 6064 (33.8) 6037 (33.7) 95 (29.3) 96 (29.6)

10-14 consultations 477 (16.7) 462 (16.1) 3000 (18.5) 3076 (19.0) 3414 (19.0) 3404 (19.0) 61 (18.8) 62 (19.1)

15-19 consultations 239 ( 8.4) 215 ( 7.5) 1577 ( 9.7) 1563 ( 9.6) 1729 ( 9.6) 1741 ( 9.7) 17 ( 5.2) 20 ( 6.2)

>=20 consultations 359 (12.5) 383 (13.4) 2092 (12.9) 2087 (12.9) 2374 (13.2) 2376 (13.2) 38 (11.7) 43 (13.3)

Missing 0 ( 0.0) 0 ( 0.0) 0 ( 0.0) 0 ( 0.0) 0 ( 0.0) 0 ( 0.0) 0 ( 0.0) 0 ( 0.0)

Systolic BP 156.9 (17.3) 157.4 (16.4) 166.7 (17.5) 166.7 (16.7) 165.3 (17.7) 165.2 (16.9) 158.5 (17.4) 159.6 (16.2)

Diastolic BP 99.0 (10.6) 97.5 ( 9.8) 91.0 (11.6) 90.8 (11.0) 92.2 (11.8) 91.6 (11.1) 96.9 (10.2) 97.7 (10.1)

Page 33 of 50


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


7

Before Matching After Matching

<55years

≥55years

Non-Black

Black

Page 34 of 50


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


8

Appendix 3: Diastolic blood pressure results

Appendix 3 Table: Difference in diastolic blood pressure since index date for CCB versus ACE-I/ARB and thiazide versus CCB.

Difference in Diastolic BP between CCB and ACE-I/ARB (99% CI)


Age <55* 20,964 -0.98 (-1.39, -0.56) -0.28 (-0.79, 0.22) 0.11 (-0.76, 0.97)

Age>55* 58,396 -1.06 (-1.39, -0.72) -0.14 (-0.47, 0.18) 0.01 (-0.47, 0.49)

Non-black** 73,726 -1.06 (-1.27, -0.85) -0.20 (-0.43, 0.02) -0.02 (-0.44, 0.40)

Black** 894 -2.46 (-4.11, -0.82) -0.96 (-3.40, 1.49) 0.99 (-2.48, 4.46)

Difference in Diastolic BP between thiazide and CCB (99% CI)

Age <55* 5,724 1.29 (0.40, 2.18) 0.62 (-0.22, 1.45) 0.83 (-0.75, 2.42)

Age>55* 32,464 1.80 (1.44, 2.15) 1.36 (1.00, 1.71) 0.80 (0.15, 1.45)

Non-black** 35,876 1.75 (1.41, 2.08) 1.24 (0.93, 1.56) 0.84 (0.29, 1.40)

Black** 648 0.82 (-1.11, 2.74) -0.26 (-2.86, 2.34) 1.92 (-3.04, 6.89)

*In a non-black, non-diabetic population ** In a non-diabetic population All analyses are 1:1 PS matched within each stratum. Each n consists of half CCB and half ACEI/ARB (or half thiazide and half CCB). For the CCB vs ACEI/ARB comparison a negative result means CCB produced larger reductions in systolic BP and a positive results indicates that ACEI/ARB produced larger reductions in systolic BP. For the thiazide vs CCB comparison a negative result means thiazides produced larger reductions in systolic BP and a positive results indicates that CCB produced larger reductions in systolic BP.

Page 35 of 50


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


9

Appendix 4: Diagrams of diastolic blood pressure results

Appendix 4 Figure 1: Achieved diastolic blood pressures for calcium channel blocker (CCB) and ACE-

inhibitor/angiotensin receptor blocker (ACE-I/ARB) in each stratum

Calcium Channel Blockers vs ACE-inhibitors/Angiotensin Receptor Blockers: Achieved diastolic BP since baseline in a non-black, non-diabetic population

Calcium Channel Blockers vs ACE-inhibitors/Angiotensin Receptor Blockers: Achieved diastolic BP since baseline in a non-diabetic population

Page 36 of 50


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


10

Appendix 4 Figure 2: Achieved diastolic blood pressures for thiazide and calcium channel blocker (CCB)

in each stratum

Thiazide vs Calcium Channel Blockers: Achieved diastolic BP since baseline in a non-black, non-diabetic population

Thiazide vs Calcium Channel Blockers: Achieved diastolic BP since baseline in a non-diabetic population

Page 37 of 50


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


11

Appendix 5: 95% CI for changes in systolic blood pressure

Appendix 5 Table: Difference in systolic blood pressure since drug initiation for calcium channel blockers (CCB) versus ACE-inhibitors/angiotensin receptor blockers

(ACEI/ARB) and thiazide versus CCB, 95% CI.



Age <55* 20,964 -1.69 (-2.32, -1.06) -0.48 (-1.24, 0.27) -0.53 (-1.62, 0.57)

Age>55* 58,396 -0.40 (-0.84, 0.04) 0.63 (0.20, 1.07) 0.85 (0.12, 1.58)

Non-black** 73,726 -0.98 (-1.37, -0.59) 0.11 (-0.30, 0.51) 0.08 (-0.60, 0.75)

Black** 894 -2.15 (-5.21, 0.91) 0.55 (-2.45, 3.55) 2.28 (-3.53, 8.09)


Age <55* 5,724 1.51 (0.34, 2.69) 0.07 (-1.28, 1.41) 0.34 (-1.74, 2.42)

Age>55* 32,464 2.16 (1.55, 2.77) 1.43 (0.77, 2.10) 0.17 (-0.68, 1.03)

Non-black** 35,876 2.10 (1.55, 2.65) 1.19 (0.58, 1.80) 0.32 (-0.49, 1.13)

Black** 648 -0.28 (-4.40, 3.84) -1.75 (-5.73, 2.23) 2.01 (-3.68, 7.69)

*In a non-black, non-diabetic population ** In a non-diabetic population All analyses are 1:1 PS matched within each stratum. Each n consists of half CCB and half ACEI/ARB (or half thiazide and half CCB). For the CCB vs ACEI/ARB comparison a negative result means CCB produced larger reductions in systolic BP and a positive results indicates that ACEI/ARB produced larger reductions in systolic BP. For the thiazide vs CCB comparison a negative result means thiazides produced larger reductions in systolic BP and a positive results indicates that CCB produced larger reductions in systolic BP.

Page 38 of 50


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


12

Appendix 6 Figure 1: Negative and positive outcomes for calcium channel blockers (CCB) versus ACE-inhibitors/angiotensin receptor blockers (ACEI/ARB) in age-based and ethnicity-based strata. ~ is used when cell sizes are <5

Appendix 6: Negative and positive outcomes in age-based and ethnicity-based strata

Page 39 of 50


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


13

Appendix 6 Figure 2: Negative and positive outcomes for thiazide versus calcium channel blockers (CCB) in age-based and ethnicity-based strata. ~ is used when cell sizes are <5

Page 40 of 50


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


14

Appendix 7: Comparison of thiazide-like diuretics with calcium channel blockers

Appendix 7 Figure: Comparison of thiazide-like diuretics (chlortalidone and indapamide) with calcium

channel blockers (CCB). Comparison of traditional thiazide diuretics (eg bendroflumethazide) with

CCB.

Data-points below the null line indicate that thiazides are better at reducing BP. Data-points above

the null line indicate that CCB are better at reducing BP.

Page 41 of 50


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


15

Appendix 8: On treatment analysis

Appendix 8 Table: Difference in systolic blood pressure since index date for CCB versus ACE-I/ARB and thiazide versus CCB in an on-treatment analysis



Age <55* 19,228 0.22 (-0.66, 1.11) 0.35 (-0.83, 1.54) 1.01 (-0.90, 2.93)

Age>55* 54,149 -0.19 (-0.95, 0.57) 0.17 (-0.51, 0.84) -0.09 (-1.21, 1.02)

Non-black** 35,876 -0.35 (-0.94, 0.24) 0.03 (-0.61, 0.67) -0.30 (-1.28, 0.67)

Black** 648 -3.65 (-7.99, 0.70) -5.16 (-11.13, 0.81) -3.36 (-12.10, 5.37)


Age <55* 5,056 3.04 (1.17, 4.90) 3.06 (0.72, 5.40) 0.40 (-3.03, 3.83)

Age>55* 29,965 4.14 (3.23, 5.04) 4.12 (3.05, 5.20), 3.91 (2.39, 5.44)

Non-black** 32,833 4.05 (3.21, 4.88) 3.93 (2.89, 4.97) 3.55 (2.03, 5.07)

Black** 555 2.73 (-3.51, 8.98) 1.46 (-4.95, 7.86) 3.47 (-8.51, 15.46) *In a non-black, non-diabetic population ** In a non-diabetic population All analyses are 1:1 PS matched within each stratum. Each n consists of half CCB and half ACE-i/ARB (or half thiazide and half CCB). For the CCB vs ACE-I/ARB comparison a negative result means CCB produced larger reductions in systolic BP and a positive results indicates that ACEI/ARB produced larger reductions in systolic BP. For the thiazide vs CCB comparison a negative result means thiazides produced larger reductions in systolic BP and a positive results indicates that CCB produced larger reductions in systolic BP. There was a median of 4 (IQR 2-6) BP measurements available during one year follow up for each patient. On treatment analysis: patients follow-up BP data were censored if the patient switched to another drug class, added a drug from another class or discontinued their anti-hypertensive treatment (drug classes: thiazides, CCB, ACE-I/ARB, beta-blockers, alpha-blockers, potassium sparing diuretics, aldosterone antagonists and other)

Page 42 of 50


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


16

Appendix 9: Complete case analysis

Appendix 9 Table: Complete case analysis



Age <55* 11,660 -1.77 (-2.77, -0.77) -0.45 (-1.63, 0.73) -0.84 (-2.40, 0.72)

Age>55* 40,890 -0.46 (-1.06, 0.14) 0.52 (-0.09, 1.12) 1.01 (-0.002, 2.02)

Non-black** 49,186 -1.06 (-1.61, -0.52) -0.12 (-0.69, 0.45) -0.04 (-0.96, 0.89)

Black** 592 -2.21 (-6.58, 2.17) 0.27 (-4.21, 4.76) 0.000 (-6.78, 6.78)


Age <55* 3,472 1.22 (-0.59, 3.03) 0.04 (-1.84, 1.92) -1.01 (-3.87, 1.85)

Age>55* 21,448 1.84 (0.98, 2.69) 0.83 (0.04, 1.63) -0.04 (-1.36, 1.28)

Non-black** 23,550 1.84 (1.01, 2.67) 0.89 (0.07, 1.70) -0.01 (-1.26, 1.25)

Black** 372 -1.11 (-7.29, 5.07) 0.88 (-5.00, 6.77) 2.75 (-6.01, 11.51)

*In a non-black, non-diabetic population

** In a non-diabetic population

All analyses are 1:1 PS matched within each stratum. Each n consists of half CCB and half ACEI/ARB (or half thiazide and half CCB).

For the CCB vs ACEI/ARB comparison a negative result means CCB produced larger reductions in systolic BP and a positive results indicates that ACE-I/ARB produced larger

reductions in systolic BP. For the thiazide vs CCB comparison a negative result means thiazides produced larger reductions in systolic BP and a positive results indicates that CCB

produced larger reductions in systolic BP. There was a median of 4 (IQR 2-6) BP measurements available during one year follow up for each patient.

Page 43 of 50


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


17

Appendix 10: RECORD-PE checklist

The RECORD statement for pharmacoepidemiology (RECORD-PE) checklist of items, extended from the STROBE and RECORD statements, which

should be reported in non-interventional pharmacoepidemiological studies using routinely collected health data

Item No STROBE items RECORD items RECORD-PE items Page No Title and abstract

1 (a) Indicate the study’s design with a

commonly used term in the title or the

abstract.

(b) Provide in the abstract an

informative and balanced summary of

what was done and what was found.

1.1: The type of data used should be

specified in the title or abstract. When

possible, the name of the databases

used should be included.

1.2: If applicable, the geographical

region and timeframe within which

the study took place should be reported in the title or abstract.

1.3: If linkage between databases was

conducted for the study, this should be

clearly stated in the title or abstract.

— 1 & 2

Introduction

Background rationale

2 Explain the scientific background and

rationale for the investigation being

reported.

— — 5

Objectives

3 State specific objectives, including

any prespecified hypotheses.

— — 5

Methods

Study design

4 Present key elements of study design

early in the paper.

— 4.a: Include details of the specific

study design (and its features) and

report the use of multiple designs if

used.

4.b: The use of a diagram(s) is recommended to illustrate key aspects

of the study design(s), including

6

Page 44 of 50


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


18

exposure, washout, lag and

observation periods, and covariate

definitions as relevant.

Setting

5 Describe the setting, locations, and

relevant dates, including periods of

recruitment, exposure, follow-up, and

data collection.

— — 6

Participants

6 (a) Cohort study—give the eligibility

criteria, and the sources and methods of selection of participants. Describe

methods of follow-up. Case-control

study—give the eligibility criteria,

and the sources and methods of case

ascertainment and control selection.

Give the rationale for the choice of

cases and controls. Cross sectional

study—give the eligibility criteria,

and the sources and methods of

selection of participants.

(b) Cohort study—for matched studies, give matching criteria and

number of exposed and unexposed.

Case-control study—for matched

studies, give matching criteria and the

number of controls per case.

6.1: The methods of study population

selection (such as codes or algorithms used to identify participants) should

be listed in detail. If this is not

possible, an explanation should be

provided.

6.2: Any validation studies of the

codes or algorithms used to select the

population should be referenced. If

validation was conducted for this

study and not published elsewhere,

detailed methods and results should be

provided. 6.3: If the study involved linkage of

databases, consider use of a flow

diagram or other graphical display to

demonstrate the data linkage process,

including the number of individuals

with linked data at each stage.

6.1.a: Describe the study entry criteria

and the order in which these criteria were applied to identify the study

population. Specify whether only

users with a specific indication were

included and whether patients were

allowed to enter the study population

once or if multiple entries were

permitted. See explanatory document

for guidance related to matched

designs.

6

Variables

7 Clearly define all outcomes,

exposures, predictors, potential

confounders, and effect modifiers.

Give diagnostic criteria, if applicable.

7.1: A complete list of codes and

algorithms used to classify exposures,

outcomes, confounders, and effect

modifiers should be provided. If these

cannot be reported, an explanation

should be provided.

7.1.a: Describe how the drug exposure

definition was developed.

7.1.b: Specify the data sources from

which drug exposure information for

individuals was obtained.

7.1.c: Describe the time window(s)

during which an individual is considered exposed to the drug(s).

The rationale for selecting a particular

6 & 7

Page 45 of 50


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


19

time window should be provided. The

extent of potential left truncation or

left censoring should be specified.

7.1.d: Justify how events are

attributed to current, prior, ever, or

cumulative drug exposure.

7.1.e: When examining drug dose and risk attribution, describe how current,

historical or time on therapy are

considered.

7.1.f: Use of any comparator groups

should be outlined and justified.

7.1.g: Outline the approach used to

handle individuals with more than one

relevant drug exposure during the

study period.

Data sources/measurement

8 For each variable of interest, give

sources of data and details of methods

of assessment (measurement).

Describe comparability of assessment methods if there is more than one

group.

— 8.a: Describe the healthcare system

and mechanisms for generating the

drug exposure records. Specify the

care setting in which the drug(s) of interest was prescribed.

6

Bias

9 Describe any efforts to address

potential sources of bias.

— — 6-8

Study size

10 Explain how the study size was

arrived at.

— — flowchart

Quantitative variables

11 Explain how quantitative variables

were handled in the analyses. If

applicable, describe which groupings

were chosen, and why.

— — 7

Statistical methods

12 (a) Describe all statistical methods,

including those used to control for

confounding.

— 12.1.a: Describe the methods used to

evaluate whether the assumptions

have been met.

7 & 8

Page 46 of 50


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


20

(b) Describe any methods used to

examine subgroups and interactions.

(c) Explain how missing data were

addressed.

(d) Cohort study—if applicable,

explain how loss to follow-up was

addressed. Case-control study—if applicable, explain how matching of

cases and controls was addressed.

Cross sectional study—if applicable,

describe analytical methods taking

account of sampling strategy.

(e) Describe any sensitivity analyses.

12.1.b: Describe and justify the use of

multiple designs, design features, or

analytical approaches.

Data access and cleaning methods

12 — 12.1: Authors should describe the

extent to which the investigators had

access to the database population used

to create the study population.

12.2: Authors should provide

information on the data cleaning

methods used in the study.

— 7

Linkage

12 — 12.3: State whether the study included

person level, institutional level, or

other data linkage across two or more

databases. The methods of linkage

and methods of linkage quality

evaluation should be provided.

— 6

Results

Participants

13 (a) Report the numbers of individuals

at each stage of the study (eg,

numbers potentially eligible,

examined for eligibility, confirmed

eligible, included in the study,

completing follow-up, and analysed).

(b) Give reasons for non-participation at each stage.

13.1: Describe in detail the selection

of the individuals included in the

study (that is, study population

selection) including filtering based on

data quality, data availability, and

linkage. The selection of included

individuals can be described in the

— See flowchart

Page 47 of 50


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


21

(c) Consider use of a flow diagram. text or by means of the study flow

diagram.

Descriptive data

14 (a) Give characteristics of study

participants (eg, demographic,

clinical, social) and information on

exposures and potential confounders.

(b) Indicate the number of participants

with missing data for each variable of interest.

(c) Cohort study—summarise follow-

up time (eg, average and total

amount).

— — See table 1

Outcome data

15 Cohort study—report numbers of

outcome events or summary measures

over time. Case-control study—report

numbers in each exposure category, or

summary measures of exposure. Cross

sectional study—report numbers of

outcome events or summary

measures.

— — See table 2

(outcome was

BP – a

continuous

outcome)

Main results

16 (a) Give unadjusted estimates and, if applicable, confounder adjusted

estimates and their precision (eg, 95%

confidence intervals). Make clear

which confounders were adjusted for

and why they were included.

(b) Report category boundaries when

continuous variables are categorised.

(c) If relevant, consider translating

estimates of relative risk into absolute

risk for a meaningful time period.

— — See Table 2

Other analyses

17 Report other analyses done—eg,

analyses of subgroups and

interactions, and sensitivity analyses.

— — 12

Page 48 of 50


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


22

Discussion

Key results

18 Summarise key results with reference

to study objectives.

— — 13

Limitations

19 Discuss limitations of the study,

taking into account sources of

potential bias or imprecision. Discuss

both direction and magnitude of any

potential bias.

19.1: Discuss the implications of

using data that were not created or

collected to answer the specific

research question(s). Include

discussion of misclassification bias,

unmeasured confounding, missing data, and changing eligibility over

time, as they pertain to the study

being reported.

19.1.a: Describe the degree to which

the chosen database(s) adequately

captures the drug exposure(s) of

interest.

13 & 14

Interpretation

20 Give a cautious overall interpretation

of results considering objectives,

limitations, multiplicity of analyses,

results from similar studies, and other

relevant evidence.

— 20.a: Discuss the potential for

confounding by indication,

contraindication or disease severity or

selection bias (healthy adherer/sick

stopper) as alternative explanations

for the study findings when relevant.

13

Generalisability

21 Discuss the generalisability (external

validity) of the study results.

— — 15

Other information

Funding

22 Give the source of funding and the

role of the funders for the present

study and, if applicable, for the original study on which the present

article is based.

— — 16

Accessibility of protocol, raw data,

and programming code

22 — 22.1: Authors should provide

information on how to access any

supplemental information such as the

study protocol, raw data, or

programming code.

— 9

Page 49 of 50


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


23

RECORD=reporting of studies conducted using observational routinely collected data; RECORD-PE=RECORD for pharmacoepidemiological research; STROBE=strengthening the

reporting of observational studies in epidemiology.

*REFERENCE: Langan SM, Schmidt S, Wing K, Ehrenstein V, Nicholls S, Filion K, Klungel O, Petersen I, Sorensen H, Guttmann A, Harron K, Hemkens L, Moher D,

Schneeweiss S, Smeeth L, Sturkenboom M, von Elm E, Wang S, Benchimol EI. The REporting of studies Conducted using Observational Routinely-collected health

Data (RECORD) Statement for Pharmacoepidemiology (RECORD-PE). BMJ 2018; 363: k3532.

Page 50 of 50


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

https://www.bmj.com/content/363/bmj.k3532



Documents

Confidential: For Review Only · 2019. 11. 11. · Confidential: For Review Only 1 Title The effectiveness of first-line antihypertensive drugs according to age and ethnicity: a cohort