JAMDA xxx (2019) 1e7

JAMDA
journal homepage: www.jamda.com

Original Studies

Discontinuation of Antihypertensive Medication, Cognitive

Complaints, and Incident Dementia
Jan Willem van Dalen PhD a, b, *, Eric P. Moll van Charante PhD c,
William A. van Gool PhD a, Edo Richard PhD b, a
a
Amsterdam UMC, University of Amsterdam, Department of Neurology, Amsterdam, the Netherlands
b
Department of Neurology, Donders Institute for Brain, Behaviour and Cognition, Radboud University Medical Center, Nijmegen, the Netherlands
c
Amsterdam UMC, University of Amsterdam, Department of General Practise, Amsterdam, the Netherlands

a b s t r a c t

Keywords: Objectives: To evaluate whether discontinuation of antihypertensive medication in community-dwelling

Older people older people is associated with a reduction in memory complaints and/or incident dementia.
hypertension Design: Prospective observational cohort study within the Prevention of Dementia by Intensive Vascular
antihypertensive drugs
Care (preDIVA) trial.
withdrawal
Setting and Participants: Community-dwelling participants (aged 70-78 years at baseline) who under-
cognition
dementia went 2-yearly assessments during 6-8 years of follow-up.
Measures: Cox regression analyses of the relation between discontinuation of antihypertensive medi-
cation during the study and change in subjective memory complaints, incident dementia, and mortality.
Results: Dementia occurred more often in participants discontinuing antihypertensive treatment (13.4%
vs 6.2%, P ¼ .02); mortality was similar (16.5% vs 13.9%, P ¼ .52). Discontinuation of antihypertensive
medication was associated with a double dementia hazard [hazard ratio (HR) (95% confidence
interval) ¼ 2.15 (1.15-4.03)], which somewhat attenuated after adjustment for sex, blood pressure,
number of antihypertensives and other medications [HR ¼ 1.92 (1.01-3.65)], and additionally for stroke,
cardiovascular disease, diabetes, smoking, memory complaints, and MMSE score [HR ¼ 1.79 (0.93-3.44)].
Antihypertensive discontinuation was associated with an approximately 50% higher hazard of dementia
and/or mortality combined [HR ¼ 1.58 (1.04-2.40); model 2: HR ¼ 1.64 (1.07-2.51); model 3: HR ¼ 1.49
(0.96-2.30)]. Antihypertensive discontinuation was not associated with change in memory complaints
[odds ratio (95% confidence interval) ¼ 0.96 (0.55-1.67)]. Subgroup and sensitivity analyses addressing
possible sources of bias and confounding gave similar results.
Conclusions/Implications: Our results suggest that antihypertensive withdrawal in community-dwelling
older people does not preserve cognition and may in fact increase dementia risk. This is not due to
reduced mortality as competing risk. Additional analyses suggest results are unlikely to be explainable by
confounding, reverse causality, or observational biases. Studies with person-specific reasons for anti-
hypertensive discontinuation may be able to exclude reverse causality completely. Given the beneficial
effects of antihypertensive medication on cardiovascular risk, observational data may be the best
currently obtainable on the pressing issue of when withdrawal of antihypertensives in older people is
acceptable and what consequences need to be weighed.
Ó 2018 AMDA e The Society for Post-Acute and Long-Term Care Medicine.

Elevated blood pressure (BP) is common in older people, affecting

This study was funded by the Dutch Ministry of Health,Welfare and Sport;
Dutch Innovation Fund of Collaborative Healthy Insurances; and the Netherlands
Organisation for Health Research and Development.
The authors declare no conflicts of interest.
* Address correspondence to Jan Willem van Dalen, PhD, Department of
Neurology, Room H2-225, Amsterdam UMC, University of Amsterdam Meibergdreef
9, Amsterdam, the Netherlands.
E-mail address: j.vandalen@amc.nl (J.W. van Dalen).
more than 70% of individuals aged 70-79.1 Midlife hypertension
(systolic BP >140 mmHg) is an important risk factor for dementia.2
This association is however unclear for hypertension in later life.3
Population-based cohort studies suggest that a J-shaped relation ex-
ists in old age: both lower and higher BP convey an increased risk of
cognitive decline and mortality compared to average levels.3,4 Hypo-
thetically, increased cerebrovascular resistance and faltering cerebral

https://doi.org/10.1016/j.jamda.2018.12.006
1525-8610/Ó 2018 AMDA e The Society for Post-Acute and Long-Term Care Medicine.
2 J.W. van Dalen et al. / JAMDA xxx (2019) 1e7
autoregulation associated with aging may cause BP lowering in older
people and evoke hypoperfusion, resulting in cerebral damage and
cognitive dysfunction.5 Cognitive symptoms might be the direct
consequence of hypoperfusion or may develop over time with accu-
mulating cerebrovascular damage resulting from hypoperfusion.3,5,6
Next to cognitive decline, low BP and antihypertensive drug (AHD)
in older people have been associated with adverse events, depression,
and activities of daily living impairment.7e10 Consequently, it has been
hypothesized that BP lowering in older people may be detrimental
and withdrawal of AHD treatment is warranted.5,11,12
To address this hypothesis, a few relatively small RCTs have been
conducted in patients with stroke and mild cognitive impairment.5,13
These did not show any beneficial effects on cognition. However, they
assessed short-term effects and short withdrawal periods only.
Consequently, a Cochrane meta-analysis concluded there was insuf-
ficient evidence regarding the effects of AHD withdrawal on cognitive
functioning.12 Given the beneficial effects of AHD in older people on
incident cardiovascular disease (CVD),14 ethical objections hinder the
conduct of long-term RCTs with extended withdrawal periods.
Therefore, observational data may present the best opportunity
available to study this relation and provide guidance on this matter,
although observational biases will need to be extensively addressed.
For this purpose, well-delineated cohorts with complete outcome data
over a long follow-up are required.
In this study, we evaluate the relation between withdrawal of AHD
in community-dwelling older people and subjective memory com-
plaints (SMCs), incident dementia and mortality, using the detailed
data from the Prevention of Dementia by Intensive Vascular care
(preDIVA) trial.15
Methods
Population
The preDIVA cluster-RCT evaluated the efficacy of an cardiovas-
cular risk intervention compared to usual care in community-dwelling
older people (aged 70-78 years at baseline).15 The intervention
comprised 4-monthly appointments with a practice nurse who
addressed cardiovascular risk factors. Exclusion criteria at baseline
were dementia and disorders likely to impede successful long-term
follow-up according to the general practitioner (GP), for example,
terminal illness or alcoholism. The trial ran from 2006 to 2015. Follow-
up ranged from 6 years for participants randomized in 2009 up to
8 years for those randomized in 2006-2007. Participants were
assessed at baseline and during 2-yearly assessments. At these visits,
data were collected on medical history, medication use, cardiovascular
risk factors, cognitive status [Mini Mental-State Examination
(MMSE)], depression [Geriatric Depression Scale (GDS-15)], and
disability. In both treatment groups, BP levels were reported back to
both the participant and the GP; whether AHD treatment was initiated
was left to the GP’s discretion. The trial found no difference between
intervention and control groups in incident dementia or mortality.15
For the current study, we evaluated the RCT population as 1 single
cohort. A sensitivity analysis was conducted adjusting for the
intervention.
Antihypertensive Withdrawal
At each assessment, AHD use was derived from the GP prescription
list. To limit bias introduced by comparing participants with long-
standing hypertension to those who only developed hypertension
during the course of the study, we only included participants using
AHD at baseline. AHD withdrawal was defined as discontinuation of
AHD treatment and remaining without AHD prescription for the
remainder of the study participation.
Memory Complaints, Dementia, and Mortality
SMCs were defined according to item 10 (“Do you feel you have
more problems with your memory than most people?”) of the 15-item
Geriatric Depression Scale.16 Dementia was defined as a clinical
diagnosis according to Diagnostic and Statistical Manual of Mental
Disorders, Fourth Edition, criteria,17 confirmed by 2 members of an
independent outcome adjudication committee, blinded to treatment
allocation, based on all available clinical information. Diagnoses of
dementia were re-evaluated after 1 year of additional follow-up to
avoid false-positive diagnoses. For participants who dropped out of
the study, the dementia status was retrieved by a research nurse from
electronic health records and/or contact with the GP at the end of the
study and presented to the blinded outcome adjudication committee.
The dementia diagnostic process is described in more detail else-
where.15 Survival status and date of death were obtained directly from
the GP records supplemented with data from the national municipal
health registry.
Statistics
Analyses were performed using lme and coxme packages in R
studio with R version 3.5.1.18e21 Groups were compared using Fisher
exact tests for proportions, and Student t tests for normal and Mann-
Whitney U tests for nonparametric distributed variables.
The relation between AHD withdrawal and incident dementia and/
or mortality was assessed using Cox regression. We assessed dementia
and mortality separately and combined as 1 outcome to evaluate the
competing risk of death and because both can be considered unfa-
vorable. Age was used as time-scale and baseline age included as co-
variate. Model 2 was additionally adjusted for sex and baseline values
of systolic BP and total number of prescribed AHD as proxies for hy-
pertension severity, and total number of prescribed medications as
proxy for comorbidity and polypharmacy. Model 3 was additionally
adjusted for factors deemed likely to influence both the decision to
withdraw AHD treatment and the risk of adverse outcomes according
to treatment guidelines.22 Stroke, CVD, and diabetes were included as
comorbidities included in AHD treatment guidelines that increase the
risk of poor outcome and reduce the likelihood of AHD withdrawal.
Body mass index and smoking were included as lifestyle factors
associated with high BP that may also reduce the chance of AHD
withdrawal and increase the risk of poor outcome. SMC and MMSE
score were included as markers of cognitive complaints to account for
potential reverse causality by GPs withdrawing AHD in response to
cognitive symptoms. Interactions between AHD withdrawal and these
confounders were assessed for all outcomes.
To further examine reverse causality by AHD withdrawal in
response to memory complaints, we performed subgroup analyses
based on ever reporting SMCs during the study. To further address
confounding by severe medical conditions leading to AHD withdrawal,
we performed subgroup analyses based on whether participants used
other medication when withdrawing from AHD. Finally, in a previous
qualitative study among GPs participating in preDIVA into motivations
for deprescribing AHD, the reasons given were prolonged achieve-
ment of target blood pressure, side effects, patient preference (non-
adherence), terminal illness, and expected quality of life.23 Based on
this, because withdrawing AHD in participants whose hypertension
was not under control (ie, >140 mmHg) is more likely to have occurred
in the context of potentially confounding reasons such as non-
adherence, unacceptable side effects, or severe disease, we performed
analyses in subgroups of participants who did and did not attain a
systolic BP 140 during the course of the study.
To evaluate whether SMCs were associated with concurrent AHD
use and whether AHD withdrawal was associated with improvement
of SMCs, we assessed SMCs and AHD use in the total preDIVA
 J.W. van Dalen et al. / JAMDA xxx (2019) 1e7
population, regardless of AHD use at baseline. Mixed models were
employed, including all available measurements during the study,
with a random intercept per participant to adjust for multiple mea-
surements per individual. The cross-sectional relation between SMCs
and concurrent AHD was assessed with AHD use as predictor and
SMCs at the same visit as outcome. The relation between AHD with-
drawal and improvement of SMC was assessed with SMC as outcome
and AHD withdrawal compared to the preceding visit as predictor,
only including measurements with SMC present at the preceding visit.
Ethical Approval and Registration
The preDIVA study was approved by the ethics committee of the
Academic Medical Center Amsterdam, and all participants gave
written informed consent prior to the study. The preDIVA trial was
registered with ISRCTN, number ISRCTN29711771.
Results
In total, 1541 of 1934 participants (79.7%) on AHD at baseline
attended at least 1 follow-up assessment (Figure 1). Of those, 85
participants (5.5%) discontinued AHD during the study. Follow-up for
dementia was complete for 1512/1541 participants (98.1%) and for
mortality for 1537/1541 participants (99.7%). Characteristics of par-
ticipants who did and did not discontinue AHD during the study are
compared in Table 1. At baseline, participants who discontinued AHD
generally used less antihypertensive drugs [median (interquartile
range): 1 (1-2) vs 2 (1-2), P < .001], had a lower systolic BP
Fig. 1. Flow-chart of participants included in the study. “AHD stop” indicates cessation of AHD use for the remainder of the study with at least 1 additional subsequent visit
attended.
3
(mean  standard deviation: 147.9  21.9 vs 156.1  20.9, P ¼ .001),
lower mean body mass index (27.0  3.6 vs 28.5  4.3, P ¼ .001), and
less often a history of CVD (28.2% vs 43.0%, P ¼ .01). Compared with
AHD continuation, AHD withdrawal was more common in the lowest
(55.3% vs 33.5%, P ¼ .01) and less common in the highest tertile (17.6%
vs 32.9%, P ¼ .03) of baseline systolic BP. During the course of the study,
67.1% in the AHD withdrawal group attained a systolic BP 140 mmHg
compared to 55.6% in the nonwithdrawal group (P ¼ .04). Participants
discontinuing AHD more often reported SMC at least once during the
study (44.7% vs 33.7%, P ¼ .045), and 25.9% reported SMC at the visit
preceding AHD withdrawal. At the time of discontinuing AHD, 65.9%
used other medication. Dementia occurred more often in participants
discontinuing AHD treatment compared to those who did not (13.4%
vs 6.2%, P ¼ .02); mortality was not significantly different (16.5% vs
13.9%, P ¼ .52). Most incident cases of dementia (7/11, 63.6%) and
mortality (11/14, 78.6%) occurred more than 2 years after AHD with-
drawal (Supplementary Figure 1). There were no significant differ-
ences in cause of death between participants with and without AHD
withdrawal, but subgroups were small (Supplementary Table 1).
Results of Cox regression of incident dementia and/or mortality are
listed in Table 2. Withdrawal of AHD treatment was associated with an
approximately 2-fold higher hazard of dementia in model 1 [hazard
ratio (HR) (95% confidence interval [CI]) ¼ 2.15 (1.15-4.03)]. This was
somewhat attenuated after adjustment for sex, systolic BP, total
number of prescribed medications, and antihypertensives in model 2
[HR ¼ 1.92 (1.01-3.65)] and additional adjustment for history of stroke,
CVD, diabetes, smoking, SMC, and MMSE score in model 3 [HR ¼ 1.79
(0.93-3.44)]. AHD withdrawal was associated with an approximately
4 J.W. van Dalen et al. / JAMDA xxx (2019) 1e7

Table 1
Characteristics of Participants With Antihypertensive Drugs at the Start of the Study Who Did and Did Not Discontinue Treatment During the Study

AHD Stop* (n ¼ 85) No AHD Stop (n ¼ 1451) P

Baseline
Age, y, mean (SD) 74.5 (2.5) 74.3 (2.5) .50
Women 50 (58.8) 778 (53.6) .37
Systolic BP, mmHg
Mean (SD) 147.9 (21.9) 156.1 (20.9) .001
99-146 47 (55.3) 486 (33.5) .01
146-164 23 (27.1) 487 (33.6) .42
164-232 15 (17.6) 478 (32.9) .03
Systolic BP preceding AHD stop, mmHg
Diastolic BP, mmHg .20
Mean (SD) 79.8 (11.5) 81.5 (10.8)
49e76.5 35 (41.2) 488 (33.6) .33
76.5e85.5 26 (30.6) 484 (33.4) .82
85.5e131 24 (28.2) 479 (33.0) .57
Diastolic BP preceding AHD stop, mmHg
Number of AHDs <.001
Median (IQR) 1 (1-2) 2 (1-2)
1 58 (68.2) 582 (40.1) .003
2 23 (27.1) 544 (37.5) .18
3 3 (3.5) 261 (18.0) .001
4 1 (1.2) 64 (4.4) .26
Number of prescriptions other than AHD, median (IQR) 4 (3-7) 5 (3-7) .06
BMI, mean (SD) 27.0 (3.6) 28.5 (4.3) .001
Diabetes 17 (2) 403 (27.8) .13
Smoking 14 (16.5) 161 (11.1) .16
History of CVD 24 (28.2) 620 (43.0) .01
History of stroke 12 (14.1) 184 (12.8) .74
MMSE score, median (IQR) 28 (27-29) 29 (27-29) .50
VAT score <6 390 (27.0) 387 (26.9) .01
During study
Systolic BP 140 mmHg achieved 57 (67.1) 809 (55.6) .04
Medication other than AHD during study 76 (89.4) 1313 (90.2) .85
Medication concurrent with AHD stop 56 (65.9)
Subjective memory complaints (GDS 10y) during study 38 (44.7) 488 (33.7) .045
Subjective memory complaints preceding AHD stop 22 (25.9)
Incident dementia during study 11 (13.4) 89 (6.2) .02
Incident mortality during study 14 (16.5) 201 (13.9) .52
Incident dementia or mortality during study 24 (29.3) 268 (18.8) .03

Note. Bold values are statistically significant (P < .05).

BMI, body mass index; IQR, interquartile range; SD, standard deviation; VAT, visual association test.
The table lists characteristics at baseline unless stated otherwise. Missing values (AHD stop/no AHD stop): BMI, 0/1; smoking, 0/3; CVD, 0/9; stroke, 0/16; MMSE, 1/0; GDS 10,
0/1; dementia, 3/26; mortality, 0/4; dementia/mortality, 3/28. Values presented are n (%) unless otherwise noted.
*Cessation of AHD use for the remainder of the study with at least 1 additional subsequent visit attended.
y
GDS 10: Geriatric Depression Scale question 10: “Do you feel you have more problems with your memory than most people?”

60% increased hazard for dementia and/or mortality combined in
model 1 [HR ¼ 1.58 (1.04-2.40)] and model 2 [HR ¼ 1.64 (1.07-2.51)],
which was slightly attenuated in model 3 [HR ¼ 1.49 (0.96-2.30),
P ¼ .07]. The hazard for mortality for participants who discontinued
AHD was 16% to 21% higher but not significantly in any of the models
(all P  .38).
HRs were not significantly different between subgroup pairs for
any of the outcomes. HRs for dementia in participants discontinuing
AHD appeared slightly lower in those with concurrent medication at
the time of AHD withdrawal compared to those without [model 3
HR ¼ 1.54 (0.69-3.45) vs 2.60 (0.93-7.24)] and for participants with
SMCs during the course of the study compared to those without
[HR ¼ 1.57 (0.68-3.59) vs 2.17 (0.71-6.60)]. For dementia and/or
mortality combined, HRs appeared higher in subgroups of participants
without other medication at the time of AHD withdrawal [HR ¼ 2.33
(1.22-4.42) vs 1.18 (0.68-2.06)] and those not attaining a systolic BP of
140 mmHg during the study [HR ¼ 2.07 (1.02-4.21) vs 1.18 (0.68-
2.05)]. HRs for mortality for AHD withdrawal appeared higher in
participants without concurrent medication [2.30 (1.07-4.95) vs 0.80
(0.37-1.73)] and higher in participants who did not attain a BP 140
[1.86 (0.78-4.42) vs 0.89 (0.78-4.42)]. HRs in the other subgroup pairs
were similar.
There was a significant interaction between AHD withdrawal and
baseline systolic BP in the risk for mortality [HR interaction ¼ 1.29
(1.04-1.59) per 10 mmHg increase, P ¼ .02], suggesting the HR for
discontinuing AHD treatment increased by approximately 30% per
10 mmHg higher systolic BP. Other interactions between AHD with-
drawal and confounders included in the models were not significant
(all P > .13).
Sensitivity analyses excluding participants who did not attain a BP
140 mmHg during the course of the study (Supplementary Table 1)
gave largely similar results regarding dementia [model 3 overall
HR ¼ 1.73 (0.79-3.83)]. HRs were generally lower for dementia/mor-
tality combined [overall HR ¼ 1.16 (0.66-2.02)] and mortality [overall
HR ¼ 0.89 (0.66-2.02)]. Results of sensitivity analyses excluding par-
ticipants in whom dementia/mortality occurred within 2 years of
follow-up (Supplementary Table 2) were also largely similar, with a
somewhat increased HR for dementia [2.08 (1.00-4.33), P ¼ .049], a
similar HR for dementia/mortality combined [HR ¼ 1.42 (0.86-2.36)],
and a neutral HR for mortality [HR ¼ 1.00 (0.51-1.95)]. Subgroup pairs
showed corresponding trends, and none of the subgroup pairs were
significantly different. Sensitivity analysis adjusting for the interven-
tion gave results similar to the main analysis.
Table 3 lists the results of mixed model analyses of the association
between AHD use and SMCs at any time point during the study.
Combining 11,252 observations in 3518 individual participants, par-
ticipants with SMCs were more often concurrently on AHD compared
to participants without memory complaints [64.2% vs 60.5%, odds
 J.W. van Dalen et al. / JAMDA xxx (2019) 1e7 5

Table 2
Cox Regression of Dementia and/or Mortality on Discontinuation of AHD Treatment in Participants With AHD at Baseline

Model 1 Model 2 Model 3

Event/Total HR (95% CI) P Event/Total HR (95% CI) P Event/Total HR (95% CI) P

Outcome: dementia
AHD stop 100/1507 2.15 (1.15-4.03) .02 100/1507 1.92 (1.01-3.65) .047 98/1467 1.79 (0.93-3.44) .08
Concurrent medication
AHD stop without other medication 93/1453 2.70 (0.99-7.39) .053 93/1453 2.59 (0.94-7.11) .07 91/1415 2.60 (0.93-7.24) .07
AHD stop with other medication 96/1479 1.96 (0.91-4.25) .09 96/1479 1.70 (0.77-3.77) .19 94/1439 1.54 (0.69-3.45) .29
Memory complaints during study
AHD stop without complaints 34/985 3.04 (1.07-8.69) .04 34/985 2.33 (0.79-6.89) .13 33/961 2.17 (0.71-6.60) .17
AHD stop with complaints 66/521 1.50 (0.68-3.30) .31 66/521 1.39 (0.62-3.13) .42 65/509 1.57 (0.68-3.59) .29
Systolic BP on target during study
AHD stop systolic BP 140 mmHg 62/903 2.25 (1.07-4.73) .03 62/903 1.94 (0.90-4.18) .09 61/884 1.73 (0.79-3.83) .17
AHD stop systolic BP >140 mmHg 38/604 2.20 (0.67-7.18) .19 38/604 2.05 (0.61-6.89) .25 37/583 1.93 (0.56-6.65) .30
Outcome: dementia/mortality
AHD stop 292/1505 1.58 (1.04-2.40) .03 292/1505 1.64 (1.07-2.51) .02 287/1475 1.49 (0.96-2.30) .07
Concurrent medication
AHD stop without other medication 278/1451 2.24 (1.19-4.22) .01 278/1451 2.42 (1.28-4.57) .01 272/1413 2.36 (1.24-4.48) .01
AHD stop with other medication 282/1477 1.31 (0.76-2.24) .33 282/1477 1.32 (0.76-2.29) .32 276/1437 1.22 (0.70-2.12) .49
Memory complaints during study
AHD stop without complaints 157/984 1.36 (0.70-2.67) .37 157/984 1.43 (0.72-2.85) .30 153/960 1.34 (0.67-2.70) .40
AHD stop with complaints 135/520 1.55 (0.90-2.65) .11 135/520 1.60 (0.92-2.79) .10 133/505 1.60 (0.91-2.81) .11
Systolic BP on target during study
AHD stop systolic BP 140 mmHg 173/901 1.42 (0.83-2.41) .20 173/901 1.32 (0.77-2.27) .32 170/882 1.20 (0.69-2.09) .52
AHD stop systolic BP >140 mmHg 119/604 2.04 (1.03-4.05) .04 119/604 2.31 (1.15-4.64) .02 116/583 2.15 (1.06-4.39) .04
Outcome: mortality
AHD stop 215/1532 1.16 (0.68-2.00) .59 215/1532 1.28 (0.74-2.22) .38 210/1491 1.21 (0.69-2.11) .51
Concurrent medication
AHD stop without other medication 208/1476 2.00 (0.94-4.26) .07 208/1476 2.33 (1.09-4.97) .03 203/1437 2.30 (1.07-4.95) .03
AHD stop with other medication 208/1503 0.82 (0.38-1.73) .60 208/1503 0.86 (0.4-1.86) .70 203/1462 0.80 (0.37-1.73) .57
Memory complaints during study
AHD stop without complaints 133/1007 1.02 (0.45-2.31) .96 133/1007 1.18 (0.51-2.72) .69 129/982 1.24 (0.53-2.87) .62
AHD stop with complaints 82/524 1.25 (0.60-2.60) .55 82/524 1.38 (0.65-2.93) .40 81/509 1.22 (0.57-2.62) .61
Systolic BP on target during study
AHD stop systolic BP 140 mmHg 121/915 0.99 (0.48-2.03) .98 121/915 0.98 (0.47-2.04) .96 119/896 0.89 (0.42-1.86) .75
AHD stop systolic BP >140 mmHg 94/617 1.57 (0.68-3.62) .29 94/617 1.89 (0.81-4.42) .14 91/595 1.86 (0.78-4.42) .16

HR, hazard ratio.

Model 1: adjusted for age; model 2: additionally adjusted for sex, baseline mean systolic blood pressure, total number of prescribed medications and total number of pre-
scribed antihypertensive drugs; model 3: additionally adjusted for baseline history of stroke, cardiovascular disease, diabetes, current smoking, subjective memory com-
plaints, and MMSE score.

ratio (95% CI) ¼ 1.28 (1.06-1.55), P ¼ .01]. However, withdrawal of AHD
treatment was not associated with co-occurring improvement of
memory complaints [odds ratio (95% CI) ¼ 0.96 (0.55-1.67), P ¼ .89].
Discussion
Our results suggest AHD withdrawal in community-dwelling older
people does not convey clinical benefits regarding incident dementia
or mortality. Conversely, participants discontinuing AHD treatment
seemed to be at higher risk of adverse outcome, particularly dementia.
This increased risk appeared greater in participants without other
medication at the time of AHD withdrawal and in those without SMCs.
Dementia risk was similar for participants who did and did not attain a
systolic BP 140 mmHg during the study. Results were not markedly
different after adjustment for confounders relating to comorbidity,
polypharmacy, cardiovascular risk, or cognitive symptoms. Partici-
pants on AHD more often reported SMCs, but there was no association
between withdrawal of AHD treatment and change in memory
complaints.
Our data are observational, precluding firm conclusions regarding
causality. We tried to address bias and confounding through extensive
adjustment and subgroup analyses. Adjustment for SMC and MMSE
scores together with SMC subgroup analyses suggest little risk of
reverse causality caused by clinicians withdrawing AHD treatment in

Table 3
Repeated Measurements Analyses of the Association Between AHD Use and Co-occurring SMCs

Observations/Individuals SMCs, No SMCs, OR (95% CI) P

n (%) n (%)
(n ¼ 2108) (n ¼ 9144)

Concurrent AHD use 11,252/3518 1354 (64.2) 5528 (60.5) 1.28 (1.06-1.55) .01

Observations/Individuals Improved SMCs, Continuing SMCs, OR (95% CI) P

n (%) n (%)
(n ¼ 539) (n ¼ 728)

Co-occurring AHD stop 1267/739 32 (5.9) 43 (5.9) 0.96 (0.55-1.67) .89

OR, odds ratio.

Mixed models combining all observations during the study with random intercept to account for clustering within individuals. Incident SMCs operationalized as new SMCs
compared to previous visit only including measurements without SMCs at the preceding visit. Improved SMCs operationalized as no longer having SMCs only including
measurements with SMCs present at the preceding visit.
6 J.W. van Dalen et al. / JAMDA xxx (2019) 1e7
response to cognitive complaints. Adjustment for number of pre-
scriptions and comorbidity and subgroup analyses based on medica-
tion use make reverse causality due to AHD withdrawal in response to
polypharmacy and comorbidity unlikely. Results of the sensitivity
analysis excluding cases within the first 2 years of follow-up and
following AHD withdrawal further diminish reverse causality risk.
Risk of observation bias in relation to outcome is negated by the high
completeness of follow-up. We do not know the reasons for dis-
continuing AHD for each participant individually. In a qualitative
study among GPs who participated in the preDIVA trial, the reasons
for AHD deprescription named were BP target achievement, side ef-
fects, hypotension, risk of falling, patient preference, terminal illness,
and anticipated benefits on quality of life.23 We tried to address the
potential confounding and bias these may cause in our analyses, but
because exact motivations for AHD withdrawal per individual are
unknown, we cannot completely exclude residual confounding.
Our study was not equipped to evaluate the relation between AHD
withdrawal and mortality in detail. It is unlikely that GPs readily
withdrew AHD in participants with severe hypertension or CVD,23
which may have led to the relatively low overall mortality HR. This
is supported by the low proportion of participants with AHD with-
drawal in the higher tertiles of baseline BP and by AHD withdrawal
being more strongly associated with mortality in participants with
higher baseline systolic BP. In addition, in the few participants with
high baseline BP that discontinued AHD, GPs must have had
compelling reasons to withdraw treatment, which may relate to
mortality risk and could cause bias.23
In using prescription lists obtained through electronic health re-
cords we did not assess participant adherence nor do we know the
AHD status of study dropouts from the time they left the study.
However, these potential sources of misclassification would likely be
nondifferential, leading to dilution and thereby underestimation of
the true effect sizes.24 In the maximally adjusted model including 11
covariates, our study may be underpowered, as illustrated by the 49%
higher hazard for combined dementia and/or mortality having 95% CI
including 1. Furthermore, we were unable to evaluate the effect of
AHD withdrawal on specific cause of death as a result of small
subgroups.
Regarding generalizability, preDIVA participants likely form a
relatively healthy selection of older people with hypertension.
Baseline age ranged 70-78 years. Terminal illness and dementia were
exclusion criteria at baseline, and participants with severe medical
conditions are more likely to have forgone the 2-yearly assess-
ments.25 Our results may therefore not be directly translatable to frail
geriatric populations, or those with extensive comorbidity and pol-
ypharmacy, in whom other treatment considerations may apply.26e28
Also, our study only evaluated relations with memory complaints,
incident dementia, and mortality, and does not address any other
potential benefits of AHD withdrawal, such as alleviation from side
effects.
Little evidence exists regarding the effects of AHD withdrawal on
prevention of cognitive decline.11,12 A recent Cochrane meta-analysis
concluded that data were too sparse to make inferences.12 Results
regarding initiation of AHD treatment in older people on cognition
are inconclusive, but suggest therapy may have beneficial
effects.29e32 A recent review concluded that AHD withdrawal in older
people is relatively well tolerated, and 25% of hypertension cases
remain under control 2 years, but did not report on effects on
cognition.11 Regarding memory complaints, our results show that
withdrawal of AHD is not associated with changes in SMCs, corre-
sponding to results from RCTs that suggest withdrawal of AHD has no
short-term cognitive benefits.5,13 The cross-sectional association be-
tween AHD and memory complaints in our study may be due to
indication bias: having hypertension being a risk factor for cognitive
impairment.2
Conclusions/Relevance
AHD treatment in older people is currently a highly debated sub-
ject. The recent large-scale SPRINT trial suggested that a systolic
treatment target of <120 mmHg compared to 140 mmHg in in-
dividuals with elevated CVD risk is efficacious for the prevention of
CVD and CVD mortality, also in high age and frailty groups. Following
this, recent American College of Cardiology/American Heart Associa-
tion hypertension guidelines have lowered the systolic BP threshold
for antihypertensive treatment in non-institutionalized older people
to <130 mmHg, also for high age categories.33,34 This has raised
concerns about adverse effects, medicalization, and the great increase
in hypertension prevalence.35e39 The J-shaped relation for BP and risk
of cognitive decline and mortality adds to these concerns.3,4 Weighing
expected harms and benefits, the American College of Physicians and
American Academy of Family Physicians’ latest guideline recommends
a treatment threshold of 150 mmHg for individuals aged >60 years.37
The European Society of Cardiology and European Society of Hyper-
tension guidelines and those of Hypertension Canada recommend a
threshold of 140 mmHg for older people, the European Society of
Cardiology additionally specifying BP should not drop below 130/
70 mmHg.38,39 Adjustment of treatment targets may be necessary
according to CVD risk, comorbidity, activities of daily living, cognitive
impairment, high age, and limited life expectancy.40e42 A recent
prospective study in 90-year-olds suggested the oldest old with
elevated BP on antihypertensive medication might have higher risk of
mortality than those without, regardless of comorbidity or frailty.42
Regarding cognition, a recent systematic review and meta-analysis
concluded that lowering BP in older people does not significantly
reduce dementia risk.43 Although SPRINT trial authors announced
significant beneficial effects on incident mild cognitive impairment,
these data remain to be published following peer review.44 A large
recent prospective observational cohort study involving >12,000 in-
dividuals (median age 81 years) suggested a J-shaped risk-curve, with
a BP range 120 to 130 mmHg conveying the lowest risk of incident
cognitive impairment, but did not report on AHD treatment nor the
competing risk of mortality.45 The J-shaped risk curve has led to the
hypothesis that BP lowering in older age groups may cause cognitive
symptoms, and given the controversial benefit and other adverse ef-
fects of AHD treatment and polypharmacy in older people, withdrawal
of AHD treatment is warranted.3,5e12
Our results contribute to this discussion by suggesting that
regardless of the effects of initiating AHD treatment, withdrawal of
AHD is not beneficial and may in fact increase dementia risk. Subgroup
analyses in participants attaining a systolic BP 140 mmHg suggest
this is regardless of whether target BP <140 mmHg has been achieved.
The increased risk of dementia in our study is not due to a reduction in
the competing risk of mortality, because the combined HR of these
outcomes was also higher in participants discontinuing AHD. Our
analyses make it unlikely that results can be merely attributed to
confounding, reverse causality, or observational biases, but replication
is needed. Although observational data are limited, it is important that
they are made available because they are currently lacking and may
help answer an important clinical question: When is withdrawal of
AHD in older people acceptable and what beneficial and detrimental
consequences need to be weighed? Observational data are essential
for the design of RCTs addressing this subject and for the question of
whether such trials are ethical and necessary.
References
1. Mills KT, Bundy JD, Kelly TN, et al. Global disparities of hypertension prevalence
and control. Circulation 2016;134:441e450.
2. Livingston G, Sommerlad A, Orgeta V, et al. Dementia prevention, intervention,
and care. Lancet 2017;390:2673e2734.
 J.W. van Dalen et al. / JAMDA xxx (2019) 1e7
3. Iadecola C, Yaffe K, Biller J, et al. Impact of hypertension on cognitive function:
A scientific statement from the American Heart Association. Hypertension
2016;68:e67ee94.
4. Tadic M, Cuspidi C, Hering D. Hypertension and cognitive dysfunction in
elderly: Blood pressure management for this global burden. BMC Cardiovasc
Disord 2016;16:208.
5. Moonen JEF, Foster-Dingley JC, de Ruijter W, et al. Effect of discontinuation of
antihypertensive treatment in elderly people on cognitive functioningdThe
DANTE Study Leiden. JAMA Intern Med 2015;175:1622.
6. Yang T, Sun Y, Lu Z, et al. The impact of cerebrovascular aging on vascular
cognitive impairment and dementia. Ageing Res Rev 2017;34:15e29.
7. Lenoir H, Lacombe J-M, Dufouil C, et al. Relationship between blood pressure
and depression in the elderly. The Three-City Study. J Hypertens 2008;26:
1765e1772.
8. Sabayan B, Oleksik AM, Maier AB, et al. High blood pressure and resilience to
physical and cognitive decline in the oldest old: The Leiden 85-Plus study. J Am
Geriatr Soc 2012;60:2014e2019.
9. Bangalore S, Toklu B, Gianos E, et al. Optimal systolic blood pressure target after
SPRINT: Insights from a network meta-analysis of randomized trials. Am J Med
2017;130:707e719.e8.
10. Tinetti ME, Han L, Lee DSH, et al. Antihypertensive medications and serious fall
injuries in a nationally representative sample of older adults. JAMA Intern Med
2014;174:588e595.
11. van der Wardt V, Harrison JK, Welsh T, et al. Withdrawal of antihypertensive
medication. J Hypertens 2017;35:1742e1749.
12. Jongstra S, Harrison JK, Quinn TJ, Richard E. Antihypertensive withdrawal for
the prevention of cognitive decline. Cochrane Database Syst Rev 2016;11:
CD011971.
13. Bath PMW, Woodhouse L, Scutt P, et al. Efficacy of nitric oxide, with or without
continuing antihypertensive treatment, for management of high blood pres-
sure in acute stroke (ENOS): A partial-factorial randomised controlled trial.
Lancet 2015;385:617e628.
14. Musini VM, Tejani AM, Bassett K, Wright JM. Pharmacotherapy for hyperten-
sion in the elderly. Cochrane Database Syst Rev 2009;4:CD000028.
15. Moll van Charante EP, Richard E, Eurelings LS, et al. Effectiveness of a 6-year
multidomain vascular care intervention to prevent dementia (preDIVA): A
cluster-randomised controlled trial. Lancet 2016;388:797e805.
16. Sheikh JI, Yesavage JA. Geriatric Depression Scale (GDS): Recent evidence and
development of a shorter version. Clin Gerontol 1986;5:165e173.
17. American Psychiatric Association. Diagnostic and Statistical Manual of Mental
Disorders. 4th ed. Washington, DC: American Psychiatric Association; 2000.
18. Therneau TM. coxme: Mixed effects Cox models. R Packag version 22-10.
Available at: 2018. https://cran.r-project.org/package¼coxme; 2018.
19. Bates D, Maechler M, Bolder B, Walker S. Fitting linear mixed-effects models
using lme4. J Stat Softw 2015;67.
20. RStudio Team. RStudio: Integrated Development for R. Boston, MA: RStudio,
Inc. Available at:http://www.rstudio.com; 2015.
21. R Core Team. R: A language and environment for statistical computing. Vienna,
Austria: R Foundation for Statistical Computinghttps://www.R-project.org/;
2018.
22. Wiersma TJ, Walma EP, Thomas S, Assendelft WJJ. Summary of the practice
guideline “Hypertension” (third revision) from the Dutch College of General
Practitioners. [Dutch]yrSamenvatting van de standaard “Hypertensie” (derde
herziening) van het Nederlands Huisartsen Genootschap. Ned Tijdschr Gen-
eeskd 2004;148:923e931.
23. van Middelaar T, Ivens SD, van Peet PG, et al. Prescribing and deprescribing
antihypertensive medication in older people by Dutch general practitioners: A
qualitative study. BMJ Open 2018;8:e020871.
24. Höfler M. The effect of misclassification on the estimation of association: A
review. Int J Methods Psychiatr Res 2005;14:92e101.
25. Beishuizen CRL, Coley N, Moll van Charante EP, et al. Determinants of dropout
and nonadherence in a dementia prevention randomized controlled trial: The
7
prevention of dementia by intensive vascular care trial. J Am Geriatr Soc 2017;
65:1505e1513.
26. Benetos A, Bulpitt CJ, Petrovic M, et al. An expert opinion from the European
Society of Hypertension-European Union Geriatric Medicine Society Working
Group on the management of hypertension in very old, frail subjects. Hyper-
tension 2016;67:820e825.
27. Benetos A, Rossignol P, Cherubini A, et al. Polypharmacy in the aging patient:
Management of hypertension in octogenarians. JAMA 2015;314:170e180.
28. Odden MC, Beilby PR, Peralta CA. Blood pressure in older adults: The impor-
tance of frailty. Curr Hypertens Rep 2015;17:55.
29. Tzourio C, Levy C, Dufouil C, et al. Low cerebral blood flow velocity and risk of
white matter hyperintensities. Ann Neurol 2001;49:411e414.
30. Skoog I, Lithell H, Hansson L, et al. Effect of baseline cognitive function and
antihypertensive treatment on cognitive and cardiovascular outcomes: Study
on COgnition and Prognosis in the Elderly (SCOPE). Am J Hypertens 2005;18:
1052e1059.
31. Di Bari M, Pahor M, Franse LV, et al. Dementia and disability outcomes in large
hypertension trials: Lessons learned from the Systolic Hypertension in the
Elderly Program (SHEP) trial. Am J Epidemiol 2001;153:72e78.
32. Peters R, Beckett N, Forette F, et al. Incident dementia and blood pressure
lowering in the Hypertension in the Very Elderly Trial cognitive function
assessment (HYVET-COG): A double-blind, placebo controlled trial. Lancet
Neurol 2008;7:683e689.
33. Williamson JD, Supiano MA, Applegate WB, et al. Intensive vs standard blood
pressure control and cardiovascular disease outcomes in adults aged 75
years: A randomized clinical trial. JAMA 2016;315:2673e2682.
34. 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. J Am Coll Cardiol
2018;71:2199e2269.
35. Khera R, Lu Y, Lu J, et al. Impact of 2017 ACC/AHA guidelines on prevalence of
hypertension and eligibility for antihypertensive treatment in United States and
China: Nationally representative cross sectional study. BMJ 2018;362:k2357.
36. Lamprea-Montealegre JA, Zelnick LR, Hall YN, et al. Prevalence of hypertension
and cardiovascular risk according to blood pressure thresholds used for diag-
nosis. Hypertension 2018;72:602e609.
37. Qaseem A, Wilt TJ, Rich R, et al. Pharmacologic treatment of hypertension in
adults aged 60 years or older to higher versus lower blood pressure targets: A
clinical practice guideline from the american college of physicians and the
american academy of family physicians. Ann Intern Med 2017;166:430e437.
38. Nerenberg KA, Zarnke KB, Leung AA, et al. Hypertension Canada’s 2018
guidelines for diagnosis, risk assessment, prevention, and treatment of hy-
pertension in adults and children. Can J Cardiol 2018;34:506e525.
39. Williams B, Mancia G, Spiering W, et al. 2018 ESC/ESH guidelines for the
management of arterial hypertension. Eur Heart J 2018;39:3021e3104.
40. Phillips RA, Arnold RM, Peterson LE. Hypertension guidelines: The threads that
bind them. J Am Coll Cardiol 2018;72:1246e1251.
41. Cushman WC, Johnson KC. The 2017 U.S. hypertension guidelines: What is
important for older adults? J Am Geriatr Soc 2018;66:1062e1067.
42. Stessman J, Bursztyn M, Gershinsky Y, et al. Hypertension and its treatment at
age 90 years: Is there an association with 5-year mortality? J Am Med Dir Assoc
2017;18:277.e13e277.e19.
43. van Middelaar T, van Vught LA, van Gool WA, et al. Blood pressure-lowering
interventions to prevent dementia. J Hypertens 2018;36:1780e1787.
44. Press release, AAIC conference. Study shows intensive blood pressure control
reduces risk of mild cognitive impairment (MCI) and the combined risk of MCI
and dementia. Available at: 2018. https://www.alz.org/aaic/releases_2018/
AAIC18-Wed-developing-topics.asp. Accessed October 17, 2018.
45. Yuan JQ, Lv YB, Chen HS, et al. Association between late-life blood pressure and
the incidence of cognitive impairment: A community-based prospective cohort
study. J Am Med Dir Assoc; 2018. https://doi.org/10.1016/j.jamda.2018.05.029
[Epub ahead of print].
7.e1 J.W. van Dalen et al. / JAMDA xxx (2019) 1e7

Appendix

Supplementary Table 1
Causes of Death in Participants Discontinuing and Not Discontinuing Antihyper-
tensive Drug (AHD) Treatment

No AHD Stop AHD Stop P

Event/n % Event/n %

Vascular 54/201 26.9 1/14 7.1 .48

Possibly vascular 38/201 18.9 0/14 0.0 .62
Malignancy 21/201 10.4 3/14 21.4 .77
Other, nonvascular 59/201 29.4 5/14 35.7 .15
Missing 29/201 14.4 5/14 35.7 .57

Supplementary Table 2
Sensitivity Analysis Excluding Participants With AHD at Baseline Who Did Not Attain a Systolic BP 140 mmHg During the Course of the Study

Model 1 Model 2 Model 3

Event/Total HR (95% CI) P Event/Total HR (95% CI) P Event/Total HR (95% CI) P

Outcome: dementia
AHD stop 62/903 2.25 (1.07-4.73) .03 62/903 1.94 (0.90-4.18) .09 61/884 1.73 (0.79-3.83) .17
Concurrent medication
AHD stop without other medication 56/862 2.48 (0.6-10.28) .21 56/862 2.39 (0.57-9.99) .23 55/844 2.47 (0.58-10.57) .22
AHD stop with other medication 60/888 2.20 (0.95-5.13) .07 60/888 1.86 (0.77-4.48) .17 59/869 1.60 (0.64-3.96) .31
Memory complaints during study
AHD stop without complaints 22/569 4.25 (1.41-12.75) .01 22/569 3.61 (1.14-11.37) .03 22/558 2.54 (0.75-8.62) .14
AHD stop with complaints 40/333 1.33 (0.47-3.75) .59 40/333 1.18 (0.40-3.42) .77 39/328 1.25 (0.41-3.79) .69
Outcome: dementia/mortality
AHD stop 173/901 1.42 (0.83-2.41) .20 173/901 1.32 (0.77-2.27) .32 171/888 1.16 (0.66-2.02) .61
Concurrent medication
AHD stop without other medication 162/860 1.63 (0.60-4.42) .34 162/860 1.86 (0.68-5.07) .22 159/842 1.77 (0.65-4.85) .27
AHD stop with other medication 169/886 1.36 (0.74-2.51) .32 169/886 1.19 (0.64-2.23) .59 166/867 1.08 (0.57-2.04) .82
Memory complaints during study
AHD stop without complaints 93/568 1.47 (0.68-3.18) .33 93/568 1.46 (0.66-3.21) .35 91/557 1.22 (0.54-2.73) .63
AHD stop with complaints 80/332 1.27 (0.61-2.63) .53 80/332 1.08 (0.51-2.31) .83 79/325 1.21 (0.56-2.59) .63
Outcome: mortality
AHD stop 121/915 0.99 (0.48-2.03) .98 121/915 0.98 (0.47-2.04) .96 119/896 0.89 (0.42-1.86) .75
Concurrent medication
AHD stop without other medication 116/873 1.65 (0.52-5.21) .40 116/873 2.07 (0.65-6.60) .22 114/855 1.99 (0.62-6.38) .25
AHD stop with other medication 118/899 0.80 (0.33-1.97) .63 118/899 0.73 (0.29-1.82) .50 116/880 0.65 (0.26-1.64) .36
Memory complaints during study
AHD stop without complaints 75/579 0.97 (0.35-2.65) .95 75/579 1.05 (0.37-2.93) .93 73/568 0.96 (0.34-2.73) .94
AHD stop with complaints 46/335 1.03 (0.37-2.88) .95 46/335 0.88 (0.30-2.52) .81 46/328 0.89 (0.30-2.63) .84

AHD, antihypertensive drug; BP, blood pressure; HR, hazard ratio.

 J.W. van Dalen et al. / JAMDA xxx (2019) 1e7 7.e2

Supplementary Table 3
Cox Regression of Dementia and/or Mortality on Discontinuation of AHD Treatment in Participants With AHD at Baseline Who Attended Visit 2, Excluding Events <2 Years of
Follow-up or <2 Years After AHD Withdrawal

Model 1 Model 2 Model 3

Event/Total HR (95% CI) P Event/Total HR (95% CI) P Event/Total HR (95% CI) P

Outcome: dementia
AHD stop 83/1286 2.08 (1.00-4.33) .049 83/1286 1.86 (0.88-3.92) .10 82/1253 2.08 (1.00-4.33) .049
Concurrent medication
AHD stop without other medication 79/1247 3.30 (1.20-9.07) .02 79/1247 3.16 (1.15-8.72) .03 78/1216 3.30 (1.20-9.07) .02
AHD stop with other medication 79/1266 1.55 (0.57-4.25) .39 79/1266 1.33 (0.48-3.72) .59 78/1233 1.55 (0.57-4.25) .39
Memory complaints during study
AHD stop without complaints 28/839 3.88 (1.15-13.05) .03 28/839 3.09 (0.9-10.65) .07 27/819 3.88 (1.15-13.05) .03
AHD stop with complaints 55/446 1.20 (0.48-3.01) .70 55/446 1.08 (0.42-2.76) .88 55/436 1.20 (0.48-3.01) .70
Systolic BP on target during study
AHD stop systolic BP 140 mmHg 49/774 2.00 (0.79-5.06) .14 49/774 1.76 (0.68-4.53) .25 49/758 2.00 (0.79-5.06) .14
AHD stop systolic BP >140 mmHg 34/512 2.65 (0.80-8.81) .11 34/512 2.37 (0.68-8.20) .17 33/495 2.65 (0.80-8.81) .11
Outcome: dementia/mortality
AHD stop 240/1284 1.42 (0.86-2.36) .18 240/1284 1.47 (0.88-2.47) .14 236/1258 1.42 (0.86-2.36) .18
Concurrent medication
AHD stop without other medication 231/1245 2.06 (0.97-4.38) .06 231/1245 2.26 (1.06-4.81) .04 226/1214 2.06 (0.97-4.38) .06
AHD stop with other medication 233/1264 1.15 (0.59-2.25) .68 233/1264 1.15 (0.59-2.28) .68 228/1231 1.15 (0.59-2.25) .68
Memory complaints during study
AHD stop without complaints 127/838 1.53 (0.67-3.49) .31 127/838 1.63 (0.71-3.77) .25 123/818 1.53 (0.67-3.49) .31
AHD stop with complaints 113/445 1.13 (0.59-2.17) .71 113/445 1.15 (0.59-2.24) .67 112/433 1.13 (0.59-2.17) .71
Systolic BP on target during study
AHD stop systolic BP 140 mmHg 145/772 1.27 (0.67-2.42) .46 145/772 1.16 (0.60-2.24) .65 143/756 1.27 (0.67-2.42) .46
AHD stop systolic BP >140 mmHg 95/512 1.88 (0.82-4.35) .14 95/512 2.23 (0.95-5.25) .07 92/495 1.88 (0.82-4.35) .14
Outcome: mortality
AHD stop 180/1310 1.00 (0.51-1.95) .99 180/1310 1.08 (0.55-2.13) .83 175/1276 1.00 (0.51-1.95) .99
Concurrent medication
AHD stop without other medication 175/1269 1.47 (0.54-3.95) .45 175/1269 1.72 (0.64-4.66) .29 170/1237 1.47 (0.54-3.95) .45
AHD stop with other medication 176/1289 0.79 (0.32-1.92) .60 176/1289 0.82 (0.33-2.02) .66 171/1255 0.79 (0.32-1.92) .60
Memory complaints during study
AHD stop without complaints 109/860 1.10 (0.41-2.99) .85 109/860 1.30 (0.47-3.58) .61 105/839 1.10 (0.41-2.99) .85
AHD stop with complaints 71/449 0.86 (0.35-2.13) .74 71/449 0.93 (0.37-2.36) .88 70/437 0.86 (0.35-2.13) .74
Systolic BP on target during study
AHD stop systolic BP 140 mmHg 106/785 0.99 (0.43-2.25) .98 106/785 0.93 (0.40-2.16) .87 104/769 0.99 (0.43-2.25) .98
AHD stop systolic BP >140 mmHg 74/525 1.07 (0.33-3.45) .91 74/525 1.38 (0.42-4.51) .60 71/507 1.07 (0.33-3.45) .91

AHD, antihypertensive drug; BP, blood pressure; HR, hazard ratio.

Model 1: adjusted for age; model 2: additionally adjusted for sex, baseline mean systolic blood pressure, total number of prescribed medications, and total number of
prescribed antihypertensive drugs; model 3: additionally adjusted for baseline history of stroke, cardiovascular disease, diabetes, current smoking, subjective memory
complaints, and Mini-Mental State Examination score.
7.e3 J.W. van Dalen et al. / JAMDA xxx (2019) 1e7

Supplementary Table 4
Cox Regression of Dementia and/or Mortality on Discontinuation of AHD Treatment in Participants With AHD at Baseline Adjusting for the preDIVA Intervention

Event/Total HR (95% CI) P Event/Total HR (95% CI) P Event/Total HR (95% CI) P

Outcome: dementia
AHD stop 100/1507 2.15 (1.15-4.02) .02 100/1507 1.90 (1.00-3.62) .05 98/1467 1.75 (0.90-3.37) .10
Concurrent medication
AHD stop without other medication 93/1453 2.70 (0.99-7.40) .05 93/1453 2.61 (0.95-7.15) .06 91/1415 2.71 (0.97-7.57) .06
AHD stop with other medication 96/1479 1.96 (0.90-4.24) .09 96/1479 1.68 (0.76-3.73) .20 94/1439 1.48 (0.66-3.33) .35
Memory complaints during study
AHD stop without complaints 34/985 3.15 (1.10-9.02) .03 34/985 2.44 (0.83-7.20) .11 33/961 2.21 (0.73-6.72) .16
AHD stop with complaints 66/521 1.50 (0.68-3.29) .31 66/521 1.39 (0.62-3.12) .43 65/509 1.57 (0.68-3.59) .29
Systolic BP on target during study
AHD stop systolic BP 140 mmHg 62/903 2.23 (1.06-4.68) .04 62/903 1.91 (0.89-4.11) .10 61/884 1.63 (0.73-3.63) .23
AHD stop systolic BP >140 mmHg 38/604 2.23 (0.68-7.32) .19 38/604 2.08 (0.62-7.00) .24 37/583 1.93 (0.56-6.65) .30
Outcome: dementia/mortality
AHD stop 292/1505 1.58 (1.04-2.39) .03 292/1505 1.63 (1.06-2.50) .03 287/1475 1.48 (0.96-2.28) .08
Concurrent medication
AHD stop without other medication 278/1451 2.24 (1.19-4.22) .01 278/1451 2.42 (1.28-4.56) .01 272/1413 2.35 (1.23-4.46) .01
AHD stop with other medication 282/1477 1.31 (0.76-2.24) .33 282/1477 1.31 (0.75-2.27) .34 276/1437 1.20 (0.69-2.10) .52
Memory complaints during study
AHD stop without complaints 157/984 1.37 (0.70-2.68) .36 157/984 1.42 (0.71-2.82) .32 153/960 1.33 (0.66-2.67) .42
AHD stop with complaints 135/520 1.55 (0.90-2.65) .11 135/520 1.61 (0.92-2.80) .10 133/505 1.59 (0.90-2.81) .11
Systolic BP on target during study
AHD stop systolic BP 140 mmHg 173/901 1.41 (0.83-2.39) .21 173/901 1.30 (0.76-2.24) .34 170/882 1.17 (0.67-2.04) .58
AHD stop systolic BP >140 mmHg 119/604 2.04 (1.03-4.04) .04 119/604 2.32 (1.16-4.66) .02 116/583 2.16 (1.06-4.41) .03
Outcome: mortality
AHD stop 215/1532 1.16 (0.68-2.00) .59 215/1532 1.28 (0.74-2.23) .38 210/1491 1.21 (0.69-2.11) .51
Concurrent medication
AHD stop without other medication 208/1476 2.00 (0.94-4.26) .07 208/1476 2.33 (1.09-4.97) .03 203/1437 2.30 (1.07-4.95) .03
AHD stop with other medication 208/1503 0.82 (0.38-1.73) .60 208/1503 0.86 (0.40-1.86) .70 203/1462 0.79 (0.37-1.72) .56
Memory complaints during study
AHD stop without complaints 133/1007 1.02 (0.45-2.32) .96 133/1007 1.18 (0.51-2.73) .69 129/982 1.24 (0.53-2.87) .62
AHD stop with complaints 82/524 1.25 (0.60-2.60) .55 82/524 1.39 (0.65-2.94) .40 81/509 1.21 (0.56-2.60) .63
Systolic BP on target during study
AHD stop systolic BP 140 mmHg 121/915 0.99 (0.48-2.02) .97 121/915 0.97 (0.47-2.03) .94 119/896 0.88 (0.42-1.85) .73
AHD stop systolic BP >140 mmHg 94/617 1.56 (0.68-3.59) .30 94/617 1.90 (0.81-4.45) .14 91/595 1.86 (0.78-4.43) .16

AHD, antihypertensive drug; BP, blood pressure; HR, hazard ratio.

Model 1: adjusted for age; model 2: additionally adjusted for sex, baseline mean systolic blood pressure, total number of prescribed medications, and total number of
prescribed antihypertensive drugs; model 3: additionally adjusted for baseline history of stroke, cardiovascular disease, diabetes, current smoking, subjective memory
complaints, and Mini-Mental State Examination score.

Supplementary Figure 1. Time until death/dementia from the time of antihyperten-

sive drug (AHD) withdrawal. Y-axis: number of cases, x-axis: time in years.