|
|
| |||||||||||||||||||||||||||||||||||||||
Background We conducted a multicenter trial comparing two angiotensin-converting-enzyme inhibitors to determine whether effects on quality of life during antihypertensive therapy are uniform within this pharmacologic class of agents, and to relate the effects of the drugs on quality of life to objective adverse events, such as the loss of a job or the death of a spouse.
Methods After a four-week washout period when they received placebo, 379 men with mild-to-moderately-severe hypertension were randomly assigned to receive captopril (25 to 50 mg twice daily, with or without hydrochlorothiazide) or enalapril (5 to 20 mg per day, with or without hydrochlorothiazide) for 24 weeks. Blood pressure, quality of life, and life events were monitored. Differences between treatments were evaluated by calibrating measures of quality of life with objective life events.
Results Throughout the treatment period, no differences were found in blood pressure, frequency of withdrawal of patients from the study, or major side effects. Patients treated with captopril had more favorable changes in overall quality of life, general perceived health, vitality, health status, sleep, and emotional control (P<0.05 for each). The changes varied according to the quality of life at base line (P<0.001); patients with a low quality of life at base line remained stable or improved with either drug, whereas those with a higher quality of life remained stable with captopril but worsened with enalapril. The quality-of-life scales correlated with life events and symptom distress (P<0.001), and calibration analysis indicated that differences between treatments were clinically important.
Conclusions Two angiotensin-converting-enzyme inhibitors, captopril and enalapril, indistinguishable according to clinical assessments of efficacy and safety, had different effects on quality of life. Calibration with life events showed that drug-induced changes are substantial and that the different effects of these two agents on quality of life can be clinically meaningful.
Antihypertensive therapy is a special situation, in which the primary process being treated usually results in few symptoms. Antihypertensive agents of the same pharmacologic class typically produce similar clinical responses with regard to efficacy and major side effects. However, changes in the quality of life during therapy depend on more subtle reports by patients of drug effects on overall well-being and functioning and may therefore be more sensitive to differences between drugs. In this study, we sought, first, to ascertain whether two agents in the same general pharmacologic class would exert a similar influence on quality of life and, second, to translate the effects of drug therapy on quality of life into an easily appreciated index that is meaningful to clinicians. For that reason, we designed a randomized trial comparing the responses to two angiotensin-converting-enzyme inhibitors, captopril and enalapril, and attempted to relate changes in quality of life to unanticipated life events, using a procedure of calibration and validation that has been employed in previous studies16,17,18,19,20,21,22,23.
In a study by Croog et al.13 antihypertensive therapy with methyldopa and propranolol induced the expected frequency and severity of side effects and the corresponding negative effects on quality of life. The response to captopril, however, was surprising. Global measures of quality of life and many of the individual components of this measurement actually improved from base line during the study. Subsequent studies raised the interesting possibility that captopril had a positive influence on a number of central nervous system functions that are immediately relevant to quality of life24,25,26,27,28,29. We attempted in addition to assess in greater depth the nature of this putative positive response by investigating whether the patients with the lowest quality of life at the start of therapy would have the greatest potential for gain.
Methods
Clinical Trial
We screened 479 active men with hypertension who were 55 to 79 years of age at 25 clinical centers. Approval was secured from the relevant institutional review boards, and informed consent was obtained from all the patients. Hypertension considered appropriate for entry into this trial was defined as a diastolic blood pressure of between 90 mm Hg and 115 mm Hg, inclusive, measured with the patient seated, on two consecutive visits during an initial, four-week phase in which the patient received placebo. Patients had to be free of substantial cardiac, hematologic, renal, hepatic, metabolic, neoplastic, psychiatric, or symptomatic peripheral vascular disease. Patients were also excluded if they had a history of noncompliance or previous intolerance or known hypersensitivity to captopril, enalapril, or hydrochlorothiazide.
The trial comprised three phases: the initial phase, in which the patient received placebo in a blinded fashion (4 weeks), a double-blind phase involving titration of the study drug (10 weeks), and a double-blind phase of maintenance therapy (14 weeks). During the initial phase, a medical history, physical examination, screening biochemical tests, and quality-of-life assessment (including life events) were obtained. At week 0 (the beginning of the titration phase), qualifying patients were randomly assigned to receive the active drug or matching placebo. At each follow-up visit, three readings of blood pressure were made with the patient seated; they were obtained within 12 ±3 hours after the most recent dose. Repeated clinical, laboratory, and quality-of-life assessments were also obtained. The patients were not informed of their blood pressures or drug doses.
During the titration phase, the drug doses were titrated upward at weeks 2, 4, and 8 if the mean diastolic blood pressure exceeded the goal of 90 mm Hg. Four dose levels were used for enalapril given once per day: 5 mg, 10 mg, 20 mg, and 20 mg plus hydrochlorothiazide. Three dose levels were used for captopril given twice per day: 25 mg, 50 mg, and 50 mg plus hydrochlorothiazide. The diastolic blood pressure had to exceed 95 mm Hg for a patient to receive hydrochlorothiazide (50 mg per day), administered in blinded fashion, at the fourth follow-up visit (week 8). If the blood pressure rose above 100 mm Hg on two consecutive readings during week 10, the patient was excluded from the study. All the other patients continued to receive their titrated doses between weeks 10 and 24 at the levels used at week 8. Clinical follow-up was performed at weeks 8, 10, 14, 18, and 24, and quality-of-life follow-up at weeks 8, 18, and 24. Clinical and quality-of-life assessments were obtained whenever a patient withdrew prematurely from the study. If a patient's serum potassium level was less than 3.4 mmol per deciliter, liquid potassium supplements were prescribed.
The quality-of-life assessment included a comprehensive, self-administered questionnaire with the following scales: Psychological Distress (anxiety, depression, and loss of behavioral or emotional control), Psychological Well-Being (emotional ties, general positive affect, and life satisfaction), General Perceived Health (vitality, general health status, and sleep disturbance), Well-Being at Work or in Daily Routine, and Sexual-Symptom Distress. The patient also reported the degree of distress due to side effects and symptoms, the level of stress (on an analogue scale), and important adverse life events during the month before the visit. The contents of the specific scales and items are described in Appendix 213,14,30,31,32,33,34,35,36,37,38,39,40,41. The conceptualization of quality of life and the rationale for choosing these scales were based on previous studies13,14,15,36.
Statistical Analysis
The validity, reliability, and responsiveness42,43,44,45,46 of the quality-of-life scales were analyzed. Cronbach's alpha43 (measured each week when there was a visit) and test-retest reliability44 (measured at weeks 18 and 24) were found to be consistently high (>0.85). (More complete data on these measures are available elsewhere.*) This study was designed to detect changes in quality of life between treatment groups from base line to the end point (at 24 weeks or the last visit at which the patient could be evaluated). To standardize scores across subscales with different ranges of raw scores, reliability, and variances, we used a standard procedure, the responsiveness index45,46. Between weeks 18 and 24, both treatment and the response to treatment remained stable, and thus the standard deviation of the change during this period reflected the underlying instability of the measurement. The responsiveness index was computed by dividing the raw score by this measure of instability; it was subsequently indexed to life events. Except when units are specified for raw scores, all differences between treatment groups are expressed in responsiveness-index units (referred to simply as units). The multiple, correlated quality-of-life end points were analyzed with the SPSS computer package by (1) univariate linear models that used the average of all the subscales (equal weighting) as a single measure, the Overall Quality of Life scale, and (2) multivariate linear models to account for the covariance structure among correlated scales47,48. (The quality-of-life measures are described in Appendix 2.) Age, base-line quality of life, and titration level (monotherapy or combination therapy) were chosen as a priori covariates in the model on the basis of previous studies13,14. Once the difference between treatments on the overall and profile scales was determined to be statistically significant, the univariate model was used to identify the quality-of-life scales that were most affected by the different treatments. The study was designed to detect an unadjusted change in the Overall Quality of Life scale from base line that corresponded to 0.2 responsiveness-index unit and an unadjusted difference between treatments of 0.35 responsiveness-index unit, with a two-tailed test of significance (alpha = 0.05, power = 0.95). The chi-square statistic was used in the analysis of proportions, Pearson's correlation coefficient was used for tests of association, and ordinary least-squares linear and polynomial regression analyses were used to derive the calibration metric47,48. Means are presented with standard errors. P values were unadjusted for multiple comparisons and were all based on two-tailed tests of significance.
Results
Base-Line Characteristics
One hundred ninety-two eligible patients were randomly assigned to receive captopril, and 187 patients to receive enalapril. At base line, there were no significant differences between the two study groups in demographic variables, clinical characteristics, or quality-of-life scales (Table 1). Mean quality-of-life measures were comparable between the two groups, with scores similar to those found in previous studies,13,14 reflecting a relatively healthy population. Arthritis was the most commonly reported concomitant disease (29 percent), and diabetes the second most common (9 percent). Twelve percent of the patients reported that illness, disability, or a related condition restricted their daily activities to some degree.
|
There were no statistically significant differences between captopril and enalapril in their effects on systolic or diastolic blood pressure (Figure 1). By the end of the titration phase (at week 10), the distribution of patients according to dose level was comparable between the two treatments. Only 4 of the 136 types of adverse events encountered were reported in more than 5 percent of the patients, and there were no differences between the captopril and enalapril groups in the incidence of these adverse events: headache (15 percent vs. 16 percent), asthenia (9 percent vs. 10 percent), dizziness (8 percent vs. 7 percent), and cough (5 percent vs. 6 percent). Treatment led to no clinically important changes in laboratory values. Of the patients originally randomized, 40 (21 percent) withdrew from the captopril group, and 38 (20 percent) withdrew from the enalapril group before the completion of the 24 weeks of active therapy. The primary reasons for withdrawal were nearly identical, and the largest number in both groups withdrew because of adverse events: 18 (9 percent) in the captopril group, and 15 (8 percent) in the enalapril group. Other reasons for withdrawal, including lack of blood-pressure control, unrelated illnesses, cardiovascular events, violations of protocol, poor compliance, and loss to follow-up, each accounted for less than 3 percent of the overall number of patients in the two study groups.
|
Changes in quality of life were estimated for all patients (studied on an intention-to-treat basis) who had an end-point visit that could be evaluated, regardless of whether they withdrew early. Scores representing changes in end points were available for 95.8 percent of the patients assigned to captopril (184 patients) and 95.2 percent of the patients assigned to enalapril (178 patients). There was a statistically significant treatment effect on the Overall Quality of Life scale, with a positive change for captopril (+0.11 unit) and a negative change for enalapril (-0.11 unit; P = 0.040) from base line to the most recent available follow-up visit. The corresponding multivariate test statistic for the three major scales was also statistically significant (P = 0.003), with a consistently more favorable quality-of-life profile for captopril (Figure 2). For captopril, there were positive changes of 0.10 unit or more on the Side Effects and Symptoms Distress Index (+0.10), the major scales for General Perceived Health (+0.12) and Psychological Well-Being (+0.24), and on the four subscales of vitality (+0.15), emotional ties (+0.18), general positive affect (+0.20), and behavioral or emotional control (+0.15). There was a negative change of more than 0.10 unit on one scale, Sexual-Symptom Distress (-0.19). In contrast, there were more mixed effects and more negative changes for enalapril: the Psychological Well-Being scale (+0.12) and the subscales for emotional ties (+0.14) and anxiety (+0.11) changed positively by 0.10 unit or more; however, negative changes of 0.10 unit or more were found for Well-Being at Work or in Daily Routine (-0.15), Sexual-Symptom Distress (-0.18), General Perceived Health (-0.21), and the subscales of vitality (-0.20), general health status (-0.20), and behavioral or emotional control (-0.10).
|
The score on the base-line Quality of Life scale was a significant predictor of the change in quality of life (P<0.001) in these univariate and multivariate analyses. For descriptive purposes, we grouped patients into thirds (low, medium, and high) according to the ranking of their scores on the base-line Quality of Life scale (Table 2). There was a significant difference among the three groups with regard to the change from base line on the Overall Quality of Life scale (P<0.001) and on the General Perceived Health (P = 0.002), Psychological Well-Being (P = 0.014), and Psychological Distress (P<0.001) scales. For patients with low base-line scores for quality of life, there were fairly large increases on all scales with captopril therapy; patients in the medium and high groups, on the other hand, had mixed and smaller changes. With enalapril, there was approximately the same magnitude of improvement in the group with low scores on the Psychological Well-Being scale, but there was more worsening among the patients who had high scores.
|
To evaluate the clinical meaning and relevance of the treatment effects, we pooled the data for both treatment groups and, for each visit, correlated the quality-of-life scales with the three most objective indexes: the Side Effects and Symptoms Distress Index, the Stress Index (consisting of an overall rating on an analogue scale), and the Life Events Index (the sum of the points for stress for all negative life events during the past month). At week 8, the strength of the associations was highest on the Side Effects and Symptoms Distress Index, and it explained up to 46 percent of the total variability in scores on the General Perceived Health scale (r = 0.68, P<0.001). This index was also correlated with the Psychological Well-Being (r = 0.50, P<0.001) and Psychological Distress (r = 0.58, P<0.001) scales, indicating that the quality-of-life scales were consistent with the patients' reports of common side effects and symptoms. The corresponding correlations with the Stress Index and the Life Events Index ranged from 0.36 to 0.57 (P<0.001).
Since the impact of stressful life events on health is an easily understood concept, a calibration model was constructed that related the longitudinal change in quality-of-life scores to the longitudinal change in scores for stressful life events (Figure 3). The association between these scores was also statistically significant (P<0.001). Models were developed for the Overall Quality of Life scale and the three major scales, General Perceived Health, Psychological Well-Being, and Psychological Distress. Calibrations among these scales were fairly comparable, with a change of 0.10 unit that corresponded to a change in the Life Events Index of 27 points for overall quality of life, 28 points for psychological distress, 37 points for general perceived health, and 39 points for psychological well-being.
|
Although there was broad variability in responses from person to person, values ranging from 0.10 to 0.20 unit can be considered clinically meaningful and can be considered the lower bound of what constitutes a minimally important response to treatment as reflected in quality of life. The results also show that these quality-of-life measures are responsive and sensitive to stressful life events that have been documented to have a negative effect on health.
Discussion
The most striking finding in this study was that two angiotensin-converting-enzyme inhibitors that acted identically with regard to efficacy, adverse events, and laboratory outcomes acted quite differently with regard to quality of life. This difference, although already apparent in an analysis with unadjusted base-line measures, was more dramatic when patients were categorized according to base-line quality of life.
Therapeutic clinical trials typically balance the gain from relief of symptoms with the loss due to the adverse effects of the treatment. In the case of antihypertensive therapy, the studies of quality of life appear to have been interpreted with the premise that hypertension, in itself, is a process free of symptoms. The quality-of-life comparisons were therefore thought to involve primarily the frequency and severity of adverse effects. In one of the more compelling studies performed to date, Croog et al. found that captopril, an angiotensin-converting-enzyme inhibitor, improved the quality of life from base line, whereas the agents with which it was compared, methyldopa and propranolol, had a negative influence13. A common interpretation of that study was that the progressive improvement in quality of life with captopril reflected the anticipated sequence of events if an agent that was free from adverse effects on the central nervous system was used to control hypertension. As an alternative, it has been suggested that captopril has a more active positive influence on elements that enter into quality of life, including mood and cognitive function24,25,26,27,28,29.
In this study, the positive benefits of both captopril and enalapril were restricted primarily to patients with low quality-of-life scores at base line. In contrast, the negative effects of treatment on General Perceived Health scores, particularly for vitality and general health, that were observed among the enalapril-treated patients, but not the captopril-treated patients, were concentrated in the patients with the highest quality-of-life scores at base line. The influence of the drug was evident even with the influence of regression to the mean. That is, all the significant changes from base line in the enalapril-treated group were on the negative side in those whose base-line scores were high, whereas the significantly positive changes all occurred only in the captopril-treated patients whose base-line scores were low. As a possible explanation for a positive effect of captopril on the central nervous system, an influence on peptide degradation or on cholinergic mechanisms has been postulated28,29.
We used an extensive quality-of-life questionnaire (requiring 30 to 40 minutes to complete) and were thus able to detect changes that corresponded to moderately stressful life events. Caution should be taken in the design of studies using brief quality-of-life forms, which may be less sensitive and less responsive to therapeutic effects. If a correspondence is established between quality of life and more objective measures, the responsiveness or lack thereof of the instruments used can be established in the context of each clinical trial. In a previous study, changes in quality of life due to treatment that were comparable to those observed here were thought to be clinically meaningful if they corresponded in severity to "being laid off or fired from a job"13. This claim, however, was not supported by rigorous data or substantive analyses. The results of the calibration of quality of life with negative life events presented here are consistent with the claim that quality-of-life changes and differences between treatments are clinically meaningful. In this study, the Overall Quality of Life scores shifted positively with captopril by 0.11 unit and negatively with enalapril by 0.11 unit: negative shifts of this magnitude correspond to life events such as "major change in work responsibilities," "in-law troubles," or "mortgage foreclosure." The resulting difference of 0.22 between treatments appears to have clinical relevance. We indexed the changes in quality of life to life events in a calibration model to validate the responsiveness of the quality-of-life scales to treatment effects and to determine what degree of change in quality of life could be considered clinically important. The results indicated that the threshold for a clinically important change is to be found between 0.1 and 0.2 responsiveness-index unit and that greater changes in quality of life are associated with more stressful life events. Further research is needed to develop complete prediction models that are useful for the one-to-one translation of each individual difference between treatments.
In summary, our findings indicate that drug-induced changes in the quality of life can be substantial and clinically meaningful even when they involve drugs in the same pharmacologic class. Rates of withdrawal appear to be too crude a marker for the more subtle side effects and symptoms experienced by the patient. Furthermore, the instruments used in quality-of-life assessment are sufficiently sensitive and specific to show a close correlation with stressful life events. Most studies assessing quality of life and the effects of antihypertensive therapy have concluded that treatments were equivalent49 when many had neither a sufficiently large sample nor adequately sensitive quality-of-life measures to detect a meaningful difference. Calibration with life events or other objective measures can aid in determining when two treatments are equivalent within a meaningful range. Finally, since base-line quality of life has such a profound effect on response to therapy, caution should be exercised in the estimation of sample size and analyses should always include adjustment for base-line quality of life.
Supported in part by a grant from Bristol-Myers Squibb, Princeton, N.J.
Source Information
From the Department of Biostatistics, Harvard School of Public Health, Boston (M.A.T.); Phase V Technologies, Inc., Wellesley Hills, Mass. (R.B.A., J.F.N.); and the Departments of Radiology and Medicine, Brigham and Women's Hospital and Harvard Medical School, Boston (N.K.H.). Dr. Anderson and Ms. Nackley are employees of Phase V Technologies, which served as the data coordinating center for the trial. Drs. Testa and Hollenberg are paid consultants for Phase V Technologies. Dr. Hollenberg has served as a paid consultant to Bristol-Myers Squibb and Merck Sharp & Dohme, the companies that manufacture the drugs used in this study.The members of the study group are listed in Appendix 1.
Address reprint requests to Dr. Testa at the Department of Biostatistics, Harvard School of Public Health, 677 Huntington Ave., Boston, MA 02115.
References
The members of the Quality-of-Life Hypertension Study Group are as follows: Medical advisor: G.H. Williams, Brigham and Women's Hospital and Harvard Medical School, Boston; Clinical investigators: M.A. Moore, Southern Drug Research, Birmingham, Ala.; J.E. Angelo, New Orleans Institute of Clinical Investigations, New Orleans; W. Brown and H. Ayvazyan, Clinical Programs, Limited, Providence, R.I.; G.A. Behrens, St. Louis Center for Clinical Research, St. Louis; D. Boushea, Jackson Foundation, Madison, Wis.; J. Burris, Cardiovascular Center of Northern Virginia, Alexandria; T.T. Crouch, Kansas City, Mo.; C.A. DeAbate, Metairie, La.; R. Fiddes, Whittier, Calif.; E. Fishman, Los Angeles; T. Gaarder, Whittier, Calif.; J. Garofalo, West Caldwell, N.J.; L.I. Gilderman, Pembroke Pines, Fla.; N.K. Imes, Baptist Medical Foundation, Oklahoma City; J. Kern, Metropolitan Clinic, Portland, Oreg.; T.W. Littlejohn III, Piedmont Research Associates, Winston-Salem, N.C.; C.P. Lucas, Beaumont Preventative and Nutritional Medical Clinic, Birmingham, Mich.; D. Mitchell, Mitchell Clinic and Research Center, McKinney, Tex.; R. Motley, Lansdale Medical Group, Lansdale, Pa.; D. O'Connor, Veterans Affairs Medical Center, San Diego, Calif.; P. Ogden, Ross Valley Medical Center, Greenbrae, Calif.; V. Papademetriou, Washington Veterans Affairs Hospital, Washington, D.C.; D. Sugimoto, Chicago Center for Clinical Research, Chicago; A. Taranto, Atlanta; and P.D. Toth, Midwest Research Institute, Indianapolis.
Quality-of-Life Scales Used in This Study
The Psychological Distress scale contains 24 items, with corresponding subscales for anxiety, depression, and behavioral or emotional control30. The Psychological Well-Being scale contains 14 items, with subscales for general positive affect, emotional ties, and life satisfaction30. The General Perceived Health scale contains 11 items, with subscales for vitality (4 items), general health status including bodily disorders (3 items),31,32 and sleep disturbance (4 items)33,34,35.
The base-line Quality of Life scale contains all 49 items included in the three major scales described above, which are combined to form the base-line adjustment for quality-of-life response.
The Well-Being at Work or in Daily Routine scale contains 11 items relating to the patient's daily work responsibilities or performance in a daily role or routine if the patient was primarily retired, a student, or not employed outside the home. These items include satisfaction with work or daily routine, enthusiasm, performance, challenge, keeping up, tiredness or alertness, concentration, physical well-being, general feelings at the beginning and end of the workday, and the quality of leisure-time activities14.
The Sexual-Symptom Distress scale contains five items related to the degree of distress with sexual dysfunction13,14,35,36.
The Overall Quality of Life scale is a summary measure with multiple end points that is based on the mean score for the 11 subscales (anxiety, depression, behavioral or emotional control, general positive affect, emotional ties, life satisfaction, vitality, sleep disturbance, general health status, well-being at work or in daily routine, and sexual-symptom distress).
The Side Effects and Symptoms Distress Index contains 50 items relating to common symptoms and side effects of antihypertensive medications about which patients were asked in a survey for the presence or absence, frequency, and degree of general distress14,36.
The Life Events Index is adapted from the Social Readjustment Rating scale37 and consists of 42 scales for major life events rated from 0 to 100 according to their level of stress. The very stressful events included the death of a spouse, divorce, marital separation, and personal illness or injury, with ratings of 100, 73, 65, and 53, respectively. Moderately stressful events included retirement (assigned a rating of 45), change in the health of a family member (44), and change to a different line of work (36). Less stressful events included trouble with in-laws (29), trouble with the boss (23), and minor violations of the law (11). The patient was allowed to specify "other significant life events," which were coded and scored accordingly. Only events with a negative effect were used in the computation of the Life Events Index, because previous research indicates the importance of directionality39,40,41.
The Stress Index is an analogue visual-rating scale (ranging from no stress to unbearable stress, on a scale from 1.00 to 7.00) of overall stress during the month before the current visit.
| |||||||||||||||||||||||||||||||||||||||
Related Letters:
Captopril, Enalapril, and Quality of Life
Santanello N. C., Guess H., Heyse J. F., Kaplan N. M., Waud D. R., Fletcher A., Ware J. E., Testa M. A., Hollenberg N. K.
Extract |
Full Text
N Engl J Med 1993;
329:505-507, Aug 12, 1993.
Correspondence
This article has been cited by other articles:
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2008 Massachusetts Medical Society. All rights reserved. |
Previous
