Effects of Reduction in Drugs or Dosage After Long-Term Control of Systemic Hypertension Edward D. Freis, MD, J.R. Thomas, MD, Susan G. Fisher, MS, Robert Hamburger, MD, Richard E. Borreson, MD, Kalman C. Mezey, MD, Bangshi Mukherji, MD, William W. Neal, MD, H. Mitchell Perry, MD, and James T. Taguchi, MD, for the Veterans Administration Cooperative Study Group on Antihypertensive Agents* The possibility of discontinuing+ompared to reducing- antihypertensive drug treatment was in- vestigated in 606 male hypertensive patients with entry diastolic blood pressure (BP) in the range of 50 to 114 mm Hg. Diastolic BP was controlled at 14 days. Candidates com- pleting this 6-month baseline phase with maintenance of BP control (n = 606) were included in the core study sample. Randomization and postrandomization phases: Two-thirds of the study patients were randomly as- signed to either dose reduction or drug discontinuation depending on drug regimen; the other one-third of the patients continued to receive their unchanged drug regi- men as a control group (Figure 1). This 2:l randomiza- tion scheme ensured adequate sample size for a second randomization of the dose reduction group later in the study. Patients in the low dose HCTZ group had the diuretic discontinued while high dose HCTZ patients were switched to low dose HCTZ. Low dose step II pa- tients had the step II drug discontinued but remained on high dose HCTZ and high dose step II patients re- ceived a low dose of the step II medication and contin- ued on high dose HCTZ. The appearance of all medica- tions and placebos was kept uniform to maintain the double-blinded design. This first postrandomization lasted 6 months, as did all subsequent study phases. Patients maintaining goal BP were continuously treated and followed. Patients not maintaining goal BP Screened for Eligibility (1316) ----- lnellglble (444) + Titrated on Study Drugs (672) ----w Excluded* 1 (147) Entered Baseline (725) ----- Excluded o (119) I I I I LD HCTZ (209, HD "C,Z 1146) LD step II 1135, HD step II 11161 FIGURE 1. Flow sheet showing the prerandomization drug titration phase using the 4 step regimens as needed to control dii- stolic blood pressure to <66 mm Hg followed by 6-month treatment maintenance phase. Eligible patients then entered the post- randomization period where drugs were reduced in two-thirds of patients and remained unchanged in one-third. After 6 months, two-thirds of eligible patients underwent a second medication reduction. *Excluded patients include those in whom blood pres- sure was not controlled within study limits and patients with other medical or nomnediil problems. **Reasons for postran- domization terminations are similar to prerandomization reasons with the addiion of patients being unable to complete the 6- month phase because the study ended (see text). LD = low dose. THE AMERICAN JOURNAL OF CARDIOLOGY MARCH 15,1989 703 REDUCTION VERSUS DISCONTINUATION OF ANTIHYPERTENSIVE DRUGS TABLE I Background Characteristics of Randomized Patients Pts (n) Mean Age Ws) Race White Black Other Before Randomizatron BP* (mm Hg) Prior Rx Yes No Low dose HCTZ No change 72 58f8 31 38 3 126/80 i 13/6 67 5 Dose reduced 137 57f9 60 75 2 129/82 zk 13/6 124 13 High dose HCTZ No change 47 56i9 22 25 0 128/84&13/5 45 2 Dose reduced 99 59k8 44 54 1 129/82 k 15/5 88 11 Low dose step II No change 48 56zk 10 24 22 2 126/82 zk 10/6 44 4 Dose reduced 87 56% 10 40 47 0 126/83 f 13/5 84 3 Hugh dose step II No change 34 56i9 19 15 0 125/84f11/5 32 2 Dose reduced 82 55* 10 43 38 1 128/84 f 14/5 81 1 All treatment arms 606 57f9 283 314 9 127/83 f 136 565 41 ' Durq treatment to lower BP. All f values are mean * standard d&&on. No slgnlflcant differences I" background characteristics between treatment groups. BP = blood pressure; HCTZ = hydrochlorothwide. TABLE II Percent of Patients Remaining Under Blood Pressure Control at Various Times During 30 Months After Randomization Percent Controlled After Randomization (months) Group Low dose HCTZ unchanged Low dose HCTZ drscontinued High dose HCTZ unchanged High dose HCTZ reduced to LD High dose HCTZ second stepdown Low dose STEP II unchanged Low dose STEP II discontinued High dose STEP II unchanged High dose STEP II reduced to LD Hugh dose STEP II second stepdown 6 18 94f3 89i4 55f4 35zk4 79*7 55k9 84zk4 58+9 * 20f6 96i3 84k6 56k5 37i6 85i6 70fll 62i6 47f9 * 17 i 6 30 86f4 23f5 12f7 84i6 16f8 70f 11 47 f 9 * Pabents were randomized to Mial drug reduction for first 6 months after ran- domlzatlon. Second reduction began after this period. iSome data were omitted at the Dmonth follow-up because the number of patents were too few to provide a reliable statement. I were considered to have reached a study endpoint; their drugs were unblinded and appropriate increases were made in study medication to achieve control of BP. Pa- tients unable to maintain goal BP on the maximum study regimen were treated in hypertension clinics on nonprotocol regimens. Following the first 6-month postrandomization phase, patients maintaining BP control in whom medi- cation was originally reduced but not discontinued-i.e., the high dose HCTZ and the high dose step II regi- mens-were randomized a second time. Thus, two- thirds received a second dose reduction/drug discontin- uation while one-third remained on the same dosage. In that way, two-thirds of the high dose HCTZ patients initially reduced to low dose HCTZ now had it discon- tinued. High dose step II patients, in whom the step II medication had been previously reduced, now had the step II medication discontinued although these patients continued to receive high dose HCTZ. After this second randomization no further changes in medication were made unless a study endpoint was reached. Patients 704 THE AMERICAN JOURNAL OF CARDIOLOGY VOLUME 63 who maintained control BP were followed for up to 30 months after randomization. Throughout the study medication compliance was measured. Patients had to return their medication bot- tles containing the remaining unused tablets (we delib- erately provided a small excess of tablets). Patients were considered compliant if they returned <30% of the number of tablets that should have been taken without exceeding 110% of the number to be taken. The return of an empty bottle was considered a violation. Tablet counts were done on every visit throughout the trial. In cases in which therapy was stopped because of illness, surgery, trauma, etc., the code-labeled medica- tions could be reinstituted if the interruption was 17 days during the prerandomization phases or 114 days during the postrandomization period. Otherwise, pa- tients were terminated from the trial. Data analysis: Descriptive statistics including 95% confidence intervals were used to describe the percent- age of patients maintaining BP control in each study phase. Long-term maintenance of BP was estimated us- ing Kaplan-Meier techniques. Differences between background characteristics, laboratory values and side effects were examined by the chi-square test of homoge- neity and the Student t test. Logistic regression proce- dures were used to identify variables as potential predic- tors of a patient's ability to maintain BP control. In all cases a 2-tailed p value of 0.05 was considered statisti- cally significant. All patients entering the trial signed an informed consent statement after the procedure had been fully explained. RESULTS After the 6-month baseline treatment period 606 pa- tients maintained BP control and were therefore eligible for randomization. Of these, 209 were receiving the low dose HCTZ, 146 the high dose HCTZ, 135 the low dose step II regimen and 116 the high dose step II regi- men. The major characteristics of randomized patients are listed in Table I. There were no significant differences between treatment groups. The mean age of the sample was 56.9 years. The racial distribution was 52% black and 47% white. There was no significant difference in pretreatment BP between the control group and the drug reduction group within each drug regimen cate- gory. Ninety-three percent of the patients entering the trial had been receiving antihypertensive therapy previ- ously. First six-month postrandomization phase: Depend- ing on the therapeutic regimen 55 to 84% of patients on reduced dosage remained within the acceptable levels of diastolic BP during the first 6 months (Table II). Low dose HCTZ was the only regimen in which all antihy- pertensive drug therapy had been discontinued. Al- though 55% of these patients maintained goal BP throughout this treatment phase (Figure 2), this propor- tion was significantly less than the fully treated low dose HCTZ control patients (p 114 mm Hg on 2 visits or >119 mm Hg on 1 visit (2). Long-term follow-up: Patients were followed for a maximum of 30 months after randomization. Table III lists the number of patients available for entry into the first 3 postrandomization phases by regimen. Patient at- trition during these phases is also shown. The 3 main reasons for a constantly decreasing pool of patients are elevated BP, which was the major study endpoint; ter- -NO CHANGE `I ----STEP 2 REDUCED -- STEP 2 REDUCED ONCE TWICE 0 0 12 18 24 MONTHS FIGURE 5. Percent of patients on high dose step II regimens remaining under diastolic BP control during 20 months after randomization. During the first reduction (line with frequent interruptions) the step II drugs were reduced but not discon- tinued. HCTZ dosage remained unchanged. At 6 months some of the patients were subjected to a second reduction in whiih the step II drug was discontinued. Discontinuation of step II drug, despite continuation of HCTZ, was followed by a steep decline of patients remaining under diastolic BP control. 706 THE AMERICAN JOURNAL OF CARDIOLOGY VOLUME 63 such as cardiovascular complications, or administrative reasons such as refusal to return to clinic; and inability to complete all phases before the end of the study be- cause of sequential enrollment. Although 30-month re- sults are listed in Table II, it should be recognized that due to attrition these results are based on a very small sample of patients. Table II lists the decreasing percent of patients re- maining under BP control at 6-month intervals up to 30 months from the time of randomization. Among the low and high dose HCTZ groups the patients most effec- tively maintaining their BP over time were those in whom doses were reduced rather than discontinued. In fact, up to 18 months after randomization, the percent of high dose HCTZ patients remaining normotensive was essentially the same between the dose reduction and the dose unchanged control groups. Also, among the step II patients in whom dosage was reduced but not discontinued, nearly half maintained BP control up to 30 months. The long-term results in the patients whose drug was discontinued were not as favorable. Among patients having low dose HCTZ discontinued 23% were able to maintain BP control as long as 30 months (Table II). The patients originally given high dose HCTZ and who after 2 dose reductions were receiving no active drug fared worse: the percentage with BP control decreased to 12% after 30 months of follow-up (Table II, Figure 3). Results of discontinuation of low dose step II drug were also poor (Table II), even though patients contin- ued to take HCTZ. Some of these patients had begun prerandomization treatment with step II regimens. Many others who had begun with HCTZ failed to show BP control on the single drug and were, therefore, ad- vanced to step II. Over 30 months, major complications averaged 3% (n = 6) in the control group and 2% (n = 8) in the reduced dosage group. Thus, dose reduction did not appear to increase the risk of major complica- tions over the period of the study. Reestablishment of blood pressure control: During the course of the study there were 202 patients on re- duced drug dosages whose BP increased above the per- missible limits of goal BP and at least 5 mm Hg above the prerandomization level. These patients had their dosages retitrated. Of these 202 patients, 172 were re- controlled on their prerandomization medication and dosage. Thus, the great majority of the patients re- TABLE IV Biochemical Changes After Six Months of Study Treatment Low Dose HCTZ High Dose HCTZ Low Dose Step II High Dose Step II No No No No Change Reduction Change Reduction Change Reduction Change Reduction Glucose (mg/dl) (X 0 05551 mmol/llter) -3.7 -7.7 3.8 2.5 -7.0 3.4 -1.9 0.5 Potassium (mEq/L) X 1.0 mmol/llter) -0 1 0 6+ 0.2 0.2 0.1 -0.1 0.1 -0 1 Cholesterol (mg/dl) (X 0 02586 mmol/llter) -15.4 -19.9 -2.0 0.3 -18.9 -3.o* -21.9 1.3 Triglycerides (mg/dl) (X 3.397 mmol/llter) -35.2 -43.1 14.0 1.0 -1.2 32.7 31.2 -54.6 Unc Acid (mg/dl) (X 59.48 pmol/liter) -0.2 -0.9+ -0.2 -0.6 -0.1 -0.0 0 -0.2 Alkalme phosphatase (unlts/llter) (X 7.1 U/liter) -1.1 7.5' 2.1 2.4 -2.2 1.8 -2.1 -3.0 * p <0.05. f p l visit by 4% of patients. Gout and dizziness occurred in- termittently in 3% and 2% of the patients, respectively. There were no significant differences in the incidence of side effects between the drug reduction and the control groups or between the different regimens. No particular side effect was associated with any specific drug or regi- men. As estimated by pill counts there was no differ- ence in the level of compliance between the various treatment groups. Factors predictive of successful dose reduction: Characteristics predictive of successful reduction in medication were investigated using univariate and mul- tivariate analyses. Characteristics differing (p <0.20) between patients in whom medication was successfully reduced or discontinued and those with unsuccessful drug reduction were selected for inclusion in a logistic regression analysis. These characteristics included mean diastolic BP during baseline, variability of BP during baseline, patient compliance, time since diagnosis, he- moglobin, hematocrit, alkaline phosphatase, potassium, white blood cell count and pulse rate. Other factors con- sidered as potential covariates included in the statistical model were race, age, each drug regimen and treatment assignment, i.e., reduction of drug dosage or discontinu- ation. Variables that contributed significantly to the model included treatment assignment (p 89 mm Hg were those with higher levels THE AMERICAN JOURNAL OF CARDIOLOGY MARCH 15.1989 707 REDUCTION VERSUS DISCONTINUATION OF ANTIHYPERTENSIVE DRUGS of diastolic BP at randomization. By the time of the second dose reduction most patients were receiving min- imal or no therapy. In the patients receiving HCTZ alone and randomized to medication reduction, HCTZ had been stopped, while in the patients on step II treat- ment only HCTZ was maintained. No further reduc- tions in dosage were made after the second randomiza- tion. Despite minimal or no medication the rate at which patients exhibited return of hypertension dimin- ished after the first 12 months following randomization. This was probably because no further dosage reductions occurred during that period and also because patients most susceptible to lose BP control had experienced BP elevations during their first year in the trial. However, following the second reduction, at 6 months many of the patients failed to maintain normal BP when most anti- hypertensive drugs were discontinued (Figures 3 and 5). The critical factor, therefore, seemed to be whether the medication was reduced or completely discontinued. It is possible that after the first 6 months following the initial reduction, the patients remaining in the trial represented a population that for various reasons were better able to maintain their BP within normal limits following treatment discontinuation. Although not insti- tuted in this study, weight reduction and sodium restric- tion have been used with some success in preventing the return of elevated blood pressure after drug withdraw- al." The relative lack of purely drug-related side effects in this study applied to all drug regimens including re- serpine plus HCTZ. However, more than half of the patients received only 0.1 mg reserpine daily. These re- sults confirm our former study of small doses of reser- pine with HCTZ.`O The present results suggest that degree of elevation of the prerandomization diastolic BP was significantly related to the return of hypertension. The patient's BP variability before randomization also directly correlated with return of hypertension. It is also possible that some of the mildly hypertensive patients had only temporary elevations of BP when admitted to the trial and they returned spontaneously to normal after repeated visits.`* The present results suggest that after long-term treat- ment, dosage reduction is more effective in preserving antihypertensive control than is discontinuation of medi- cations. Dose reduction should be advantageous over the long term because smaller doses reduce the risk of ad- verse effects and decrease the cost of treatment. REFERENCES 1. Perry HM Jr, Schroeder HA. Studies in thecontrol of hypertension. VI. Some evidence for reversal of the process during hexamethonium and hydraluine thera- py. Circulation /956:13:528-536. 2. Dustan HP, Page IH, Tarazi RC, Frohlich ED. Arterial pressure responses to discontinuing antihypertensive drugs. Cimdation /968;37:370-379. 3. Thurm RH, Smith WM. On resetting of barostats in hypertensive patients. JAMA 1967:201:301-304. 4. Veterans Administration Cooperative Study Group on Antihypertensive Agents. Return of elevated blood pressure after withdrawal of antihypertensive drugs. Circulation 1975:5/:1107-1113. 5. Levinson PD. Khatri IM, Freis ED. Persistence of normal BP after withdrawal of drug treatment in mild hypertension. Arch intern Med 1982;142.2265-2268. 6. Maland LJ, Lutz LJ, Castle CJ. Effect of withdrawing diuretic therapy or blood pressure in mild hypertension. Hypertension 19835539-544. 7. Finnerty FA. Step-down therapy in hypertension: its importance in long-term management. JAMA 1981;246:2593-2596. 8. Finnerty FA Jr. Step-down treatment of mild systemx hypertension. ilnz J Cardiol 1984;53:1304-1307. 9. Veterans Administration Cooperative Study Group. Low doses YS standard doses of reserpine. A randomized double-blind multiclinic trial in patlcnts taking chlorthalidone. JAMA 1982;248:247/-2477. 10. Freis ED. Short-term versus long-term changes in serum cholesterol aith thiazide diuretics alone (letter). Lmret /984:/:14/4-1415. 11. Langford HG, Blaufox D, Oberman A, Hawkins CM. Curb JD. Cutter CR, Wassertheil-Smaller S, Pressel S, Babcock C, Abernethy JD. Hotchkiss J. Tyler M. Dietary therapy slows the return of hypertension after stopping prolonged medication. JAMA 19RS:2S3:657-664. 12. The Management Committee of the Australian Therapeutic Trial in Mild Hypertension. Untreated mild hypertension. Lancer /982;/:/85-19/ APPENDIX The following persons participated in the study. Chairman's Office: Barbara Gregory, BSN, Washington, DC. Consultant: Barry J. Materson, MD, Miami, Florida. Clinic Associates: Debbie Ohlen, PA, Patricia Kershen, PA, Dallas, Texas; Ruth Collins, RN, Houston; Phyllis Mangas, FNP, Dayton, Ohio; Lorretta Hoerman, PA, St. Louis, Missou- ri; Mary Jo Barsotti-Tracey, MS, Barbara Lesniak, East Or- ange, New Jersey; Anne M. Faiella, BSN, Catherine Cum- mings, RN, Boston, Massachusetts. Cooperative Studies Program Coordinating Center: Wil- liam G. Henderson, PhD, Jean Rowe, Laura Weber, MS, Mary Ellen Vitek, Hines, Illinois. Cooperative Studies Program Clinical Research Pharma- cy: Larry M. Young. RPh, Jane Weber-San-Hamel, PharmD, Dennis Toussaint, RPh, Mike Sather, RPh, Albuquerque, New Mexico. Operations Committee: Morten H. Maxwell, MD, Chair- man, Walter M. Kirkendall, MD, Theodore Colton, ScD. Cooperative Studies Program Central Administration: Daniel Deykin, MD, Janet Gold, Boston; James A. Hagans, MD, PhD, Ping Huang, PhD, Washington, DC. Hines Cooperative Studies Program Coordinating Center's Human Rights Committee: Lauren Lawson, PhD (Chairper- son), Hines Veterans Administration Medical Center, Hines, Illinois; Edgard Perez, Horace C. Dudley, PhD, Nancy Cahill, Paul Peterson, MD, William Upholt, PhD. 708 THE AMERICAN JOURNAL OF CARDIOLOGY VOLUME 63