Objective: To examine whether women who were using postmenopausal hormone therapy (HT) before publication of Women's Health Initiative findings about risks associated with HT had been informed about the findings once published; and to estimate how knowledge of these findings was associated with their decision to discontinue HT.

Methods: We performed a telephone survey of 670 female members of a large health maintenance organization, aged 50-69 yr, who had regularly used HT from July 1, 2001, through June 30, 2002.

Results: Most women (93%) reported hearing about the new findings; however, only 57% considered the quality of this information to be good, regardless of its source: mass media (21%), the health plan (32%), or a health care practitioner (34%). Women's knowledge of Women's Health Initiative findings was generally poor; 64% did not know what the findings were, 7% were unsure of their knowledge, 6% had incorrect knowledge, and 23% had correct knowledge of Women's Health Initiative findings. On a simple, five-question, true-or-false quiz about HT risks, 30% of respondents answered four to five questions correctly. Although not well informed, 56% reported attempting to discontinue HT in the 6 to 8 months after July 2002. Our multivariable model included five statistically significant predictors of attempting to stop HT: having been sent a letter about Women's Health Initiative findings (odds ratio [OR] 2.7; 95% confidence interval [CI] 1.8, 3.9), reporting good-quality information from media (OR 2.1; 95% CI 1.3, 3.3), having started HT for health promotion (OR 2.0; 95% CI 1.2, 3.3), using a lower-than-standard dosage of estrogen (OR 1.9; 95% CI 1.1, 3.1), and correctly answering four or more items on the HT quiz (OR 1.9; 95% CI 1.2, 2.8).

Conclusion: During the 6-8 months after publication of Women's Health Initiative trial findings, most regular postmenopausal HT users tried to stop using HT, despite not being well informed about the Women's Health Initiative findings.

Financial Disclosure: This research was supported by a grant from Eli Lilly and Company Inc. and by Grant AG 12262, which is cofunded by the National Institutes on Aging and the National Institute for Arthritis, Musculoskeletal, and Skin Disease.

^aDivision of Research, Kaiser Permanente Northern California, Oakland, Oakland, California, USA
^bWomen's Health Clinical Research Center, Mount Zion Medical Center, University of California, San Francisco, California, USA
^cDepartments of Medicine and Community and Family Medicine, Clinical Research Division, Dartmouth Medical School, Hanover, New Hampshire, USA

(Obstet Gynecol 2003:102:1225-1232. © 2003 by The American College of Obstetricians and Gynecologists.)

In July 2002, the Women's Health Initiative published principal findings of a randomized trial that explored whether postmenopausal hormone therapy (HT) affects risk of several diseases.¹ The estrogen-plus-progestin arm of this landmark trial was stopped prematurely because women treated with this form of HT had an increased risk of coronary events, stroke, venous thromboembolic events, and breast cancer.¹ These findings have led many physicians to encourage postmenopausal women without vasomotor symptoms to stop HT and to limit use to short-term treatment for menopausal symptoms.² Publication of the Women's Health Initiative findings engendered extensive media coverage that generally conveyed a message of warning and recommended that women using HT discuss the findings of the Women's Health Initiative with their health care providers.

Kaiser Permanente of Northern California is a large health maintenance organization, which provides health care for 3.3 million members. In July 2002, approximately 100,000 health plan members were receiving HT through Kaiser Permanente pharmacies. Within weeks after publication of the Women's Health Initiative results, the health plan described them in a detailed mass mailing to women using postmenopausal estrogen with progestin. To evaluate the response of HT users to efforts by mass media and the health plan to provide information about the effects of HT, we conducted a telephone survey of recipients of the mass mailing. The survey asked about women's key sources of information regarding the new Women's Health Initiative findings on HT, the extent to which women were well informed on this topic, their degree of satisfaction with the quality and quantity of information, and the role of this information in women's subsequent decision to continue or stop HT.

We hypothesized that women who were well informed about HT research findings and those who were satisfied with this knowledge would be more likely to make a decision to stop. We further reasoned that women using long-term HT would be less likely to decide to stop.

Materials and Methods

Women aged 50-69 years who were members of Kaiser Permanente of Northern California for at least 3 consecutive years and who regularly filled prescriptions for HT during the year before publication of the Women's Health Initiative findings were identified from membership rolls and from a computerized pharmacy database. We used the pharmacy database to obtain information on type of estrogen used (conjugated equine estrogens, micronized estradiol, or esterified estrogens), hormone regimen (estrogen or estrogen plus progestin), dosage of estrogen (1.0 mg micronized estradiol and 0.625 mg esterified estrogens were considered equivalent to 0.625 mg conjugated equine estrogens), and specialty of HT prescriber (obstetrics and gynecology, primary care, or other specialty). To ensure that a cohort of regular users was selected, we limited selection to health plan members who showed high compliance with the HT regimen. High compliance was defined as filling three prescriptions for 100 pills of estrogen from July 1, 2001, through June 30, 2002. We included in this cohort women using dosages equivalent to at least 0.625 mg conjugated estrogen. We stratified this cohort of 35,275 eligible women by age group (50-54, 55-59, 60-64, and 65-69 years), and we randomly selected 250 women from each of the four age groups.

We mailed these 1000 women a letter explaining the purpose of the study and a prepaid postcard that could be returned to refuse further contact; if we did not receive this refusal card within 3 weeks after sending it, we attempted to contact the women by telephone for an interview. The protocol and methods were approved by the Kaiser Permanente of Northern California Institutional Review Board.

During July and August 2002, Kaiser Permanente administration launched an initiative to give information about the Women's Health Initiative findings to all women receiving prescriptions for postmenopausal estrogen with progestin (but not to those receiving estrogen alone): HT users were mailed a detailed explanation of the study findings³ with recommendations about continuing HT. These letters were approved by individual prescribers of HT or, if the prescriber was not available, by Kaiser Permanente administration. Letters from clinicians could advise either consultation or that it was safe to stop HT without consultation. The letter from Kaiser Permanente administration recommended consultation with a health care practitioner unless coronary disease was present; if coronary disease was present, discontinuation was recommended. The quality and operations support department provided us with data regarding which of our study participants were sent this letter and the type of letter sent.

A survey instrument was developed to collect information on a number of variables to determine women's reasons for starting, stopping, or continuing HT. Six to eight months after the Women's Health Initiative findings were published, during the months of January through March 2003, five trained interviewers using standardized scripts contacted potential participants and carried out interviews lasting a mean of 20 minutes. No attempt was made to validate that recipients had received the letter or which type they had received. The interviewers collected information on demography, health habits, and medical history. Perceived health status was obtained by asking women to rate their current health as excellent, very good, good, fair, or poor. Perceptions of risk for being affected by breast cancer, colon cancer, coronary heart disease, stroke, venous thromboembolism, and osteoporosis (relative to an average woman's risk) were assessed with a Likert scale of 1 ("much lower") to 5 ("much higher"). Most of these survey items had been used in prior studies.^4,5 New survey questions asked about sources of information, as well as quality, quantity, and importance of information about HT in each woman's decision to continue or discontinue HT. These questions were modeled on questions contained in the Health Employer Data Information System measure on menopause counseling⁶ and used a Likert response scale of 0 ("none or not at all") to 10 ("best or most important").

To characterize knowledge of risks of HT, correct responses to the following five true-or-false questions were summed: using HT reduces risk of breast cancer (correct answer: false); using HT reduces risk of heart attack (false); using HT reduces risk of stroke (false); using HT increases risk of blood clots in the veins of the leg and lung (true); overall, for the average woman, using HT improves health (false).

Regarding knowledge of Women's Health Initiative study results, we asked the following series of questions: 1) "Do you know about the overall results about HT from a research study called 'The Women's Health Initiative'?"; If yes, "Did it show that a) HT was good overall for the women in the study? b) HT was bad overall for the women in the study? c) HT made no difference for women in the study? or d) don't know." "Bad overall" or "made no difference" were considered correct answers, "good" an incorrect answer.

We asked, "Since July of last year, have you stopped or tried to stop taking HT?" and asked standardized follow-up questions to add detail to responses. Responses were categorized as "attempted to stop [using] HT" if a woman stated that she intended to stop and that she either had stopped (for at least 1 week) or that she had reduced the dosage of HT. This category did not include responses of women who reduced the dosage without intending to stop soon.

We used ² tests to compare the characteristics of women who did and did not attempt to discontinue HT. For questions relating to quality and quantity of information, as well as for questions relating to importance of information in the subsequent decision to discontinue HT, we compared Likert scores of 8 or more with lower scores. We considered answering four or more true-or-false questions correctly as an indicator of good knowledge of HT risks. Variables associated with attempting to stop (difference significant at P < .05) were entered into a multivariable logistic regression model for each of the four age strata that were used in the random selection of participants. Three predictive variables were highly correlated with each other: use of estrogen alone, hysterectomy, and not being sent a letter from the health plan about Women's Health Initiative findings. Of these three predictors, we elected to use only the variable that the letter had been sent in the multivariable model. All statistical analyses were performed with SAS 8.2 statistical software (SAS Institute, Cary, NC).

Results

Of the 1000 women invited, 670 (67.0%) completed the interview (Figure 1). The 230 women who refused to be interviewed were similar to those interviewed in mean age, frequency of HT regimen, and frequency of HT prescriber type. Characteristics of the women interviewed are summarized in Tables 1 and 2. Their mean age was 59 years, about three quarters were white, and more than one third were college graduates. Few reported ever having a myocardial infarction (2.6%), stroke (2.4%), breast cancer (1.1%), colon cancer (0.8%), venous thromboembolism (1.5%), or osteoporosis (7.3%). Most (88%) of the women believed that their health status was "good to excellent," and fewer than one third believed that they were at higher-than-average risk for myocardial infarction, breast or colon cancer, venous thromboembolism, or osteoporosis. Estrogen alone was used by 39% and estrogen with progestin by 61%. Of women using estrogen with progestin, two thirds were using continuous combined HT (Table 3). Conjugated equine estrogens at the standard daily dosage of 0.625 mg was used by 85%. Most were long-term users; the median duration of use was 9 years, 9% had used HT for less than 3 years, and 9% reported more than 20 years' use. One third of the women reported that they started HT mainly for vasomotor symptom relief; only 17% reported that they started HT mainly to improve health. Symptom relief was a common secondary reason for starting HT: 63% of women report symptom relief as a secondary reason. Hysterectomy was reported by 40%, and 52% of these women considered this surgery the main reason for starting HT. Nearly all hysterectomized women took estrogen alone (97%), and 8% of women with a uterus used estrogen alone. Of 408 women who used estrogen with progestin, 401 (98%) were sent a letter by the health plan.

Figure 1. Flowchart showing process of screening and enrolling subjects. Ettinger. Decisions About Hormone Therapy. Obstet Gynecol 2003.

Some awareness of new findings about HT since July 2002 was reported by almost all women (93%), but fewer (57%) considered the quality or quantity of this information from any source to be good: women did not consider the quality of information to be good, whether coming from mass media (21%), from the health plan (32%), or from health care providers (34%) (Table 4). Compared with women who were not sent a health plan letter of information, women who were sent letters more frequently reported receiving good-quality information (41% versus 18%, P < .001).

Knowledge of recent findings from HT trials was generally poor. Regarding the Women's Health Initiative trial main results, 64% had no knowledge, 7% were unsure of their knowledge, 6% had incorrect knowledge, and only 23% had correct knowledge. Women who had been sent a letter were better informed of Women's Health Initiative findings than those who had not (29.7% versus 12.3% correct). In response to the five-question quiz about HT risks, 30% answered four to five questions correctly. Compared with women who were not sent the health plan letter, those sent the letter were more likely to answer four or more correctly (37.4% versus 18.6%, P < .001).

Despite the apparent lack of information or of satisfaction with counseling, most women (56.3%) reported that they had tried to stop using HT since July 2002. Several factors were statistically significantly associated with participants attempting to stop (Table 5). Among women who attempted to stop, 79% rated information from mass media as important in their decision, compared with 21% of women who did not attempt to stop (P < .001); 67% of women who attempted to stop rated health plan information important, compared with 33% of those who did not attempt to stop (P < .001). Women who were sent the health plan letter (explaining the Women's Health Initiative findings) were about twice as likely to attempt to stop therapy compared with those who were not sent the letter (Figure 2). Furthermore, the effect of knowledge (about HT risks and benefits) on likelihood of attempting to stop was additive. Each of the two types of letter sent by the health plan (either suggesting that the woman stop HT or suggesting that she continue HT until a counseling appointment was completed) was associated with a similar percentage of women who attempted to stop HT (data not shown). Although the following variables showed statistically significant associations with the decision to stop HT in univariate models, they failed to reach significance in the multivariable model: higher level of education, higher level of perceived health status, average or lower-than-average self-perceived risk of colon cancer, and receiving prescription of HT from a gynecologist. The following variables were not associated with the decision to stop: body mass index, ethnicity, starting HT for vasomotor symptoms, perceived health risks other than colon cancer, and self-reported quality of information received from health professionals.

Figure 2. Percentages of women who attempted to stop hormone therapy (HT), by their knowledge of HT risks and benefits (measured by number of correct answers on quiz), and by their having been sent a letter explaining Women's Health Initiative results. Ettinger. Decisions About Hormone Therapy. Obstet Gynecol 2003.

Among the 248 (44%) women who chose to continue HT, the main reasons given were relief of hot flashes (25.7%) and protection against osteoporosis (19.2%). A variety of other main reasons were given by 47.2% of women, including relief from mood swings, vaginal dryness, urinary incontinence, or depression. Only 5.5% stated that they had not considered the issue, and 2.4% were continuing HT while awaiting a counseling visit. Estrogen dosage reduction was reported by 17.7% of women who did not attempt to stop.

Discussion

In a random sample of women using postmenopausal HT, most tried to stop therapy in the 6-8 months after release of the Women's Health Initiative results. For these women, the mass media were the most commonly reported sources of good-quality new information about HT. However, the strongest predictor of attempts to stop therapy was being sent a letter that summarized Women's Health Initiative results and that suggested that women reconsider their decision to use HT. In most cases, attempts to stop therapy were not made on the basis of accurate knowledge of HT risks; however, women who were better informed were more likely to choose to stop. Women using the standard dosage for a longer duration were more likely to continue therapy, but women who were taking HT primarily for health promotion reasons were more likely to try stopping.

Among women who start taking HT, half discontinue therapy within 1 year, but the annual discontinuation rate decreases to about 10% in both the second^4,7 and third year after starting.⁷ Although we selected our sample at random from the group of women who had been taking HT regularly for a minimum of 1 year before July 2002, most of these women had been using HT for much longer periods (median duration 9 years). Thus, the high proportion of women who decided to discontinue HT in the 6-8 months after publication of Women's Health Initiative trial results can reasonably be ascribed to release of this new information on HT.

It seems that the mass media and Kaiser Permanente succeeded in creating an environment of concern about use of HT; in this environment, the majority of users attempted to stop—in many cases without clear knowledge of HT risks. We do not know whether the therapeutic decisions made by these women on the basis of less-than-good-quality knowledge will be rescinded over time.

The health plan sent informational letters to women using estrogen with progestin but not to those receiving estrogen alone. Thus, we are unable to evaluate the effect of information provided by the health plan among women using estrogen alone. In addition, we did not measure whether patient or provider beliefs about the risks and benefits of HT differ for estrogen taken alone compared with estrogen taken in combination with progestin. Given the overall low level of interviewees' knowledge about Women's Health Initiative trial results, regardless of HT regimen, we suspect that women surveyed were unaware of the potential differences in Women's Health Initiative outcomes related to hormone regimen. Expert groups, such as the North American Menopause Society⁸ and the American College of Obstetricians and Gynecologists³ have cautioned that until proven safe, all forms of HT should be considered to have the same adverse effect on health as combined estrogen and progestin. However, some experts have suggested that because the estrogen arm of the Women's Health Initiative has not been stopped, estrogen alone might not have the same adverse effects as estrogen with progestin.⁹ In January 2003, the US Food and Drug Administration required both Premarin (estrogen alone) and Prempro (estrogen with progestin) to carry the same "black box" label warning of increased risk associated with treatment.^10,11

Our study has several strengths. The study cohort represents a random sample of women using postmenopausal hormone therapy in a large health maintenance organization. Most of our questionnaires were used in previous research and were designed to avoid leading questions. Interviewers were trained and used standardized scripts to elicit responses. Although we relied on self-report and recollection, we were able to interview women only 6-8 months after new information about the risks of HT was published in the biomedical literature and presented in the mass media. However, we were unable to interview one third of the selected random sample—in most cases because they returned a refusal postcard (included with the mailed invitation) or because these women refused to be interviewed once they were contacted by telephone. Characteristics of those who were not interviewed did not differ from those who were interviewed.

Our results, however, might not be generalizable to all patient populations. Although the health plan membership reflects US national averages in ethnic makeup, the members are somewhat wealthier and better educated¹² than the national average. In our cohort, the percentage of women who had a hysterectomy was about twice that in the US population. Hysterectomized women are two to three times more likely to choose and continue taking HT than are women with an intact uterus.¹³ Beliefs and values regarding use of HT could differ between women in northern California and women in the rest of the United States. However, our finding that approximately half of women using HT have tried to stop is consistent with the 25-30% reduction in sales of Premarin products reported between July and September 2002.¹⁴ In addition, our report does not describe the cohort's experiences subsequent to stopping, particularly withdrawal symptom patterns and recidivism.

How should results of this study be used by clinicians? First, despite the prolific media response to publication of the results of the Women's Health Initiative and other clinical trials of HT in 2002, we must not assume that women have obtained the requisite knowledge to make informed decisions. Improving women's knowledge about HT through better counseling is the aim of the Health Employer Data Information System menopause counseling measure, instituted in 2000.⁶ Managed care organizations throughout the United States are required to report percentages of randomly surveyed women (aged 47-55 years) who report satisfaction with menopause counseling and who have basic knowledge about risks and benefits of HT. Similar to our findings, the results of the Health Employer Data Information System surveys in 1999 and 2000 indicated that only 33% of women members reported receiving counseling from health care providers about menopause and HT that they considered of good to excellent quality.¹⁵ Thus, we must recognize that women are, in general, not satisfied with information about HT provided by the mass media or by health care providers. However, the impact of a simple letter of information from the health plan was impressive, both in increasing patients' knowledge of HT and in affecting their decision to stop HT. On the basis of our finding that approximately one in four women who were correctly informed about the Women's Health Initiative trial results decided not to stop HT, we must assume that many still have a strong commitment to continuing therapy.

We conclude that in the 6-8 months after publication of the Women's Health Initiative trial results, most regular HT users attempted to stop using HT despite not being well informed about new research findings. Efforts are still required to provide women with adequate information about risks and benefits of HT.

References

1. Rossouw JE, Anderson GL, Prentice RL, LaCroix AZ, Kooperbeg C, Stefanick ML, et al. Writing Group for the Women's Health Initiative Investigators. Risks and benefits of estrogen plus progestin in healthy postmenopausal women: Principal results from the Women's Health Initiative randomized controlled trial. JAMA 2002;288:321-33.

2. Grady D. A 60-year-old woman trying to discontinue hormone replacement therapy. JAMA 2002;287:2130-7.

3. American College of Obstetricians and Gynecologists. Questions and answers on hormone therapy. Available at: www.acog.com/from_home/publications/press_releases/nr08-30-02.cfm. Accessed 2003 Mar 29.

4. Ettinger B, Pressman A, Silver P. Effect of age on reasons for initiation and discontinuation of hormone replacement therapy [published erratum appears in Menopause 2000;7:135]. Menopause 1999;6:282-9.

5. Tosteson AN, Grove MR, Hammond CS, Moncur MM, Ray GT, Hebert GM, et al. Early discontinuation of treatment for osteoporosis. Am J Med 2003;115:209-16.

6. National Committee for Quality Assurance. HEDIS 2000. Informed health care choices. Volume 3. Washington: National Committee for Quality Assurance, 2000, 133-59.

7. Ettinger B, Li DK, Klein R. Continuation of postmenopausal hormone replacement therapy: Comparison of cyclic versus continuous combined schedules. Menopause 1996;3:185-9.

8. North American Menopause Society. Amended report from the NAMS Advisory Panel on Postmenopausal Hormone Therapy. Menopause 2003;10:6-12.

9. Schneider HP, Executive Committee of the IMS. The view of The International Menopause Society on the Women's Health Initiative. Climacteric 2002;5:211-6.

10. Wyeth Pharmaceuticals. Prempro/Premphase full revised label. Available at: http://www.fda.gov/medwatch/SAFETY/2003/prempro_PI.pdf. Accessed 2003 May 16.

11. Wyeth Pharmaceuticals. Premarin full revised label. Available at: http://www.fda.gov/medwatch/SAFETY/2003/premarin_PI.pdf. Accessed 2003 May 16.

12. Krieger N. Overcoming the absence of socioeconomic data in medical records: Validation and application of a census-based methodology. Am J Public Health 1992;82:703-10.

13. Lancaster T, Surman G, Lawrence M, Mant D, Vessey M, Thorogood M, et al. Hormone replacement therapy: Characteristics of users and non-users in a British general practice cohort identified through computerised prescribing records. J Epidemiol Community Health 1995;49:389-94.

14. Colata G. Rush to fill void in menopause-drug market. New York Times. September 1, 2002;sect 1:1

15. National Committee for Quality Assurance. Management of menopause: Informed choices. Available at: http://www.ncqa.org/somc2001/MENOPAUSE/SOMC_2001_MOM.html. Accessed 2003 May 16.

Address reprint requests to: Bruce Ettinger, MD, Kaiser Permanente Medical Care Program, Division of Research, 3505 Broadway, Oakland, CA 94611-5714; E-mail: bxe@dor.kaiser.org

Received May 29, 2003.
Received in revised form July 15, 2003.
Accepted August 1, 2003.