Novel Approach to Examining First Cardiovascular Events After Hypertension Onset

Transcription

1 Novel Approach to Examining First Cardiovascular Events After Hypertension Onset Donald M. Lloyd-Jones, Eric P. Leip, Martin G. Larson, Ramachandran S. Vasan, Daniel Levy Abstract Hypertension confers risk for multiple types of cardiovascular events, but competing risks for these outcomes are unknown. We estimated the competing risks over 12 years after hypertension onset among cases and age-, sex-, and examination-matched controls using competing Cox cumulative incidence and proportional hazards models. We included all Framingham Heart Study subjects examined after 1977 with new-onset hypertension who were free of cardiovascular disease. There were 645 men and 702 women with new-onset hypertension (mean age: men, years; women, years). Compared with matched nonhypertensive controls, subjects with new-onset hypertension were more likely to experience a cardiovascular event first rather than noncardiovascular death. Among new-onset hypertensives, the 12-year competing cumulative incidence of any cardiovascular end point as a first event in men was 24.7%, compared with 9.8% for noncardiovascular death (hazards ratio [HR], 2.53; 95% confidence interval [CI], 1.83 to 3.50); in women, the competing incidences were 16.0% versus 10.1%, respectively (HR, 1.58; 95% CI, 1.13 to 2.20). The most common first major cardiovascular events were hard coronary disease (8.2%) in men and stroke (5.2%) in women. Type and incidence of first cardiovascular events varied by age and severity of hypertension at onset, with stroke predominating among older subjects with new-onset hypertension. After hypertension onset, cardiovascular events are more likely to occur first as opposed to noncardiovascular death. Types of initial events differ by gender, age, and severity of hypertension at onset. These results represent a novel approach to understanding the complications of hypertension and may help target therapies for patients with new-onset hypertension to optimize prevention strategies. (Hypertension. 2005;45:39-45.) Key Words: hypertension, detection and control risk factors cardiovascular diseases More than 65 million adults in the United States currently have hypertension. 1 Hypertension is a risk factor for a number of cardiovascular diseases (CVD), including stroke, congestive heart failure (CHF), coronary heart disease (CHD), and peripheral arterial disease, as well as for endorgan damage such as renal disease, and mortality from all causes. 2,3 Observational data have established independent associations between hypertension and each of these outcomes, and placebo-controlled clinical trials of antihypertensive therapy have documented impressive reductions in CVD events. 4 8 To date, however, there have been no studies examining which CVD events occur as a first event after hypertension onset in a competing risks framework. This distinction is important because the occurrence of one type of CVD event (eg, stroke) may be associated with greater morbidity and mortality than another type of CVD (eg, angina), and because the occurrence of one CVD event (eg, myocardial infarction [MI]) may markedly increase the risk for subsequent CVD events (CHF, stroke). Furthermore, different strategies may be needed to prevent different outcomes. Thus, understanding which of the competing CVD events occurs first after hypertension onset may have important implications for prognosis and prevention. Using data from the Framingham Heart Study, we sought to quantitate the competing risks for different first CVD events after hypertension onset, relative to each other and to non-cvd death, and to identify characteristics affecting these competing risks. We compared competing risks among men and women with new-onset hypertension, and between subjects with new-onset hypertension and a nonhypertensive control group. Methods Study Sample and Definitions Study design and entry criteria for the Framingham Heart Study have been detailed elsewhere. 9,10 All examinations and procedures were approved by the institutional review board of Boston Medical Center, and all participants have provided informed consent in accordance with institutional guidelines. For the present analysis, we Received September 15, 2004; first decision October 5, 2004; revision accepted October 20, From the Department of Preventive Medicine (D.M.L.-J.), Feinberg School of Medicine, Northwestern University, Chicago, Ill; National Heart, Lung, and Blood Institute s Framingham Heart Study (E.P.L., M.G.L., R.S.V., D.L.), National Institutes of Health, Framingham, Mass; the Division of Epidemiology and Preventive Medicine (M.G.L., R.S.V., D.L.), Boston University School of Medicine, Boston, Mass; and the National Heart, Lung, and Blood Institute (D.L.), Bethesda, Md. Correspondence to Donald M. Lloyd-Jones, MD, ScM, Department of Preventive Medicine, Feinberg School of Medicine, Northwestern University, 680 N Lake Shore Dr, Suite 1120, Chicago, IL dlj@northwestern.edu 2004 American Heart Association, Inc. Hypertension is available at DOI: /01.HYP

2 40 Hypertension January 2005 TABLE 1. Baseline Characteristics at Examination of Hypertension Diagnosis Men Women Baseline Characteristic New-Onset Hypertension New-Onset Hypertension Age at hypertension diagnosis, y Stage 1 hypertension, % Stage 2 hypertension or treated, % Antihypertensive therapy, % Mean systolic blood pressure, mm Hg Mean diastolic blood pressure, mm Hg Current cigarette smoking, % Mean total cholesterol, mg/dl Diabetes, % Mean body mass index, kg/m included all subjects who were examined after September 1, 1977 and were free of any pre-existing CVD. At each examination, blood pressure was measured twice in the left arm by a physician, and the average of the 2 values was used as described previously. 11 Subjects with prevalent CVD, blood pressure 160 mm Hg systolic or 100 mm Hg diastolic, or receiving antihypertensive therapy before the study period were excluded. To select only those subjects who were truly hypertensive, we defined new-onset hypertension as blood pressure 160 mm Hg systolic or 100 mm Hg diastolic, or new antihypertensive therapy at one Framingham Heart Study clinic examination; or blood pressure 140 mm Hg systolic or 90 mm Hg diastolic at 2 consecutive Heart Study examinations (4 years apart). We also selected control subjects matched 1:1 with cases on age (within 5 years), sex, and examination cycle at diagnosis of hypertension. New-onset hypertensive subjects and controls were followed-up for up to 12 years after the diagnosis of new-onset hypertension; follow-up was censored at the time of death, a CVD event, the end of the study period (1999), or (for controls) the occurrence of new-onset hypertension. All events were reviewed and adjudicated by a panel of 3 trained physicians after review of all Framingham Heart Study examinations, as well as physician notes and hospital charts, using previously published criteria. 12 We defined a CVD event ( all CVD ) as the occurrence of a hard coronary heart disease event (coronary death, MI, or coronary insufficiency [hospitalized unstable angina]), stroke, congestive heart failure, angina pectoris, intermittent claudication, or other cause of CVD death. A hard CVD event was defined as the aforementioned excluding angina and claudication. Non-CVD death was defined as the occurrence of death attributable to causes other than CVD (eg, cancer). Statistical Analysis We first estimated the overall Kaplan Meier 13 cumulative incidence for the occurrence of any CVD or non-cvd death at any time during 12 years after hypertension onset. We then determined the first event occurring after hypertension onset, whether it was CVD or non-cvd death, separately for cases and controls. We estimated the hazards and competing cumulative incidences of CVD events compared with non-cvd death as the first event after hypertension onset in a competing risks framework, as described previously. 14,15 Briefly, we used the method of data duplication as described by Lunn and McNeil 14 to fit Cox proportional hazards models 16 for all CVD events combined compared with non-cvd death as a first event after hypertension onset, separately for men and women. This analysis was repeated for the comparison between hard CVD events and non-cvd death. Similar analyses were performed examining the competing risks with individual CVD events and non-cvd death in the models, using non-cvd death as the referent. We then estimated the cumulative incidence of competing first CVD events (combined and individually) and non-cvd death using the methods described by Tai et al, 15 which extend the approach of Lunn and McNeil, 14 separately for cases and controls. In sex-specific Cox models, we also evaluated the hazards for a first CVD event versus non-cvd death among new-onset hypertensives compared with nonhypertensive controls by entering hypertensive status as a covariate in models containing both hypertensive subjects and controls. To check for effect modification by age and severity of hypertension, we stratified subjects into those aged younger than 60 years compared with those 60 years or older at diagnosis of hypertension and also stratified subjects into those with stage 1 hypertension (140 to 159 mm Hg systolic or 90 to 99 mm Hg diastolic) compared with stage 2 (systolic 160 mm Hg or diastolic 100 mm Hg or treated) hypertension at diagnosis. Competing Cox models and cumulative incidences were then repeated as mentioned. If a CVD event occurred on the same day as death, then the CVD event was coded as occurring first. When CHF and MI were diagnosed simultaneously, we arbitrarily assigned the MI as occurring first. Otherwise, when 2 CVD events were diagnosed simultaneously, we used random selection to break the tie. All analyses were performed using SAS statistical software. 17 Results Baseline Characteristics and Outcomes There were 645 men and 702 women with new-onset hypertension and no previous CVD. Baseline characteristics of hypertensive subjects and age-, sex-, and examinationmatched control subjects are shown in Table 1. As expected, hypertensive subjects and controls differed with respect to blood pressure levels and antihypertensive therapy, but they were similar with regard to age, cardiovascular risk factor profiles, and rates of non-cvd death. Among men with new-onset hypertension, 129 CVD events and 51 non-cvd deaths occurred during follow-up; male controls experienced 90 CVD events and 57 non-cvd deaths. Among hypertensive women, 77 CVD events and 50 non-cvd deaths occurred during follow-up; female controls experienced 50 CVD events and 49 non-cvd deaths. Competing Risks for All CVD Events After Hypertension Onset In men and women with new-onset hypertension, a CVD event was significantly more likely as a first event than non-cvd death. As shown in Table 2, the competing cumulative incidence of any CVD as a first event in hypertensive

3 Lloyd-Jones et al Risks of CVD After Hypertension Onset 41 TABLE 2. Hazards Ratios and Competing Cumulative Incidences for First CVD Events Compared With Non-CVD Death Through 12 Years Among Subjects With New-Onset Hypertension and Among Age-, Sex-, and Examination-Matched Control Subjects First Event During Follow-Up: All CVD Men Women Hazards ratio (all CVD vs non-cvd death) 2.53 ( ) 1.54 ( ) 1.58 ( ) 1.02 ( ) Competing cumulative incidences Non-CVD death (n 51 men and 57 women), % Any CVD event (n 129 men and 90 women), % Hard CHD (n 43 men and 17 women), % Stroke (n 25 men and 29 women), % CHF (n 12 men and 12 women), % Other CVD death (n 3 men and 2 women), % AP/IC (n 46 men and 30 women), % HTN indicates hypertension; AP/IC, angina pectoris or intermittent claudication. men was 24.7% compared with 9.8% for non-cvd death as a first event (hazards ratio [HR], 2.53; 95% confidence interval [CI], 1.83 to 3.50); in hypertensive women, the incidences were 16.0% versus 10.1%, respectively (HR, 1.58; 95% CI, 1.13 to 2.20). Nonhypertensive men were also more likely to experience a CVD event first, but at a much lower rate (15.7% versus 10.2%; HR, 1.54; 95% CI, 1.08 to 2.20). Among nonhypertensive women, the hazards for CVD and non-cvd death were the same. Competing cumulative incidence curves for all CVD events compared with non-cvd death are shown in Figure 1 for those with new-onset hypertension and for controls, separately for men and women. When all individual CVD events were placed in the model competing with non-cvd death (Table 2), the most common first CVD events in hypertensive men were angina/claudication (8.8%), followed by hard CHD (8.2%) in men. These patterns were similar but at lower rates among male controls. In hypertensive women, the most common first CVD events were angina/claudication (5.3%), followed by stroke (5.2%). In female control subjects, the rates of stroke as a first event were substantially lower (1.7%). In an analysis comparing those with new-onset hypertension with controls, male hypertensives were more likely to experience a CVD event than non-cvd death first compared with controls (HR for CVD first in cases versus controls 1.64; 95% CI, 1.02 to 2.66). For female hypertensives compared with controls, the HR for having a CVD event first was 1.53 (95% CI, 0.93 to 2.59). Competing Risks for Hard CVD Events After Hypertension Onset When we excluded angina/claudication as a first event (Table 3), the competing cumulative incidence of hard CVD as a first event in men was 19.5%, compared with 10.8% for non-cvd death (HR, 1.80; 95% CI, 1.30 to 2.50); in women, the incidences were 12.6% versus 10.9%, respectively (HR, 1.15; 95% CI, 0.81 to 1.63). Control subjects were not more likely to experience a hard CVD event first. Competing cumulative incidence curves for hard CVD events compared Figure 1. Competing cumulative incidences of first events of CVD or non- CVD death in subjects with new-onset hypertension and age-, sex-, and examination-matched control subjects. For nonhypertensive women, the curves for initial CVD event and initial non-cvd death are nearly superimposed. HTN indicates new-onset hypertension.

4 42 Hypertension January 2005 TABLE 3. Hazards Ratios and Competing Cumulative Incidences for First Hard CVD Events Compared With Non-CVD Death Through 12 Years Among New-Onset Hypertensive Subjects and Among Age-, Sex-, and Examination-Matched Control Subjects First Event During Follow-Up: Hard CVD Men Women Hazards ratio (hard CVD vs non-cvd death) 1.80 ( ) 1.14 ( ) 1.15 ( ) 0.70 ( ) Competing cumulative incidences Non-CVD death (n 55 men and 60 women), % Any CVD event (n 99 men and 69 women), % Hard CHD (n 54 men/22 women), % Stroke (n 27 men and 32 women), % CHF (n 15 men and 13 women), % Other CVD death (n 3 men and 2 women), % with non-cvd death are shown in Figure 2. Table 3 also shows the results for the model containing individual hard CVD events as a first event competing with non-cvd death. In this analysis, the most common first events were hard CHD (10.6%) in hypertensive men and stroke (5.8%) in hypertensive women. The median times to event were 6.4 years for hard CHD in hypertensive men and 6.8 years for stroke in hypertensive women. Competing Risks for CVD by Age and Severity of Hypertension at Onset Among subjects with new-onset hypertension, the incidence of both CVD events and non-cvd death increased for subjects with onset at age 60 years or older compared with those younger than 60 years (Table 4). At all ages after the onset of hypertension, CVD was more likely to occur as a first event than non-cvd death. The contrast between risk for CVD and non-cvd death was most striking in men with hypertension onset at age younger than 60 years; in this group, the 12-year risk of CVD was 19.9% compared with 3.5% for non-cvd death (HR, 5.67; 95% CI, 3.07 to 10.46). For men and women, there was a dramatic increase in stroke and CHF as first CVD events with hypertension onset at age 60 years or older compared with younger than 60 years. Stroke became the most common type of all first CVD events after hypertension onset at age 60 years or older in both men and women (Table 4). As shown in Table 5, there was somewhat higher CVD incidence as a first event in subjects with stage 2 or treated hypertension compared with stage 1 hypertension. In contrast, the 12-year incidence of non-cvd death was not strikingly different between subjects with higher versus lower blood pressure. For both men and women with stage 2 or treated hypertension, angina/claudication was the most common first CVD event. There were lower rates of stroke as a first CVD event among men with stage 2 or treated hypertension compared with stage 1 hypertension. In a secondary analysis, when we excluded subjects receiving antihypertensive therapy and only analyzed subjects with untreated stage 2 hypertension, the results did not change materially (data not shown). There were no other substantial differences in the type of first CVD event between subjects with higher versus lower hypertension stage. Figure 2. Competing cumulative incidences of first events of hard CVD or non-cvd death in subjects with newonset hypertension and age-, sex-, and examination-matched control subjects.

5 Lloyd-Jones et al Risks of CVD After Hypertension Onset 43 TABLE 4. Cox Cumulative Incidence for Competing First CVD Events, Through 12 Years After Hypertension Onset, Stratified by Sex and Age at Diagnosis All CVD Events Hard CVD Events Competing Cumulative Incidences (n 417) Men Women Men Women (n 228) (n 363) (n 339) (n 417) (n 228) (n 363) (n 339) Non-CVD death, % Any CVD event, % Hard CHD, % Stroke, % CHF, % Other CVD death, % AP/IC, % AP/IC indicates angina pectoris or intermittent claudication. Discussion Principal Findings In our largely middle-aged sample of men and women with new-onset hypertension, we observed that a CVD event was 1.6 to 2.5 times more likely to occur as a first event than was a non-cvd death over 12 years of follow-up. Compared with age-, sex-, and examination-matched control subjects, men and women with new-onset hypertension were more likely to experience a first CVD event than non-cvd death, although this finding did not achieve statistical significance in women. Men with new-onset hypertension at younger than 60 years of age were almost 6 times more likely to experience a CVD event first than die from non-cvd causes. Types of first hard CVD events differed by gender and age at hypertension onset. In younger men, hard CHD events were the most common first events, whereas in older men and in women at all ages, stroke was the most common first event. Implications Hypertension is known to increase the risks for stroke, CHF, CHD, peripheral vascular disease, CVD death, end-stage renal disease, and mortality from all causes. 2,3 Rates of treatment and of blood pressure control to goal levels among hypertensive subjects remain suboptimal. 18,19 With the aging of the population, the overall prevalence of hypertension will increase, and its impact as a major risk factor will grow, unless major public health and clinical efforts can focus on prevention and achievement of greater control. The current results may have important implications for primary prevention strategies. For example, in women and older subjects with new-onset hypertension, hypertensiverelated CVD events (stroke, CHF) predominate as first events. Stroke is the most common first hard CVD event after new-onset hypertension in women and in all patients 60 years of age or older. Thus, clinicians and patients may want to focus their major efforts on achieving blood pressure control specifically aiming to lower the risk of stroke and CHF in these subgroups. Previous epidemiologic data have shown that hypertension confers higher relative risks for stroke and CHF than for other CVD outcomes, such as CHD. 3,20 Further, clinical trials have documented striking reductions in stroke and CHF incidence with antihypertensive therapy. 7,21,22 Global risk reduction is clearly indicated for these patients, but other strategies such as antiplatelet therapy, lipid- TABLE 5. Cox Cumulative Incidence for Competing First CVD Events, Through 12 Years After Hypertension Onset, Stratified by Sex and Stage of Hypertension at Diagnosis Competing Cumulative Incidences (n 246) All CVD Events Hard CVD Events Men Women Men Women (n 399) (n 188) (n 514) (n 246) (n 399) (n 188) (n 514) Non-CVD death, % Any CVD event, % Hard CHD, % Stroke, % CHF, % Other CVD death, % AP/IC, % AP/IC indicates angina pectoris or intermittent claudication. *Stage 1 indicates systolic 140 to 159 mm Hg or diastolic 90 to 99 mm Hg at diagnosis of hypertension. Stage 2 indicates systolic 160 mm Hg or diastolic 100 mm Hg or receiving antihypertensive therapy.

6 44 Hypertension January 2005 lowering, and smoking cessation have not been shown to reduce the risk of stroke to the same degree Conversely, in men younger than 60 years of age, who were almost 6 times more likely to experience a CVD event than non-cvd death, the most common first hard CVD events were related to atherosclerotic CHD (eg, MI, unstable angina). Thus, in younger men with new-onset hypertension, in addition to blood pressure control, significant emphasis might be placed on use of cholesterol-lowering therapy and aspirin, for which there are impressive data on preventing CHD. Recent clinical trials support such an approach. As part of the Anglo-Scandinavian Cardiac Outcomes Trial (AS- COT), 27 a trial of newer versus older antihypertensive therapies, the investigators randomized a subset of hypertensive individuals with mean total cholesterol of 213 mg/dl (5.5 mmol/l) to atorvastatin 10 mg or placebo daily. The cholesterol-lowering portion of the trial was terminated prematurely, after a median follow-up of only 3.3 years, because of a significant 36% reduction in fatal CHD and nonfatal MI in the atorvastatin group, even among those without elevated cholesterol. 27 Furthermore, in the Hypertension Optimal Treatment trial, treatment with 75 mg per day of aspirin was associated with a 36% reduction in MI compared with placebo among hypertensive individuals. 28 Some of the sex-associated differences in patterns of first events that we observed may have been related to baseline differences in smoking, cholesterol, and antihypertensive therapy (Table 1). Our observation that there were few differences in first event patterns between those with stage 1 hypertension and those with stage 2 or treated hypertension is of interest. It is possible that individuals with stage 1 hypertension remained untreated for a longer period during follow-up, thereby placing them at higher risk. The current study design did not allow us to address these issues definitively. Our results may be useful in the design of clinical trials examining antihypertensive therapy in new-onset hypertensives at similar ages. Clinical trials typically only analyze follow-up data on participants until the occurrence of a first event of interest. Because we focused on first events, the event rates we report may be useful in generating power and sample-size calculations for CVD and non-cvd death endpoints in such trials. In particular, our examination of all CVD and hard CVD endpoints, as well as individual CVD event types, may be useful in light of the trend for trialists to choose composite CVD outcomes as primary endpoints. Potential Limitations The Framingham Heart Study cohort is composed almost exclusively of white individuals, which may limit the generalizability of our findings to other ethnic groups, in whom the competing risks of CVD, CVD subtypes, and non-cvd death may differ. It is difficult, if not impossible, to assign an exact date of hypertension onset, given the lability of blood pressure readings within individuals during the day and over time. We required the presence of stage 2 hypertension or new antihypertensive therapy at 1 examination, or stage 1 hypertension at 2 consecutive examinations (4 years apart). We chose our conservative definition of hypertension based only on blood pressure readings obtained according to protocol at Heart Study examinations in an attempt to include only subjects who were truly hypertensive. Thus, we may have diagnosed new-onset hypertension somewhat later than would have been the case had subjects been examined more frequently, possibly leading to somewhat shorter times to event for CVD outcomes. Conversely, a high proportion (46%) of our subjects with new-onset hypertension diagnosed was receiving antihypertensive therapy, which would have delayed or prevented CVD events. Furthermore, the mean blood pressure among new-onset hypertensive subjects was only modestly elevated, which likely reduced risk of CVD. We had limited power to detect differences between subgroups in our stratified analyses based on age or hypertensive stage, given a relatively small number of events. Current Study in Context The present analysis represents a novel approach to understanding the risks of different outcomes associated with new-onset hypertension. Traditional epidemiologic methods using Kaplan Meier analysis have several limitations. First, they do not account for the competing and concurrent risks of multiple different outcomes. Kaplan Meier methods directly estimate the probability of an event if the risk of other outcomes is reduced to zero, an assumption that does not accurately reflect the clinical situation. Further, Kaplan Meier methods assume independence of different event types, which is inappropriate, because a first CVD event (eg, MI) may markedly enhance the risk for a second (eg, stroke or CHF), thus ignoring the conditional and time-dependent nature of the associations between outcomes. Finally, Kaplan Meier analysis also assumes independence of the censoring mechanism, which may introduce unpredictable biases. 14,15 For example, patients lost to follow-up after hypertension onset may have been less likely to return for an examination if they had a severe CVD event (such as a large stroke), or more likely to return if they had a mild CVD event (angina). Our study benefited from methods that do not assume independence of event types or censoring mechanisms, but do account for joint and competing risks between CVD events and non-cvd death. Acknowledgments D.M.L.-J. is supported by grant K23 HL04253 from the National Institutes of Health, Bethesda, Md. The Framingham Heart Study is supported by NIH/NHLBI Contract N01-HC References 1. Fields LE, Burt VL, Cutler JA, Hughes J, Roccella EJ, Sorlie P. The burden of adult hypertension in the United States 1999 to Hypertension. 2004;44: Heart Disease and Stroke Statistics 2004 Update. Dallas, TX: American Heart Association; Prospective Studies Collaboration. Age-specific relevance of usual blood pressure to vascular mortality: a meta-analysis of individual data for one million adults in 61 prospective studies. Lancet. 2002;360: Seventh Report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure. Hypertension. 2003;42: Gueyffier F, Boutitie F, Boissel JP, Pocock S, Coope J, Cutler J, Ekbom T, Fagard R, Friedman L, Perry M, Prineas R, Schron E. Effect of antihypertensive drug treatment on cardiovascular outcomes in women and men. Ann Intern Med. 1997;126:

7 Lloyd-Jones et al Risks of CVD After Hypertension Onset Psaty BM, Lumley T, Furberg CD, Schellenbaum G, Pahor M, Alderman MH, Weiss NS. Health outcomes associated with various antihypertensive therapies used as first-line agents: a network meta-analysis. JAMA. 2003;289: Staessen JA, Gasowski J, Wang JG, Thijs L, Den Hond E, Boissel JP, Coope J, Ekbom T, Gueyffier F, Liu L, Kerlikowske K, Pocock S, Fagard RH. Risks of untreated and treated isolated systolic hypertension in the elderly: meta-analysis of outcome trials. Lancet. 2000;355: Turnbull F. Effects of different blood-pressure-lowering regimens on major cardiovascular events: results of prospectively-designed overviews of randomised trials. Lancet. 2003;362: Dawber TR, Kannel WB, Lyell LP. An approach to longitudinal studies in a community: the Framingham study. Ann N Y Acad Sci. 1963;107: Kannel WB, Feinleib M, McNamara PM, Garrison RJ, Castelli WP. An investigation of coronary heart disease in families. The Framingham offspring study. Am J Epidemiol. 1979;110: Lloyd-Jones DM, Evans JC, Larson MG, O Donnell CJ, Levy D. Differential impact of systolic and diastolic blood pressure level on JNC-VI staging. Hypertension. 1999;34: Abbott RD, McGee DL. The Framingham Study: An Epidemiological Investigation of Cardiovascular Disease, Section 37: The Probability of Developing Certain Cardiovascular Diseases in Eight Years at Specified Values of Some Characteristics. Bethesda, MD: National Heart, Lung, and Blood Institute; Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. J Am Stat Assoc. 1958;53: Lunn M, McNeil D. Applying Cox regression to competing risks. Biometrics. 1995;51: Tai B-C, Machin D, White I, Gebski V. Competing risks analysis of patients with osteosarcoma: a comparison of four different approaches. Stat Med. 2001;20: Cox DR. Regression models and life tables (with discussions). J R Stat Soc B. 1972;34: SAS Institute. SAS/STAT User s Guide, version 8. Cary, NC: SAS Institute, Inc; Wolf-Maier K, Cooper RS, Kramer H, Banegas JR, Giampaoli S, Joffres MR, Poulter N, Primatesta P, Stegmayr B, Thamm M. Hypertension treatment and control in five European countries, Canada, and the United States. Hypertension. 2004;43: Hajjar I, Kotchen TA. Trends in prevalence, awareness, treatment, and control of hypertension in the United States, JAMA. 2003; 290: Kannel WB. Blood pressure as a cardiovascular risk factor: prevention and treatment. JAMA. 1996;275: SHEP Cooperative Research Group. Prevention of stroke by antihypertensive drug treatment in older persons with isolated systolic hypertension. JAMA. 1991;265: The Systolic Hypertension in Europe (Syst-Eur) Trial Investigators. Randomised double-blind comparison of placebo and active treatment for older patients with isolated systolic hypertension. Lancet. 1997;350: Antiplatelet Trialists Collaboration. Collaborative overview of randomised trials of antiplatelet therapy-i: prevention of death, myocardial infarction, and stroke by prolonged antiplatelet therapy in various categories of patients. BMJ. 1994;308: Corvol JC, Bouzamondo A, Sirol M, Hulot JS, Sanchez P, Lechat P. Differential effects of lipid-lowering therapies on stroke prevention: a meta-analysis of randomized trials. Arch Intern Med. 2003;163: Kawachi I, Colditz GA, Stampfer MJ, Willett WC, Manson JE, Rosner B, Speizer FE, Hennekens CH. Smoking cessation and decreased risk of stroke in women. JAMA. 1993;269: Straus SE, Majumdar SR, McAlister FA. New evidence for stroke prevention: scientific review. JAMA. 2002;288: ASCOT Investigators. Prevention of coronary and stroke events with atorvastatin in hypertensive patients who have average or lower-thanaverage cholesterol concentrations, in the Anglo-Scandinavian Cardiac Outcomes Trial - Lipid Lowering Arm (ASCOT-LLA). Lancet. 2003; 361: Hansson L, Zanchetti A, Carruthers SG, Dahlof B, Elmfeldt D, Julius S, Menard J, Rahn KH, Wedel H, Westerling S. Effects of intensive bloodpressure lowering and low-dose aspirin in patients with hypertension: principal results of the Hypertension Optimal Treatment (HOT) randomised trial. Lancet. 1998;351: