Long‐Term Follow‐Up of Moderately Hypercholesterolemic Hypertensive Patients Following Randomization to Pravastatin vs Usual Care: The Antihypertensive and Lipid‐Lowering Treatment to Prevent Heart Attack Trial (ALLHAT‐LLT)

Abstract

The authors conducted a randomized, controlled, multicenter trial, in which they assigned well‐controlled hypertensive participants aged 55 years and older with moderate hypercholesterolemia to receive pravastatin (n=5170) or usual care (n=5185) for 4 to 8 years, when trial therapy was discontinued. Passive surveillance using national databases to ascertain deaths and hospitalizations continued for a total follow‐up of 8 to 13 years to assess whether mortality and morbidity differences persisted or new differences developed. During the post‐trial period, fatal and nonfatal outcomes were available for 98% and 64% of participants, respectively. The primary outcome was all‐cause mortality and the secondary outcomes included cardiovascular mortality, coronary heart disease (CHD), stroke, heart failure, cardiovascular disease, and end‐stage renal disease. No significant differences appeared in mortality for pravastatin vs usual care (hazard ratio [HR], 0.96; 95% confidence interval [CI], 0.89–1.03) or other secondary outcomes. Similar to the previously reported in‐trial result, there was a significant treatment effect for CHD in black patients (HR, 0.79; 95% CI, 0.64–0.98). However, the in‐trial result showing a significant treatment by race effect did not remain significant during the entire follow‐up (P=.08). These findings are consistent with evidence from other large trials that show statins prevent CHD and add evidence that they are effective for CHD prevention in black patients.

The Lipid‐Lowering Trial component of the Antihypertensive and Lipid‐Lowering Treatment to Prevent Heart Attack Trial (ALLHAT‐LLT) was a multicenter randomized trial comparing usual care (UC) vs treatment with an HMG CoA‐reductase inhibitor (pravastatin) in a moderately hypercholesterolemic, treated hypertensive cohort with other coronary heart disease (CHD) risk factors, including preexisting cardiovascular disease (CVD).1 After an average of 4.8 years of follow‐up, there was no difference in the primary endpoint of all‐cause mortality (hazard ratio [HR], 0.99; 95% confidence interval [CI], 0.89–1.11; P=.88) and a nonsignificant lowering of secondary CHD endpoints of nonfatal myocardial infarction (MI) and CHD death (HR, 0.91; 95% CI, 0.79–1.04; P=.16).2 Fatal and nonfatal stroke was also nonsignificantly reduced in the pravastatin group (HR, 0.91; 95% CI, 0.75–1.09; P=.31). These results, which contrasted with those of most statin trials,3 were attributed to the diminished power resulting from treatment crossovers and a consequently smaller‐than‐expected differential reduction in cholesterol levels.

There was no heterogeneity of ALLHAT‐LLT primary and secondary outcomes by age, sex, previous history of CHD, history of diabetes, or baseline level of cholesterol.2 However, significant differences were observed between black and non‐black patients for secondary outcomes of CHD and stroke. For stroke, pravastatin showed no benefit in blacks but was protective in non‐blacks (HR, 1.12; 95% CI, 0.86–1.48; vs 0.74; 95% CI, 0.57–0.96; P for interaction=.03),4 whereas for CHD events, pravastatin was protective in black but not in non‐black patients (HR, 0.71; 95% CI, 0.57–0.90; P=.005 vs 1.00; 95% CI, 0.85–1.19; P=.95, P for interaction=.02).4, 5 For CHD events, in‐depth analyses have shown that these differences were not accounted for by baseline differences in risk factors, adherence during trial, or achieved blood pressure (BP) and lipid lowering.5 Because black patients have been under‐represented in statin trials and ALLHAT‐LLT remains one of the largest statin trials with a substantial proportion of black participants (38%), further exploration of these unexplained findings is of great scientific interest.

The purpose of this paper is to report 10‐year mortality and morbidity of ALLHAT‐LLT participants using in‐trial data plus post‐trial data from administrative databases (National Death Index, Social Security Administration, Center for Medicare and Medicaid Services, and US Renal Data System)4, 6 to assess long‐term effects of treatment with pravastatin vs UC. We will focus in particular on whether black vs non‐black racial differences in CHD and stroke observed during the trial persisted or attenuated. Further, we will examine whether differences emerge for outcomes that were not statistically different at the trial's end.

Methods

ALLHAT‐LLT was a randomized and nonblinded large simple trial with 10,355 participants randomly assigned to either open‐label pravastatin (40 mg/d) or UC.1, 2, 7Participants were drawn exclusively from ALLHAT, a 4‐armed antihypertensive treatment trial that compared amlodipine, lisinopril, and doxazosin each with chlorthalidone. Major eligibility criteria included fasting low‐density lipoprotein cholesterol (LDL‐C) of 120 to 189 mg/dL for those with no known CHD or 100 to129 mg/dL for those with known CHD. Total cholesterol (TC) was measured at baseline and at the year 2, 4, and 6 visits. A fasting lipid profile was obtained at baseline in all participants and in a randomly selected 5% sample of UC participants and a 10% sample of pravastatin participants. Primary outcome for LLT was all‐cause mortality. Prespecified secondary endpoints included a composite of fatal CHD or nonfatal MI, cause‐specific mortality, and total and site‐specific cancers. Other outcomes of interest included stroke, heart failure (HF), and end‐stage renal disease (ESRD).8

Extended Follow‐Up Outcomes: Definitions and Determination

For the post‐trial period, data are not available on treatments, lipid levels, BP levels, or other laboratory values.

Mortality‐Only Endpoints

Mortality data were available for all non‐Canadian participants in in‐trial and post‐trial periods (N=10,199) (Figure 1a). In‐trial cause of death was determined by the investigator and confirmed by death certificates. If cause of death was unknown, the National Death Index (NDI) Plus database was used to provide cause‐specific mortality. Post‐trial all‐cause mortality data were obtained from the NDI and Social Security Administration (SSA), using Social Security number, name, sex, and date of birth as matching criteria, and cause of death was ascertained from the NDIPlus database.

Figure 1.

(a) CONSORT diagram for mortality. (b) CONSORT diagram for fatal and nonfatal cardiovascular disease. VA indicates Veterans Administration; CVD, cardiovascular disease.

Consistent with the original trial, total mortality was the primary outcome for the LLT extension study. Cardiovascular (CV) mortality (death caused by coronary heart disease, stroke, HF, or other CV disease) was considered the most important secondary outcome of extended follow‐up. CHD death and other cause‐specific mortality outcomes were prespecified outcomes.8

A death identified through the NDI or SSA was verified at the Coordinating Center after receipt of a death certificate from the state, Commonwealth, or other jurisdictions. Death certificates were used for verification of decedent identity only. Causes of death (International Classification of Diseases—10th Revision [ICD‐10] coding) were obtained from NDIPlus and collapsed into 11 categories used in these analyses. NDIPlus (fact of death plus cause) were initally provided under International Classification of Diseases—9th Revision (ICD‐9); for deaths occurring in 1999 forward this was changed to ICD‐10. To continue with the original conversion scheme of ICD‐9 to ALLHAT categories, the World Health Organization (WHO)'s Two‐way Translator for the Ninth and Tenth Revisions (1997) was used to convert ICD‐10 codes to ICD‐9.9

Combined Fatal and Nonfatal Endpoints

Hospitalization data were available for both in‐trial and post‐trial periods for 6638 participants, due to exclusion of Canadian, VA, and non‐Medicare participants (Figure 1b). During the in‐trial period, events were ascertained and classified by investigators and confirmed by the Coordinating Center Endpoints Department based on discharge summaries. For quality‐control purposes, a sample of CHD and stroke events was centrally reviewed by a Steering Committee subcommittee. Nonfatal events were also ascertained from Centers for Medicare and Medicaid Services (CMS) and the US Renal Data System (USRDS). During the post‐trial period, nonfatal events were ascertained from the CMS and USRDS. However, during post‐trial, these outcome data, except for ESRD (dialysis or transplant), were only available for participants who were eligible for and enrolled in Medicare (thus excluding Canadian participants). In order to obtain complete event data for the entire post‐trial period, we restricted these analyses to participants who were eligible for Medicare and had valid Medicare or Social Security numbers at the end of the ALLHAT main trial period (80% of all randomized participants). Because of lack of access to post‐trial information from the Veterans Administration (VA), participants randomized at VA sites could not be included in analyses (lowering to 64% of all randomized participants). Fatal/nonfatal outcomes were prespecified as secondary endpoints: CV disease (CVD mortality; or hospitalized nonfatal MI, stroke, or HF), coronary heart disease (CHD mortality or hospitalized nonfatal MI), stroke (fatal or nonfatal hospitalized), HF (fatal or nonfatal hospitalized), and ESRD.

Nonfatal events were identified through CMS data and classified using ICD‐9 codes. Renal endpoints were determined from USRDS for both in‐trial and post‐trial periods, and as such exclude only Canadian participants. The USRDS had not been previously used to determine in‐trial renal endpoints.

Statistical Analysis

To compare baseline characteristics of participants assigned to UC and pravastatin, contingency table and z tests were used. Intermediate outcomes of drug adherence and laboratory measures at baseline, year 2, year 4, year 6, and end of trial were examined. Evaluations of the effect of assigned treatment on risk for primary and secondary outcomes were performed using Cox regression. Prespecified tests for interactions were conducted to determine whether effects of treatment intervention differ between black/non‐black subgroups. Time‐dependent Cox regression was used to estimate hazard ratios (HRs) associated with treatment assignment separately for the trial and post‐trial follow‐up periods. Given the many multivariate, subgroup, and interaction analyses performed, statistical significance at the .05 level should be interpreted with caution.

For purposes of power calculations, estimated 10‐year event rates for mortality in the pravastatin group were calculated using a Weibull survival model of observed results in the original study projected to 10 years. Statistical power for this analysis was obtained using these rates and sample sizes within two treatment groups of ALLHAT. For primary outcome, we estimated a 90% power with an α=.05 to detect a reduction in risk of 9.3% (HR=0.91) for the UC group compared with the pravastatin group (10‐year mortality rate of 32%).4

In addition to analyses specified in the extension study protocol, we conducted additional sensitivity analyses to examine events captured entirely from national databases during both in‐trial and post‐trial periods (see online supplement description and tables). The purpose of these analyses was to ensure that means of ascertainment of outcomes (in‐trial methods or national databases) would provide consistent treatment difference results.

Results

Follow‐Up of Cohort

Figure 1 shows numbers of participants randomized and followed as part of the extension study. Of the original 10,199 non‐Canadian ALLHAT‐LLT participants, 8860 survived until the study end and were followed for mortality and ESRD through the end of 2006. A subset of the 8860 (n=5683) participants who were not enrolled at VA sites and were enrolled in Medicare at the end of the main trial were followed for fatal/nonfatal CVD, CHD, stroke, HF, and gastrointestinal bleeding through the end of 2006. Mean length of follow‐up including post‐trial period was 8.8 years, and 99% of expected person‐years were observed. Maximum follow‐up was 12.7 years.

Patient Characteristics

Table 1 presents characteristics of patients at baseline and conclusion of the trial who were included in analyses of extension data. Mean baseline age was 67 years; 49% were women; 38% self‐identified as black and 19% self‐identified as Hispanic; and 14% had a history of CHD. Participants who survived to enter the post‐trial period were slightly younger (mean age, 66 years), slightly more likely to be women (50%), and were less likely to have CV risk factors or comorbid conditions (such as smoking, diabetes, or a history of CHD). Participants who were eligible for follow‐up for combined fatal/nonfatal outcomes were older (mean age, 68 years), more likely to be women (58%), and less likely to smoke. There were generally similar distributions of baseline factors in pravastatin and UC treatment groups for those included in post‐trial analyses.

Table 1.

Baseline Characteristics of the ALLHAT Lipid Component Participants (Excluding Canadian Participants)

Characteristic	Baseline	Surviving to Conclusion of Trial
		VA Data Included for Mortality Cohort a			VA Data Excluded for Mortality/Morbidity Cohort b
		Pravastatin	Usual Care	P Value	Pravastatin	Usual Care	P Value
Number	10,199	4428	4432		2832	2851
Age, mean (SD), y	66.6 (7.61)	66.0 (7.36)	66.0 (7.27)	.94	68.4 (6.69)	68.4 (6.62)	.93
55–64, No. (%)	4473 (43.9)	2074 (46.8)	2086 (47.1)	.046	930 (32.8)	930 (32.6)	.86
65+, No. (%)	5726 (56.1)	2354 (53.2)	2346 (52.9)		1902 (67.2)	1921 (67.4)
Race, No. (%)
Black	3908 (38.2)	1702 (38.4)	1635 (36.9)		1076 (38.0)	1032 (36.2)
Non‐black	6291 (61.7)	2726 (61.6)	2797 (63.1)		1756 (62.0)	1819 (63.8)
White, non‐Hispanic	4101 (40.2)	1760 (39.7)	1761 (39.7)	.40	1100 (38.8)	1098 (38.5)	.37
Black, non‐Hispanic	3517 (34.5)	1525 (34.4)	1471 (33.2)		960 (33.9)	925 (32.4)
White Hispanic	1562 (15.3)	697 (15.7)	739 (16.7)		472 (16.7)	527 (18.5)
Black Hispanic	391 (3.8)	177 (4.0)	164 (3.7)		116 (4.1)	107 (3.7)
Other	628 (6.2)	269 (6.1)	297 (6.7)		184 (6.5)	194 (6.8)
Women	4983 (48.9)	2215 (50.0)	2228 (50.3)	.82	1648 (58.2)	1658 (58.2)	.98
Education, mean (SD), y	10.7 (4.09)	10.8 (4.1)	10.8 (4.1)	.70	10.4 (4.3)	10.5 (4.3)	.51
Antihypertensive treatment, No. (%)
Treated	9172 (89.9)	3977 (89.8)	3973 (89.6)	.79	2543 (89.8)	2566 (90.0)	.79
Untreated	1027 (10.1)	451 (10.2)	459 (10.4)		289 (10.2)	285 (10.0)
Blood pressure, mean (SD), mm HG
SBP	146.3 (15.6)	146.3 (15.6)	145.7 (15.4)	.10	147.0 (15.4)	146.6 (15.2)	.31
DBP	84.7 (10.0)	84.9 (9.9)	84.9 (9.9)	.87	84.4 (9.9)	84.4 (9.8)	.94
Treated at baseline
SBP	145.2 (15.5)	145.1 (15.5)	144.5 (15.2)	.07	145.9 (15.3)	145.5 (15.1)	.33
DBP	84.1 (9.9)	84.4 (9.9)	84.3 (9.8)	.59	84.0 (9.8)	83.9 (9.8)	.72
Untreated at baseline
SBP	156.4 (12.5)	156.1 (12.6)	156.0 (12.6)	.93	156.6 (12.5)	156.5 (12.4)	.89
DBP	89.8 (9.4)	89.5 (9.5)	90.1 (9.1)	.33	88.6 (9.2)	89.3 (9.0)	.31
Cholesterol, mean (SD), mg/dL
Total	223.7 (26.8)	223.9 (26.6)	224.0 (26.5)	.86	224.9 (26.5)	225.3 (26.6)	.54
HDL‐C	47.5 (13.5)	47.7 (13.2)	47.5 (13.4)	.35	48.8 (13.5)	48.6 (13.8)	.60
LDL‐C	145.5 (21.3)	145.6 (21.2)	145.7 (21.3)	.84	145.7 (21.2)	145.9 (21.4)	.64
Fasting triglycerides, mean (SD)	151.4 (77.5)	151.0 (73.4)	152.4 (72.2)	.42	150.2 (72.7)	152.2 (72.3)	.34
Glucose, mean (SD), mg/dL
Total	123.3 (59.3)	122.8 (58.5)	121.6 (57.0)	.31	122.0 (57.3)	120.1 (56.0)	.21
Fasting	122.2 (56.0)	121.5 (54.5)	121.0 (54.3)	.71	121.3 (54.4)	119.6 (52.9)	.30
Eligibility risk factors, c No. (%)
Cigarette smoker	2351 (23.0)	995 (22.5)	993 (22.4)	.94	530 (18.7)	523 (18.3)	.72
Atherosclerotic cardiovascular disease d	4558 (44.7)	1912 (43.2)	1924 (43.4)	.83	1327 (46.9)	1328 (46.6)	.83
History of MI or stroke	1755 (17.2)	695 (15.7)	700 (15.8)	.90	477 (16.8)	454 (15.9)	.35
History of coronary revascularization	711 (7.0)	279 (6.3)	305 (6.9)	.27	189 (6.7)	215 (7.5)	.20
Other ASCVD	2088 (20.5)	875 (19.8)	883 (19.9)	.85	651 (23.0)	660 (23.1)	.88
ST‐T wave	1163 (11.5)	519 (11.8)	487 (11.2)	.29	336 (12.0)	320 (11.3)	.44
HDL‐C <35 mg/dL	1085 (10.6)	476 (10.7)	486 (11.0)	.74	294 (10.4)	283 (9.9)	.57
LVH by ECG or echocardiography	2379 (23.3)	1018 (23.0)	1037 (23.4)	.65	655 (23.1)	687 (24.1)	.39
No diabetes, No. (%)	5553 (56.2)	2454 (57.0)	2448 (56.9)	.12	1572 (57.1)	1569 (57.0)	.45
Impaired fasting glucose, No. (%)	445 (4.5)	179 (4.2)	218 (5.1)		114 (4.1)	133 (4.8)
Diabetes, No. (%)	3890 (39.3)	1669 (38.8)	1634 (38.0)		1066 (38.7)	1050 (38.1)
History CHD at baseline, No. (%)	1446 (14.2)	566 (12.8)	633 (14.3)	.04	380 (13.4)	430 (15.1)	.07
Body mass index, mean (SD), kg/m2	29.9 (6.1)	29.9 (5.8)	30.1 (6.2)	.22	29.8 (5.8)	29.9 (6.2)	.39
Atrial fibrillation/flutter, No. (%)	76 (0.84)	25 (0.64)	24 (0.61)	.84	20 (0.81)	22 (0.9)	.80
Estimated GFR (mL/min/1.73 m2), mean (SD)	78.6 (19.0)	79.4 (18.5)	79.1 (18.5)	.56	77.7 (18.7)	77.04 (18.2)	.20
Chlorthalidone, No. (%)	3706 (36.3)	1605 (36.2)	1615 (36.4)	.94	1047 (37.0)	1052 (36.9)	.58
Amlodipine, No. (%)	2194 (21.5)	949 (21.4)	951 (21.5)		611 (21.6)	607 (21.3)
Lisinopril, No. (%)	2136 (20.9)	935 (21.1)	913 (20.6)		603 (21.3)	579 (20.3)
Doxazosin, No. (%)	2163 (21.2)	939 (21.2)	953 (21.5)		571 (20.2)	613 (21.5)

Abbreviations: ALLHAT, Antihypertensive and Lipid‐Lowering Treatment to Prevent Heart Attack Trial; ASCVD, atherosclerotic cardiovascular disease; CHD, coronary heart disease; DBP, diastolic blood pressure; GFR, glomerular filtration rate; HDL‐C, high‐density lipoprotein cholesterol; LDL‐C, low‐density lipoprotein cholesterol; LVH, left ventricular hypertrophy; MI, myocardial infarction; SBP, systolic blood pressure; SD, standard deviation; VA, Veterans Administration. ^aParticipants alive at end of trial. ^bParticipants alive at end of trial and on Medicare or with Social Security number and age 65 years or older; excludes VA data. ^cFor trial eligibility, participants had to have at least one other risk factor in addition to hypertension. Thus, the indicated risk factors are not mutually exclusive or exhaustive and may not represent prevalence. ^dHistory of MI or stroke; history of coronary revascularization; major ST‐segment depression on T‐wave inversion documented by electrocardiography in the past 2 years; or other ASCVD (history of angina pectoris; history of intermittent claudication, gangrene, or ischemic ulcers; history of transient ischemic attack; coronary, peripheral vascular, or carotid stenosis ≥50% documented by angiography or Doppler studies; ischemic heart disease documented by reversible or fixed ischemia on stress thallium or dipyridamole thallium, ST depression ≥1 mm for ≥1 minute on exercise testing or Holter monitoring; reversible wall motion abnormality on stress echocardiography; ankle‐arm index <0.9; abdominal aortic aneurysm detected by ultrasonography, computed tomography, or radiography; or carotid or femoral bruits).

Intermediate Outcomes

Data on visit adherence, medication adherence, BP, and lipid levels by treatment group over time during the in‐trial phase have been previously reported.2 Briefly, for those assigned to pravastatin, adherence to the study drug dropped from 87% at year 2 to 80% at year 4 and 77% at year 6. In the UC group, crossovers to statin treatment increased from 8% at year 2 to 17% at year 4 and 26% at year 6. At year 4, LDL‐C levels decreased by 28% in the pravastatin group and 11% in the UC group.

Primary and Secondary Outcomes

As shown in Table 2 and Figure 2a, 10‐year rates of mortality did not differ between the pravastatin and UC groups (HR, 0.96; 95% CI, 0.89–1.03; P=.24), nor were there any overall significant differences in cause‐specific mortality. However, CHD mortality was significantly lower in blacks in the pravastatin group compared with the UC group (HR, 0.74; 95% CI, 0.58–0.94; P=.01), while there was no difference in CHD mortality between the treatment groups in non‐black participants (P for interaction=.052). There was no significant difference between treatment groups in stroke mortality in either black or non‐black participants, although number of events was much lower for strokes than CHD events (Table 2).

Table 2.

Outcomes by Lipid‐Lowering Treatment Group

	Usual Care	Pravastatin	HR (95% CI)	P Value
	10‐y Rate per 100 Persons (SE)	10‐y Rate per 100 Persons (SE)
Mortality outcomes
Total analyzed	5110	5089	10199
All‐cause mortality	31.0 (0.7)	30.1 (0.7)	0.96 (0.89–1.03)	.24
CV mortality	15.1 (0.6)	14.2 (0.5)	0.93 (0.84–1.04)	.19
CHD	7.9 (0.4)	7.0 (0.4)	0.89 (0.76–1.03)	.11 a
Black	8.6 (0.7)	6.7 (0.6)	0.74 (0.58–0.94)	.01
Non‐black	7.4 (0.5)	7.2 (0.5)	1.00 (0.83–1.21)	1.00
Stroke	2.4 (0.2)	2.4 (0.2)	1.02 (0.78–1.33)	.89 b
Black	2.2 (0.4)	2.8 (0.4)	1.33 (0.87–2.04)	.19
Non‐black	2.4 (0.3)	2.0 (0.3)	0.85 (0.60–1.21)	.37
Heart failure	1.6 (0.2)	1.6 (0.2)	0.95 (0.69–1.31)	.74
Other CVD	4.1 (0.3)	4.0 (0.3)	0.97 (0.79–1.19)	.76
Non‐CV mortality	17.7 (0.6)	17.5 (0.6)	0.98 (0.89–1.08)	.71
Cancer	7.5 (0.4)	8.0 (0.4)	1.04 (0.90–1.20)	.62
Kidney disease	0.9 (0.2)	1.0 (0.2)	1.10 (0.72–1.69)	.66 c
Black	0.9 (0.2)	1.5 (0.3)	1.40 (0.76–2.60)	.28
Non‐black	1.0 (0.2)	0.7 (0.2)	0.85 (0.46–1.56)	.60
Accident, suicide, homicide	0.8 (0.1)	0.7 (0.1)	0.83 (0.51–1.35)	.46
Other non‐CVD	9.4 (0.5)	8.7 (0.4)	0.94 (0.81–1.07)	.34
Unknown cause	1.2 (0.2)	1.3 (0.2)	0.97 (0.66–1.41)	.86
Combined fatal or nonfatal hospitalized events
Total analyzed	3325	3313	6638
CHD	17.5 (0.8)	15.9 (0.7)	0.92 (0.81–1.04)	.19 d
Black	16.7 (1.2)	14.2 (1.1)	0.79 (0.64–0.98)	.032
Non‐black	17.9 (1.0)	17.0 (1.0)	1.01 (0.86–1.18)	.95
CVD	36.1 (1.0)	34.8 (1.0)	0.95 (0.87–1.04)	.28
Heart failure	13.8 (0.7)	14.0 (0.7)	0.99 (0.85–1.14)	.86
Stroke	12.9 (0.7)	11.7 (0.7)	0.90 (0.77–1.05)	.18 e
Black	13.6 (1.2)	12.8 (1.1)	0.99 (0.78–1.26)	.95
Non‐black	12.6 (0.9)	10.9 (0.8)	0.84 (0.69–1.03)	.09
GI bleed	21.1 (0.9)	20.6 (0.9)	0.99 (0.88–1.12)	.92
ESRD from USRDS
Total analyzed	5110	5089	10199
ESRD	2.4 (0.2)	2.6 (0.3)	1.09 (0.83–1.41)	.54 f
Black	3.0 (0.4)	4.6 (0.6)	1.36 (0.96–1.94)	.085
Non‐black	2.0 (0.3)	1.3 (0.2)	0.77 (0.51–1.17)	.22

Abbreviations: CHD, coronary heart disease; CI, confidence interval; CV, cardiovascular; CVD, cardiovascular disease; ESRD, end‐stage renal disease; GI, gastrointestinal; HR, hazard ratio; SD, standard deviation; SE, standard error; USRDS, US Renal Data System; VA, Veterans Administration. ^aCHD mortality treatment × race interaction, P=.052. ^bStroke mortality treatment × race interaction, P=.11. ^cKidney disease mortality treatment × race interaction, P=.26. ^dCombined fatal and nonfatal CHD events treatment × race interaction, P=.08. ^eCombined fatal and nonfatal stroke events treatment × race interaction, P=.30. ^fUSRDS ESRD events treatment × race interaction, P=.041.

Figure 2.

(a) Cumulative mortality rates by treatment group in the Antihypertensive and Lipid‐Lowering Treatment to Prevent Heart Attack Trial (ALLHAT) Lipid‐Lowering Trial. (b) Cumulative combined morbidity and mortality rates by treatment group in the ALLHAT Lipid‐Lowering Trial. CHD indicates coronary heart disease.

The 10‐year rate of combined fatal/nonfatal CHD was nonsignificantly lower in the pravastatin group compared with the UC group, with HR (HR, 0.92; 95% CI, 0.81–1.04; P=.19; Table 2) quite similar to that observed during the in‐trial period (HR, 0.94; 95% CI, 0.80–1.12; P=.50, Table 3). Treatment groups also did not differ in 10‐year rates of CVD, HF, stroke, or ESRD (Figure 2b). However, there were differences in treatment effects by race. Similar to results observed during the in‐trial period, the 10‐year rate of combined fatal/nonfatal CHD was significantly lower in blacks in the pravastatin group compared with the UC group (HR, 0.79; 95% CI, 0.64–0.98; P=.03), while there was no difference between treatment groups in non‐blacks (P for interaction=.08, Table 2). Although there was a suggestion of a protective effect of pravastatin on combined fatal/nonfatal stroke in non‐blacks in this reduced‐size cohort of 6638, HR was not significantly different from null, nor was there a significant interaction in treatment effects between racial groups (P for interaction=.30, Table 2). This differed from the results observed during the in‐trial period in the entire cohort of 10,355 patients, where a significant race‐by‐treatment interaction (P=.03) was observed for stroke and the benefit in non‐blacks was statistically significant (HR, 0.74; 95% CI, 0.57–0.96).2, 4

Table 3.

Cox Regression Analyses of Outcomes by Time Period, Overall, and by Race

Outcomes	Pravastatin vs Usual Care		P Value for Treatment × Period Interaction
	HR (95% CI) P Value
	In‐Trial	Post‐Trial
Mortality outcomes
All‐cause mortality	0.97 (0.87–1.07) .52	0.95 (0.87–1.05) .31	.85
Black	0.99 (0.84–1.17) .92	0.93 (0.80–1.07) .30	.54
Non‐black	0.97 (0.84–1.11) .63	0.95 (0.85–1.07) .44	.90
CV mortality	0.96 (0.83–1.13) .65	0.91 (0.79–1.04) .17	.56
Black	0.93 (0.74–1.18) .56	0.88 (0.70–1.10) .25	.71
Non‐black	1.01 (0.82–1.24) .92	0.91 (0.76–1.09) .29	.44
CHD	0.92 (0.75–1.14) .47	0.85 (0.69–1.05) .13	.60
Black	0.71 (0.51–1.00) .05	0.76 (0.55–1.07) .11	.78
Non‐black	1.13 (0.85–1.49) .41	0.90 (0.69–1.17) .43	.25
Stroke	1.02 (0.71–1.47) .91	1.02 (0.69–1.51) .93	.99
Black	1.32 (0.76–2.29) .32	1.35 (0.68–2.67) .39	.96
Non‐black	0.85 (0.51–1.40) .51	0.86 (0.53–1.40) .54	.96
Heart failure	1.10 (0.68–1.76) .70	0.83 (0.54–1.30) .42	.41
Black	1.17 (0.58–2.38) .66	1.00 (0.52–1.90) 1.00	.74
Non‐black	1.03 (0.54–1.94) .93	0.70 (0.38–1.30) .26	.39
Other CVD	0.95 (0.66–1.38) .79	0.98 (0.76–1.25) .85	.91
Black	1.09 (0.62–1.92) .75	0.91 (0.62–1.34) .64	.60
Non‐black	0.84 (0.51–1.40) .50	1.02 (0.73–1.41) .91	.53
Non‐CV mortality	0.98 (0.84–1.15) .82	0.98 (0.87–1.11) .76	.99
Black	1.04 (0.82–1.33) .73	1.00 (0.82–1.22) .73	.77
Non‐black	0.95 (0.77–1.16) .61	0.97 (0.83–1.13) .67	.88
Cancer	1.10 (0.88–1.36) .41	0.99 (0.81–1.21) .95	.52
Black	1.20 (0.86–1.67) .28	1.19 (0.88–1.61) .26	.98
Non‐black	0.98 (0.73–1.31) .91	0.88 (0.68–1.15) .35	.58
Kidney disease	1.15 (0.56–2.35) .71	1.07 (0.63–1.83) .79	.89
Black	2.12 (0.74–6.10) .16	1.10 (0.51–2.39) .80	.33
Non‐black	0.57 (0.19–1.70) .31	1.02 (0.49–2.16) .94	.38
Accident, suicide, or homicide	1.00 (0.51–1.97) .99	0.68 (0.34–1.38) .29	.44
Black	0.97 (0.34–2.77) .96	1.27 (0.28–5.68) .75	.77
Non‐black	1.03 (0.43–2.47) .95	0.58 (0.26–1.31) .19	.35
Other non‐CVD	0.84 (0.66–1.08) .17	0.98 (0.83–1.16) .82	.31
Black	0.80 (0.53–1.20) .28	0.81 (0.61–1.08) .15	.96
Non‐black	0.87 (0.63–1.19) .38	1.09 (0.89–1.34) .42	.24
Unknown cause	0.91 (0.58–1.41) .66	1.14 (0.56–2.34) .72	.59
Black	0.74 (0.38–1.41) .36	1.11 (0.37–3.30) .85	.53
Non‐black	1.08 (0.58–1.99) .81	1.16 (0.45–3.00) .76	.90
Combined fatal or nonfatal hospitalized events
CHD	0.94 (0.80–1.12) .50	0.89 (0.73–1.08) .23	.64
Black	0.71 (0.53–0.94) .02	0.92 (0.66–1.27) .61	.24
Non‐black	1.12 (0.90–1.38) .31	0.87 (0.68–1.11) .27	.13
CVD	0.97 (0.87–1.10) .68	0.93 (0.82–1.05) .26	.60
Black	0.91 (0.76–1.10) .35	1.02 (0.83–1.25) .88	.46
Non‐black	1.02 (0.87–1.19) .80	0.88 (0.75–1.03) .12	.19
Heart failure	0.98 (0.80–1.22) .89	0.99 (0.81–1.20) .91	.97
Black	1.16 (0.84–1.62) .36	1.07 (0.79–1.44) .68	.69
Non‐black	0.87 (0.65–1.16) .34	0.92 (0.71–1.19) .52	.78
Stroke	0.89 (0.71–1.11) .30	0.91 (0.74–1.13) .40	.86
Black	1.02 (0.74–1.42) .90	0.96 (0.68–1.36) .81	.79
Non‐black	0.78 (0.57–1.06) .11	0.89 (0.68–1.16) .39	.51
GI bleed	0.95 (0.80–1.14) .60	1.03 (0.88–1.20) .74	.54
Black	0.93 (0.70–1.23) .60	1.10 (0.86–1.41) .45	.37
Non‐black	0.97 (0.77–1.23) .82	0.97 (0.80–1.19) .80	>.99
ESRD	0.90 (0.60–1.35) .62	1.25 (0.88–1.78) .21	.23
Black	0.94 (0.57–1.55) .80	1.96 (1.18–3.26) .01	.04
Non‐black	0.81 (0.41–1.59) .54	0.75 (0.45–1.26) .28	.87

Abbreviations: CHD, coronary heart disease; CI, confidence interval; CV, cardiovascular; ESRD, end‐stage renal disease; GI, gastrointestinal; HR, hazard ratio; SD, standard deviation; VA, Veterans Administration.

There was also a nominally significant race by treatment interaction for the ESRD endpoint (P for interaction=.04). Compared with the UC group, black participants in the pravastatin group had a borderline higher risk of ESRD (HR, 1.36; 95% CI, 0.96–1.94; P=.085), while treatment effects did not differ significantly in non‐blacks (HR, 0.77; 95% CI, 0.51–1.17; P=.22). Overall, rates of ESRD were low (on the order of 2.5 per 100 persons over 10 years). This endpoint was not originally reported for the in‐trial period.10 Using the USRDS database to examine the in‐trial period, there were no overall differences (HR, 0.90; 95% CI, 0.60–1.35; P=.62) or differences in blacks (HR, 0.94; 95% CI, 0.57–1.55; P=.80) and non‐blacks (HR, 0.81; 95% CI, 0.41–1.59; P=.54) between treatment groups. However, in the post‐trial period (Table 3) there was a significantly higher risk of ESRD in blacks in the pravastatin group (HR, 1.96; 95% CI, 1.18–3.26; P=.01), but not in non‐blacks (HR, 0.75; 95% CI, 0.45–1.26; P=.28). For blacks, there was a significant difference in in‐trial and post‐trial HR for ESRD (P for interaction=.04). There were no significant differences in HRs for any other outcome between in‐trial and post‐trial periods, either overall or in black or non‐black subgroups (Table 3). Sensitivity analyses to compare different outcome ascertainment methods yielded results that were similar in almost all respects to protocol‐specified analyses (online Supplemental Tables 1, 2). The only difference was a statistically significant lowering of CHD mortality in the pravastatin group compared with the UC group (Supplemental Table 2). The 10‐year CHD mortality rates were 6.2% for the pravastatin and 7.5% for UC (HR, 0.81; 95% CI, 0.69–0.95; P<.01). The CHD mortality rate in blacks remained significantly lower in the pravastatin vs UC group (HR, 0.70; 95% CI, 0.55–0.91; P<.01), but there was not a significant difference in effect between blacks and non‐blacks (P for interaction=.16).

Discussion

With a few exceptions, the results of extended 10‐year follow‐up of the cohort enrolled in the ALLHAT‐LLT trial show similar long‐term effects of treatment with pravastatin vs UC as were observed in‐trial. Specifically, there was no significant effect of pravastatin on total mortality, cause‐specific mortality, or combined fatal/nonfatal events including CHD, CVD, HF, stroke, or ESRD. In extended period analyses, as in in‐trial analyses, there was evidence of a protective effect of pravastatin for CHD mortality and combined fatal/nonfatal CHD in blacks, but not in non‐blacks. The in‐trial finding of a protective effect of pravastatin on stroke in non‐blacks but not in blacks was not confirmed in the reduced‐size cohort available for the extended morbidity follow‐up. However, there was a finding of a possible adverse effect of pravastatin on ESRD in blacks that emerged only during post‐trial follow‐up.

Based on a large number of trials and meta‐analyses, statins have been shown to lower CVD morbidity and mortality overall and in numerous subgroups based on age, sex, comorbidities (diabetes, hypertension, renal disease), and baseline cholesterol levels without increasing cancer or other causes of serious morbidity.3, 11, 12 There is also a fairly extensive literature on long‐term effects of statins from trial extension studies.13, 14, 15, 16, 17, 18 These have included patients across a broad spectrum of cholesterol levels and CV risk, ranging from relatively low‐risk populations to those with established CHD or renal disease. In general, despite provision of post‐trial open‐label therapy to all groups, these showed either continued benefits of statins or emergence of benefit when none was shown during the trial.15 Only one study, in a cohort of patients with microalbuminuria, showed no benefit during the trial or its extension, but the trial was quite small.18 Notably, ALLHAT was the only lipid‐lowering trial in an all‐hypertensive cohort with a relatively well‐controlled BP at the time of enrollment (mean 146/84 mm Hg).

Current results from the ALLHAT‐LLT extension study are consistent with these studies in that the protective effect of pravastatin on CHD in blacks during the trial was also observed during long‐term follow‐up. We were not able to discover a definite mechanism for this race‐specific result, but were able to rule out a number of possible explanations, including differences in adherence and risk factor levels.5 It is possible that in a predominantly white hypertensive population with moderately elevated cholesterol levels and no other indications for lipid‐lowering treatment, once a reasonable level of BP control is achieved, there is no further benefit for CHD prevention from statin treatment to levels recommended by the Third Report of the Adult Treatment Panel (mean LDL‐C and total cholesterol levels in the pravastatin arm at year 6 of follow‐up were 104 mg/dL and 178 mg/dL, respectively). This may not be true in the black population, potentially because of a different pathology of CHD, genetic factors, or both. For example, a possible explanation for the observed greater protective effect of statins in blacks vs non‐blacks could be the higher prevalence in blacks of the CC variant of MTHFR 677 C>T and the CBS844 ins64 I+variants, which are associated with greater protective effect of pravastatin against CHD.19

There is no clear explanation for our current results of an increased risk of ESRD emerging during long‐term follow‐up in the group treated with pravastatin. Statin therapy has been shown to reduce proteinuria and slow the decline of GFR, and, to our knowledge, has not been shown to increase the risk of ESRD.20, 21 Notably, there were 110 analyses performed in Table 3 for in‐trial and post‐trial periods and an additional 55 examining interactions by time period. Whether this finding is simply by chance a result of multiple comparisons performed in these analyses is unclear.

Study Limitations

Several limitations of ALLHAT‐LLT and its extension study deserve mention. We believe that modest cholesterol differences achieved in the study explain the overall null results. Extension study outcome results were gathered from national research administrative databases, so no information is available for nonfatal CVD outcomes in participants who were not covered by Medicare. We also did not have information about statin use or lipid levels in either treatment group after trial end. Finally, we were unable to obtain follow‐up data on the small number of Canadian participants or on nonfatal events among participants enrolled at the VA sites.

Conclusions

In the context of the modest cholesterol differences between treatment groups observed in ALLHAT‐LLT, results are consistent with evidence from other large trials that statins prevent CHD. ALLHAT‐LLT enrolled the largest number of blacks in any lipid‐lowering trial with long‐term follow‐up. Our results provide further evidence that statins are effective for prevention of fatal and nonfatal CHD in blacks. Although the proportion of blacks who are being treated for hypercholesterolemia according to current guidelines is improving, racial disparities persist in treatment and cholesterol lowering.22, 23, 24 The ALLHAT‐LLT findings support the need for intensive efforts to eliminate these disparities.

Funding

This study was supported by contracts NO1‐HC‐35130 and HHSN268201100036C with the National Heart, Lung, and Blood Institute (NHLBI). ALLHAT investigators received contributions of study medications supplied by Pfizer (amlodipine besylate and doxazosin mesylate), AstraZeneca (atenolol and lisinopril), and Bristol‐Myers Squibb (pravastatin), and financial support provided by Pfizer Inc.

Role of the Sponsors

The NHLBI sponsored the study and was involved in all aspects other than direct operations of the study centers. This included collection, analysis, and interpretation of the data plus the decision to submit the manuscript for publication. Pfizer Inc, AstraZeneca, and Bristol‐Myers Squibb had no role in the design and conduct of the study, the collection, analysis, and interpretation of the data, or the preparation or approval of the manuscript.

Disclosures

Karen Margolis has received research support from Bristol‐Myers Squibb. Charles Baimbridge has held a financial interest in Johnson & Johnson. Jerry Ciocon has consulted for Forest Pharmaceuticals and Novartis. Barry Davis has consulted for Amgen, Forest Pharmaceuticals, and Takeda. L. Julian Haywood has held a financial interest in Pfizer. Suzanne Oparil has received honoraria from Boehringer Ingelheim, Daiichi Sankyo, Eli Lilly, Forest Laboratories, Forest Pharmaceuticals, Merck, NicOx, Novartis, Omron Healthcare, Pfizer, Salt Institute, Sanofi Aventis, and Schering Plough; and has received research support from Amgen, Daiichi Sankyo, Eisai Medical Research, Gilead, Merck, and Takeda. Jeffrey Probstfield has received honoraria from Boehringer Ingelheim and has received research support from Abbott Laboratories, Boehringer Ingelheim, GlaxcoSmithKline, KOS, and Sanofi Aventis. John Stokes has received honoraria from GlaxoSmithKline. Drs Cuyjet, Dart, Einhorn, Ford, Gordon, Hartney, Holtzman, Mathis, Simpson, Wiegmann, and Williamson have no financial interests to disclose.

Supporting information

Table S1. (a) Number of events in‐trial and post‐trial by treatment group (primary analyses). (b) Number of events in‐trial and post‐trial by treatment group (sensitivity analyses).

Table S2. Primary and sensitivity analyses of outcomes by lipid‐lowering treatment group overall and by race for selected outcomes.

J Clin Hypertens (Greenwich). 2013;15:542–554. ©2013 Wiley Periodicals, Inc.23889716