It is generally accepted that hypertension is associated

1605 Hypertension and Risk of Stroke Recurrence Milton Alter, MD, PhD; Gary Friday, MD; Sue Min Lai, PhD; John O'Connell, MD; Eugene Sobel, PhD Background and Purpose Hypertension is a risk factor for initial stroke, but its relation to stroke recurrence is unclear. Therefore, we sought to analyze the effect of hypertension and its control on risk of stroke recurrence. Methods Within 1 month of onset, a population-based cohort of 662 patients from the Lehigh Valley with an initial stroke were enrolled. Hypertension was determined at enrollment by history. Blood pressure was also measured at enrollment and at each follow-up at 4- to 6-month intervals for up to 48 months (mean, 24 months). Stroke recurrence was verified by history, examination, and review of medical reports. Various criteria for control of blood pressure were defined. History of hypertension, measured blood pressure, and its control were analyzed in relation to stroke recurrence frequency using Kaplan-Meier and univariate, multivariate, and time-dependent Cox proportional hazards models. Results At enrollment, 59.4% of the cohort had a history of hypertension and 81 patients had a second stroke. Those with a history of hypertension had a significantly higher stroke recurrence rate than those without such a history (P=.01). Among those with measured diastolic blood pressure at enrollment ^95 mm Hg, 43% had a stroke recurrence by the end of the study compared with only 19% below this cutoff (P=.005). Recurrence risk was reduced in a multivariate analysis as quality of diastolic blood pressure control increased (relative risk=8.4, 3.9, and 2.0 among those with poor, fair, and good control, respectively, compared with nonhypertensive subjects). Systolic blood pressure and its control appeared less or not significantly associated with stroke recurrence. Condusions History of hypertension and elevated measured diastolic pressure after the initial stroke were associated with an increased risk of second stroke. Controlling diastolic pressure substantially reduced this risk. (Stroke. 1994^5:1605-1610.) Key Words epidemiology hypertension risk factors stroke prevention It is generally accepted that hypertension is associated with increased risk of initial stroke and that control of hypertension reduces this risk. However, it is unclear that the same associations hold for recurrent stroke. Although in some case series control of blood pressure reduced recurrent stroke, 14 no reduction in stroke recurrence was noted when hypertension was controlled in two population-based studies. 5 ' 6 We investigated the association of hypertension and recurrent stroke using the Lehigh Valley Stroke Cohort. Our study was based on patients admitted to the hospital with an initial stroke and was specifically designed to determine the frequency of stroke recurrence prospectively. We also studied the effect of blood pressure control on risk of stroke recurrence and took into account comorbidities, time since initial stroke, and demographic variables such as age and sex in calculating stroke recurrence rates. Subjects and Methods Subjects were identified from among individuals admitted with a diagnosis of stroke to any of the eight hospitals in the Lehigh Valley. This region in northeastern Pennsylvania includes approximately 600 000 inhabitants. Using this popula- Received January 21, 1994; final revision received May 4, 1994; accepted May 4, 1994. From the Division of Neuroepidemiology, Medical College of Pennsylvania, Philadelphia (M.A., G.F., J.O.); the Department of Preventive Medicine, The University of Kansas, Wichita (S.M.L.); and the Department of Biometry and Preventive Medicine, University of Southern California, Los Angeles (E.S.). Reprint requests to Milton Alter, MD, PhD, Medical College Hospitals, Main Clinical Campus, Division of Neuroepidemiology, 3300 Henry Ave, Philadelphia, PA 19129. O 1994 American Heart Association, Inc. tion, we have previously reported information on the epidemiology of stroke 7 as well as on stroke risk factors. 8-9 The patients in the present study were admitted to the hospital between July 1987 and August 1989. Those whose medical history, examination, and computed tomogram (CT) confirmed that there was no prior stroke were diagnosed as having an acute initial stroke and were invited to enroll. Reasons for exclusion based on an analysis of a consecutive sample of 200 screened subjects who were not enrolled have been described elsewhere. 10 We enrolled 692 subjects with initial stroke within 4 weeks of onset (93% within 2 weeks). After excluding 8 patients subsequently found not to have been eligible and 22 who died during the initial hospital admission from causes other than a recurrent stroke, 662 patients constituted the study cohort, of whom 621 had an ischemic stroke. We followed up this group and analyzed the stroke recurrence rate in relation to history of hypertension, measured blood pressure in the hospital, and measured blood pressure during follow-up. Trained study nurses interviewed patients or surrogates and reviewed hospital records and attending physician reports to determine whether there was a history of hypertension at the time of enrollment. History of hypertension was used for some analyses because measured blood pressure may be transiently elevated after a stroke. However, only measured blood pressure at enrollment and during follow-up was used to determine whether the blood pressure was controlled. Age, sex, stroke type, and presence of selected comorbidities at enrollment were also noted. These comorbidities included myocardial infarction (MI) by electrocardiogram (ECG), cardiac arrhythmia by ECG, diabetes mellitus (DM) by history, and transient ischemic attack (TLA) by history. The type of stroke was diagnosed using a modification of Stroke Data Bank criteria as described elsewhere. 10 Regular follow-up visits at home or nursing home by a study nurse were conducted first at 4 months and subsequently at about 6-month intervals until July 1991 for up to seven visits or

1606 Stroke Vol 25, No 8 August 1994 until a second stroke occurred or the patient died, left the area, or refused follow-up. At each visit the patient's blood pressure was measured. The method involved three consecutive measurements with a cuff mercury sphygmomanometer with the patient seated. The last two readings were averaged. The nurses who measured blood pressure were certified by the Pennsylvania High Blood Pressure Control Program of the Health Promotion Council of Southeastern Pennsylvania Department of Health. In this study several criteria were used to define hypertension: a measured systolic blood pressure a 150 mm Hg for those <65 years of age or 160 mm Hg for those 265 years of age was considered hypertensive; alternatively, a measured blood pressure >150 mmhg systolic for all patients was used. For diastolic blood pressure, 295 mm Hg or, alternatively, 290 mm Hg was used to define diastolic hypertension. History of a stroke recurrence or TLA was queried, and the patient was examined at each visit by the trained study nurses. If any new neurological signs were suspected, the patient was queried and reexamined by a study neurologist (G.F. or MA.). In addition, all available interim medical information was reviewed, including hospital records, CT, and assessments by physicians. Patients whose new signs and symptoms were confirmed by the study neurologist as compatible with a new stroke were listed as having had a stroke recurrence. Information regarding comorbidities (MI, arrhythmia, DM, TLA.) and an ECG were also obtained by the nurse during the initial and all subsequent visits using standardized methods described elsewhere. 10 These data were reviewed by a neurologist (G.F.). For patients known to be diabetic or if symptoms suggestive of DM were elicited on interview during follow-up, blood for a grycosylated hemoglobin assay was obtained. In Table 1 we show the demographic characteristics, medical comorbidities, and stroke types for the cohort of 662 discharged from the hospital after their initial stroke and for the 537 with at least one follow-up visit. A multifactorial analysis of these data has been reported, using age, sex, and comorbidities as variables. 11 Stroke recurrence frequency in patients with an ischemic stroke who were hypertensive at enrollment by history was compared with that in patients with an ischemic stroke who were not hypertensive by history using Kaplan-Meier survival curves. We also analyzed differences in stroke recurrence in the entire cohort using measured blood pressure and various cutoff points to define hypertension as described above, again using the Kaplan-Meier survival curves. "Survival" was defined as the interval from the initial stroke until stroke recurrence for as long as the subject remained in the study. Significance of differences in survival among those above and below the cutoff in these analyses was determined using the log-rank test. In addition to analyzing stroke recurrence frequency based on blood pressure status by history and by measured blood pressure at enrollment, we also attempted to determine the effect of controlling blood pressure during follow-up. Several levels of control were established. Those who were hypertensive by history at enrollment but had a measured blood pressure ^150 mm Hg (160 mm Hg if aged 265 years) or a measured diastolic pressure <95 mm Hg on 75% or more of follow-up visits were considered to be in good control. Those with measured blood pressure under the cutoff levels as defined above during follow-up on 50% to 74% of visits were considered to be in fair control; if measured blood pressure was below the cutoffs on less than 50% of follow-up visits, control was considered to be poor. Patients with no follow-up visits were censored, and these Kaplan-Meier curves were based on 537 patients with at least one follow-up visit. Since the chance of being in control using the above method could be affected by length of follow-up, we also performed a time-dependent, Cox proportional hazards analysis taking the last recorded systolic and diastolic blood pressure as the most TABLE 1. Demographic Characteristics, Comorbidities, and Stroke Type at Enrollment In the Lehlgh Valley Stroke Cohort Age at onset, y (mean) Men Women Male sex Race White Black Other Medical comorbidities Hypertension Myocardial infarction Cardiac arrhythmia Diabetes mellitus Transient ischemic attack Stroke type at enrollment Thrombosis Embolus Lacune Nonspecific infarct Intracerebral hemorrhage Entire Cohort (n=662) 69.6±10.6 74.3±11.1 51.4 97.2 2.0 0.8 59.4 25.1 46.8 29.3 17.8 14.4 22.8 8.8 47.9 6.2 Values (except age) are percentages. Cohort With at Least One Follow-up Visit (n=537) 69.1 ±10.2 73.6±10.8 51.0 97.2 2.0 0.8 59.0 23.5 46.2 27.6 16.8 14.7 22.5 8.4 48.4 current blood pressure in the analysis of risk of recurrent stroke. The independent effect of blood pressure control was tested using this model with forward selection taking into account age, sex, and comorbidities at enrollment by history (DM and TLA) or by ECG (MI and arrhythmia). This analysis was performed using the 537 patients with at least one follow-up visit. Results In this cohort, history of hypertension was present at enrollment in 59.4%. In addition, we were able to measure blood pressure in the hospital on all but 6 patients (99%). The distribution of measured systolic and diastolic blood pressure at enrollment in this cohort is shown in Fig 1. Using measured diastolic blood pressure of 95 mm Hg as a criterion, 2.6% of the cohort was hypertensive in the hospital, whereas when &90 mm Hg was used, 7.2% of the cohort was hypertensive. For systolic blood pressure, 23.3% had hypertension using a blood pressure of s 150 mm Hg as the criterion, and 13.0% were hypertensive using the criterion of ^150 mm Hg for those <65 years of age and 2160 mm Hg for those >65 years of age (Table 2). During the study there were 138 (20.8%) deaths, 29 (4.4%) who moved out of the study area, and 70 (10.6%) who refused to be followed up. The follow-up period averaged 24 months (range, 1 to 48 months), and during this period a total of 81 (12%) of the 662 enrolled had a recurrent stroke. The rate of recurrent 6.0

Alter et al Hypertension and Risk of Stroke Recurrence 1607 Number of patient FIG 1. Bar graphs show distribution of measured blood pressure at enrollment (n=656). DBP indicates diastolic blood pressure; SBP, systolic blood pressure. SO 100 ito 120 130 140 150 16O 170 180 180 200 210 220230 stroke in the cohort was 8% in the first year, 5% in year 2, 3% in year 3, and 5% in year 4. Using a Kaplan-Meier analysis, the frequency of stroke recurrence was analyzed in the 621 patients with an ischemic stroke; those with a history of hypertension were compared with those without such a history (Fig 2). The difference between these two groups was statistically significant (/>=.O1). Similar analyses were carried out using the alternative criteria for defining hypertension and comparing those above and below the cutoff points for measured blood pressure at enrollment. These analyses showed that only diastolic blood pressure in the hypertensive range but not systolic blood pressure was significantly associated with risk of stroke recurrence (Figs 3 and 4). Control of diastolic blood pressure was also significant in reducing risk of recurrent stroke (Fig 5). We also used a Cox proportional hazards model to examine the effect of controlling blood pressure on the stroke recurrence rate. Both controlled diastolic and controlled systolic pressure appeared to be associated with reduced risk of stroke recurrence in the univariate Cox analysis (Table 2). However, in the multivariate analysis of risk factors, taking into account age, sex, and comorbidities, only diastolic blood pressure emerged as a significant and independent risk factor for stroke recurrence (Table 2). As shown in Table 2, the risk of stroke recurrence was approximately eightfold higher in those with poorly controlled diastolic pressure compared with nonhypertensive subjects (P=.O1). It was almost fourfold higher in those with fair control of diastolic pressure (P=.O7), and in those with good diastolic pressure control stroke recurrence was twofold higher (P=.O3). Thus, although there was a clear trend indicating reduced risk of stroke recurrence with each level of improvement in diastolic pressure control, the presence of diastolic hypertension, as defined, was still associated with increased risk of stroke recurrence. Systolic blood pressure, on the other hand, was not a significant independent risk factor for stroke recurrence in the multivariate analysis (P=.15) even after taking age, sex, and comorbidities into account. Other results TABLE 2. Relative Risk of Stroke Recurrence Using Cox Proportional Hazards Models for Nonhypertensive and Hypertensive Subjects in Good, Fair, and Poor Control* Blood Pressure Risk Ratio 95% Cl DBP >95 mm Hgt Nonhypertensive Good control Fair control Poor control 1.0 2.0 3.9 7.7.04.08.01 1.04-3.68 0.87-17.09 1.73-34.55 SBP >150 mm Hg if age <65 y; >160 if age >65 yt Nonhypertensive 1.0 Good control 2.5.01 1.20-5.01 Fair control 2.8.02 1.20-6.63 Poor control 1.3.67 0.36-4.80 DBP >95 mm Hg* Nonhypertensive 1.0 Good control 2.0.03 1.06-3.74 Fair control 3.9.07 0.88-17.31 Poor control 8.4.01 1.07-37.49 DBP >95 mm Hg 3.9.004 1.55-10.22 Cl indicates confidence interval; DBP, diastolic blood pressure; and SBP, systolic blood pressure. *See text for definition of control. tunivariate risk ratio from Cox proportional hazards model. JRisk ratio from stepwise forward selection used including age, sex, and comorbidities at enrollment (see Table 1). SBP was not significant (P=.15) and therefore is not shown. Risk ratio from time-dependent Cox proportional hazards model with stepwise forward selection as with immediately preceding section. DBP and SBP were the time-dependent variables; again SBP was not significant.

1608 Stroke Vol 25, No 8 August 1994 No HTN (n = 252) + HTN (n = 369) 100 - DBP 96 mm Hg - DBP > 96 mm Hg so (n 630) (n 17) 60 12 18 24 30 Months following the Initial stroke 36 42 FIG 2. Line graph shows Kaplan-Meier estimates of cumulative risk of second stroke in those with an ischemic stroke and history of hypertension (HTN) at enrollment in the Lehigh Valley Stroke Cohort (n=621). of this multifactorial analysis are reported elsewhere. 11 A time-dependent Cox proportional hazards model in which diastolic and systolic blood pressures at the last follow-up visit before stroke recurrence were the timedependent variables yielded essentially similar results (Table 2). Again, control of diastolic but not control of systolic blood pressure was associated with reduced risk of stroke recurrence. Discussion In the present study a history of hypertension at enrollment was significantly associated with increased risk of stroke recurrence. Also, level of diastolic (but not systolic) blood pressure over the course of the study was a significant risk factor for stroke recurrence. Good control of diastolic pressure, as defined in this study, appeared to be associated with a substantially reduced risk of stroke recurrence. This association persisted even when the effects of other variables such as age, sex, and comorbidities were taken into account in a multivariate analysis and when several different cutoff levels were used to define quality of control. Similar results were seen using the Kaplan-Meier method as well as univariate, multivariate, and time-dependent Cox pro- 80 60... SBP < 160 mmhg if age < 66 yra or SBP < 160 mmhg If age > 66 yre * - SBP > 160 mmhg If age ' 66 yre or SBP > 160 mmhg if age» 66 yi 12 18 24 30 36 42 Months following the initial stroke FIG 4. Line graph shows Kaplan-Meier estimates of cumulative risk of second stroke in the entire cohort (n=656) by measured diastolic blood pressure (DBP) at enrollment. portional hazards models. Thus, our study provides additional support for the belief that diastolic blood pressure control is important in reducing risk of stroke recurrence. MacMahon et al, 12 in an analysis of nine prospective observational studies involving 420 000 subjects, reported that the higher the diastolic blood pressure, the higher the risk of stroke, with no threshold effect. Level of diastolic blood pressure was strongly and positively related to risk of stroke even among those whose blood pressure was considered to be in the normotensive range. This study did not clearly differentiate between initial and recurrent stroke risk, but given our results, which showed a clear trend between better control and reduced risk of stroke recurrence, the relation between risk and quality of control of diastolic pressure is likely to exist for both initial stroke and stroke recurrence. Collins et al 13 analyzed 14 studies in which therapies for controlling hypertension were evaluated as preventives of cardiovascular disease, including stroke. A 42% reduction in stroke, both fatal and nonfatal, was observed. Again, there was no clear distinction between Cua. 2nd strok* 100 No HTN (n=211) +Poor (n=6) ^Falr(n=9) - -Good (n=311) 80 60 40 40 Months following the initial stroke FIG 3. Line graph shows Kaplan-Meier estimates of cumulative risk of second stroke in the entire cohort (n=656) by measured systolic blood pressure (SBP) at enrollment. 12 18 24 30 36 42 Months following the Initial stroke OBP < ss raw Hg: In < 50% vltlts (Peer): In 50% 73% vur (Fair); In 75% + vttltt (Good) FIG 5. Line graph shows Kaplan-Meier estimates of cumulative risk of second stroke in the entire cohort by hypertensive control status during follow-up. HTN indicates hypertension; DBP, diastolic blood pressure. See text for definition of good, fair, and poor control.

Alter et al Hypertension and Risk of Stroke Recurrence 1609 initial and recurrent stroke in this study. The designs in the studies reviewed by Collins et al were not all identical, yet the variance in their results was statistically insignificant. Moreover, in the Framingham Study, 14 the beneficial effect of blood pressure control on stroke reduction in the 1950s and 1960s, when less effective treatments for hypertension were available, was similar to the effect in studies done in the 1970s, when more effective antihypertensive therapies were used. Similar results despite differences in treatment suggest that the benefit of blood pressure reduction in the prevention of initial and recurrent stroke is not due simply to recently available antihypertensive therapies with more protective antistroke mechanisms. A substantial reduction in risk of stroke recurrence appears likely in those whose diastolic blood pressure control is successful. When those with poorest control were compared with those with best control in a multivariate analysis, there was a more than fourfold reduction in relative risk of stroke recurrence among the latter, from 8.4 to 2.0, although those with good control still had almost double the risk of nonhypertensive subjects. Our observation that diastolic rather than systolic blood pressure control is significantly associated with reduction in stroke recurrence has been noted previously. For example, Beever et al 1 showed that hypertensive subjects whose diastolic blood pressure was under good control (<100 mm Hg) had a lower rate of stroke recurrence than other patients with only fair control (100 to 109 mmhg) or poor control (>110 mm Hg). Sacco et al, 4 using Stroke Data Bank information, analyzed the 30-day cumulative risk of stroke recurrence. An increased risk of recurrence was significant for diastolic but not systolic blood pressure. In Carter's 15 study of patients who survived an acute stroke, hypertensive subjects were defined as those with a diastolic pressure >110 mmhg or a systolic pressure >160 mmhg. In those aged >65 years, control of systolic blood pressure, as in our study, was of no benefit in reducing stroke recurrence rate. Control of diastolic pressure in Carter's study was marginally beneficial compared with those with uncontrolled blood pressure. The finding of only marginally significant benefit may have been due to his having studied only 97 hypertensive patients; with a larger number, the benefit might have reached statistical significance. In a somewhat larger study, Johnson et al 2 included 124 patients with prior stroke. A lower stroke recurrence rate was found both among those with systolic pressure <160 mmhg and diastolic pressure <90 mm Hg. Using the Framingham Study data, Wolf et al 14 described a health risk appraisal model for evaluating risk of stroke. Their model included systolic but not diastolic blood pressure. However, they did note that among borderline hypertensive subjects with blood pressure between 140/90 and 160/95 mm Hg (values that include both systolic and diastolic blood pressure), treatment of hypertension was associated with reduced risk of stroke in both men and women. Not all studies have shown a benefit in reduction of risk of stroke recurrence for those whose blood pressure is controlled. For example, in the population-based study of stroke recurrence reported from Rochester, Minn by Meissner et al 5 during the period 1950 to 1979, level of blood pressure before the initial stroke and the degree of control of blood pressure among hypertensive subjects had no apparent effect on risk of stroke recurrence. In the Hypertension-Stroke Cooperative Study 6 452 stroke survivors were included. No significant difference in frequency of stroke recurrence was found between controlled and uncontrolled hypertensive subjects during 3.5 years of follow-up. The latter study differed in a number of respects from ours, not the least of which was that patients were enrolled up to a year after the initial stroke, whereas all of our patients were enrolled within 1 month of onset. Also, the Hypertension-Stroke Cooperative Study included a large number of black patients, whereas in our study 97% of patients were white. Frequency of hypertension differs by race, 16 and control of hypertension may have different effects on recurrent stroke rate in different races. A possible explanation for our observation that diastolic rather than systolic blood pressure control influences stroke recurrence rate is our definition of good control for systolic blood pressure (ie, >150 mm Hg on 75% of follow-up visits). Also, reducing systolic blood pressure may have an effect only on initial 17 but not on recurrent stroke. No standard definition of what constitutes optimal control for systolic and diastolic pressure exists, although Fletcher and Bulpitt 18 recommended recently that the systolic pressure be lowered to <125 mm Hg and diastolic pressure to <85 mm Hg for good control. Also, the latest report of the Joint National Committee on Detection, Evaluation and Treatment of High Blood Pressure classified normal blood pressure as < 130/85 mm Hg. 19 These levels are lower than those we used to define good control, and possibly our cutoff point for defining good control of systolic pressure may have been too high to see an effect on stroke recurrence rate. The stroke recurrence rate in our cohort was similar to that observed by other investigators. For example, in the recently reported study by Hier et al 3 involving 1273 patients with an ischemic stroke, history of hypertension was significantly associated with risk of stroke recurrence, with a 2-year recurrence frequency of 14%. In our study the frequency was 13% in 2 years. The estimated relative risk for recurrent stroke in relation to concurrent medical conditions has been calculated previously for a different cohort of patients in the Lehigh Valley. 8 In that stroke cohort, hypertension was associated with a 4.5-fold increase in the risk of stroke recurrence (P<.001). The significance levels for the age-specific relative risks were 5.01 except for the risk among those aged <65 years (P=.05). The risk of stroke recurrence did not vary much with age except in those aged ^85 years, when it increased. However, when logistic regression modeling was conducted on the simultaneous action of risk factors in that study, hypertension did not emerge as an independent risk factor for ischemic stroke recurrence (P=.55). 9 A retrospective design was used in that study. Many records had missing data on hypertensive status, and blood pressure control was not evaluated. In the present study, with an improved prospective design in which blood pressure status was specifically queried in every case and determined at enrollment in 656 of 662 (99%) of the cohort, diastolic but not systolic blood pressure did emerge as a significant and independent risk factor for stroke recurrence even when comorbidities were considered. In

1610 Stroke Vol 25, No 8 August 1994 addition, control of diastolic blood pressure significantly reduced risk of recurrence. It is apparent from our study that standardized definitions of hypertension and control of hypertension are needed. More detailed modeling of various cutoff points for blood pressure control could be instructive in providing information on the most effective levels. For example, lowering blood pressure too much might increase stroke recurrence, especially in older persons. Moreover, the period of enrollment after a stroke during which a patient would be eligible for enrollment must be specified and standardized if recurrence rates are to be compared in different studies. Confounding risk factors such as stroke subtype, treatments, and presence of additional risk factors, such as smoking history, lipid status, exercise, and alcohol consumption, should be considered. However, even with these caveats in interpreting our results, uniformity among available studies, including ours, in showing that blood pressure control (especially diastolic blood pressure) reduces stroke recurrence suggests that the effect is real and offers hope that aggressive efforts to lower blood pressure in hypertensive subjects will substantially lower the risk of recurrent stroke. Acknowledgments The authors are grateful to Valerie Permint and Linda Goldsmith, who typed the manuscript; Pei-Jiuan Lee and Hewa Saranadassa, PhD, who assisted in the statistical analysis; Steven Scheiner, MD, who assisted in the literature search; and Jiji Mani, MBBS, who assisted in data management. References 1. Beevers DG, Fairman MJ, Hamilton M, Harpur JE. Antihypertensive treatment and the course of established cerebral vascular disease. Lancet 1973;l:14O7-14O9. 2. Johnston JH, Beevers DG, Dunn FG, Larkin H, Titterington DM. The importance of good blood pressure control in the prevention of stroke recurrence in hypertensive patients. Postgrad Med J. 1981^7:690-693. 3. Hier DB, FouUces MA, Swiontoniowski M, Sacco RL, Gorelick PB, Mohr JP, Price TR, Wolf PA. Stroke recurrence within 2 years after ischemic infarction. Stroke. 1991;22:155-161. 4. Sacco RL, Foulkes MA, Mohr JP, Wolf PA, Hier DB, Price TR. Determinants of early recurrence of cerebral infarction: the Stroke Data Bank. Stroke. 1989;20:983-989. 5. Meissner I, Whisnant JP, Garraway WM. Hypertension management and stroke recurrence in a community (Rochester, Minnesota, 1950-1979). Stroke. 1988;19:459-463. 6. Hypertension-Stroke Cooperative Study Group. Effect of antihypertensive treatment on stroke recurrence. JAMA. 1974;229: 409-418. 7. Alter M, Sobel E, McCoy RL, Francis ME, Shofer F, Levitt LP, Meehan EF. Stroke in the Lehigh Valley: incidence based on a community-wide, hospital register. Neuroepidemiology. 1985; 4:1-15. 8. Alter M, Sobel E, McCoy RL, Francis ME, Davanipour Z, Shofer F, Levitt LP, Meehan EF. Stroke in the Lehigh Valley: risk factors for recurrent stroke. Neurology. 1987^37:503-507. 9. Sobel E, Alter M, Davanipour Z, Friday G, McCoy R, Levitt LP, Isack T. Stroke in the Lehigh Valley: combined risk factors for recurrent ischemic stroke. Neurology. 1989^39:669-672. 10. Alter M, Friday G, Sobel E, Lai S-M. The Lehigh Valley recurrent stroke study: description of design and methods. Neuroepidemiology. 1993;12:241-248. 11. Lai SM, Alter M, Friday G, Sobel E. A multifactorial analysis of risk factor for recurrence of ischemic stroke. Stroke. 1994;25: 958-962. 12. MacMahon S, Peto R, Cutler J, Collins R, Sorlie P, Neaton J, Abbott R, Godwin J, Dyer A, Stamler J. Blood pressure, stroke, and coronary heart disease, part I: prolonged differences in blood pressure: prospective observational studies corrected for the regression dilution bias. Lancet. 1990-335:765-774. 13. Collins R, Peto R, MacMahon S, Hebert P, Fiebach NH, Eberlein KA, Godwin J, Qizilbash N, Taylor JO, Hennekens CH. Blood pressure, stroke, and coronary heart disease, part II: short-term reductions in blood pressure: overview of randomized drug trials in their epidemiological context. Lancet. 1990^35:827-838. 14. Wolf PA, D'Agostino RB, Belanger AJ, Kannel WB. Probability of stroke: a risk profile from the Framingham Study. Stroke. 1991;22: 312-318. 15. Carter AB. Hypotensive therapy in stroke survivors. Lancet. 1970; 1:485-489. 16. Friday G, Lai SM, Alter M, Sobel E, LaRue L, McCoy RL, Levitt LP, Isack T, Peralta AG. Stroke in the Lehigh Valley: racial/ethnic differences. Neurology. 1989^9:1165-1168. 17. SHEP Cooperative Research Group. Prevention of stroke by antihypertensive drug treatment in older persons with isolated systolic hypertension: final results of the Systolic Hypertension in the Elderly Program (SHEP). JAMA. 1991;265:3255-3264. 18. Fletcher AE, Bulpitt CJ. How far should blood pressure be lowered? N Engl J Med. 1992^326:251-254. 19. Joint National Committee on Detection, Evaluation and Treatment of High Blood Pressure. The Fifth Report of the National Committee on Detection, Evaluation and Treatment of High Blood Pressure (]ticv).arch Intern Med. 1993;153:154-183.