Quality-of-Life and Depressive Symptoms in Postmenopausal Women After Receiving Hormone Therapy:  Results From the Heart and Estrogen/Progestin Replacement Study (HERS) Trial FREE

Survey results indicate that 38% of US women between 50 and 74 years of age are taking postmenopausal hormones.¹ The reasons that women and their physicians initiate hormone therapy are varied, but the potential to prevent disease, such as osteoporosis and heart disease,^2- 3 is a strong consideration. There is also a widespread belief that postmenopausal hormones may have general positive effects on the health of older women, reflected in the common use of the term hormone replacement, which suggests correction of an abnormal deficiency. While large randomized trials of postmenopausal hormone therapy are ongoing,^4- 5 many women are struggling with decisions about whether to begin or continue hormone therapy.⁶

Postmenopausal hormone therapy may affect future morbidity and mortality through possible beneficial effects on the incidence of hip fracture,⁷ myocardial infarction (MI),^8- 9 and stroke⁸ and adverse effects on the incidence of breast cancer,¹⁰ endometrial cancer,¹¹ and venous thromboembolism.¹² The net effect of hormone therapy on these disease processes may well be to increase life expectancy,^2,6,13 yet length of life is not the only consideration for patients making decisions about medical treatments. Indeed, quality of life may be more important than length of life to many patients. Thus, rational decisions about treatments, especially preventive therapies, should consider effects on quality of life.¹⁴ The effects of postmenopausal hormone therapy on quality of life have not been well documented, especially when prescribed specifically for the purpose of disease prevention. The purpose of this study was to compare the quality of life of patients assigned to postmenopausal hormone therapy or placebo in a multicenter, double-blind, randomized controlled clinical trial.¹⁵

The design¹⁶ and main findings¹⁵ of the Heart and Estrogen/Progestin Replacement Study (HERS) have been published previously. Briefly, postmenopausal women younger than 80 years with documented coronary artery disease were eligible for entry. Postmenopausal status was established by age (≥55 years) and no natural menses for at least 5 years; by no natural menses for at least 1 year and serum follicle-stimulating hormone (FSH) level higher than 40 mIU/mL; by documented bilateral oophorectomy; or by reported bilateral oophorectomy with FSH level higher than 40 mIU/mL and estradiol level lower than 25 pg/mL (92 pmol/L). Coronary disease was established either by a previous MI, by coronary angiography demonstrating more than 50% luminal diameter narrowing of a major vessel or by a prior coronary revascularization procedure. Patients were not eligible if they had had an MI or a coronary revascularization procedure in the previous 6 months. Patients with a prior hysterectomy, contraindications to hormone therapy, or hormone therapy in the previous 3 months were excluded.¹⁶ Patients with other life-threatening illnesses were also excluded, as were those who were unable to return for follow-up visits.

The study was a randomized, double-blind, placebo-controlled clinical trial. The sample size was established to provide 90% power to detect a 24% reduction in nonfatal MI or death due to coronary disease.¹⁵ Eligible participants were assigned with equal probability to the 2 treatment groups using tamper-proof blocked randomization, stratified by clinical center. Study medication consisted of 0.625 mg of conjugated equine estrogens and 2.5 mg of medroxyprogesterone acetate (Prempro; Wyeth-Ayerst, Radnor, Pa) or placebo. All participants provided written informed consent. The study was approved by the institutional review boards of the University of California, San Francisco and the clinical sites.

Baseline data were collected prior to randomization according to a standardized manual of operations. Quality of life questionnaires were completed by participants at study entry, at 4 months of follow-up, at 1 year of follow-up, and annually thereafter. Subjects were randomly assigned between January 1993 and September 1994 and followed up to study closeout between April and July 1998. Four-year follow-up data were available in only some participants due to planned study closeout, so analyses were restricted to the first 3 years of follow-up.

The quality-of-life questionnaire assessed functional capacity, emotional health, vitality, and depression. Physical function was assessed using the 12-item Duke Activity Status Index¹⁷ on a scale that ranged from 0 (worst) to 58.2 (best). Energy/fatigue was measured using a 4-item RAND scale graded from 0 (worst) to 100 (best).¹⁸ Mental health was measured by the RAND Mental Health Inventory, a 5-item scale that assesses anxiety and depression, with scores ranging from 0 (worst) to 100 (best).^19- 20 Depressive symptoms were assessed using an 8-item scale developed by Burnam et al²¹ to screen for depression in the National Study of Medical Outcomes that ranges from 4 (worst) to −8.2 (best).

The primary contrasts between treatment groups were done on an intention-to-treat basis. We used mixed linear regression models fit by maximum likelihood²² to test for differences in serial quality-of-life scores over follow-up between groups. This analysis tests for a between-group difference in scores at baseline, for within-group linear changes in scores over follow-up, and for a between-group difference in the 2 group-specific rates of change. Model results were summarized by the estimated change in mean score from baseline to year 3. We also performed multivariable models that included clinical characteristics documented at the time of study entry. All analyses were performed using SAS version 6.12 (SAS Institute Inc, Cary, NC).

A total of 2763 patients were randomized: 1383 to placebo and 1380 to hormone therapy (Figure 1). Baseline clinical characteristics were similar in these randomly assigned groups, as reported previously.¹⁵ Quality-of-life data were complete at all 5 time points for 1138 patients assigned to placebo and 1108 patients assigned to hormone therapy. Quality-of-life scores were missing at 1 or more time points due to death (168 patients), withdrawal from the study (41 patients), or other reasons (307 patients). Patients with some missing scores were older and more likely to have diabetes and lower health ratings (all P<.001) but were not significantly different in treatment assignment (P = .19) (Table 1). We performed the analyses using all patients and repeated them using only patients with quality-of-life data at every time point. Since the results were essentially unchanged, we report the analyses based on all patients.

Patients assigned to receive hormone therapy were less likely to have taken 80% or more of their assigned treatment (70% vs 79% at 3 years).¹⁵ At the end of 3 years, 1027 patients assigned to receive hormone therapy and 1099 patients assigned to receive placebo were still taking study medication. In the first 3 years of follow-up, the number of women who had adverse events after assignment to hormone therapy vs placebo, respectively, were: death, 92 (6.7%) vs 76 (5.5%), P = .19; nonfatal MI, 96 (7.0%) vs 95 (6.9%), P = .97; unstable angina or coronary revascularization, 224 (16.2%) vs 238 (17.2%), P = .53; and venous thromboembolism, 28 (2.0%) vs 9 (0.7%), P = .003.¹⁵ As reported in detail elsewhere, assignment to hormone therapy had no significant effect on stroke,²³ peripheral vascular disease,²⁴ and clinical fractures²⁵ and was associated with a significant increase in biliary tract surgery²⁶ and urinary incontinence.²⁷ Changes in patient weight over follow-up were small and not significantly different between women assigned respectively to hormone therapy and to placebo: 0.02 kg vs 0.03 kg at 1 year (P = .25), 0.02 vs 0.02 kg at 2 years (P = .57), and 0.02 vs 0.01 kg at 3 years (P = .10). Assignment to hormone therapy was associated with a significant reduction in the frequency of flushing, vaginal dryness, and trouble sleeping but also was associated with an increased frequency in vaginal discharge, uterine bleeding, and breast symptoms (unpublished observations).

At study entry, the physical function scores of the randomly assigned groups were not significantly different (mean Duke Activity Status Index of 25.2 placebo vs 25.5 hormone therapy) and consistent with moderate levels of physical activity. Over follow-up, the scores in both groups declined progressively (Figure 2). At 3 years of follow-up, the physical function scores of patients assigned to placebo had changed by a mean of −3.1 units vs −4.4 units for patients assigned to hormone treatment (median change, −2.7 vs −3.5). The mixed linear regression model showed a significant decline over time in all patients in physical function scores (P<.001), no difference in baseline scores according to treatment assignment (P = .77), and a significant interaction between treatment assignment and follow-up time (P = .03), indicating a more rapid decline in physical function scores among women assigned to hormone therapy.

Energy/fatigue scores at study entry were in a range typical of the general population of similar age and sex, with mean scores of 55.3 in the placebo group and 55.8 in the hormone group. Energy/fatigue scores declined overall during follow-up (Figure 2). At 3 years of follow-up, the scores had changed by a mean of −3.0 in the placebo assigned vs −4.6 in the hormone assigned patients (median change, 0 vs −5). The mixed linear regression model showed a significant decrease in energy/fatigue scores over time in all patients (P<.001), no difference in baseline scores according to treatment assigned (P = .88), and a trend toward faster declines in energy/fatigue in women assigned to hormone therapy (P = .05).

Mental health as assessed by the RAND Mental Health Inventory-5 was typical of the general population at study entry in both the placebo (75.7) and hormone (75.9) groups. Both groups showed small changes from baseline during subsequent follow-up (Figure 2). At 3 years, changes of −0.9 vs −0.2 from baseline were observed in the placebo and hormone treated groups, respectively (median change 0 in each group). The mixed linear regression model showed a significant decline over time in all patients in mental health scores (P = .05), no difference in baseline scores according to treatment assignment (P = .82), and no significant interaction between treatment assignment and follow-up time (P = .10).

Depressive symptoms were assessed using an 8-item screening test at baseline and follow-up. In the mixed linear regression model of the raw scores, there were no differences in depression over time (P = .20) or at baseline according to treatment assigned (P = .16). But there was a significant interaction between follow-up time and treatment assignment, such that depressive symptoms decreased more over time among the patients assigned to hormone therapy (P = .005).

We divided patients into 2 groups based on their response at the time of study entry to the question, "During the past week, including today, did hot flashes or flushes bother you or interfere with your life?" Patients who answered "all," "most," "a good bit," or "some of the time" were classified as having symptoms (n = 434), whereas those who responded "a little" or "none of the time" were classified as being without symptoms (n = 2325). Patients with flushing symptoms were younger (mean 63.0 vs 67.3 years, P<.001) and had a shorter interval since their last menstrual period (mean, 13.7 vs 18.7 years, P<.001) than those without flushing symptoms. The prevalence of flushing among patients in the hormone-therapy group was reduced significantly compared with patients in the placebo group: 16.7% vs 14.7% at study entry (P = .17), 7.1% vs 14.5% at 1 year (P<.001), 6.3% vs 11.7% at 2 years (P<.001), and 5.8% vs 11.2% at 3 years (P<.001), respectively.

All 4 quality-of-life scores were significantly worse at baseline and throughout follow-up in patients with flushing symptoms (P<.001) than in patients without them (Figure 3). The rate of change in physical function (P = .05), energy/fatigue (P = .06), and depressive symptoms (P = .001) showed interactions with treatment assignment and presence of flushing symptoms. Among women who reported flushing at study entry, assignment to hormone therapy was associated with improved mental health (+2.6 vs −0.5, P = .04) and depressive symptoms (−0.5 vs + 0.007, P = .01) over 3 years' follow-up but no significant change in physical function (−3.1 vs −2.2, P = .42) or energy/fatigue (−2.3 vs −2.4, P = .99) scores compared with placebo. In contrast, among women without flushing at entry, assignment to hormone therapy was associated with a greater decline in physical function (−4.2 vs −3.3, P = .04) and in energy/fatigue scores (−4.6 vs −3.1, P = .03) over 3 years' follow-up but no significant change in mental health (−0.6 vs − 1.1, P = .40), or depressive symptoms (−0.08 vs + 0.06, P = .08) compared with placebo.

Several baseline clinical factors were associated with effects on quality of life, both at study entry and throughout follow-up, that were larger than the effects of assignment to hormone therapy (Table 2). Chest pain and lower levels of education each had substantial negative effects on all 4 quality-of-life measures throughout follow-up.

The effect of age on quality of life was complex (Table 2). Older age was associated with lower physical function scores (P<.001) but had no significant effect on energy/fatigue scores (P = .65). In sharp contrast, older patients had significantly better mental health scores (P<.001) and significantly less depressive symptoms (P<.001) throughout follow-up.

Patients with a history of heart failure, diabetes, and hypertension each had worse physical function (P<.001) and lower energy levels (P<.001) throughout follow-up but smaller and less consistent effects on mental health and depressive symptoms (Table 2).

Patients with prior coronary angioplasty or coronary bypass graft surgery were not significantly different in any of the 4 quality-of-life measures from patients without prior coronary revascularization (Table 2). Prior MI had no significant effect on physical function or energy/fatigue but better mental health scores and fewer depressive symptoms (Table 2).

This large trial of clinically stable, postmenopausal women with documented coronary artery disease shows that quality of life generally declined during follow-up in the entire cohort and that there were large differences in quality of life according to clinical characteristics. Hormone replacement had overall significant negative effects on physical function, but it improved depressive symptoms. The effect of hormone therapy on quality of life was significantly modified, however, by the presence or absence of postmenopausal symptoms. Among women with flushing, hormone therapy significantly improved mental health and depressive symptoms (Figure 3). Although women without flushing had generally better quality-of-life scores, hormone therapy led to faster declines in their physical function and energy/fatigue scores (Figure 3). These mixed results suggest that hormone therapy does not have a general benefit for postmenopausal women with heart disease; rather, it improves quality of life only for women with menopausal symptoms.

These findings are consistent with previous studies that examined the effects of hormone therapy on quality of life. Wiklund et al²⁸ randomly assigned 223 postmenopausal women (mean age, 53 years) with climacteric symptoms to transdermal estradiol or placebo. After 3 months of follow-up, women assigned to estradiol had significantly greater improvements on the Nottingham Health Profile and the Psychologic General Well-Being Scale, as well as better relief of vasomotor symptoms.²⁸ Limouzin-Lamothe et al²⁹ randomly assigned 499 postmenopausal women (mean age, 51 years) with moderate to severe perimenopausal symptoms to transdermal estradiol or symptomatic treatment in an unblinded trial.²⁹ They found that hormone replacement was superior to symptomatic treatment and that the effects on quality of life were more pronounced among women with more frequent flushing.²⁹ The Postmenopausal Estrogen/Progestin Interventions trial randomly assigned 875 women (mean age, 56 years) to either placebo or 1 of 4 active therapies. They found that patients assigned to active treatment had significantly fewer vasomotor symptoms but no change in anxiety, cognitive, or affective symptoms.³⁰ The women in these earlier studies were significantly younger than those in HERS (mean age, 67 years) and more often had menopausal symptoms. Our study confirms that the beneficial effect of hormone therapy on menopausal symptoms improves quality of life, even among older women.

A conceptual model of health-related quality of life suggests that symptom status is the key intermediary between underlying physiologic variables and quality of life.³¹ In previous studies, the severity of symptoms responsive to therapy have been shown to affect the degree of changes in quality of life after treatment,^32- 33 and therapy of asymptomatic conditions, such as hypertension, often decreases quality of life.³⁴ Based on these earlier studies we hypothesized that symptoms potentially responsive to hormone therapy would identify women more likely to have improved quality of life. A priori we chose flushing as the symptom at baseline most likely to improve with hormone therapy and examined only this single variable in a treatment-by-covariate interaction test. Consistent with our hypothesis, we found a significant interaction of flushing symptoms with hormone therapy, with generally positive effects of hormone therapy on quality of life among women with flushing and generally negative effects among women without menopausal symptoms. These observations are consistent with the concept that among symptomatic patients therapy typically improves quality of life, whereas among asymptomatic patients therapy often adversely affects quality of life.

The variations in quality of life according to clinical characteristics of the women in HERS were substantial and generally exceeded the magnitude of the effect of hormone therapy (Table 2). In other studies, differences in 2 or more units on the Duke Activity Status Index, the Mental Health Index, and the energy/fatigue scale and 0.1 unit on the depressive symptoms scale are clinically meaningful. Most of the baseline factors we examined had effects of this magnitude on quality-of-life scores.

The negative effect of increasing age on physical function (Table 2) is not surprising, but it is interesting that older women had significantly better levels of emotional health than younger women (Table 2), similar to the findings of other studies.³⁵ It is possible that the same level of physical limitation causes more emotional distress in younger patients who have higher expectations for good health than do older patients.

Patients with chest pain, heart failure, diabetes, or hypertension had substantially worse quality-of-life scores throughout this study. The deleterious effect of such chronic conditions on quality of life has been documented in previous studies of the general population.²⁰ The negative impact of angina or heart failure on quality of life underscores the potential value of therapies aimed specifically at reducing such symptoms.

This study has several limitations that should be considered in interpreting our findings. Quality-of-life data were missing at 1 or more time points in 19% of participants, and we cannot entirely exclude the possibility that differential loss of participants may have affected our results. However, we consider this unlikely because treatment assignment was not significantly different among women with complete vs incomplete data and our results were similar whether the analysis included or excluded patients with some missing data. HERS was conducted in an older population of women (mean age, 67 years), all of whom had heart disease. The results of this investigation do not necessarily apply to younger women or women without heart disease. We studied a specific combination estrogen/progestin, and our results may not apply to other regimens, such as unopposed estrogen. Although the HERS trial was large and conducted in 20 centers across the United States, the population enrolled may not be completely representative of women with heart disease. In addition to meeting inclusion and exclusion criteria, all HERS participants were volunteers and may consequently differ in quality of life from the general population of women with heart disease.

In conclusion, estrogen/progestin therapy may have either positive or negative effects on quality of life depending on the presence or absence of postmenopausal symptoms. Women with flushing have improvements in emotional dimensions of quality of life when given hormone therapy, but women without menopausal symptoms have net negative effects on physical dimensions of quality of life.