1. PS Sever, PM Rothwell, SC Howard, JE Dobson, B Dahlöf,
The Anglo-Scandinavian Cardiac Outcomes Trial – Blood Pressure Lowering Arm (ASCOT-BPLA) Blood Pressure Variability and Cardiovascular Outcomes
PS Sever, PM Rothwell, SC Howard, JE Dobson, B Dahlöf,
H Wedel, NR Poulter, for the ASCOT Investigators
International Centre for Circulatory Health, Imperial College London
and
Stroke Prevention Research Unit, University of Oxford

2. A randomised controlled trial of the prevention of CHD and other vascular events by BP and cholesterol lowering in a factorial study design

3. Study design 19,257 hypertensive patients ASCOT-BPLA
Stopped after 5.5 yrs
atenolol ± bendroflumethiazide
PROBE
design
amlodipine ± perindopril
Investigator-led, multinational randomised controlled trial
conducted in hypertensive patients, yrs, with no prior history
of CHD, but with 3 additional cardiovascular risk factors (male sex,
> 55 yrs, smoking etc )

4. Treatment algorithm to BP targets < 140/90 mmHg
or < 130/80 mmHg in patients with diabetes
amlodipine 5-10 mg
atenolol mg
add
add
bendroflumethiazide-K mg
perindopril 4-8 mg
add
doxazosin GITS 4-8 mg
add
additional drugs, eg, moxonidine/spironolactone
Median follow up was for 5.5 years 3

5. Baseline characteristics
Amlodipine ± perindopril n = 9639
Atenolol ± thiazide n = 9618
Demographic and clinical characteristics
Male
7381 (76.6%)
7361 (76.5%)
White
9187 (95.3%)
9170 (95.3%)
Current smoker
3168 (32.9%)
3110 (32.3%)
Age (years)
63.0 (8.5)
SBP (mmHg)
164.1 (18.1)
163.9 (18.0)
DBP (mmHg)
94.8 (10.4)
94.5 (10.4)
Heart rate (bpm)
71.9 (12.7)
71.8 (12.6)
BMI (kg/m2)
28.7 (4.6)
28.7 (4.5)
Diabetes
2567 (27%)
2578 (27%)
Other vascular disease
2169 (23%)
2162 (22%)
Total cholesterol (mmol/L)
5.9 (1.1)
Drug therapy
Previous antihypertensive treatments
1841 (19.1%)
1825 (19.0%) 1
4280 (44.4%)
4283 (44.5%) ≥2
3518 (36.5%)
3510 (36.5%)
Lipid-lowering therapy
1046 (10.9%)
1004 (10.4%)
Aspirin
1851 (19.2%)
1837 (19.1%)
Values are number of patients (%), or mean (SD)

6. Systolic and diastolic blood pressure
amlodipine  perindopril
atenolol  bendroflumethiazide
180
164.1
SBP
160
163.9
Mean difference 2.7
137.7
140
136.1
Blood pressure (mmHg)
120
DBP
100
94.8
Mean difference 1.9
94.5
79.2 80
77.4 60
Baseline
0.5 1
1.5 2
2.5 3
3.5 4
4.5 5
5.5
Last visit
Follow-up (years)

7. ASCOT-BPLA: summary of all endpoints
Unadjusted hazard ratio (95% CI)
0.90 ( )
0.87 ( )
0.87 ( )
0.84 ( )
0.89 ( )
0.76 ( )
0.77 ( )
0.84 ( )
1.27 ( )
0.68 ( )
0.98 ( )
0.65 ( )
1.07 ( )
0.70 ( )
0.85 ( )
0.86 ( )
0.84 ( )
Primary Non-fatal MI (incl. silent) + fatal CHD
Secondary Non-fatal MI (excl. silent) + fatal CHD
Total coronary endpoint Total CV events and procedures All-cause mortality Cardiovascular mortality Fatal and non-fatal stroke Fatal and non-fatal heart failure
Tertiary Silent MI
Unstable angina Chronic stable angina Peripheral arterial disease Life-threatening arrhythmias New-onset diabetes mellitus New-onset renal impairment
Post hoc
Primary endpoint + coronary revasc procs
CV death + MI + stroke
0.50
0.70
1.00
1.45
2.00
Amlodipine  perindopril better
Atenolol  thiazide better
The area of the blue square is proportional to the amount of statistical information

8. Conclusions
Amlodipine  perindopril-based therapy conferred an advantage over atenolol  thiazide-based therapy on all major CV endpoints, all-cause mortality and new-onset diabetes
Additional statistical analyses demonstrated that adjusting for blood pressure differences between treatment groups early on in the trial, did not account for the observed differences in cardiovascular outcomes

9. ASCOT-Blood pressure variability: methods (based on over 1 million BP readings)
Of 19,257 patients, 18,530 had ≥ 2 follow-up visits (median = 10) from 6 months onwards until the end of the trial
3 blood pressure measurements were recorded at each visit, using standardised techniques, at 6 monthly intervals for a median follow up of 5.5 years
From 6 months onwards there were 350 strokes and 704 coronary events (non-fatal MI, fatal CHD, new onset angina, non-fatal and fatal heart failure) in the atenolol-based group and 279 and 611 respectively in the amlodipine-based group

10. Blood pressure variability: methods
Visit-to-visit variability of SBP and DBP during follow-up, from 6 months after randomisation to the end of the trial, were expressed as the standard deviation (SD), coefficient of variation (CV), and a transformation of SD uncorrelated with mean BP (variability independent of mean – VIM)
Within-visit variability was expressed as the SD of the three measurements taken at each visit averaged across all follow-up visits
Among 1905 patients, mean BP and variability were also determined with annual 24 hour ambulatory monitoring (ABPM)
Cox models were used to determine associations with risks of vascular events during follow-up, and whether an effect on variability in BP could account for the reduction in events in the amlodipine group

11. Means and measures of variability of clinic SBP by treatment group
Parameters calculated using all measurements from 6 months onwards
Amlodipine-based regimen n = 9302
Atenolol-based regimen n = 9228
Difference (95% CI)
Parameter
Mean (SD)
Mean SBP
139.1 (11.1)
141.8 (13.0)
2.68 (2.58–2.78)
Maximum SBP
157.4 (16.1)
164.2 (18.9)
6.80 (6.68–6.92)
Any SBP ≥180 mmHg
9.1% (851)
19.2% (1776)
10.1% (9.1–11.1)
Any SBP ≥200 mmHg
1.8% (164)
4.7% (438)
3.0% (2.5–3.5)
Visit-to-visit variability
SD SBP
10.99 (4.79)
13.42 (5.77)
2.43 (2.36–2.50)
CV SBP
7.87 (3.23)
9.41 (3.78)
1.54 (1.49–1.59)
VIM SBP
11.14 (4.52)
13.13 (5.21)
1.99 (1.93–2.05)
Within-visit variability
5.42 (0.02)
5.91 (0.02)
0.49 (0.44–0.54)
SD, standard deviation; CV, coefficient of variation; VIM, variability independent of mean

12. Means and measures of variability of clinic DBP by treatment group
Parameters calculated using all measurements from 6 months onwards
Amlodipine-based regimen n = 9302
Atenolol-based regimen n = 9228
Difference (95% CI)
Parameter
Mean (SD)
Mean DBP
80.2 (7.4)
82.1 (7.6)
1.98 (1.90–2.06)
Maximum DBP
90.4 (9.0)
93.5 (9.6)
3.10 (3.00–3.20)
Any DBP ≥100 mmHg
14.3% (1326)
24.5% (2257)
10.2% (9.1–11.3)
Any DBP ≥105 mmHg
6.1% 568)
11.6% (1071)
5.5% (4.7–6.3)
Visit-to-visit variability
SD DBP
6.26 (2.42)
6.98 (2.72)
0.72 (0.67–0.77)
CV DBP
7.86 (3.04)
8.54 (3.30)
0.68 (0.63–0.73)
VIM DBP
6.30 (2.41)
6.95 (2.66)
0.65 (0.60–0.70)
SD, standard deviation; CV, coefficient of variation; VIM, variability independent of mean

13. Visit-to-visit mean systolic blood pressure expressed in deciles, hazard ratios (95% CI) and number of stroke and coronary events in each decile
Mean SBP
Stroke risk
Coronary risk

14. Stroke and coronary risk expressed by decile of measure of visit-to-visit SBP variability
Stroke Risk
Coronary Risk
Standard deviation of SBP
Atenolol
Amlodipine
Coefficient of variation of SBP
Variation independent of mean SBP
Decile of measure
Decile of measure

15. Group distribution (SD and CV) of measures of SBP at baseline and at each follow-up visit in the two treatment groups

16. Number of patients in each decile of within-visit SD SBP
Stroke risk and coronary risk expressed by decile of within-visit SBP variability
Number of patients in each
decile of within-visit SD SBP
Stroke risk
(HR, 95% CI)
Coronary risk
(HR, 95% CI)

17. Average within-visit CV in the two treatment groups during follow-up

18. Systolic blood pressure
Hazard ratios (95% CI) for the effect of treatment (amlodipine versus atenolol) on risk of stroke
Parameters calculated using all BP measurements from 6 months onwards. Mean, SD, CV, and VIM are entered into the model as deciles
Stroke
Systolic blood pressure
Variables in model
HR (95% CI)
p value
Treatment (Rx)
0.78 (0.67–0.90)
0.001
Usual BP
Rx + mean
0.84 (0.72–0.98)
0.025
Visit-to-visit BP variability
Rx + mean + SD
0.96 (0.82–1.12)
0.59
Rx + mean + CV
0.95 (0.82–1.11)
0.55
Rx + mean + VIM
0.58
Within-visit and visit-to-visit BP variability
Rx + within-visit SD
0.024
Rx + mean + VIM + WVSD
0.99 (0.85–1.16)
0.89
SD, standard deviation; CV, coefficient of variation; VIM, variability independent of mean; WVSD, within-visit standard deviation

19. Systolic blood pressure
Hazard ratios (95% CI) for the effect of treatment (amlodipine versus atenolol) on risk of coronary events
Parameters calculated using all BP measurements from 6 months onwards. Mean, SD, CV, and VIM are entered into the model as deciles
Coronary Events
Systolic blood pressure
Variables in model
HR (95% CI)
p value
Treatment (Rx)
0.85 (0.77–0.94)
0.002
Usual BP
Rx + mean
0.88 (0.80–0.98)
0.019
Visit-to-visit BP variability
Rx + mean + SD
1.00 (0.90–1.11)
0.98
Rx + mean + CV
0.99
Rx + mean + VIM
1.00 (0.90–1.10)
Within-visit and visit-to-visit BP variability
Rx + within-visit SD
0.88 (0.79–0.97)
0.013
Rx + mean + VIM + WVSD
1.01 (0.91–1.12)
0.88
SD, standard deviation; CV, coefficient of variation; VIM, variability independent of mean; WVSD, within-visit standard deviation

20. Ambulatory blood pressure monitoring
1905 patients had an average of 3.25 recordings from 6 months onwards
Daytime SBP slightly higher but night-time SBP slightly lower on amlodipine-based treatment
Morning surge similar on both treatments and only weakly correlated with BP visit-to-visit variability
Intra ABPM coefficient of variation of SBP correlated with visit-to-visit variability in clinic SBP
atenolol group, r = 0.38
amlodipine group, r = 0.29, p < for both groups
Intra ABPM variability in daytime SBP predicted both stroke and coronary events (but less so than visit-to-visit variability)

21. Summary
Mean BP in trial has minimal effect on stroke outcome and no effect on CHD outcome
Various measures of visit-to-visit BP variability (SD, coefficient of variation and variation independent of mean BP) are powerful predictors of both stroke and CHD outcomes
Other measures of variability (within-visit variability and variability assessed by ABPM) also predict cardiovascular outcomes but less than visit-to-visit variability
Amlodipine reduces variability compared with atenolol
Variability increased with age, diabetes, smoking, and in those with established vascular disease
Adjusting for BP variability completely explains differences in stroke and CHD outcomes between amlodipine-based and atenolol-based treatment in ASCOT