Effect of Hysterectomy vs Medical Treatment on Health-Related Quality of Life and Sexual Functioning:  The Medicine or Surgery (Ms) Randomized Trial FREE

Hysterectomy is the most common major surgical procedure performed in the United States for nonobstetric reasons.¹ In 2000, approximately 633 000 hysterectomies were performed, and US women have an estimated 25% risk of having their uterus removed.² Ninety percent of hysterectomies are elective and performed before menopause for abnormal uterine bleeding and other non–life-threatening reasons.³ Large regional variations in the use of hysterectomy in the United States^4- 5 and lower hysterectomy rates in other developed nations⁶ have called into question the frequency with which this procedure is used.

A woman with abnormal uterine bleeding who decides to have a hysterectomy does so because she believes that this operation will be more effective than medicine in improving the quality of her life.⁷ Observational studies have reported improvements in health-related quality of life and sexual functioning.^8- 12 In addition to the discomfort and inability to function normally in the weeks after surgery,¹³ however, hysterectomy is associated with early ovarian failure,^14- 15 incontinence many years later,^16- 18 and a small mortality risk.^13,19 As a result, the initial approach to treating abnormal uterine bleeding is with medicines²⁰ such as progestins,²¹ combinations of estrogen and progestin,²² prostaglandin synthetase inhibitors,²³ and antifibrinolytics.²⁴ However, medical approaches sometimes fail to relieve symptoms or cause adverse effects, leading to the clinical dilemma of whether to continue efforts with medical approaches or to elect hysterectomy. The benefits and harms of these 2 approaches at this stage of decision making have not been studied in a randomized clinical trial.

Women who enrolled in the Medicine or Surgery (Ms) randomized trial had symptoms for a median of 4 years despite a variety of medical treatments, including medroxyprogesterone acetate.²⁵ These women were randomly assigned to receive either expanded medical treatment or hysterectomy. The specific approaches within each category were left to their physicians. We focused on overall mental health as the primary outcome, using the Mental Component Summary (MCS) of the 36-Item Short-Form Health Survey (SF-36) as our measure, and we examined several other aspects of health-related quality of life, including physical health, symptom resolution and satisfaction, body image and sexual functioning, specific aspects of mental health, and general health perceptions.

The design of our study is described in detail elsewhere.^25- 26 Briefly, we recruited premenopausal women, aged 30 to 50 years, who were under care for abnormal uterine bleeding and willing to be randomized to undergo hysterectomy or receive selected medical treatments. Current standard practice favors a trial of medical treatment before hysterectomy. Eligibility criteria therefore included abnormal uterine bleeding, which we defined as more than 7 days of flow per month, or anemia (hematocrit ≤32%) with heavy menstrual flow and no other cause, and previous unsuccessful treatment with cyclic medroxyprogesterone. Potentially eligible participants who had not previously been treated with medroxyprogesterone were invited to partici in a medroxyprogesterone trial.²⁵ Women completing this trial who were not satisfied with treatment for control of abnormal uterine bleeding were then eligible for the Ms randomized trial. Patients were recruited from the gynecology clinics and practices of the University of Alabama, Birmingham; the University of Tennessee, Memphis; and Wayne State University, Detroit, Mich; and clinics affiliated with the University of California, San Diego. Institutional review board approval was obtained at each clinical site and at the University of California, San Francisco, which served as the coordinating center.

Randomization was stratified by clinical center, using randomly permuted blocks of randomly varying sizes of 4, 6, and 8. Participating gynecologists were free to select the choice of medical treatment for women in the medicine group; the preferred regimen was a combination of 21 days of low-dose oral contraceptive and 7 days of placebo during each 4-week period and use of a prostaglandin synthetase inhibitor starting on the first day of menses and continuing for 5 days. Other therapeutic options included continuously administered oral contraceptives, alternative progestogens (oral or intramuscular), and cyclic estrogen-progestin regimens. If results were unsatisfactory, other medical therapies could be substituted. Women assigned to undergo hysterectomy had the surgery performed by a participating gynecologist, who selected the type and route of hysterectomy using clinical judgment.

Each woman was followed up for 2 years after randomization. We measured baseline sociodemographic, clinical, and health-related quality-of-life variables at the randomization visit and conducted a telephone interview 4 weeks after randomization to the medicine group or 4 weeks after undergoing hysterectomy to assess short-term, health-related quality-of-life outcomes. Other contacts, timed from the date of randomization for both groups, included detailed clinic-based interviews every 6 months and brief telephone interviews at 3, 9, 15, and 21 months. Interim monitoring to assess safety was performed by an independent data and safety monitoring board.

To assess health-related quality-of-life outcomes, we relied primarily on measures from the Medical Outcomes Study (MOS),^27- 28 including the MOS SF-36,^29- 31 which has been used extensively in studies of women's health,^32- 34 and several scales from the complete set of the MOS functioning and well-being measures.²⁸ The SF-36 has 2 summary indexes, the MCS and the Physical Component Summary (PCS), which are derived using weighted averages of the individual domain scores.³⁵ We used the MCS to test our primary hypothesis because it addresses the areas we believed to be of greatest importance to women with abnormal bleeding, and we relied on the PCS as our measure of physical health, a secondary outcome.

Measures used for the other secondary outcomes included 3 single items adapted from the Maine⁸ and Maryland³⁶ Women's Health studies to assess symptom resolution and satisfaction with symptoms and with overall health. We also used 5 scales measuring body image and sexual functioning that we developed for the Ms randomized trial using a combination of new questions and adapted items from the Body Attitudes Questionnaire,³⁷ the MOS Sexual Problems Index,^{28(pp194-204)} and the Maryland study.¹¹ To more thoroughly explore the impact of the interventions on mental health, we used the 17-item MOS measure of psychological distress and well-being,^{28(pp102-142)} the MOS Sleep Problems Index,^{28(pp,235-239)} and a shortened version of the MOS health distress scale.³⁸ Overall health was assessed by combining the traditional, single, self-rated health item (excellent, very good, good, fair, poor) with the EuroQol visual analog scale score.³⁹ All measures were transformed to a 0-to-100 scale, with 100 always representing optimal health, functioning, or well-being. Details of scale construction and psychometric properties are reported elsewhere.²⁶

A sample size of at least 60 participants would allow us to reject the null hypothesis of no difference in MCS improvements between hysterectomy and medical treatment with 90% power in a 2-sided test with an α of .05 if the true effect size was 6.8 units or greater. This difference of 6.8 units represents approximately 70% of an SD and, although large, is similar to 1-year MCS change scores for patients who were classified as having improved in 1 of 5 tracer conditions in the MOS study.³⁵ A larger sample powered to detect a smaller effect was originally planned, but major difficulties in the early stages of recruitment and a failed effort to randomize women in a similar trial⁴⁰ led our data and safety monitoring board to endorse a decision to reduce the recruitment goal to 60.

We analyzed study outcomes according to original treatment assignment (intention to treat). We began by plotting mean outcomes at 4 weeks after hysterectomy or after randomization (for women in the medicine group) and at 6-month intervals after randomization. We then assessed differences between the hysterectomy and medicine groups using linear regression models for the effect of treatment on changes from baseline to 6 months and 2 years after randomization, controlling for the baseline value. To add precision to the 2-year outcomes, we analyzed changes from baseline to 18 and 24 months as repeated measures, using a random effect for each participant to account for within-participant correlation. We also examined several models in which the intention-to-treat comparison of outcomes was adjusted for race and for those baseline symptoms that were least equitably distributed between randomized groups.

We concluded with analyses comparing women in the medicine group who had underwent hysterectomy by the end of the 2-year follow-up period with those who had not. We examined both baseline and time-dependent predictors of crossing over using single-predictor Cox models. We then graphically compared the mean values of the outcome measures over time in the 2 as-treated medicine groups with those for women in the hysterectomy group. Finally, we assessed within-group changes from baseline to year 2 in all 3 as-treated groups, again using random-effects models for repeated measures (18 and 24 months). All analyses were performed using SAS statistical software version 8.2 (SAS Institute Inc, Cary, NC).

Patients were randomized between August 1997 and December 2000. A total of 1557 were initially screened, 1013 of whom were found to be potentially eligible for the Ms randomized trial (Figure 1). Ninety-two had previously been treated with medroxyprogesterone and therefore were eligible to participate in the Ms randomized trial. Four hundred thirteen of the women who had not been treated previously with medroxyprogesterone agreed to participate in a study of this treatment.²⁵ Ninety-two of the 215 women who completed that study were dissatisfied with medroxyprogesterone and thus were eligible for the Ms randomized trial. Twenty-five women who had previously tried medroxyprogesterone and 38 women who completed the medroxyprogesterone trial agreed to enter the Ms randomized trial. In total, 63 women for whom treatment had failed were randomly assigned: 31 to undergo hysterectomy and 32 to receive expanded medical treatment.

The trial participants formed a relatively diverse and highly symptomatic group. Approximately half were black and had annual family incomes of not more than $25 000 (Table 1). The majority had been experiencing abnormal bleeding for several years and had long, heavy periods. The majority reported pelvic pain and pressure and had fibroids on examination. Their mean body mass index (a measure of weight in kilograms divided by the square of height in meters) was 32. Mean baseline MCS and PCS scores were 45 and 42, respectively, suggesting moderate impairment (US population norms for women aged 35-44 years are 49 and 51, respectively³⁵).

We observed a nominally significant difference among 30 comparisons of the baseline characteristics of the randomized groups at P<.05, indicating that the randomization was technically successful. Twenty-eight (90%) of the 31 women in the hysterectomy group underwent the procedure: 10 (36%) abdominally and 18 (64%) vaginally. The 32 women in the medicine group received a variety of treatment regimens. Twenty-nine (91%) received hormonal treatment with or without a prostaglandin inhibitor while 3 (9%) were treated only with a prostaglandin inhibitor. Specifically, 12 (38%) received combined oral contraceptive pills, 5 (16%) received a cyclic progestin, 2 (6%) received a continuous progestin, 8 (25%) received conjugated estrogen with progestin, and 2 (6%) received conjugated estrogen without progestin. Seventeen (53%) of the 32 participants received a prostaglandin synthetase inhibitor, most commonly naproxen sodium (n = 15), that was usually combined with a hormone.

At 4 weeks after surgery, women in the hysterectomy group experienced a decrease in physical health based on the PCS (P = .004 compared with the medicine group at 4 weeks after randomization; Figure 2). Mental health based on the MCS score peaked at this point and was somewhat higher than scores for the medicine group (P = .07), suggesting that although the recovery from surgery had hampered their physical health, they were experiencing an upsurge in mental health.

At 6 months after randomization, women in the hysterectomy group reported significantly greater improvement in the primary outcome (MCS score) than did women in the medicine group (Table 2). They also reported significantly greater improvements in symptom resolution and satisfaction with degree of symptom relief, pelvic problem interference with sex, sexual desire, health distress, sleep problems, overall health, and satisfaction with health compared with women in the medicine group.

At the end of the 2-year follow-up period, these improvements were maintained in the hysterectomy group, and sizable improvements were also observed in women in the medicine group. Thus, between-group differences were no longer significant with one exception: the improvement in sexual desire remained significantly greater for those in the hysterectomy group than for those in the medicine group. Intention-to-treat comparisons of outcomes after adjusting for race and baseline pelvic or bladder pressure, heavy or very heavy bleeding, and fibroids revealed essentially the same findings.

By the end of the first year, 14 (43%) of the 32 women in the medicine group had asked for and received a hysterectomy. Three more women underwent the procedure in the first 6 months of year 2, resulting in 17 (53%) undergoing hysterectomy, with a median interval to crossing over of 6 months. Table 3 presents the 2-year change in outcomes within each as-treated group. Women in the hysterectomy group who had undergone the procedure improved significantly on all outcome measures. Women in the medicine group who crossed over to hysterectomy had significant improvements in the MCS and 11 of the 13 secondary outcomes. Women in the medicine group who did not cross over did not achieve significant improvements on the MCS, but they experienced significant improvements in 7 secondary outcomes. Figure 3 compares the MCS and PCS patterns of women in the medicine group, classified by whether they had undergone hysterectomy during the 2-year follow-up period.

This randomized trial of women with prolonged abnormal uterine bleeding refractory to medroxyprogesterone treatment demonstrated clear benefits of hysterectomy over expanded medical treatment after 6 months. At this point, women in the hysterectomy group demonstrated a consistent pattern of improvement in all but 1 of the health-related quality-of-life outcomes we explored. Significant improvements were found in overall mental health, symptom resolution, satisfaction with symptoms, sexual functioning, health distress, sleep problems, and general health perceptions, underscoring the value of this procedure for women with chronic uterine bleeding for whom a satisfactory resolution cannot be obtained with medroxyprogesterone.

The levels of improvement for women in the hysterectomy group were nearly the same after 2 years. Due to the quality-of-life improvements among women in the medicine group, however, the superiority of hysterectomy over medical treatment was no longer statistically significant at 2 years. This diminution in the relative benefit of hysterectomy over medicine was presumably influenced by the fact that half of the women in the medicine group elected to have a hysterectomy before the end of the study (despite encouragement to continue with medical treatment). Women in the medicine group who crossed over to hysterectomy experienced subsequent improvements in quality of life similar to those of the women in the hysterectomy group. Importantly, although women who continued with medical treatment did not show significant improvement in our primary outcome (the SF-36 MCS), they did experience significant improvements in many of the other outcomes, including satisfaction with symptom level and with health. This suggests that persisting efforts at medical treatment eventually lead to improved quality of life in some women with prolonged abnormal uterine bleeding refractory to medroxyprogesterone treatment. Moreover, because only 3 of the women who crossed over did so in the first 6 months of the second year of the study and none in the last 6 months of the study, it appears that for those who did not undergo hysterectomy, medical treatment may be a satisfactory and lasting solution to the medical problem.

There is no clear consensus on what constitutes a clinically meaningful improvement in health-related quality-of-life scales^41- 42 although 6 points on the 0-to-100 scale has been used as a threshold for improvement on the MCS for 5 tracer conditions from the MOS³⁵ and comparisons between patient populations suggest that a 3- to 5-point difference on the MOS measures is important.²⁷ All significant differences between randomized groups in our 6-month intention-to-treat analyses were at least this magnitude, and all 3 as-treated groups had improvements greater than 5 points from baseline to 24 months for most outcomes. Moreover, although the women we studied had scores on our 2 summary scales at baseline that were lower than those of the general US female population aged 35 to 44 years and comparable to those of patients with chronic lung disease and back pain or sciatica,³⁵ their scores at 24 months were similar to those of population norms. We conclude that women enrolling in the Ms randomized trial had a poor quality of life, reflecting the duration and severity of their abnormal uterine bleeding, and that the improvements we observed are clinically meaningful and important.

A previous report²⁵ on the prerandomization phase of our study showed that among women with long-term abnormal uterine bleeding who completed a 3- to 5-month course of cyclic medroxyprogesterone, 57% were satisfied with the outcome. The 43% who were medroxyprogesterone-resistant were candidates for randomization, and the Ms trial findings suggest that half of these women achieve satisfactory outcomes from further effort at treatment with medical regimens. For the remaining women (<25% of those treated with medroxyprogesterone), further attempts to treat medically may simply delay the ultimate decision to have a hysterectomy, suggesting that this procedure should be available to women for whom first-line medical treatment has failed. Not surprisingly, we found that women with lower symptom-resolution scores after the 3 to 5 months of medroxyprogesterone treatment were more likely to cross over to hysterectomy, suggesting that this scale may provide helpful information to women in this situation who are deciding about having a hysterectomy.

Our study has limitations. It was difficult to find women who were sufficiently indifferent to the choice of medical vs surgical intervention that they could accept being randomized to one or the other, and the degree to which findings in these select women are representative of the entire population of women choosing between medicine and surgery is unclear. Other generalizability issues include the fact that most of the women we studied lived in the southeastern United States, where hysterectomy rates are higher than in other parts of the country,² the average body mass index was substantially higher than that of US women in their 40s,⁴³ and two thirds of the women had fibroids (which may diminish the effectiveness of medical treatment although this was not the case in our medroxyprogesterone trial²⁵). Also, the specifics of surgical procedures and medical treatment were left to attending physicians, enhancing generalizability but preventing us from knowing if a specific aspect of the interventions we studied had a particular effect on our outcomes. Thus, for example, it remains possible that one of the individual medical treatments could be equivalent to hysterectomy in its effects on quality of life.

Another limitation is the small size of the study, a consequence of the difficulty in finding women willing to be randomized. This was compounded by the high rate of crossing over to hysterectomy by women in the medicine group, biasing the intention-to-treat estimate of the difference between medicine and hysterectomy toward the null (the 6-month findings are the least biased, since much of the crossing over did not occur until later). Our study was neither large enough nor long enough to examine rare or delayed consequences of the treatments we studied. Thus, we must continue to rely on observational studies to estimate any adverse effects of the medical regimens and the complications of surgery, such as perioperative mortality,¹⁹ ovarian failure,^14- 15 and incontinence many years later.^16- 18

Despite these limitations, the Ms trial is the first randomized trial comparing hysterectomy with primarily oral medical treatment and thus contributes to the body of knowledge that informs discussions between physicians and patients about options for treating prolonged abnormal uterine bleeding. Another strength is the caliber and extent of our measures of health-related quality of life, a particularly important outcome in decisions about elective surgery. We have shown that hysterectomy results in substantial improvement in health-related quality of life within 6 months for women who have not responded to medroxyprogesterone. Persisting efforts at medical treatment can also produce benefits for many of these women throughout the ensuing 2 years although others who prolong medical treatment at this stage will decide within a year to have a hysterectomy.