Early Predictors of 15-Year End-Stage Renal Disease in Hypertensive Patients
Abstract There has been a continuing increase in the incidence of end-stage renal disease (ESRD) in the United States, including the fraction that has been attributed to hypertension. This study was done to seek relationships between ESRD and pretreatment clinical data and between ESRD and early treated blood pressure data in a population of hypertensive veterans. We identified a total of 5730 black and 6182 nonblack male veterans as hypertensive from 1974 through 1976 in 32 Veterans Administration Hypertension Screening and Treatment Program clinics. Their mean age was 52.5±10.2 years, and their mean pretreatment blood pressure was 154.3±19.0/100.8±9.8 mm Hg. During a minimum of 13.9 years of follow-up, 5337 (44.8%) of these patients died and 245 developed ESRD. For 1055 of these subjects, pretreatment systolic blood pressure (SBP) was greater than 180 mm Hg; 901 were diabetic; 1471 had a history of urinary tract problems; and 2358 of the 9644 who were treated had an early fall in SBP of more than 20 mm Hg. We used proportional hazards modeling to fit multivariate survival models to determine the effect of the available pretreatment data and early treated blood pressure levels on ESRD. This model demonstrated the independent increased risk of ESRD associated with being black or diabetic (risk ratio, 2.2 or 1.8), having a history of urinary tract problems (risk ratio, 2.2), or having high pretreatment SBP (for SBP 165 to 180 mm Hg, risk ratio was 2.8; for SBP >180 mm Hg, risk ratio was 7.6). In addition, myocardial infarction during follow-up increased the risk of subsequent ESRD almost twofold, and congestive heart failure increased it more than fivefold. The rate of ESRD in those whose SBP fell more than 20 mm Hg decreased by two thirds.
The number of patients entering the Medicare end-stage renal disease (ESRD) program in the United States increased from 15 195 in 1978 to 28 944 in 1985, representing an annual growth rate of 9.6% for new patients requiring dialysis.1 Hypertension and diabetes, each associated with about one third of new ESRD cases, are almost entirely responsible for this increase. From 1983 to 1987, the rates of change for diabetic ESRD and hypertensive ESRD were 200% and 150%, respectively, accounting for the total rate of change for ESRD. The increase in hypertension-related ESRD is in sharp contrast to the trends for two major complications of hypertension, myocardial infarction (MI) and stroke, both of which decreased by about 25% during the same decade.2 These declines have been attributed at least in part to effective blood pressure control.3
There are apparent major racial differences in the incidence of new cases of ESRD. During 1988, the age- and sex-adjusted incidence of ESRD was approximately 100 cases per million people in whites and 415 cases per million in blacks.4 There are also age-related differences in the new cases of ESRD. During 1988, there were approximately 300 new cases per million in the age range of 45 to 64 years, 435 cases per million in the range of 65 to 74 years, and 535 cases per million in those 75 years old and older.4 The continuing increase in ESRD poses a major and rapidly growing public health problem: it is estimated that the continued growth in the population needing dialysis will increase the cost to Medicare, which funds 80% of the expense, by $1 billion every 5 years.2
It is not known to what level hypertensive blood pressure should be reduced for maximal protection of the kidney. Uncertainty also exists regarding the quantitative relationships between blood pressure levels and progressive renal failure. Epidemiological evidence suggests that even subclinical renal dysfunction (as identified by hypercreatininemia) is associated with elevated blood pressure both in general populations5 6 and in people being treated for hypertension.7 A 5-year increase of at least 25% in baseline creatinine to 2.0 mg/dL or greater was observed in approximately 1.5% of patients with baseline diastolic blood pressure (DBP) of 90 to 104 mm Hg, in 3.5% of those with DBP of 105 to 114 mm Hg, and in 5.5% of those with DBP of 115 mm Hg or greater.8 Despite the limitations of the data relating to the benefit of treatment, it has been widely accepted that control of hypertension to levels below 140/90 mm Hg will significantly reduce the incidence of ESRD. Because in the Hypertension Detection and Follow-up Program (HDFP), stepped-care treatment approximately halved the 5-year increase in hypercreatininemia seen in patients receiving referred care who had baseline creatinine levels of 1.5 to 1.7 mg/dL,8 further reductions of blood pressure to 120/85 mm Hg or below have been suggested for patients with renal impairment.
The overall object of this report is to examine factors that might be associated with the incidence of ESRD during a 15-year follow-up of a cohort of hypertensive veterans. We examined the available data for a cohort of 11 912 male veterans who had uncomplicated hypertension when they began treatment in special Veterans Administration (VA) hypertension clinics during the mid-1970s. The baseline data include demographic and pretreatment clinical data obtained from brief history and physical examination, both of which emphasized the cardiovascular system. In addition to pretreatment characteristics, the relationships of ESRD to early treated blood pressure and to the treatment-induced decrease in blood pressure were examined. The relationships of other common complications of hypertension (MI, congestive heart failure [CHF], and stroke), if diagnosed during VA hospital admission, were examined with respect to subsequent ESRD. Finally, a relationship between ESRD and the location of the treating clinic was sought because of the “Stroke Belt” effect on all-cause mortality previously reported for this population.9 10
Recruitment of Cohort
This report involves 11 912 male veterans who began antihypertensive treatment in the Hypertension Screening and Treatment Program (HSTP) clinics during the mid-1970s. Extensive screening at HSTP clinics identified subjects with elevated DBP who were then scheduled for evaluation visits. For some, several screening visits were scheduled to define marginally elevated blood pressure. For 72% of the patients discussed here, the evaluation visit occurred within 3 months after screening, but the actual timing depended on the patient’s DBP and the clinic load. In addition to blood pressure measurements, the evaluation visit included a medical history and a physical examination, both of which emphasized the cardiovascular system.
For patients with severe or moderate diastolic hypertension (DBP 105 to 124 mm Hg) at the evaluation visit, as well as for those with less serious diastolic hypertension (DBP 90 to 104 mm Hg) but with additional risk factors, including being black or less than 35 years of age, the treatment visit was scheduled within 1 month after evaluation. For those considered to have very severe diastolic hypertension (DBP ≥125 mm Hg), this period was shortened to 1 week or less; in some instances, the treatment and evaluation visits were on the same day. For relatively low-risk patients with mild hypertension, the treatment visit might have been scheduled for 3 months after evaluation or, rarely, for even later, depending on clinic load. For 55% of the patients, the first treatment visit was within 30 days of the evaluation visit, and for 72% it was within 90 days.
Demographics and Medical History
The computer records for the 11 912 patients were obtained from the VA processing center in St Paul, Minn. Race, age, and clinic location were noted at a screening visit. Body mass index and self-reported history of heart disease, stroke, diabetes, urinary tract problems, and smoking habits were obtained at the last evaluation visit. Treated blood pressures were obtained from the last treatment visit recorded on computer. Data on race (black or nonblack) were available for 11 759 subjects; the remaining 153 were coded as nonblack for the analyses reported here. Data on age were unavailable for 127 patients, and they were omitted from analyses that included age. The geographic location of the clinic that the patient attended was coded as being in the Expanded Stroke Belt or in the Non–Stroke Belt according to the previously used definitions.10 Data on body mass index ([weight in kilograms]/[height in meters]2) were available for 10 222 patients. Data on smoking habits were available for 10 330 patients. A definite response to self-reported history of heart disease, stroke, diabetes, or urinary tract problems was available from 10 304, 10 307, 10 190, and 10 307 patients, respectively. If no response was recorded for a continuous variable, the patient’s data were omitted from analyses including that variable; omitted responses to other questions were coded as if they had been answered in the negative. The results of any laboratory tests done during the screening process were not included as part of the computer database and thus were not available for analysis.
The 11 912 previously untreated male veterans had a total of 19 457 screening, 10 349 evaluation, and 10 961 first treatment visits, which were recorded on computer and included in the analysis presented here. Data from all visits included at least a visit date and measurements of systolic blood pressure (SBP) and DBP. The usual visit pattern, observed in three fourths of patients (77.2%), included a screening visit that was followed by an evaluation visit and then by a series of treatment visits. A total of 85 581 treatment visits were recorded on computer; 12% of the patients had only one treatment visit, 10% had two, 10% had three, and 69% had four or more. The first HSTP screening visit used for this report was on August 22, 1972. Treatment visits were not scheduled until after January 1, 1974, when outpatient treatment was authorized. The last treatment visit recorded on computer was on January 31, 1977. After computerization of their clinical data was discontinued, the patients in the HSTP continued to be treated in HSTP clinics under treatment guidelines recommended by successive reports of the Joint National Committee on Detection, Evaluation, and Treatment of High Blood Pressure.
Blood Pressure Measurements
All blood pressure measurements were done with a mercury sphygmomanometer while the patient was seated. At screening visits, conditions varied widely. At evaluation and treatment visits, however, efforts were made to measure the blood pressure under standard conditions defined by the American Heart Association.11 The HSTP guidelines called for the blood pressure to be taken twice at each visit, with an interval of at least 1 minute between readings, with the patient having been seated and at rest for at least 5 minutes. The two blood pressures were averaged to obtain the visit blood pressure. Each patient’s pretreatment blood pressure was considered to be the average of his visit blood pressures at the first treatment visit and all earlier visits. For the 951 patients with no treatment visits, the visit blood pressures of all screening and evaluation visits were averaged. For 0.4% of the patients, the pretreatment blood pressure was based on one visit only; for 2.9% it was based on two visits; for 55.5% it was based on three visits; and for 41.2% it was based on four or more visits. (The average pretreatment blood pressure for the patients discussed here decreased at successive visits. For the first visit [regardless of visit type], blood pressure averaged 159.6/103.8 mm Hg; for the second, 154.0/100.6 mm Hg; for the third, 149.8/98.0 mm Hg; and for the fourth, 148.9/98.0 mm Hg. The blood pressure pattern was very similar if the averages were limited to patients who had four or more visits.) Each patient’s pulse pressure (PP=SBP−DBP) and mean arterial pressure (MAP=DBP+PP/3) were computed for each visit from the visit SBP and DBP.
Treated blood pressure was defined as the blood pressure taken at a subject’s last treatment visit before discontinuation of the computerized monitoring of the HSTP care in January 1977 (up to 3 years after the start of treatment), provided that the subject had had at least two treatment visits. Data on treated blood pressure were available for 9644 patients. (The average blood pressure at the last treatment visit was 138.9/90.7 mm Hg for the 9722 patients who had at least two treatment visits and 138.6/90.6 mm Hg for the 8606 patients with at least three treatment visits.) The long-range goal of HSTP treatment was control of DBP below 95 mm Hg with a minimum of adverse effects. This early phase of the HSTP, including the treatment guidelines, has been described elsewhere,12 and additional details on currently available data from the mid-1970s for the cohort discussed here have been previously reported.10 Although data were not collected on computer after January 1977, the HSTP continued unchanged, and the patients who were being treated in HSTP clinics were encouraged to continue to be followed by the program.
Collection of Mortality Data
Date of death was obtained from the Beneficiary Identification Records Location System (BIRLS) of the Department of Veterans Affairs or the National Death Index (NDI) of the National Center for Health Statistics. The BIRLS identified 5191 patients as having died during 1990 or earlier. The NDI identified 146 additional deaths during 1990 or earlier. Thus, dates of death were known for 5337 of the 11 912 patients (44.8%). No attempt was made to obtain the actual death certificates. The remaining 6575 patients were assumed to be alive as of December 31, 1990.
Collection of Morbidity Data
Data from VA hospitalizations occurring after antihypertensive treatment had begun were obtained by merging the HSTP clinical data with records from the VA’s Patient Treatment File (PTF). Patients were identified as having ESRD if the PTF indicated a discharge diagnosis with International Classification of Diseases (ICD) 8 or 9 Procedure Code 39.95–Hemodialysis, 54.98–Peritoneal Dialysis, or 39.27–Arteriovenostomy for Renal Dialysis. The HSTP clinical data were similarly merged with the records of the United States Renal Data System (USRDS) Dialysis and Transplant Registry. The date of the qualifying hospital admission from the PTF or the first service date from the USRDS, whichever occurred first, was used as the date of the ESRD event. Among the 11 912 patients, 245 were known to have had an ESRD end point after their first HSTP treatment visit and before January 1, 1991. These 245 ESRD end points were then used as the primary end points for the current analysis.
Occurrences of other specific cardiovascular disease events were identified from the PTF by the discharge ICD codes: MI, ICD8 or ICD9 code of 410–Acute Myocardial Infarction; CHF, ICD8 code of 402–Hypertensive Heart Disease, 427.0–Congestive Heart Disease, or 427.1–Left Ventricular Failure or ICD9 code 428–Heart Failure or code 402.0 or 402.9 in which the fifth digit was 1; and cerebrovascular accident [CVA], ICD9 code 430–Subarachnoid Hemorrhage, 431–Intracerebral Hemorrhage, 434–Cerebral Artery Occlusion, or 432–Other Unspecified Intracranial Hemorrhage or ICD8 code 433–Cerebral Thrombosis.
ESRD-free time was defined as the time from the first treatment visit to the first ESRD event, the date of death, or December 31, 1990, whichever occurred first. For the 951 patients without a valid treatment visit, the date of the last visit (screening or evaluation) was used. The median follow-up for those with ESRD was 9.4 years, and for those who did not have ESRD before January 1, 1991, it was 14.2 years.
The occurrence of ESRD was initially explored using Kaplan-Meier curves for the entire cohort; the data were then broken down by individual putative risk factors: race, age, pretreatment blood pressure, treated blood pressure, body mass index, smoking habits, geographic location of the clinic, and self-reported history of diabetes, heart trouble, stroke, or urinary tract problems. Death was used as a censoring event. Overall, the estimated 15-year ESRD cumulative rate was 0.0279 (95% confidence interval, 0.0243-0.0315).
Table 1⇓ summarizes the univariate relationships between putative risk factors and ESRD event rates. Similar results were obtained whether data from subjects who had no recorded response to a categorical question were excluded or were coded as if the patient did not experience the particular problem. For the results reported here the latter approach was used. Increased rates of ESRD were observed for blacks, diabetics, and those reporting a history of stroke or urinary tract problems and those attending clinics in the Stroke Belt. No systematic relationship with age or body mass index was detected. Fig 1⇓ shows the increased risk of ESRD for blacks throughout the period of surveillance.
To allow for a nonlinear relationship between pretreatment blood pressure and ESRD risk and to aid interpretation, blood pressures were categorized. They were initially broken down into groups roughly corresponding to quartiles of the cohort (four groups, each approximately the same size). Because only subjects with blood pressure in the highest quartile appeared to be at increased risk, this quartile was subsequently split, resulting in five levels of blood pressure. MAP and PP were each broken down into four groups consisting of approximately 75%, 15%, 5%, and 5% of the cohort, a distribution indicating successively increasing risk.
Table 2⇓ shows the relationship between baseline blood pressure and subsequent ESRD events. The number of events as a percentage of the number of subjects provides a crude event rate; the 15-year rate represents the Kaplan-Meier estimate at 15 years; the probability value for the log-rank test provides a test of the null hypothesis that all groups have equivalent time courses of ESRD events. Risk ratios were calculated for each method of blood pressure categorization by fitting a Cox proportional hazards model, with the lowest blood pressure being the reference group. Whether the cohort was broken down by DBP or SBP, the increased risk was confined to the group with the highest pressure. Moreover, the patients in the reference group were mildly hypertensive; ie, the comparison was not with normotensive subjects but with subjects who were borderline or mildly hypertensive. Table 2⇓ also breaks down the group according to MAP and PP, as was done in our previous report.10 Fig 2⇓ shows the relationship between pretreatment SBP and the subsequent occurrence of ESRD. The increased risk at the two highest pressure levels was apparent throughout the period of risk.
Table 3⇓ provides the analogous information for the patients whose high blood pressure was treated, expressed both as absolute value of the treated blood pressure and as a change from the pretreatment value. Subjects were divided into approximate quartiles for display and analysis. For those whose blood pressure (SBP or DBP) decreased 2 to 20 mm Hg after treatment, the risk decreased as was anticipated; however, for those with the largest change in pressure, the unexpected finding was that no decrease in risk was observed.
For estimation of the rate of new ESRD events as a function of time since the first treatment visit, the hazard function was calculated for each 2-year period with life table methods, and the results were plotted (Fig 3⇓). Clearly, the rate of new events increased in a roughly linear fashion over time. Note that this is in contrast to the lack of a systematic increase in rate with age in this hypertensive population.
Fig 4⇓ shows the event rates from the VA’s PTF for the MI, CVA, and CHF morbidity end points. Overall, 834 (7.0%) patients experienced an MI, 424 (3.6%) a CVA, and 1000 (8.4%) a CHF hospitalization before an ESRD endpoint.
Cox proportional hazards modeling was used to fit multivariate survival models for examination of the independent effect of the terms in Tables 1 through 3⇑⇑⇑. Terms for race, age, and blood pressure were used to fit a basic model. When pretreatment SBP was used, pretreatment DBP did not provide any significant additional predictive power for ESRD. In contrast, when MAP and PP were used, as by Miller et al,10 both the MAP and PP terms were significant. We have elected to present the results of using the more parsimonious SBP model here.
Further terms representing smoking history, body mass index, geographic location, and self-reported history of diabetes, stroke, or urinary tract problems were then considered for inclusion in the model. Age was modeled by a linear term. Age and race were retained in all models whether or not they had a significant effect. First, a model that retained only those risk factors available before treatment began and showing independent change in risk was fit. The model was then expanded to consider whether a hospitalization for MI, CVA, or CHF after treatment had begun increased the risk of subsequent ESRD; time-dependent covariates representing each of these events were included in the proportional hazards model. Table 4⇓ shows the estimated partial risk ratios for this model. CVA was not related to increased risk of ESRD. Analyses were repeated excluding MIs occurring fewer than 30 days before death and resulted in a very similar model. Similarly, CHF episodes during the same admission as for an MI were eliminated and resulted in essentially the same result. In summary, this model demonstrates the independent increased risk ratio of ESRD associated with being black (risk ratio, 2.2) or diabetic (risk ratio, 1.9), having a history of urinary tract problems (risk ratio, 2.3), or having high pretreatment SBP (165 to 180 mm Hg, risk ratio, 1.9; >180 mm Hg, risk ratio 4.6). In addition, the risk was increased by a factor of 2 after a hospitalization that included an MI and by a factor of 5.4 after a hospitalization that included CHF.
For assessment of the effect of early treatment, the model was further expanded to include terms for treated blood pressure (Table 4⇑). The risk ratios seen in the pretreatment-only model did not change. In contrast to the univariate relationship between the change in blood pressure and ESRD, these results demonstrate a monotonic relationship between the amount of decrease in SBP and the decrease in risk of ESRD. The rate of ESRD in those whose SBP dropped by more than 20 mm Hg was only one third that in those whose SBP did not change.
Additional multivariate models were fit to explore the possibility of interactions between the variables in the model, including interactions with the time-dependent covariates. No such interaction was identified. The adequacy of the model was assessed using martingale and deviance residuals. No lack of fit of the model to individual observations was indicated, because no outliers were identified in plots of the residuals against the linear predictor. Careful analysis of the residuals provided no evidence of departure from linearity by the age effect. Examination of the hazard functions, stratified by the risk factors, indicated that while the rate of ESRD increased in this cohort over time, the assumption of proportional hazards was not inappropriate.
ESRD as a complication of hypertension is associated epidemiologically with both demographic characteristics and environmental influences. Related characteristics include age, sex, severity and duration of the hypertension, racial and genetic factors, diabetes, and renal parenchymal disease. Other influences that may contribute to renal insufficiency include medications and nutrition.
Before effective medical treatment was available, most cases of very severe hypertension (ie, malignant hypertension) proceeded to fatal renal failure, with the median survival time being less than 6 months.13 Initially, malignant hypertension occurred in both men and women and in both whites and blacks, but after treatment became available in the early 1950s and malignant hypertension gradually decreased in frequency, the decrease was more rapid among whites and women than among black men. Moreover, there seems to be a regional effect, with the incidence of malignant hypertension decreasing more slowly in the southeast United States.
The univariate models of treated blood pressure show an interesting phenomenon suggesting an increased risk of ESRD in subjects in whom SBP decreased by more than 20 mm Hg compared with subjects in whom SBP decreased by 2 to 20 mm Hg (Table 3⇑). This phenomenon is reminiscent of the so-called J-shaped curve, suggesting that excessive fall in treated blood pressure may compromise coronary circulation and harm those with preexisting ischemic heart disease.1 2 It is intriguing to postulate that excessive fall in SBP may similarly compromise the renal circulation and renal function over the long term. It is noteworthy, however, that the relationship is no longer apparent in multivariate models (Table 4⇑). The data presented in Table 4⇑ suggest that a high level of pretreatment SBP is a strong predictor of ESRD, and the data presented in Fig 3⇑ suggest that the predictive effect persists for at least 15 years without obvious diminution. Pretreatment DBP provides no independent prediction but by itself it is a strong predictor (Table 2⇑). It is noteworthy that we find no evidence that risk is related to SBP less than 165 mm Hg or DBP less than 106 mm Hg; however, our data deal only with hypertensive subjects and do not indicate whether risk would be lower for the subjects with blood pressure in the normotensive range.
The data in Table 4⇑ also suggest that reduction of SBP early in treatment diminishes risk, with protection increasing as the drop in pressure increases. Additional analyses failed to detect any interaction between baseline SBP and SBP decrease during therapy. This predictive value of the early treatment on later cardiovascular events has been noted.5 Although the effect of race persisted in the multivariate models, the regional difference (ie, the Stroke Belt effect) did not (data not shown).
According to current data from the Health Care Financing Administration, which oversees the ESRD reimbursement program, the incidence of ESRD rises with advancing age. The incidence increases from 58 cases per million people per year in the age range of 25 to 34 years to 241 cases per million per year in the age range of 65 to 74 years.14 Race is also a potent risk factor for the development of ESRD. Data from Georgia show that the annual incidence of ESRD from 1979 to 1984 was 101.5 cases per million for blacks and 12 cases per million for whites.15 Adjusting for age and sex and taking into account the higher prevalence of hypertension among blacks compared with whites, the incidence of ESRD is still six times greater in blacks than in whites.16 However, this risk does not take into account either the severity and duration of hypertension or the quality of antihypertensive therapy. Few data compare type and intensity of antihypertensive therapy among races or between the Southeast and the rest of the United States.
Duration of hypertension also appears to have an effect on decline in renal function. In one retrospective study of 115 male veterans followed up for an average of 9.8 years, the time-averaged blood pressure was predictive of the change in serum creatinine concentration.17 In patients with chronic renal insufficiency, there is an association between a DBP less than 90 mm Hg before dialysis and a slower rate of decline in renal function compared with patients with higher DBP values.
Diabetes and hypertension frequently are comorbid conditions, and these diseases, as well as renal parenchymal disease, interact with other host characteristics to predispose a person to the development of renal insufficiency. Unfortunately, our database does not contain documentation about the severity and subsequent medical management of the self-reported history of diabetes, but Tierney et al7 have shown that lack of glucose control is associated with the development of renal dysfunction in a cohort of treated hypertensive patients.
Most of the clinical trials of hypertensive patients have focused on cardiovascular and cerebrovascular end points. The development of ESRD in all these studies is too infrequent to provide the statistical power to demonstrate a beneficial treatment effect. However, a retrospective review of the HDFP clinical trial shows that the decline in renal function was greater in men, blacks, and older subjects as well as in those with a higher entry DBP.8 Moreover, Walker et al18 reported a racial difference in patients with well-controlled hypertension and elevated creatinine levels; blacks had ongoing renal damage, but whites had no further increase in creatinine levels. Because of the size of the cohort and the duration of follow-up, the present study provides important new information about the occurrence of ESRD in hypertensive patients.
This study was supported by National Heart, Lung, and Blood Institute grant R03 HL-47677, the Department of Veterans Affairs, and Miles Pharmaceutical. Data reported here have been supplied by the United States Renal Data System (USRDS). The interpretation and reporting of these data are the responsibility of the authors and in no way should be seen as an official policy or interpretation of the US government. The authors gratefully acknowledge the personnel at the following HSTP clinics who collected the data for this article: Allen Park, Mich; Birmingham, Ala; Buffalo, NY; Chicago (Lakeside), Ill; Cleveland, Ohio; Dallas, Tex; Dayton, Ohio; East Orange, NJ; Houston, Tex; Indianapolis, Ind; Iowa City, Iowa; Jackson, Miss; Memphis, Tenn; Miami, Fla; Milwaukee, Wis; Minneapolis, Minn; New Orleans, La; Oklahoma City, Okla; Omaha, Neb; Philadelphia, Pa; Pittsburgh, Pa; Richmond, Va; St Louis, Mo; Salt Lake City, Utah; San Francisco, Calif; San Juan, Puerto Rico; Seattle, Wash; Sepulveda, Calif; Topeka, Kan; Tucson, Ariz; Washington, DC; and West Haven, Conn.
- Received May 19, 1994.
- Revision received July 14, 1994.
- Accepted November 7, 1994.
MacMahon S, Cutler JA, Stamler J. Antihypertensive drug treatment: potential, expected and observed effects on stroke and on coronary heart disease. Hypertension. 1989;13(suppl I):I-45-I-50.
US Renal Data System. USRDS 1990 Annual Data Report. Bethesda, Md: National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases; 1990.
Tell GS, Rutan GH, Kronmal RA, Bild DE, Polak JF, Wong ND, Borhani NO, for the Cardiovascular Health Study (CHS) Collaborative Research Group. Correlates of blood pressure in community-dwelling older adults: the Cardiovascular Health Study. Hypertension. 1994;23:59-67.
Tierney WM, Harris LE, Copley JB, Luft FC. Effect of hypertension and type II diabetes on renal function in an urban population. Am J Hypertens. 1990;3:69-75.
Shulman NB, Ford CE, Hall WD, Blaufox MD, Simon S, Langford HG, Schneider KA, on behalf of the Hypertension Detection and Follow-up Program Cooperative Group. Prognostic value of serum creatinine and effect of treatment of hypertension on renal function: results from the Hypertension Detection and Follow-up Program. Hypertension. 1989;13(suppl I):I-80-I-93.
Miller JP, Perry HM Jr, Rossiter JE, Baty JD, Carmody SE, Sambhi MP. Regional differences in mortality during 15-year follow-up of 11 936 hypertensive veterans. Hypertension. 1994;23:431-438.
American Heart Association. Recommendations for Human Blood Pressure Determination by Sphygmomanometers. New York, NY: American Heart Association; 1967.
Whelton PK, Klag MJ. Hypertension as a risk factor for renal disease: review of clinical and epidemiological evidence. Hypertension. 1989;13(suppl I):I-19-I-27.
Rosansky SH, Hoover DR, King L, Gibson J. The association of blood pressure levels and change in renal function in hypertensive and nonhypertensive subjects. Arch Intern Med. 1990;150: 2073-2076.
Walker WG, Neaton JD, Cutler JA, Neuwirth R, Cohen JD. Renal function change in hypertensive members of the Multiple Risk Factor Intervention Trial. JAMA. 1992;286:3085-3091.