Chinese Trial on Isolated Systolic Hypertension in the Elderly FREE

IN 1988, THE Collaborative Group Coordinating Center started the placebo-controlled Systolic Hypertension in China (Syst-China) trial to test the hypothesis whether antihypertensive drug treatment could prevent stroke in older Chinese patients with isolated systolic hypertension.^1- 2 Active treatment was initiated with the dihydropyridine calcium channel blocker nitrendipine,³ with the possible addition of captopril, hydrochlorothiazide, or both drugs. In the recently published intention-to-treat analysis of the Syst-China trial,⁴ active treatment decreased the stroke rate by 38% (P = .01) from 20.8 to 13.0 end points per 1000 patient-years. In addition, all cardiovascular end points decreased by 37% (P = .004) and all-cause mortality by 39% (P = .003). At the rates observed in the placebo group, treatment of 1000 Chinese patients for 5 years could prevent 39 strokes, 59 major cardiovascular complications, or 55 deaths.⁴

This article expands the first Syst-China report.⁴ It explores (1) whether the benefits of active treatment were evenly distributed across 4 strata, prospectively defined according to sex and previous cardiovascular complications,^2,4 and (2) whether the morbidity and mortality results were influenced by age, level of systolic or diastolic blood pressure (BP), smoking or drinking habits, or diabetes mellitus at enrollment.

The Chinese Ministry of Health approved the Syst-China trial. Its design is described in detail elsewhere.^1- 2,4 Eligible patients consented to be enrolled in the study and were at least 60 years old. While receiving single-blind placebo during the run-in phase, their sitting systolic BP had to range from 160 to 219 mm Hg, with diastolic BP below 95 mm Hg. The sitting BP determining eligibility was the average of 6 readings, 2 at 3 baseline visits 1 month apart. Patients whose serum creatinine concentration exceeded 180 µmol/L (2 mg/dL) and patients with severe concomitant cardiovascular or noncardiovascular disorders were excluded.^1- 2,4

The 31 Syst-China centers received supplies of the 3 active study drugs, all labeled A, and the 3 matching placebos, all labeled B. After stratification for sex and cardiovascular complications, eligible patients were alternatingly assigned to type A or type B medication. The first patient of each of the 4 strata always received type A medication. Active treatment was initiated with nitrendipine, with the possible addition of captopril, hydrochlorothiazide, or both drugs. The dosage steps for nitrendipine were 10 mg or 20 mg once daily in the evening, or 20 mg twice daily. For captopril and hydrochlorothiazide, the doses were 12.5 mg in the morning, or 12.5 mg or 25 mg twice daily. The study drugs were stepwise titrated and combined to reduce the sitting systolic BP by 20 mm Hg or more to less than 150 mm Hg.^1- 2,4 Drugs causing intolerable side effects were substituted by the next-line study medication. Placebos and active drugs were used in the same way. Patients who withdrew from the study medication remained in open follow-up. Patients without any report within the year before the trial ended in each center were classified as lost to follow-up, but were included in the analysis up to the most recent evaluation of their health status.^1- 2,4

Fatal and nonfatal stroke, the primary end point, and the other secondary end points were defined as in the Systolic Hypertension in Europe (Syst-Eur) trial.⁵ Stroke was clinically diagnosed as a neurologic deficit with symptoms continuing for more than 24 hours or leading to death with no apparent cause other than vascular. The diagnosis of acute myocardial infarction rested on 2 of the following 3 disorders: typical chest pain, electrocardiographic changes, or increased cardiac enzyme levels. Myocardial infarction excluded silent myocardial infarction. Congestive heart failure required the presence of 3 conditions, namely, symptoms, such as dyspnea; clinical signs, such as ankle edema or rales; and the necessity of treatment with diuretics, vasodilators, or antihypertensive drugs. Sudden death included any death of unknown origin occurring instantly or within 24 hours of the onset of acute symptoms, as well as unattended death for which no likely cause could be established by autopsy or recent medical history. Cardiac end points included fatal and nonfatal heart failure, fatal and nonfatal myocardial infarction, and sudden death.

During the course of the trials end points were logged at the coordinating office in Beijing, People's Republic of China, as reported by the investigators. In 1996 and 1997 the blinded End Point Committee validated all end points by reviewing the local patient files and other source documents, by requesting detailed written information from the investigators, or by both approaches. For the final analysis, the whole Syst-China database was transferred to Leuven, Belgium, in October 1997. The source documents certifying the stroke cases were translated into English and were again reviewed by 2 independent and blinded Syst-Eur⁶ investigators.

Diabetes mellitus at enrollment was defined according to the criteria of the World Health Organization.⁷ The diagnosis required the presence of a history of diabetes reported by the clinical investigator, a fasting blood glucose level of 7.8 mmol/L (140 mg/dL) or more, or the use of antidiabetic medication. Smokers and patients who consumed alcohol regularly were identified by questionnaire administered at enrollment.

The statistical analysis by intention-to-treat was done with a commericially available software (SAS, Version 6.12; SAS Institute Inc, Cary, NC), using 2-sided tests. Means and proportions were contrasted by the standard normal z test and the χ² statistic, respectively. Relative hazard rates were assessed by single and multiple Cox regression.⁸ Survival curves were compared using the log rank test.

In the 2394 enrolled patients the median follow-up by intention-to-treat was 3.0 years (follow-up range, 1-94 months). The number of patient-years in the placebo and active treatment groups were 2892 and 3510, respectively. At 2 years of follow-up, the sitting systolic and diastolic BP had fallen by a mean (±SD) of 11 ± 17 mm Hg and 2 ± 8 mm Hg, respectively, in the placebo group and by 20 ± 16 mm Hg and 5 ± 8 mm Hg, respectively, in the active treatment group. The differences in systolic and diastolic BP between the 2 groups then averaged 9.1 mm Hg (95% confidence interval [CI], 7.6-10.7 mm Hg) and 3.2 mm Hg (95% CI, 2.4-4.0), respectively. The analysis by intention to treat showed that both before (Table 1) and after (Table 2) adjustment for significant covariates, active treatment had reduced the incidence of total mortality (P<.01), fatal and nonfatal stroke (P<.05), and all cardiovascular end points (P<.01). Among the incident cases of stroke, the diagnosis was confirmed by computed tomographic brain scans in 12 (40%) of 30 patients with a fatal event and in 55 (72%) of 76 patients with a nonfatal stroke.

Of the enrolled patients, 223 assigned to placebo and 240 of the active treatment group did not comply with 1 or more of the eligibility criteria. Of these 463 patients (criteria are not mutually exclusive so that numbers do not add up), 43 and 110 had been recruited after only 1 or 2 run-in visits, 89 were 50 to 59 years old, 45 had grade 3 retinopathy, 189 had a sitting systolic BP from 140 to 159 mm Hg, and 91 had a sitting diastolic BP of 95 mm Hg or higher. Noncompliance with all enrollment criteria did not affect the results. For instance, for all fatal and nonfatal cardiovascular end points the relative hazard rate of active vs placebo treatment was 0.67 (95% CI, 0.48-0.94; P = .02) in the 1931 compliant patients and 0.39 (95% CI, 0.17-0.88; P = .02) in the 463 noncompliant patients. The P value for the treatment-by-compliance interaction term was .38.

Before enrollment, the patients had been prospectively stratified for sex and previous cardiovascular complications. The trial included 1541 men (64.4%), 853 women (35.6%), 269 patients with previous cardiovascular complications (11.2%), and 2125 patients without such complications (88.8%). The 4 strata were equally represented in the 2 arms of the trial. Of the 269 patients with cardiovascular complications at enrollment, 224 (9.4%) had symptoms or signs suggestive of coronary heart disease or myocardial infarction, 34 (1.4%) had a history of stroke, and 11 (0.5%) had coronary and cerebrovascular disorders.

Male sex was a significant and strong predictor of all end points. Patients with cardiovascular complications at enrollment experienced significantly more fatal and nonfatal cardiovascular end points (Table 1). With adjustments applied for the covariates listed in Table 2, these associations remained statistically significant. Further analysis showed that in general the benefits of active treatment were evenly distributed across men and women and across patients with and without previous cardiovascular complications. Only for total mortality (P = .06) and stroke (P = .07), the treatment-by-sex interaction terms approached statistical significance. For all-cause mortality, the relative hazard rates associated with active treatment were 0.74 (95% CI, 0.50-1.09; P = .12) in men and 0.34 (95% CI, 0.16-0.72; P = .005) in women; for stroke these hazard rates were 0.54 (95% CI, 0.34-0.85; P = .008) and 1.24 (95% CI, 0.56-2.74; P = .60), respectively. Thus, these treatment-by-sex interaction terms suggested that active treatment may reduce total mortality more in women and that it may prevent stroke especially in men. The P values for the other treatment-by-sex interactions ranged from .13 to .52; those for the interactions between treatment and previous cardiovascular complications ranged from .09 to .69.

Age at baseline averaged 66.5 (5.5) years (mean [SD]). In single (Table 1) and multiple (Table 2) Cox regression, age was a strong predictor of all end points with the exception of stroke. Plots of the crude relative hazard rates in 3 age strata (≤64, 65-69, and ≥70 years) suggested that for cardiovascular mortality (Figure 1) the effect of active treatment weakened in the oldest (≥70 years) age group. However, this was not the case for all fatal and nonfatal cardiovascular end points (Figure 1).

Furthermore, Cox regression with adjustments applied for the covariates listed in Table 2 demonstrated that age did not influence the effects of active treatment on outcome. Even for cardiovascular mortality (Figure 2), the relative hazard rate for the treatment-by-age interaction term was only 1.03 (95% CI, 0.95-1.11; P = .48); for the other end points the P values of these interaction terms ranged from .45 to .83.

At baseline, the sitting BP averaged 170.5 ± 11.1 mm Hg systolic and 86.0 ± 6.8 mm Hg diastolic BP. In single (Table 1) and multiple (Table 2) regression, all end points with the exception of fatal and nonfatal stroke were positively correlated with systolic BP. Plots of the crude relative hazard rates in 3 strata according to the systolic BP at enrollment (≤169, 170-179, and ≥180 mm Hg) did not show a consistent trend (Figure 1). After adjustment for the covariates listed in Table 2, the interaction terms between treatment and systolic BP were nonsignificant for all end points (Figure 2); the P values ranged from .16 to .78.

Both before and after adjustment (Table 1 and Table 2), a higher diastolic BP was associated with a lower incidence of end points. The latter associations were not influenced by treatment status; after adjustment, the P values for the interactions between treatment and diastolic BP ranged from .27 to .99.

Of the 2394 patients, 1579 (66.1%) were residents of northern China. Compared with 815 (33.9%) southern Chinese patients, the northern Chinese were at significantly greater risk of stroke both before (Table 1) and after (Table 2) adjustment for the covariates listed in Table 2.However, with these adjustments applied, there was no statistically significant difference between northern and southern Chinese patients in the effects of treatment on stroke prevention (P = .92). This was also the case for all other end points, for which the P values of the treatment-by-residence interaction terms ranged from .42 to .58.

At baseline, 737 patients (30.8%) (626 men [85%] and 111 women [15%]) smoked. In single regression (Table 1), smoking predicted outcome except for all cardiac end points. The proportion of smokers was similar among patients who suffered a cardiac end point and those who did not (29.1% vs 30.8%). After adjustment for the covariates listed in Table 2, smoking was still a significant predictor of fatal and nonfatal stroke. With adjustments for the covariates in Table 2 applied, Cox regression showed a significant interaction between treatment and smoking for all cardiac end points (P = .04). The relative hazard rate for active vs placebo treatment was 0.49 (95% CI, 0.25-0.96; P = .04) in nonsmokers, but it was 1.38 (95% CI, 0.51-3.71; P = .52) in smokers.

At baseline, 748 men (48.5%) and 89 women (10.4%) reported regular alcohol consumption. In single regression (Table 1), alcohol intake was positively correlated with the incidence of stroke and all fatal and nonfatal cardiovascular end points. However, after adjustment for significant covariates, these associations weakened to a nonsignificant level (P values ranging from .35 to .95). In multiple Cox regression, the P values for the interaction terms between treatment and drinking alcohol ranged from .19 to .87.

Of 1253 patients allocated to active treatment, 735 remained on monotherapy with nitrendipine and 518 proceeded to twice daily dosing; the mean daily doses of nitrendipine in these 2 groups were 13.8 mg and 36.9 mg, respectively. Compared with the whole placebo group and with adjustments applied for the covariates listed in Table 2, the benefit of active treatment tended to be slightly larger in patients receiving twice daily dosing (Figure 3). However, only for cardiac end points the P value for the difference between the 2 dosing schemes of active nitrendipine reached statistical significance; for the other end points the P values ranged from .14 to .65.

At enrollment 55 men and 43 women had diabetes mellitus. Of these 98 patients (4.1%), 21 had a history of diabetes mellitus and 90 had a fasting blood glucose level of 7.8 mmol/L (140 mg/dL); 2 patients were on a diabetic diet and 10 were taking oral antidiabetic drugs. In diabetic patients, compared with the 2296 nondiabetic patients, body mass index (calculated by the weight, in kilograms, divided by the height, in meters, squared) (25.4 vs 23.9, P<.001), blood glucose level (9.7 vs 5.3 mmol/L [174.7 vs 95.5 mg/dL]; P<.001), serum total cholesterol level (5.7 vs 5.1 mmol/L [220 vs 197 mg/dL]; P = .005) and systolic BP at enrollment (172.5 vs 170.2 mm Hg; P = .14) were higher. However, the percentage of patients with previous cardiovascular complications was similar in both groups (12.2% vs 11.2%; P = .75).

Diabetic and nondiabetic patients were treated similarly and their BP decreased to a similar extent (Table 3). At 2 years the net differences between the placebo and active treatment groups were 6.0 mm Hg systolic BP (95% CI, 1.5-13.5 mm Hg; P = .12) and 4.7 mm Hg diastolic BP (95% CI, 0.4-9.1 mm Hg; P = .04) in the diabetic patients; in the 2296 nondiabetic patients, these differences were 9.3 mm Hg (95% CI, 7.7-10.8 mm Hg; P<.001) and 3.1 mm Hg (95% CI, 2.3-3.9 mm Hg; P<.001), respectively. The P values for the differences in the net BP effects between diabetic and nondiabetic patients were .40 systolic and .47 diastolic.

After adjustment for the covariates listed in Table 2, active treatment showed a consistent trend in diabetic patients (Figure 4) in reducing total mortality (59%; P = .15), cardiovascular mortality (57%; P = .22), all cardiovascular end points (74%; P = .03), fatal and nonfatal stroke (45%; P = .42) and all cardiac end points (90%; P = .08). In the nondiabetic patients (Figure 4) active treatment significantly decreased total mortality (36%; P = .01) and all cardiovascular end points (34%, P = .01), while the same tendency was also observed for cardiovascular mortality (33%, P = .10) and total stroke (32%; P = .06). The P values of the interaction terms between active treatment and the presence of diabetes mellitus ranged from .15 to .88.

The incidence of all cardiovascular end points was also calculated separately for patients who continued to receive nitrendipine monotherapy and for those patients who had any combination regimen of nitrendipine, captopril, or hydrochlorothiazide (or matching placebos). In the placebo group (Table 3), the cardiovascular event rates in diabetic patients were approximately twice as high as in nondiabetic patients, regardless of the treatment regimen. In the actively treated diabetic patients, the excess cardiovascular risk was reduced to the rates observed in the nondiabetic patients; this was equally true in patients treated with only nitrendipine as in those on other treatments (Table 3).

Finally, within-group analyses (Table 4) showed that in the placebo group diabetes mellitus significantly enhanced the risk of all end points, with relative hazard rates ranging from 2.23 for all-cause mortality to 3.36 for cardiovascular mortality. In the active treatment group the risk associated with diabetes mellitus was reduced to nonsignificant levels for total mortality and for the combined fatal and nonfatal end points; the relative hazard rates ranged from 0.78 to 2.03.

A major finding of this study was that in the within-group analysis diabetes at enrollment significantly raised the risk of all types of end points in the placebo group, whereas in the active treatment group the CIs of the corresponding relative hazard rates were wide and included unity except for cardiovascular mortality. Cox regression in all patients combined failed to show significant treatment-by-diabetes interaction terms, because the diabetic patients represented only 4.1% of the whole trial population. However, in terms of clinical relevance, our findings suggest that both the relative and the absolute benefit of active treatment were greater for diabetic than for nondiabetic patients. Indeed, the point estimate for the prevention of all cardiovascular end points was 59% in the diabetic patients and only 33% in the nondiabetics; in terms of absolute benefit, at the rates observed in the placebo group, treating 1000 patients for 5 years, could prevent 221 major cardiovascular end points in diabetic patients, but only 51 in the nondiabetic patients.

The Syst-China trial findings must also be interpreted considering the prior probability of a greater treatment effect in hypertensive diabetic patients as observed in the Syst-Eur trial.^6,9 In the latter study 492 diabetic patients (10.5%) and 4203 nondiabetic patients (89.5%) were randomized to placebo or to active treatment initiated with the dihydropyridine nitrendipine (10-40 mg/d) with the possible addition of enalapril maleate (5-20 mg/d), hydrochlorothiazide (12.5-25.0 mg/d), or both drugs. In diabetic Syst-Eur patients, with adjustments for possible confounders applied, active treatment reduced all-cause mortality by 55%, cardiovascular mortality by 76%, all cardiovascular end points by 69%, fatal and nonfatal stroke by 73% and all cardiac end points by 63%; in the nondiabetic Syst-Eur patients, active treatment decreased all cardiovascular end points by 26% and fatal and nonfatal stroke by 38%.⁹ Active treatment reduced total mortality (P = .04), cardiovascular mortality (P = .02) and all cardiovascular endpoints (P = .01) significantly more for diabetic than for nondiabetic Syst-Eur patients. In terms of absolute benefit, the Syst-Eur trial showed that active treatment, per 1000 patients treated for 5 years, could prevent 178 major cardiovascular end points for diabetic patients, and 39 for nondiabetic patients.⁹

The Systolic Hypertension in the Elderly Program (SHEP)¹⁰ included 4736 men and women, aged 60 years or older, with isolated systolic hypertension (systolic BP 160-219 mm Hg and diastolic BP <90 mm Hg). At baseline, 583 patients (12.3%) had non–insulin-dependent diabetes mellitus, 4149 patients did not have diabetes, and 4 patients could not be classified.¹¹ In the SHEP control group, the rate of all cardiovascular complications was 63.0 events per 1000 patient-years in the diabetic patients (n = 300) and 36.8 events per 1000 patient-years in nondiabetic patients (n = 2069); in the diabetic (n = 283)and nondiabetic (n = 2080) patients randomized to active treatment (chlorthalidone, 12.5-25 mg/d, with the possible addition of atenolol, 25-50 mg/d, or reserpine, 0.05-0.1 mg/d¹⁰), these rates were reduced to 42.8 and 26.6 events per 1000 patient-years, respectively.¹¹ Thus, in the SHEP trial, active treatment decreased the incidence of cardiovascular complications to the same extent (34%) in diabetic patients (95% CI, −54% to −6%) and in nondiabetic patients (95% CI, −45% to 21%).¹¹ In the Hypertension Detection and Follow-up Program (HDFP),^12- 13 the stepped-care treatment regimen also started with a diuretic (chlorthalidone, 25-100 mg/d). In all, 1079 (9.9%) of the HDFP participants showed evidence of diabetes at enrollment. In the 9861 nondiabetic patients (90.1%), the mortality rate was significantly lower in the stepped-care (intervention) group than in the referred-care (control) group; the relative benefit amounted to 17.4% (60.1 vs 72.8 deaths per 1000 patient-years).¹⁴ In contrast, in the diabetic HDFP patients there was little (0.8%) difference between the death rates in the stepped-care and referred-care groups (105.8 vs 106.6 deaths per 1000 patient-years).¹⁴

In diabetic and nondiabetic patients, cardiovascular benefit was equally observed in the patients continuing to receive monotherapy with nitrendipine and in those progressing to combined treatment with nitrendipine plus captopril, hydrochlorothiazide, or both drugs. These findings are in keeping with the Syst-Eur report,¹⁵ in which 1327 patients remaining on monotherapy with nitrendipine were matched by sex, age, previous cardiovascular complications, and systolic BP at enrollment with an equal number of placebo patients. In this analysis,¹⁵ nitrendipine therapy reduced (P ≤.05) cardiovascular mortality by 41%, all cardiovascular end points by 33%, and fatal and nonfatal cardiac end points by 33%. Furthermore, the results of 2 independent trials conducted in different populations, ie, Syst-China and Syst-Eur,^6,9,15 when compared with the SHEP^10- 11 and HDFP^12- 14 findings, suggest that in hypertensive diabetic patients treatment initiated with a dihydropyridine calcium channel blocker may provide better cardiovascular protection than therapy starting with a thiazide. This may be due to the absence of metabolic side effects,^16- 17 such as glucose intolerance and perturbation of the serum lipid profile, to which diabetic patients treated with thiazides may be particularly vulnerable.¹⁸ In addition, calcium channel blockers may provide renal protection,^19- 20 and may have beneficial effects on the rheologic characteristics of the blood³ and on endothelial function.^21- 22 However, the hypothesis that in diabetic patients with hypertension calcium channel blockers may provide better cardiovascular protection than thiazides must be further investigated. The ongoing Antihypertensive and Lipid Lowering Treatment to Prevent Heart Attack Trial (ALLHAT) will be informative in this regard, because nearly one third of the ALLHAT patients have diabetes.^23- 24 The ALLHAT is comparing a calcium channel blocker (amlodipine), an angiotensin-converting enzyme inhibitor (lisinopril), and an α-adrenergic-receptor blocker (doxazosine) with a diuretic (chlorthalidone) for effects on the risk of nonfatal myocardial infarction or fatal coronary heart disease among high-risk patients with hypertension who are older than 55 years.²⁴ The ALLHAT Data and Safety Monitoring Board reviews outcome data periodically. One review occurred during fall 1997 and included a separate evaluation of the primary end point in the subgroup with diabetes. This analysis involved more than 7000 patient-years. The ALLHAT committee recommended that the trial should continue according to the protocol and did not accelerate the date of its next review of the data.²⁵

The prevalence of diabetes in the Syst-China trial (4.1%) was approximately only one third of that observed in the SHEP¹¹ and Syst-Eur trials.⁹ In the 1980s^26- 27 and the early 1990s²⁸ studies from the Chinese Mainland showed a low prevalence of diabetes, ranging from 0.15% in Guizhou and 0.33% in Guangdong to almost 1% in cities such as Beijing, Shanghai, and Ningxia.^26- 27 More recent epidemiological data demonstrated age-adjusted prevalence rates of 3.6% for diabetes and 4.2% for impaired glucose tolerance in 213,515 subjects aged 25 to 64 years.²⁸ These numbers confirm the rapid rise in diabetes that has occurred in China in the past 10 years²⁸ and underscore the possible implications of the present findings for the treatment of hypertensive Chinese patients with type 2 diabetes.

The present analysis identified male sex and previous cardiovascular complications as significant and independent cardiovascular risk factors. Stratification for these characteristics made it possible to test a priori whether the effects of treatment were similar in these 4 strata. Cox regression with adjustments applied for significant covariates, suggested that active treatment may reduce total mortality more (P = .06) in women, but that it may prevent stroke better (P = .07) in men. Whether these findings are influenced by the higher inclusion of men (≈66%) in the Syst-China study than in the other trials in isolated systolic hypertension (≈33%)^5,10,29 is uncertain. However, in Syst-China and in the 2 other trials in isolated systolic hypertension^5,10,29 men and women and patients with and without previous cardiovascular complications, irrespective of race, all benefited from antihypertensive drug treatment.

The question whether the effects of antihypertensive drug treatment are preserved in very old patients has important public health implications in view of the increasing longevity of populations worldwide. In our study the treatment-by-age interaction was not significant. These findings agree with those in 650 patients, aged 80 years or older, randomized in the SHEP trial,^10,29 in whom the relative risk of stroke on active treatment compared with placebo was 0.51 (95% CI, 0.29-0.89). In contrast, the Syst-Eur trial demonstrated that antihypertensive drug treatment did not postpone death in patients older than 75 to 80 years, but still prevented nonfatal cardiovascular complications in the very old. In the trial conducted by the European Working Party on High Blood Pressure in the Elderly (EWPHE), a significant (P = .05) treatment-by-age interaction suggested that cardiovascular mortality decreased less on active treatment in the oldest patients, especially beyond 80 years.³⁰ In the Swedish Trial in Old Patients with Hypertension (STOP-Hypertension),³¹ older patients randomized to active treatment, compared with the control group, experienced less reduction of stroke, myocardial infarction, or other cardiovascular death. The upper 95% confidence limit of the relative risk crossed unity even by 73 years.³¹ In view of the remaining uncertainty, the Hypertension in the Very Elderly Trial (HYVET) investigates the potential benefit of antihypertensive drug treatment on stroke and other cardiovascular end points in patients older than 80 years.³² Until this trial report, clinicians may consider the possible reduction of nonfatal stroke as the main argument to prescribe antihypertensive agents to very old patients, in whom this complication often leads to institutionalization.

The finding that an adverse outcome was positively correlated with systolic BP, but negatively with diastolic BP strongly suggests that pulse pressure, the difference between systolic and diastolic BP, is an important cardiovascular risk factor in elderly Chinese patients. Other studies^33- 35 also found that pulse pressure behaves as an independent cardiovascular risk factor. Furthermore, the benefits of antihypertensive drug treatment were observed regardless of the BP level at enrollment. In the Syst-Eur trial, total mortality declined slightly more on active treatment when the initial systolic BP was higher (P = .05).⁵ Statistical analyses of the EWPHE study³⁰ and the Medical Research Council trial,³⁶ which controlled for possible confounders, such as sex, age and previous cardiovascular complications, did not reveal any interaction between the BP level at enrollment and the benefits derived from treatment. The SHEP investigators^10,29 reported that in proportional hazard regression, using systolic BP as a continuous variable, the favorable trend in stroke incidence for the active treatment group compared with the placebo group prevailed irrespective of the baseline systolic BP. The P value for interaction was .13.^10,37 This conclusion was reinforced by similar findings for nonfatal myocardial infarction combined with death from coronary heart disease (P = .85), for coronary heart disease (P = .81), and for all cardiovascular complications (P = .44).³⁷

In our analysis, antihypertensive drug treatment similarly decreased cardiovascular risk in northern and southern Chinese patients. The finding that the occurrence of stroke was 60% higher in northern than in southern China is consistent with a comprehensive overview³⁸ of the geographical variations in the incidence and the risk factors of stroke in China. The benefit of active treatment tended to be greater in patients who took nitrendipine twice daily as compared with those who took the drug only once daily. However, these findings must be cautiously interpreted. The Syst-China trial was not designed as a dose-finding study so that the effects of twice daily dosing and those of the higher daily dose of nitrendipine administered in this way could not be separated in the analysis. Moreover, the differences between once and twice daily dosing were not significant, or for cardiac end points were only marginally significant.

Active treatment significantly decreased all cardiac end points in the nonsmokers (51%), but it did not reduce the incidence of cardiac complications in the smokers. In view of the small number of all cardiac end points, this finding may reflect random variability or confounding in a subgroup analysis. Smoking has been shown to be a risk factor for total and cardiovascular mortality and for all fatal and nonfatal cardiovascular and cardiac end points in the Syst-Eur trial, for stroke in the Medical Research Council Mild Hypertension trial,³⁶ and for cardiovascular disease in the Australian Therapeutic Trial in Mild Hypertension.³⁹ In our study, smoking increased the stroke risk by 64%. Chinese observational studies showed that passive smoking at home and at work is a risk factor for cardiovascular disease in Chinese women who never smoked.^40- 41

Some studies found a positive correlation between alcohol intake and stroke, particularly hemorrhagic stroke.^42- 43 However, prospective studies have demonstrated a U-shaped relationship,^44- 45 with nondrinkers having a higher relative risk of ischemic stroke than moderate drinkers (4-15 g/d), whereas the risk of ischemic stroke increased again in heavy drinkers. In contrast, for hemorrhagic stroke, there was an increased risk at all levels of alcohol intake.⁴⁵ A prospective Chinese study in 18,233 men, aged 45 to 64 years, showed that regular consumption of small amounts of alcohol was associated with lower overall mortality including death from ischemic heart disease, but had no effect on fatal stroke.⁴⁶ Furthermore, a study in elderly (≥65 years) Taiwanese reported that after adjustment for confounders alcohol consumption did not increase the risk of stroke.⁴⁷

In elderly Chinese patients with isolated systolic hypertension, stepwise antihypertensive drug treatment, starting with the dihydropyridine calcium channel blocker nitrendipine, improves prognosis. The benefit was particularly evident in diabetic patients; for cardiac end points it tended to be larger in nonsmokers. Otherwise, the benefit of active treatment was not significantly influenced by the characteristics of the patients at enrollment in the study.

Accepted for publication April 14, 1999.

The Syst-China Trial was supported by the State Planning Commission of the People's Republic of China. Dr Wang was the recipient of a fellowship at the University of Leuven, Belgium, sponsored by Bayer AG, Wuppertal, Germany.

The Syst-China trial was a research program conducted under the auspices of State Commission of Science and Technology and the Public Health Ministry of the People's Republic of China. The trial was carried out in consultation with the World Health Organization and the World Hypertension League. Nitrendipine and matching placebo tablets were purchased from Xinhua Pharmaceuticals at Shijiazhuang, Hebei, China. Captopril and hydrochlorothiazide and their matching placebo tablets were purchased from Changzhou Pharmaceuticals, Jiangsu, China. Collaboration between the Syst-China and the Syst-Eur investigators was established by the late Professor A. Amery, MD,who died November 2, 1994.

Writing Group

Willem H. Birkenhäger, MD; Manyi Bai, MD; Hilde Celis, MD; Robert Fagard, MD; Hong Ge, MD; Lansheng Gong, MD; Zhaoguang Hong, MD; Guozhang Liu, MD; Lisheng Liu, MD; Yuhui Liu, MD; Xinwei Pan, MD; Deren Qiao, MD; Ming Sun, MD; Jan A. Staessen, MD; Lutgarde Thijs, BSc; Ji-Guang Wang, MD; Xianyan Wang, MD; Hongxiu Zhang, MD; Shantong Zhang, MD; Yuqing Zhang, MD.

Clinical Investigators in the People's Republic of China

Anhui: Minming Bi, MD; Xuekui Chen, MD; Tongyuan Gu, MD; Yiqin Xie, MD (Anhui Medical Institute).

Beijing: Hong Ge, MD; Fangyi Qian, MD; Anzhen Hospital: Jing Li, MD; Juan Zhang, MD; Xianghao Zhou, MD (514 Hospital); Lisheng Liu, MD; Guozhang Liu, MD (Cardiovascular Institute and Fuwai Hospital); Ji-Guang Wang, MD; Dinghua Yuan, MD; Yuqing Zhang, MD; Zhiniu Song, MD; Chaoyuan Yang, MD (Beijing Hospital and Geriatric Institute); Huanwen Wang, MD; Liangchun Zhang, MD (Jiuxianqiao Hospital); Shuyu Wang, MD; Xuehai Yu, MD; Shuhe Zhang, MD; Yuqing Zhang, MD; Juan Zhou, MD(Capital Iron & Steel Complex Hospital).

Fujian: Xizhong Hu, MD; Ling Yu, MD (Fujian Cardiovascular Institute).

Guangdong: Ping Huang, MD; Peixiong Li, MD (Guangdong Geriatric Institute).

Guiyang: Dingsheng Wang, MD; Zong Yao, MD; Deyin Zhang, MD; Meixiang Zhang, MD (Guiyang Medical College).

Hebei: Shaoru Feng, MD; Deren Qiao, MD; Shouling Wu, MD; Jiufen Zhao, MD (Tangshan Kailuan Hospital); Zhongying Guo, MD; Yinzhu Li, MD; Hongxiu Zhang, MD (Hebei Academy of Medical Sciences).

Heilongjiang: Yinghua Hu, MD; Jixing Wang, MD (Daqing First Hospital).

Henan: Yangchun Chen, MD; Zhengfu Guo, MD; Jiazhen Lu, MD; Yi Xu, MD; Yuqin Yao, MD (Henan Traditional Medicine Institute); Renyu Liu, MD; Xifu Yan, MD; Wenxiu Yang, MD; Yufang Zhang, MD (Henan Medical University).

Hunan: Ming Sun, MD; Weiyi Tang, MD; Xiumei Xie, MD; Hongyan Zhou, MD (Hunan Medical University Xiangya Hospital); Zhonglin Wang, MD (Hunan Medical University 2nd Hospital).

Jiangsu: Li Chen, MD; Wenping Jiang, MD; Mingde Li, MD; Haiying Tang, MD; Dan Ye, MD (Suzhou Medical College); Minglong Chen, MD; Decheng Lu, MD; Zijie Su, MD; Guanghui Sun, MD (Zhenjiang Medical College); Fuchang Du, MD; Haiyan Wang, MD; Shiqing Zhang, MD; Jie Zu, MD (Nanjing Medical University).

Jiangxi: Guifen Liu, MD; Yidao Long, MD; Zuanzhen Wang, MD; Luyuan Yao, MD (Jiangxi Medical College).

Jilin: Guiwen Che, MD; Tongku Liu, MD; Bowen Wang, MD; Lihua Xu, MD (Jilin Medical College); Mei Fang, MD; Li Zhao, MD (Baiqiuen Medical University).

Liaoning: Xiaoou Cheng, RN; Yunxian Wang, MD; Zhongzhen Yang, MD; Peiyong Zhang, MD (Anshan Changdian Hospital); Zhaoji Han, MD; Henglu Zheng, MD; Yutian Zhou, MD (Fushun Steel Factory Hospital); Guixi Li, MD; Guirong Liu, MD; Fusheng Zhou, MD (Dashiqiao Liaomei Company Hospital); Yuhui Liu, MD; Shujiao Wang, MD; Yongbin Yan, MD; Xuying Ye, MD (Dalian Medical University 2nd Hospital).

Shandong: Shuyu Du, MD; Ruiguo Jiang, MD; Jingchun Ruan, MD; Shantong Zhang, MD (Shandong Medical Academy).

Shanghai: Haozhu Chen, MD; Jianping Fang, MD; Xinwei Pan, MD; Daidi Yu, MD (Shanghai Cardiovascular Institute); Lansheng Gong, MD; Pingmei Gu, MD; Xuemei Luo, MD; Luode Wang, MD; Xianyan Wang, MD; Xiaoming Wang, MD; Shundi Wu, MD (Shanghai Hypertension Institute).

Sichuan: Peifen Cheng, MD; Zhixiang Huang, MD; Ruilian Wen, MD; Tingjie Zhang, MD (Chengdu No. 1 Hospital).

Committees and Coordination

Trial Coordinators: Lansheng Gong, MD; Lisheng Liu, MD.

Data Monitoring Committee: Guozhang Liu, MD; Ji Guang Wang, MD; Xianyan Wang, MD; Dinghua Yuan, MD; Yuqing Zhang, MD.

Syst-China/Syst-Eur Liason Committee: Willem H. Birkenhäger, MD; Christopher J. Bulpitt, MD; Hilde Celis, MD; Robert H. Fagard, MD; Astrid E. Fletcher, PhD; Lansheng Gong, MD; Lisheng Liu, MD; Jan A. Staessen, MD; Lutgarde Thijs, BSc; Ji-Guang Wang, MD.

Drug Committee: Chenghua Dong; Guozhang Liu, MD; Xianyan Wang, MD.

End Point Committee: Yuhua Dai, MD; Erjuan Feng, MD; Zhaoguang Hong, MD; Shouyue Pu, MD; Wenhang Qi, MD; Lihui Wang, MD; Xinde Wang, MD; Zhitong Wang, MD; Wu Zhaoshu, MD; Zhao Zhongyan, MD; Zhu Zhaoyi, MD (Beijing); Hilde Celis, MD; Robert H. Fagard, MD.

Ethics Committee: Runquan Gao, PhD; Lansheng Gong, MD; Guanqing He, PhD (deceased in 1995); Lisheng Liu, MD; Ming Sun, MD; Xuying Ye, MD.

Publication Committee: Lansheng Gong, MD; Guozhang Liu, MD; Lisheng Liu, MD; Xianyan Wang, MD.

Steering Committee: Haozhu Chen, MD; Xuekui Chen, MD; Yangchun Chen, MD; Hong Ge, MD; Xizhong Hu, MD; Yinghua Hu, MD; Wenping Jiang, MD; Peixiong Li, MD; Yuhui Liu, MD; Xinwei Pan, MD; Fangyi Qian, MD; Deren Qiao, MD; Ming Sun, MD; Bowen Wang, MD; Shuyu Wang, MD; Xianyan Wang, MD; Zhonglin Wang, MD; Xifu Yan, MD; Luyuan Yao, MD; Zong Yao, MD; Xuying Ye, MD; Hongxiu Zhang, MD; Shiqing Zhang, MD; Shantong Zhang, MD; Tingjie Zhang, MD.

Coordinating Office: Manyi Bai, MD; Aimin Dang, MD; Chenghua Dong; Bing Duan; Weiqing Fang; Ming Jian, MD; Liping Li; Guozhang Liu, MD; Lisheng Liu, MD; Huali Ma, RN; Ji Guang Wang, MD; Dinghua Yuan, MD; Yuqing Zhang, MD.

Reprints: Jan A. Staessen, MD, PhD, Syst-Eur Coordinating Office, Laboratory of Hypertension, Department of Molecular and Cardiovascular Research, Campus Gasthuisberg, Herestraat 49, B-3000 Leuven, Belgium (e-mail: jan.staessen@med.kuleuven.ac.be).