ORIGINAL INVESTIGATION

Transcription

1 ORIGINAL INVESTIGATION Influence of Long-term, Low-Dose, Diuretic-Based, Antihypertensive Therapy on Glucose, Lipid, Uric Acid, and Potassium Levels in Older Men and Women With Isolated Systolic Hypertension The Systolic Hypertension in the Elderly Program Peter J. Savage, MD; Sara L. Pressel, MS; J. David Curb, MD; Eleanor B. Schron, RN, MS; William B. Applegate, MD; Henry R. Black, MD; Jerome Cohen, MD; Barry R. Davis, MD, PhD; Philip Frost, MD; W. Smith, MD; Nelly Gonzalez, MD; Gordon P. Guthrie, MD; Albert Oberman, MD; Gale Rutan, MD; Jeffrey L. Probstfield, MD; Jeremiah Stamler, MD; for the SHEP Cooperative Research Group Background: Previous studies often of short duration have raised concerns that antihypertensive therapy with diuretics and -blockers adversely alters levels of other cardiovascular disease risk factors. Methods: The Systolic Hypertension in the Elderly Program was a community-based, multicenter, randomized, double-blind, placebo-controlled clinical trial of treatment of isolated systolic hypertension in men and women aged 60 years and older. This retrospective analysis evaluated development of diabetes mellitus in all 4736 participants in the Systolic Hypertension in the Elderly Program, including changes in serum chemistry test results in a subgroup for 3 years. Patients were randomized to receive placebo or treatment with active drugs, with the dose increased in stepwise fashion if blood pressure control goals were not attained: step 1, 12.5 mg of chlorthalidone or 25.0 mg of chlorthalidone; and step 2, the addition of 25 mg of atenolol or 50 mg of atenolol or reserpine or matching placebo. Results: After 3 years, the active treatment group had a 13/4 mm Hg greater reduction in systolic and diastolic blood pressure than the placebo group (both groups, P.001). New cases of diabetes were reported by 8.6% of the participants in the active treatment group and 7.5% of the participants in the placebo group (P=.25). Small effects of active treatment compared with placebo were observed with fasting levels of glucose (+0.20 mmol/l [+3.6 mg/dl]; P.01), total cholesterol (+0.09 mmol/l [+3.5 mg/dl]; P.01), high-density lipoprotein cholesterol ( 0.02 mmol/l [ 0.77 mg/dl]; P.01) and creatinine (+2.8 µmol/l [+0.03 mg/dl]; P.001). Larger effects were seen with fasting levels of triglycerides (+0.9 mmol/l [+17 mg/dl]; P.001), uric acid (+35 µmol/l [+.06 mg/dl]; P.001), and potassium ( 0.3 mmol/l; P.001). No evidence was found for a subgroup at higher risk of risk factor changes with active treatment. Conclusions: Antihypertensive therapy with low-dose chlorthalidone (supplemented if necessary) for isolated systolic hypertension lowers blood pressure and its cardiovascular disease complications and has relatively mild effects on other cardiovascular disease risk factor levels. Arch Intern Med. 1998;158: Authors affiliations are listed in the Acknowledgment section on page 750. A complete listing of SHEP group members was published previously (JAMA. 1991;265: ). ALTHOUGH diuretics and -blockers lower blood pressure, reduce cardiovascular complications of hypertension, and are inexpensive, the recommendation that these drugs be used as primary agents for the treatment of hypertension 1 has been challenged. 2 In addition to their antihypertensive effects, diuretics have putatively adverse effects on levels of several other cardiovascular disease (CVD) risk factors, especially glucose, cholesterol, triglyceride, uric acid, and potassium It has been speculated that these effects may explain the smaller benefit of antihypertensive therapy on coronary heart disease rates than expected from epidemiological data on blood pressure and coronary heart disease. 18 Hypertension is common in patients with diabetes and a majority of hypertensive patients with diabetes die of CVDs. 19 Use of diuretics and -blockers in patients with diabetes has been questioned, since studies have reported that they cause a significant deterioration in glucose tolerance. 13,16,17 Because of the changes in CVD risk factors in hypertensive patients receiving diuretic therapy, it was speculated that the beneficial effects of lowering blood pressure may be reduced or negated and suggested that diuretics be avoided as first-line antihypertensive drugs in patients with diabetes

2 SUBJECTS AND METHODS Details of the design, eligibility criteria, and recruitment methods used in the SHEP clinical trial have been reported. 22,23 Major findings have also documented the benefits of low-dose, diuretic-based therapy in reducing rates of nonfatal plus fatal stroke (the SHEP primary end point), coronary events, and total cardiovascular events. 24 Similar beneficial effects were seen in participants with diabetes. 25 Aspects of the SHEP design, eligibility criteria, and participant recruitment relevant to the questions addressed in this article are described herein. The SHEP study was a multicenter, randomized, double-blind, placebo-controlled clinical trial conducted among 4736 ambulatory, community-dwelling men and women aged 60 years and older with isolated systolic hypertension. The primary end point was the incidence of fatal and nonfatal stroke. Secondary end points included coronary heart disease, total cardiovascular morbidity and mortality, and quality-of-life measurements. Potential participants were excluded because of history and signs of major CVD, cancer, alcoholic liver disease, renal dysfunction, or the presence of medical management problems. All persons gave informed consent prior to participation in the study. Of the 4736 participants in SHEP, 3161 were not receiving antihypertensive medication at the time of enrollment and 1575 were receiving therapy. All blood pressures were measured by trained, certified technicians. 26 To be eligible for randomization, participants had an average of 4 seated systolic blood pressure (SBP) measurements taken (2 at each of 2 baseline visits) that had to be between 160 and 219 mm Hg, with an average diastolic blood pressure (DBP) of less than 90 mm Hg. Those receiving antihypertensive medication who met initial screening criteria discontinued their antihypertensive medication and were required to meet blood pressure eligibility within an 8-week period of close follow-up while not taking medication. Participants underwent a physical examination, including measurement of weight and height, performance of a 12-lead electrocardiogram, and determination of baseline laboratory measurements, including levels of serum glucose, total and high-density lipoprotein (HDL) cholesterol, triglycerides, uric acid, potassium, and creatinine. While centers attempted to perform laboratory tests after an overnight fast, this was not mandatory. Because of this factor as well as missed follow-up visits, only a minority of randomized participants had a complete set of fasting laboratory test results at baseline and the 1- and 3-year follow-up examinations. On completion of the second baseline visit, eligible participants were randomly allocated to receive active drug treatment (n=2365) or placebo (n=2371), stratified by clinical center and by prior antihypertensive treatment status. Therapy was administered in double-blind, placebocontrolled fashion, with a once-daily dose of either an active drug or matching placebo. An average of 4 blood pressure measurements during baseline visits was used to establish a goal for blood pressure control for each participant. If SBP was at least 180 mm Hg, the goal was to reduce it to less than 160 mm Hg. If SBP was between 160 and 179 mm Hg, the goal was to reduce it by at least 20 mm Hg. Antihypertensive drug therapy was given with the objective of using the minimum amount of medication necessary to maintain SBP at or below the goal. All subjects were initially treated with chlorthalidone or matching placebo. Step 1 dose of the active drug was 12.5 mg of chlorthalidone. This dose was doubled to 25 mg of chlorthalidone if the SBP goal was not achieved. If this regimen did not achieve adequate control of participants blood pressures, 25 mg of the step 2 drug atenolol or a matching placebo was added. This dose could also be doubled if blood pressure remained above the goal. For persons unable to take atenolol, 0.05 mg of reserpine or matching placebo could be substituted and increased to 0.10 mg. If serum potassium concentrations were below 3.5 mmol/l on 2 consecutive visits, potassium supplements were added to the participant s regimen. Average follow-up for the final report was 4.5 years. Participants were seen monthly until they reached the SBP goal or until maximal drug or placebo dose was attained. Participants were then seen quarterly with annual medical history taking, physical examinations, and laboratory evaluations. More frequent visits were arranged if necessary. Blood pressure above defined escape criteria (sustained SBP 220 mm Hg or DBP 90 mm Hg) was an indication for prescribing known active therapy. Blood samples were obtained at the second baseline visit immediately before randomization. A defined protocol for venipuncture and sample processing was followed. Details of laboratory methods at the central laboratory (Met- Path, Teterboro, NJ) and quality control procedures have been published. 27 Information on how often participants reported having diabetes is available for all members of the cohort as of the last examination. A participant was defined as having diabetes at baseline if (1) a history of diabetes was reported by the participant; (2) fasting serum glucose levels were 7.8 mmol/l or higher ( 140 mg/dl); or (3) the participant reported using antidiabetic medication. Complete fasting laboratory data for baseline and follow-up for years 1 and 3 are available for 1663 individuals (715 men and 948 women). Four-year follow-up fasting laboratory data are available for 727 of these participants. The major reason for exclusion from the follow-up blood chemistry study was the absence of 1 or more fasting samples. Because it is substantially larger, the 3-year follow-up group is the primary focus of this article. Only fasting samples were used for analyses of glucose and triglyceride levels. Nonfasting samples were also used for analyses of cholesterol, HDL cholesterol, uric acid, and potassium levels. Comparability of baseline characteristics between the 2 treatment groups was assessed using 2 tests for categorical variables and standard normal (z) tests for continuous variables. Analyses were performed with participants in groups according to initial treatment assignment (intention to treat). Analyses were performed to compare differences at each time interval between the placebo and active treatment groups and to assess trends in risk factor levels over time. In addition, participants were divided into quintiles on the basis of baseline serum fasting glucose levels to enable assessment of effects of active treatment compared with placebo among those with different baseline fasting glucose levels. 742

3 Although several small studies document shortterm effects of diuretics on CVD risk factors, few data are available on the long-term impact of these drugs. Even less information is available on whether the long-term effects of diuretic therapy differ among subgroups of hypertensive patients with different levels of fasting glucose. This information is particularly important in older persons, among whom both glucose intolerance and elevated blood pressure are very common. 15,20 As expected, rates of reported diabetes were high in the Systolic Hypertension in the Elderly Program (SHEP) cohort at baseline. 21 Data from SHEP provide an opportunity to look at the effects of long-term, low-dose diuretic therapy, using a dose level that may be less prone to induce the previously described CVD risk factor changes. Three questions were addressed: (1) Does low-dose diuretic therapy increase the rate of how often participants report diabetes? (2) Does this therapy have a detrimental effect on levels of glucose, lipids, uric acid, potassium, and creatinine, and if so, how large is the effect on each? (3) Are those with higher baseline fasting levels of glucose at an increased risk of adverse effects? RESULTS RECRUITMENT AND TREATMENT ADHERENCE Recruitment was conducted at 16 clinical centers between March 1985 and January Characteristics of participants selected have been described previously. 28 Overall, 4736 (1.06% )of screened participants were entered into the study. Those randomized to receive active treatment and placebo control had comparable levels of SBP and DBP at baseline examinations. Some alteration in antihypertensive regimens occurred during the course of the study. At the 3-year visit, of all participants in the active treatment group, 806 (36.2%) were receiving step 1, dose 1 medication only (chlorthalidone) and 398 (17.9%) were receiving step 1, dose 2 medication only. In addition to chlorthalidone, 210 (9.4%) were also receiving step 2, dose 1 medication (atenolol or reserpine), 152 (6.8%) were receiving step 2, dose 2 medication, 328 (14.7%) were receiving other active medication, and 230 (10.3%) were receiving no antihypertensive drug. Most participants randomized to the placebo group received no active antihypertensive medication throughout the trial. The percentage for whom active antihypertensive drug therapy was prescribed increased progressively from 292 (13.1%) at year 1 to 727 (32.6%) at year 3. BlOOD PRESSURE RESPONSE As previously reported, active treatment successfully lowered both SBPand DBP. 24 Blood pressure decreased between baseline and the 3-year follow-up examination in both groups with the greater decline in the active treatment group (year 3 difference: SBP, 13.0 mm Hg; DBP, 4.4 mm Hg; P.001). Table 1. Prevalence of Diabetes* at, New Cases to 1 Year and 3 Years Treatment, No. (%) of Patients Active Drug P Prevalence at baseline 283/2363 (12.0) 300/2369 (12.7).47 New cases by year 1 68/1870 (3.6) 52/1810 (2.9).19 New cases by year 3 140/1631 (8.6) 118/1578 (7.5).25 *Diabetes is defined as (1) a participant report of diabetes or (2) a fasting serum glucose level of 7.8 mmol/l or higher ( 140 mg/dl), or (3) report or use of antidiabetic medication. Two participants reporting use of insulin at baseline (an exclusion criterion) and 2 participants with undefined diabetic status at baseline were excluded from the analysis. PARTICIPANTS WHO REPORTED DIABETES The prevalence of diabetes (as defined earlier) at baseline and the number of new cases identified at year 1 and year 3 examinations are shown in Table 1. As anticipated in an older cohort, rates were high and increased over time but the number of incident cases did not differ between treatment groups at year 1 (P=.19) or year 3 (P=.25). At 3 years, the number of new cases in the active treatment group slightly exceeded that reported in the placebo group (140 [8.6%] vs 118 [7.5%]).. EVALUATION OF RISK FACTOR CHANGE IN SUBGROUPS To evaluate the effect of antihypertensive treatment on CVD risk factors in SHEP, risk factor changes were assessed in subgroups of participants who had glucose or triglyceride levels measured while they were fasting, and a somewhat larger group who had total and HDL cholesterol, uric acid, potassium, and creatinine levels measured (fasting or nonfasting) at baseline and at follow-up years 1 and 3. (Table 2). Table 2 shows levels of risk factors at baseline, year 1, and year 3 examinations in participants in the active treatment and placebo groups, with each subdivided into diabetic or nondiabetic groups at the baseline examination. Since changes in the diabetic and nondiabetic participants were in the same direction and of similar magnitude, the groups were combined in Table 3. Table 3 shows the change in each risk factor level from baseline to the year 1 and year 3 examinations within the active treatment and placebo groups, a test of the change within the group from baseline to year 1 or year 3, and a test of the change within the group from year 1 to year 3. In addition, the difference of the change between groups was assessed. Randomization was successful, with no significant risk factor differences at the baseline examination between those assigned to active treatment or placebo groups. By the year 1 examination, trends over time were apparent in both groups, and several differences were observed between the active and placebo treatment groups. Fasting glucose levels increased in both groups but were significantly higher in the active treatment group. At the year 3 examination this difference persisted 743

4 Table 2. Serum Biochemical Values by Diabetic Status and Treatment Group Diabetic Patients Nondiabetic Patients Active Drug (n = 283) (n = 300) P Active Drug (n = 2080) (n = 2069) P Fasting serum glucose, n Mean (SD) 8.97 (3.4) [161.5 (60.7)] 8.58 (2.8) [154.6 (51.0)] (0.67) [99.9 (12.17)] 5.55 (0.65) [100.0 (11.8)].94 % 11.1 [200] n Mean (SD) 9.61 (3.67) [173.1 (66.1)] 9.18 (2.99) [165.4 (53.7)] (1.15) [105.7 (20.7)] 5.68 (0.92) [102.2 (16.5)].001 % 11.1 [200] n Mean (SD) 9.83 (3.96) [177.2 (71.3)] 9.24 (3.18) [166.4 (57.2)] (1.35) [107.7 (24.4)] 5.84 (1.14) [105.2 (20.5)].005 % 11.1 [200] Total serum cholesterol, n Mean (SD) 5.96 (1.19) [230.6 (46.2)] 5.91 (1.17) [228.5 (45.2)] (1.17) [237.7 (45.3)] 6.12 (1.12) [236.6 (43.2)].45 % 7.76 [300] n Mean (SD) 6.10 (1.28) [235.7 (49.4)] 5.97 (1.06) [230.7 (41.1)] (1.21) [242.6 (46.7)] 6.15 (1.14) [237.6 (44.0)].002 % 7.76 [300] n Mean (SD) 5.95 (1.19) [230.0 (45.9)] 5.89 (1.13) [227.9 (43.9)] (1.13) [235.6 (43.5)] 5.99 (1.06) [231.8 (40.8)].009 % 7.76 [300] HDL cholesterol, n Mean (SD) 1.22 (0.33) [47.2 (12.8)] 1.24 (0.34) [48.1 (13.27)] (0.38) [54.6 (14.6)] 1.40 (0.38) [54.2 (14.5)].47 %.9051 [35] n Mean (SD) 1.27 (0.34) [48.9 (13.2)] 1.31 (0.35) [50.6 (13.5)] (0.40) 55.4 (15.6)] 1.45 (0.39) [56.0 (15.1)].27 %.9051 [35] n Mean (SD) 1.18 (0.31) [45.8 (12.1)] 1.24 (0.35) [47.8 (13.5)] (0.40) [52.3 (15.4)] 1.36 (0.38) [52.7 (14.6)].40 %.9051 [35] Fasting triglycerides, n Mean (SD) 2.07 (1.40) [183.2 (123.7)] 1.94 (1.33) [171.5 (118.2)] (0.94) [139.1 (82.8)] 1.56 (0.91) [138.1 (80.4)].75 % [400] n Mean (SD) 2.35 (1.59) [207.8 (141.2)] 1.99 (1.17) [176.2 (104.0)] (1.11) [159.9 (98.5)] 1.64 (0.96) [145.3 (85.5)].001 % [400] n Mean (SD) 2.36 (1.75) [208.6 (154.7)] 2.22 (1.74) [196.5 (154.5)] (1.11) [159.8 (98.6)] 1.68 (1.00) [148.4 (88.7)].003 % [400]

5 Table 2. Serum Biochemical Values by Diabetic Status and Treatment Group (cont) Diabetic Patients Nondiabetic Patients Active Drug (n = 283) (n = 300) P Active Drug (n = 2080) (n = 2069) P Uric acid, µmol/l [mg/dl] n Mean (SD) (77.5) [5.1 (1.3)] (78.2) [5.2 (1.3)] (81.7) [5.4 (1.4)] (81.2) [5.3 (1.4)].14 % [10] * n Mean (SD) (103.1) [5.8 (1.7)] (90.1) [5.4 (1.5)] (102.0) [6.4 (1.7)] (84.7) [5.5 (1.4)].001 % [10] * n Mean (SD) (108.8) [5.9 (1.8)] (91.1) [5.4 (1.5)] (96.3) [6.3 (1.6)] (87.2) [5.7 (1.5)].001 % [10] * Potassium, mmol/l n Mean (SD) 4.5 (0.6) 4.5 (0.4) (0.5) 4.5 (0.4).97 % n Mean (SD) 4.2 (0.5) 4.4 (0.4) (0.5) 4.5 (0.4).001 % n Mean (SD) 4.2 (0.5) 4.4 (0.4) (0.5) 4.4 (0.5).001 % Creatinine, µmol/l [mg/dl] n Mean (SD) 93.8 (20.0) [1.1 (0.2)] 93.9 (20.6) [1.1 (0.2)] (20.9) [1.0 (0.2)] 91.7 (20.1) [1.0 (0.2)].38 % [2.0] * n Mean (SD) 99.0 (24.6) [1.1 (0.3)] 96.2 (27.6) [1.1 (0.3)] (24.2) [1.1 (0.3)] 92.8 (21.8) [1.0 (0.2)].001 % [2.0] * n Mean (SD) 98.6 (29.2) 1.1 (0.3)] 94.6 (27.9) [1.1 (0.3)] (26.8) [1.1 (0.3)] 92.2 (28.1) [1.0 (0.3)].003 % [2.0] * Body mass index, kg/m 2 n Mean (SD) 28.7 (5.2) 29.0 (5.4) 27.4 (4.9) 27.3 (5.0) n Mean (SD) 28.3 (5.4) 28.6 (5.1) 26.9 (4.8) 27.0 (4.9) n Mean (SD) 28.2 (5.2) 28.4 (5.0) 26.8 (4.7) 26.8 (5.0) Weight, kg n Mean (SD) 77.5 (16.0) 77.5 (16.2) 72.5 (15.0) 72.3 (15.0) n Mean (SD) 76.1 (15.7) 76.7 (15.7) 71.6 (14.8) 71.7 (15.0) n Mean (SD) 75.9 (15.2) 75.4 (15.3) 70.9 (14.8) 70.7 (15.2) * 2 Test not reliable due to small numbers. 745

6 Table 3. Change in Serum Biochemical Values, Body Mass Index, and Weight in and Active Treatment Groups* Active P Fasting serum glucose, n Mean (SD) (1.31) [+6.5 (23.7)] (1.22) [+2.5 (22.0)].001 P Mean (SD) (1.69) [+9.2 (30.5)] (1.42) [+5.6 (25.6)].009 Mean (SD) (1.67) [+2.6 (30.1)] (1.35) [+3.1 (24.3)].746 P Total serum cholesterol, n Mean (SD) (0.78) [+5.4 (30.2)] (0.85) [+1.2 (33.0)].001 P Mean (SD) 0.05 (0.86) [ 2.1 (33.3)] 0.14 (0.82) [ 5.3 (31.7)].007 P Mean (SD) 0.19 (0.78) [ 7.4 (30.3)] 0.17 (0.85) [ 6.5 (32.7)].419 HDL cholesterol, n Mean (SD) (0.23) [+1.2 (8.9)] (0.25) [+1.8 (9.8)].086 Mean (SD) 0.06 (0.25) [ 2.4 (9.8)] 0.04 (0.27) [ 1.4 (10.4)].005 Mean (SD) 0.10 (0.23) [ 3.7 (8.9)] 0.08 (0.24) [ 3.2 (9.2)].168 Fasting triglycerides, mmol/l n Mean (SD) (0.71) [+21.1 (63.2)] (0.64) [+48 (57.0)].001 P Mean (SD) (0.86) [+24.5 (76.2)] (0.72) [+7.9 (63.4)].001 Mean (SD) (0.83) [+3.8 (73.6)] (0.66) [+3.1 (58.6)].92 P Uric acid, µmol/l n Mean (SD) (69.1) [+1.0 (1.2)] +9.7 (53.2) [+0.2 (0.9)].001 Mean (SD) (73.5) [+0.9 (1.2)] (63.5) [+0.3 (1.1)].001 Mean (SD) 5.0 (75.0) [ 0.1 (1.3)] +9.0 (63.8) [+0.2 (1.1)].001 P

7 Table 3. Change in Serum Biochemical Values, Body Mass Index, and Weight in and Active Treatment Groups (cont) Active P Potassium, n Mean (SD) 0.43 (0.55) 0.07 (0.49).001 Mean (SD) 0.46 (0.57) 0.16 (0.57).001 Mean (SD) 0.02 (0.48) 0.10 (0.56).001 P Creatinine µmol/l [mg/dl] n Mean (SD) +4.3 (15.5) [+0.05 (0.2)] +1.3 (14.9) [+0.01 (0.2)].001 Mean (SD) +3.3 (19.4) [+0.04 (0.2)] +0.5 (21.7) [+0.01 (0.2)].001 P Mean (SD) 1.0 (18.5) [ 0.01 (0.2)] 0.7 (21.3) [ 0.01 (0.2)].73 P Body mass index, kg/m 2 n Mean (SD) 0.5 (1.9) 0.3 (2.4).04 Mean (SD) 0.6 (2.4) 0.6 (2.6).82 Mean (SD) 0.2 (2.0) 0.3 (2.3).07 Weight, kg n Mean (SD) 1.2 (3.2) 0.8 (3.9).001 Mean (SD) 2.0 (4.6) 1.8 (5.0).32 Mean (SD) 0.8 (4.1) 1.0 (4.3).10 *Test for significance of differences within groups over time and across groups at a given time interval. (P.01). Similar increments were seen among participants with diabetes (Table 2), although differences between active treatment and placebo groups were not significant (P=.14). To further assess the possibility that a small number of subjects receiving diuretics might have had a marked increase in fasting glucose levels, the percentage of subjects in the active and placebo groups with a greater than 20% increase between baseline and year 3 was determined. Such increases were found in 119 (13.8%) of the active treatment recipients and 91 (11.3%) of the placebo recipients (P=.12). Differences between active treatment and placebo groups were also seen in other risk factor levels at years 1 and 3. Total cholesterol levels decreased in both groups over 3 years but were slightly higher at year 3 in the active group compared with the placebo treatment group (P.01). Levels of HDL cholesterol decreased after year 1 in both groups. By the year 3 examination, levels were slightly ( 0.02 mmol/l [ 0.7 mg/dl]) but significantly lower in the active treatment group. Total triglyceride levels rose slightly (+0.09 mmol/l [+8 mg/dl]) for 3 years in the placebo group. A greater increase in triglyceride levels (+0.28 mmol/l [+24 mg/dl]) was observed in the active treatment group, and levels between groups differed significantly for both years 1and3(P.001). A similar pattern was seen in uric acid levels, with a greater increase in the active treatment group (+53 µmol/l) than in the placebo group (+19 µmol/l) (P.001). For all these variables, changes observed in participants with diabetes were similar to those seen in participants without diabetes. Because of the smaller sample size, few differences among treatment groups were significant among patients with diabetes. In the placebo group, mean potassium levels decreased slightly (-0.2 mmol/l; P.001) over 3 years. In 747

8 Fasting Glucose, mmol/l Year 0 Year 1 Year 3 Fasting Triglycerides, mmol/l Year 0 Year 1 Year Active Treatment 6 Active Treatment Fasting Glucose, mmol/l Fasting Triglycerides, mmol/l < Fasting Glucose Range, mmol/l >6.3 0 < Fasting Glucose, mmol/l >6.3 Figure 1. Concentrations of fasting glucose at baseline, year 1, and year 3 within quintiles of initial (baseline) fasting glucose concentration in the placebo and active treatment groups. To convert glucose levels to conventional units (milligrams per deciliter), divide the values given by contrast, in the active treatment group, potassium levels were lower for both years 1 ( 0.4 mmol/l; P.001) and 3 ( 0.5 mmol/l; P.001) and levels were significantly lower than those measured in the placebo group, although only a small percentage of patients had levels less than 3.5 mmol/l. At baseline, the frequency of serum potassium levels below 3.5 mmol/l was 0.4% in both treatment groups (8 active, 9 placebo). At year 3, it was 166 (8.5%) in the active treatment group and 36 (1.9%) in the placebo group (P.001). During the study, recorded use of supplemental potassium was 554 (23.4%) in the active treatment group and 292 (12.3%) in the placebo group. Serum creatinine concentrations increased in both active and placebo treatment groups with a greater increase (+2.8 µmol/l [+0.03 mg/dl]; P.001) in the active treatment group. However, at baseline and years 1 and 3 the total percentage of participants with values exceeding µmol/l (2 mg/dl) did not differ between active drug and placebo treatment groups. Weight and body mass index (measured as the weight in kilograms divided by the square of the height in meters) decreased in both the active treatment and placebo groups for 3 years, but no differences were seen between groups (Tables 2 and 3). CHANGE WITHIN GROUPS CLASSIFIED BY QUINTILE OF BASELINE FASTING GLUCOSE LEVELS Figure 2. Concentrations of fasting total triglycerides at baseline, year 1, and year 3 within quintiles of initial (baseline) fasting glucose concentrations in the placebo and active treatment groups. To convert glucose levels to conventional units (milligrams per deciliter), divide the values given by To investigate the possibility of greater adverse effects among those with higher levels of fasting glucose, the 1663 participants with complete fasting glucose data were divided into quintiles on the basis of their baseline fasting glucose level. Changes over time within these quintiles are presented for fasting glucose levels in Figure 1 and for fasting triglyceride levels in Figure 2. In Figure 1, fasting glucose levels are shown at baseline and at years 1 and 3 within quintiles of initial fasting glucose levels for both the placebo and active treatment groups. In both groups, higher initial fasting glucose levels were associated with higher subsequent fasting glucose levels (P.001). In both groups, fasting glucose levels increased over time in each of the quintiles. The increase in fasting glucose levels within each quintile was slightly greater in the active treatment than in the placebo treatment group. Much of the increase observed over 3 years of follow-up was evident at the year 1 follow-up examination. There was no evidence of a significant widening of the differences between the 2 treatment groups in any of the quintiles between years 1 and 3. While the increase in fasting glucose levels was slightly greater in the fifth quintile of the active treatment group, no subgroup at a particularly high risk of major increase in fasting glucose levels was identified. A similar pattern was seen in the 727 men and women followed up for 4 years (data not shown). Thus, although the number of subjects with complete test results is limited, there was little or no evidence for up to 4 years of follow-up that low-dose diuretic therapy with chlorthalidone produced a clinically significant increase in fasting glucose levels. In Figure 2, fasting total triglyceride levels are shown at baseline and at years 1 and 3 within quintiles of baseline fasting glucose levels for both placebo and active treatment groups. In both groups, except for 1 active treatment quintile (fasting glucose level, mmol/l [97-748

9 103 mg/dl]), higher baseline fasting glucose levels were associated with higher baseline and subsequent fasting triglyceride levels (P.001). In both placebo and active treatment groups, fasting triglyceride levels increased over time in each of the quintiles. The increase in fasting triglyceride levels within each quintile was greater in the active treatment than in the placebo group (P.001). Again, most of the increase observed during 3 years of follow-up was evident at the year 1 examination. There was no evidence of a further widening of the differences between the 2 treatment groups in any of the quintiles between years 1 and 3. While the increase in fasting triglyceride levels was greatest in the fifth quintile of the active treatment group, greater increases were seen in the active treatment than in the placebo group in all quintiles. A similar pattern was observed in the 727 participants followed up for 4 years (data not shown). Thus, the pattern of change for fasting triglyceride levels was similar to that seen for fasting glucose levels. COMMENT In population studies, elevated blood pressure is associated with an increased risk of stroke and multiple cardiac diseases. 29 Although antihypertensive therapy reduces stroke, coronary events, and other CVD, in a metaanalysis of clinical trials the expected reductions compared with observed reductions in cardiac disease are less than those found for stroke. 18 Several explanations have been proposed for this difference, including purported adverse effects of antihypertensive therapy. While clearcut, short-term effects of some antihypertensive drugs have been demonstrated, the relation of these changes to rates of clinical disease remains uncertain. These concerns may be especially pertinent for older patients, since both hypertension and glucose intolerance are common and often occur together in this age group. 30 The SHEP study provided the opportunity to look at the magnitude of the effects of low-dose, diuretic-based therapy over several years, both on risk factor levels (this article) and rates of CVD events. 24,25 The SHEP data are particularly valuable since the sample size is large, follow-up is relatively long, and evaluations could be performed on subsets of groups based on baseline fasting glucose levels to determine whether subgroups of the population may be at particular risk from the negative effects of diuretic-based therapy. At the conclusion of SHEP, participants in the active treatment group relative to the placebo group had lower blood pressures, a 36% reduction in incidence of stroke, a 27% reduction in coronary heart disease, and a 32% reduction in all major cardiovascular events. 24 In addition, in a post hoc analysis, similar reductions were found among those reporting diabetes at entry into SHEP. 25 Data from the 3- and 4-year follow-up of a subset of SHEP participants indicate that low-dose diuretic therapy had modest and no progressive effects on levels of fasting glucose, total cholesterol, and HDL cholesterol. While greater increases were observed in fasting triglyceride levels among active treatment participants, isolated hypertriglyceridemia appears to be less important as a risk factor for CVD. 31 A similar comment applies to the greater increase in uric acid levels seen in the active treatment group. In a 26-year follow-up of participants in the Framingham Heart Study, elevated uric acid levels were associated with several other CVD risk factor abnormalities and with use of diuretics but were not an independent CVD risk factor. 32 One feature of this analysis is an attempt to look for differential adverse effects of diuretic therapy in subgroups of participants with increasing levels of baseline glucose. At least among men with diabetes, the presence of other CVD risk factors increases the risk of CVD mortality more steeply than it does in men without diabetes. 33 Several groups of hypertensive patients may be at special risk of a deterioration in glucose tolerance. Patients with impaired but clinically unrecognized glucose intolerance are potentially at increased risk for an adverse effect from these drugs. In the SHEP study, no evidence of such a differential deterioration among those receiving active treatment was apparent across subgroups divided into quintiles based on baseline fasting glucose levels. Although glucose tolerance testing was not performed in SHEP, it is known that more than 50% of elderly men and women have abnormal glucose tolerance, 20 and SHEP baseline data indicate that the prevalence of diabetes was close to that expected for an older cohort. 21 Failure to detect any major increase in fasting glucose levels in SHEP indicates that these drugs can be used with little short-term chance of precipitating excess diabetes in older patients. A second potentially vulnerable group includes patients with established diabetes. Data from SHEP are more limited in this area since only 10% of subjects reported previously diagnosed diabetes at baseline. 27 Nevertheless, lack of a major change in risk factors among subjects with the upper glucose quintile suggests that this group also did not have major adverse effects from the low doses of diuretics and supplemental -blockers used in the SHEP protocol. Finally, use of diuretics has been associated with electrolyte disturbances, particularly potassium and magnesium depletion. Potassium depletion has been postulated as a mechanism by which diuretics increase insulin resistance, 34 an alteration that could contribute to several CVD risk factor abnormalities. 35 In addition, potassium depletion has also been proposed as a reason for the observed excess in sudden cardiac death seen in hypertensive patients receiving diuretic therapy. 36,37 In SHEP, although a greater decrease in potassium levels was seen in the active treatment group, severe hypokalemia was rare. In addition, SHEP participants received potassium supplementation if levels decreased to less than 3.5 mmol/l. Thus, provided that hypokalemia is appropriately treated, the notable reductions in chronic complications of hypertension in SHEP support the use of lowdose diuretic therapy in the initial treatment of hypertension. However, potential problems may be associated with higher doses of diuretics. 36 Although SHEP data indicate that low-dose diuretics can be used in groups with a high prevalence of glucose intolerance, caution must be exercised in both the interpretation and extrapolation of SHEP findings to the 749

10 general population. Since the SHEP participants were all older than 60 years and free of severe disease, high-risk subjects who either died prior to this age or developed severe disease were excluded from the cohort. Additionally, the data reported are from subjects classified according to original treatment assignment (intention to treat) and some subjects in the placebo group may have received diuretic therapy during the course of the study. Similarly, some subjects who received active treatment stopped diuretic therapy. While of theoretical concern, the maintenance of a significant separation in SBP ( 10 mm Hg) between the active and placebo treatment groups suggests that this is not a problem. Finally, the period of initial drug withdrawal in 33% of the cohort was sufficient to assess changes in blood pressure but may not have allowed the return of other risk factor levels to their untreated baseline levels. When analyses were performed by prior treatment category as well as randomization category, similar patterns were seen in those who had received prior antihypertensive therapy and those who had not. The adverse effects on CVD risk factor levels previously reported for those receiving diuretic therapy were seen in the SHEP cohort but appeared to be limited in magnitude. The SHEP antihypertensive regimen led to reductions in stroke, coronary heart disease, and all CVDs 24 in the whole cohort and those participants who reported diabetes at baseline. 25 It is necessary to consider how the SHEP cohort may differ from other studies reported during the past 3 decades. The greatest difference may be the low-dose diuretic therapy used in SHEP. Higher doses were used in several earlier studies. 38 Diuretics lower blood pressure and enhance effectiveness of other antihypertensive agents. 1 Given existing clinical trial data that indicate beneficial effects of blood pressure control with the use of diuretics on major cardiovascular complications, these SHEP data indicate that low doses of these drugs can be used safely in most elderly patients, even those with glucose intolerance or diabetes. Since some patients may be more susceptible to the adverse effects of these drugs, it is advisable to check CVD risk factor levels prior to initiating antihypertensive therapy and a few months after therapy is started. 15 These findings in SHEP, a clinical trial in a population with a high frequency of glucose intolerance, support the recommendations of the Joint National Committee on Detection, Evaluation, and Treatment of High Blood Pressure 1 that diuretics are indicated in the initial treatment of hypertension. Questions remain about optimal antihypertensive regimens. Many of those questions should be answered by the results of the ongoing Antihypertensive and Lipid Lowering Treatment to Prevent Heart Attack Trial. 39 Accepted for publication September 11, From the Division of Clinical Applications, the National Heart, Lung, and Blood Institute, National Institutes of Health, Bethesda, Md (Dr Savage and Ms Schron); the University of Texas School of Public Health, Houston (Ms Pressel and Dr Davis); John A. Burns School of Medicine, Honolulu, Hawaii (Dr Curb); University of Tennessee (Dr Applegate) and the Veterans Affairs Medical Center (Dr Rutan) Memphis; Rush-Presbyterian-St Luke s Medical Center (Dr Black); and Northwestern University Medical School (Dr Stamler) Chicago, Ill; Washington University School of Medicine, St Louis, Mo (Dr Cohen); University of California, San Francisco (Drs Frost and Smith); University of Minnesota, Minneapolis (Dr Gonzalez); University of Kentucky, Lexington (Dr Guthrie); University of Alabama, Birmingham (Dr Oberman); and University of Washington, Seattle (Dr Probstfield). This study was supported by contracts with the National Heart, Lung, and Blood Institute, and the National Institute on Aging, Bethesda, Md. Drugs were supplied by the Lemmon Co, Sellersville, Pa, Wyeth Laboratories/Ayerst Laboratories, St David s, Pa, AH Robbins Co, Richmond, Va, and Stuart Pharmaceuticals, Wilmington, Del. Reprints: SHEP Coordinating Center, University of Texas, Houston, Health Science Center, School of Public Health, 1200 Herman Pressler St, Suite 801, Houston, TX REFERENCES 1. Joint National Committee on Detection, Evaluation, and Treatment of High Blood Pressure. The Fifth Report of Joint National Committee on Detection, Evaluation, and Treatment of High Blood Pressure (JNC V). Arch Intern Med. 1993; 153: Weber MA, Laragh JH, for The Fifth Report of Joint National Committee on Detection, Evaluation, and Treatment of High Blood Pressure (JNC V). Hypertension: steps forward and steps backward. Arch Intern Med. 1993;153: Lewis PJ, Kohner EM, Petrie A, Dolleery CT. Deterioration in glucose tolerance in hypertensive patients on prolonged diuretic treatment. Lancet. 1976;1: Perez-Sabel E, Caralis PV. Thiazide-induced disturbances in carbohydrate, lipid and potassium metabolism. Am Heart J. 1983;106: Ames RP. Metabolic disturbances increasing the risk of coronary heart disease during diuretic-based antihypertensive therapy: lipid alterations and glucose intolerance. Am Heart J. 1983;106: Weinberger MH. Antihypertensive therapy and lipids: paradoxical influences on cardiovascular disease risk. Am J Med. 1986;80: The Working Group on Hypertension in Diabetes. Statement on hypertension in diabetes mellitus: final report. Arch Intern Med. 1987;147: Frishman WH. Beta adrenergic receptor blockers: adverse effects and drug interactions. Hypertension. 1988;11(suppl 2): Houston MC. The effects of antihypertensive drugs on glucose intolerance in hypertensive nondiabetics and diabetics. Am Heart J. 1988;115: Weinberger MH. Diuretics and their side effects: dilemma in the treatment of hypertension. Hypertension. 1988;11(suppl 2): Dall Aglio E, Strata A, Reaven GM. Abnormal lipid metabolism in treated hypertensive patients with non-insulin dependent diabetes mellitus. Am J Med. 1988; 84: Chait A. Effects of antihypertensive agents on serum lipids and lipoproteins. Am J Med. 1989;86(suppl 1B): Fhu MMT, Sheu WHH, Shen DC, Wu DA, Chen YDI, Reaven GM. Metabolic effects of diuretic and beta blocker treatment of hypertension in patients with non-insulin dependent diabetes mellitus. Am J Hypertens. 1990;3: Stein PP, Black HR. Drug treatment of hypertension in patients with diabetes mellitus. Diabetes Care. 1991;14: The National High Blood Pressure Education Program Working Group. The National High Blood Pressure Education Program Working Group report on hypertension in diabetes. Hypertension. 1994;23: Bengtsson C, Blohme G, Lapidis L, et al. Do antihypertensive drugs precipitate diabetes? BMJ. 1984;289: Skarfors ET, Lithell HO, Selinus I, Alber H. Do antihypertensive drugs precipitate diabetes in predisposed men? BMJ. 1989;298: Goodkin G. Mortality factors in diabetes. J Occup Environ Med. 1975;17:

11 19. Harris MI, Hadden WC, Knowler WC, Bennett PH. Prevalence of diabetes and impaired glucose tolerance and plasma glucose levels in US population aged years. Diabetes. 1987;36: Curb JD, Lee M, Jensen J, Applegate W. Systolic Hypertension in the Elderly Program (SHEP), part IV: baseline medical history findings. Hypertension. 1991;17 (suppl 2): Borhani NO, Applegate WB, Cutler JA, et al. Systolic Hypertension in the Elderly Program (SHEP): baseline characteristics of the randomized sample, I: rationale and design. Hypertension. 1991;17(suppl 2): The Systolic Hypertension in the Elderly Program (SHEP) Cooperative Research Group. Rationale and design of a randomized clinical trial on prevention of stroke in isolated systolic hypertension. J Clin Epidemiol. 1988;41: 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: Curb JD, Pressel SL, Cutler JA, et al, for the Systolic Hypertension in the Elderly Program Cooperative Research Group. Effect of diuretic-based antihypertensive treatment on cardiovascular disease risk in older diabetic patients with isolated systolic hypertension. JAMA. 1996;276: Labarthe DR, Blaufox MD, Smith WM, et al. Systolic Hypertension in the Elderly Program (SHEP): baseline characteristics of the randomized sample, II: baseline blood pressure and pulse rate measurements. Hypertension. 1991;17 (suppl 2): Hall WD, Davis BR, Frost P, et al. Systolic Hypertension in the Elderly Program (SHEP): baseline characteristics of the randomized sample, VII: baseline laboratory characteristics. Hypertension. 1991;17(suppl 2): Petrovitch H, Byington R, Bailey G, et al. Systolic Hypertension in the Elderly Program (SHEP): baseline characteristics of the randomized sample, II: screening and recruitment. Hypertension. 1991;17(suppl 2): MacMahon S, Peto R, Cutler J, et al. Blood pressure, stroke, and coronary heart disease, I: prolonged differences in blood pressure: prospective observational studies corrected for the regression dilution bias. Lancet. 1990;335: Collins R, Peto R, MacMahon S, et al. Blood pressure, stroke, and coronary heart disease, part II: short-term reductions in blood pressure: overview of randomised drug trials in their epidemiological context. Lancet. 1990;335: Kannel WB, McGee DL. Diabetes and cardiovascular risk factors: the Framingham Study. Circulation. 1979;59: NIH Consensus Conference. Triglyceride, high-density lipoprotein, and coronary heart disease. NIH Consensus Development Panel on Triglyceride, High-Density Lipoprotein, and Coronary Heart Disease. JAMA. 1993;269: Brand FN, McGee DL, Kannel WB, Stokes J III, Castelli WP. Hyperuricemia as a risk factor of coronary heart disease: The Framingham Study. Am J Epidemiol. 1985;121: Stamler J, Vaccaro O, Neaton JD, Wentworth D, for the MRFIT Intervention Trial Research Group. Diabetes, other risk factors, and 12-year cardiovascular mortality for men screened in the Multiple Risk Factor Intervention Trial. Diabetes Care. 1993;16: Helderman JH, Elahi D, Anderson DK, et al. Prevention of the glucose intolerance of thiazide diuretics by maintenance of body potassium. Diabetes. 1983; 32: Reaven GM. Role of insulin resistance in human disease. Diabetes. 1988;37: Siscovick DS, Raghunathan TE, Psaty BM, et al. Diuretic therapy for hypertension and the risk of primary cardiac arrest. N Engl J Med. 1994;330: Hoes AW, Grobbee DE, Lubsen J, Manin t Veld AJ, van der Does E, Hofman A. Diuretics, beta-blockers, and the risk for sudden cardiac death in hypertensive patients. Ann Intern Med. 1995;123: Psaty BM, Smith NL, Siscovick DS, et al. Health outcomes associated with antihypertensive therapies used as first-line agents. A systematic review and metaanalysis. JAMA. 1997;277: Davis BR, Cutler JA, Gordon DJ, et al. Rationale and design for the Antihypertensive and Lipid Lowering Treatment to Prevent Heart Attack Trial (ALLHAT). Am J Hypertens. 1996;9: