Menopausal hormone therapy use has fluctuated dramatically over the past 50 years since its introduction in the United States. By the late 1990s, approximately one quarter of all U.S. women 40 years of age and older were using some formulation of menopausal hormone therapy,1 making it one of the most commonly prescribed medications among women.2 Hormone therapy was widely prescribed to alleviate menopausal symptoms3 as well as reduce the risk of osteoporosis and heart disease—diseases found to be negatively associated with hormone therapy in observational studies.4–6 Some studies have also found hormone therapy to increase cognitive abilities and decrease dementia, although this association was less well established.7
In 2002, the Heart and Estrogen/Progestin Replacement Study concluded that there was no cardiovascular disease benefit from combined estrogen–progestin menopausal hormone therapy (HT) among women with established disease.8 Later that year, the HT treatment arm of the Women’s Health Initiative was terminated because an increased risk of breast cancer and cardiovascular disease events9 was identified. The estrogen-only therapy (ET) arm of the study was continued, because insufficient data had been gathered to identify a significant overall risk–benefit estimate at that time. However, by 2004, the ET study had been terminated by the Data Safety Monitoring Board because of increased stroke risk and the lack of perceived protection for cardiovascular disease. Based on evidence from these clinical trials, the Food and Drug Administration produced new labeling for ET and HT medications that recommends menopausal hormone use only for menopausal symptoms and for preventing osteoporosis among women at significant risk of that condition (Food and Drug Administration, Estrogen and Estrogen with Progestin Therapies for Postmenopausal Women, http://www.fda.gov/cder/drug/infopage/estrogens_progestins/default.htm).
Reports indicate that a dramatic drop in sales of menopausal hormone therapy occurred after the release of the 2002 Women’s Health Initiative findings.10 One study found a rapid drop in HT prescriptions between January 2002 and June 2003 based on a national sample of daily pharmacy sales.11 The same study also found a drop in physician visits with any mention of HT between 2002 and June 2003 based on a national sample of physician visits and associated medication use. A major limitation of that study, however, was that data on hormone therapy use was incomplete for 2002 and 2003 (estimates were annualized based on January through June data for each year). That study was also unable to stratify hormone therapy use by patient and physician characteristics.
The purpose of the current study is to examine the effects of the 2002 clinical trial results on prescription patterns among women aged 40 years and older by analyzing data on prescription practices using visits occurring throughout the entire calendar years 2001 through 2003. The analysis uses nationally representative data obtained from a sample of visits to office-based physicians and hospital outpatient departments. Changes in population-based rates of visits with either ET or HT are examined, as well as changes in characteristics of patients using each type of therapy and the physicians writing the prescriptions.
MATERIALS AND METHODS
This study uses data collected in the National Ambulatory Medical Care Survey (NAMCS) and the hospital outpatient department component of the National Hospital Ambulatory Medical Care Survey (NHAMCS). The NAMCS is a nationally representative survey of patient visits to nonfederally employed, office-based physicians, whereas the NHAMCS is a nationally representative survey of patient visits to emergency departments and outpatient departments in nonfederal, short-stay (less then 30 days) hospitals. Because the NHAMCS was designed to complement office-based physician visits from the NAMCS with visits occurring in a hospital setting, only visits under the supervision of a physician were eligible for the outpatient department component of the NHAMCS. Hospital clinics or outpatient departments providing only ancillary services (radiology, laboratory, pharmacy, physical therapy, and cancer treatment departments) were excluded. Emergency department visits are collected in the NHAMCS but were excluded from this analysis. Both surveys are conducted annually by the National Center for Health Statistics.
The NAMCS surveys visits made in the United States to the offices of nonfederally employed physicians (excluding those in the specialties of anesthesiology, radiology, and pathology) who were classified by the American Medical Association and the American Osteopathic Association as providing “office-based, patient care.” It uses a multistage probability sample design. Eligible sample physicians who agree to participate in the survey complete Patient Record forms for a systematic random sample of approximately 30 office visits occurring during a randomly assigned 1-week reporting period. The survey samples roughly 25,000 visits in about 2,000 physician offices each year.
The NHAMCS surveys patient visits to emergency and outpatient departments of nonfederal, short-stay or general hospitals. Short-stay hospitals are those with an average length of stay of less than 30 days, while general hospitals include those whose specialty is general medical or surgical or children’s general. Federal hospitals, hospital units of institutions, and hospitals with less than 6 beds are ineligible for the NHAMCS. It employs a 4-stage probability sample design involving samples of geographic primary sampling units, hospitals with emergency departments and outpatient departments within primary sampling units, emergency service areas within emergency departments, and clinics within outpatient departments. The final sampling stage involves systematic random samples of outpatient clinic and emergency department visits during randomly assigned 4-week reporting periods. Each year, the survey samples roughly 35,000 sampled outpatient department medical encounters in about 480 nonfederal, short-stay, and general hospitals.
The study population includes visits made by women aged 40 years and older to physicians in office-based practices and hospital outpatient departments during 2001 through 2003. For this population, a total of 56,440 visits were available for analysis. The number of sampled visits was 17,724 in 2001, 20,069 in 2002, and 18,667 in 2003. Menopausal hormone therapy drugs included any drug with a therapeutic class of estrogen or progestin, excluding contraceptives and infertility medications, based on the National Drug Therapeutic code. Menopausal hormone drugs were further classified as ET or HT (dual) therapy based on ingredients. Drugs recorded at visits with pregnancy, fertility, or family planning as the reason for the visit were excluded, based on the patient’s reported reason for the visit or the physician’s diagnosis. Women were classified as having current menopausal symptoms if they mentioned menopausal symptoms as a reason for the visit. Women whose menopausal symptoms were successfully alleviated by use of menopausal hormone therapy were missed in this definition. Although up to 8 drug mentions were collected in the NAMCS and NHAMCS outpatient department questionnaire in 2003, drug mentions for that year were limited to the first 6 listed to be comparable with data collected in 2001 and 2002.
The data for all 3 years were weighted to reflect the complex multistage national probability sample design of the NAMCS and NHAMCS. Sample data were weighted to produce annual national estimates. In this article, NAMCS estimates use a weight that includes a revised adjustment for nonresponse. The revised estimator was used for the first time with 2003 NAMCS data.12 For comparability, the NAMCS data for 2001 and 2002 were also analyzed with the revised estimator because the effect of the new weighting formula was to increase the magnitude of visit estimates.12
Key variables analyzed were hormone therapy visit rates and the percent of visits with any mention of HT or ET. Visit rates were based on visits with any hormone therapy mentioned by type (HT, ET). The denominators were the Census 2000-based postcensus estimates of the civilian noninstitutional population of the United States as of July 1 for the years 2001–2003. Trends in visits rates by time were tested using a modified least-squares regression analysis that weights each visit rate by the inverse of the standard error.13
Bivariate differences in the percent of HT and ET visits according to selected characteristics were examined using χ2 tests of association, and the Bonferroni test for multiple comparisons at the P = .05 level was used for specific characteristics.13 Characteristics examined included demographic characteristics, expected payment sources, presence of current menopausal symptoms, types of providers seen, and location of the provider. These characteristics were found in previous studies to be correlated with menopausal hormone use.1,2,11 Changes in HT and ET visits by time, while simultaneously controlling for patient and provider characteristics was analyzed in a multinomial logistic model with ET compared with HT therapy compared with all other nonhormone visits as the dependent variables.
Standard errors of individual visit rates and percentages, as well as analysis by χ2 tests of association and multinomial logistic regression models were computed using SUDAAN software (Research Triangle Institute, Research Triangle Park, NC) to account for the complex sample designs of the NAMCS and NHAMCS.14
During 2001–2003, women aged 40 years and older made a total of 1.1 billion visits to office-based physicians and hospital outpatient departments or an average of 363 million visits annually. These visits did not vary by demographic characteristics, expected payment sources, the percent with current menopausal symptoms, types of providers seen, or the location of the provider during the study period (data not shown). However, the number of visits with any hormone therapy prescribed declined from 26.5 million in 2001 to 16.9 million in 2003 (P < .002), and the population visit rate with any hormone therapy prescription declined from 41.4 visits per 100 women aged 40 years and older in 2001 to 26.0 visits per 100 women in 2003 (Table 1). An estimated 64.8 million visits involved a hormone therapy prescription during the whole study period.
The majority of visits with any hormone therapy prescription received ET prescription (72%), whereas 28% of these visits involved HT prescriptions. Between 2001 and 2003, the visit rate for women with HT prescriptions declined by 43.6% (from 11.7 per 100 women aged 40 years and older in 2001 to 6.7 in 2003). Significant declines in visit rates for women using HT occurred among those aged 50–59 years and 60 years and older. The visit rate for women with ET prescriptions also declined but by a smaller percentage (35%, from 29.7 per 100 women aged 40 years and older in 2001 to 19.3 in 2003). These declines were also due to lower ET use among women aged 50–59 years and 60 years and older, similar to the declines in visits with HT prescriptions.
During 2001–2003, 5.9% of visits by women aged 40 years and older involved a hormone therapy prescription. Current menopausal symptoms and type of provider were highly predictive of hormone therapy use. About 17.2% of visits by women with current menopausal symptoms received HT prescriptions compared with 1.5% without these symptoms (P < .05). Similarly, visits by women with current menopausal symptoms were 8.2 times as likely to receive ET prescriptions as women without menopausal symptoms (33.5% compared with 4.1%, P < .05). Women without menopausal symptoms, however, accounted for 92% of all HT visits and 94% of ET visits during 2001–2003. Visits to obstetric and gynecology physicians or clinics were more likely to result in prescription of these therapies than visits to other primary care physicians (general or family practice and internal medicine physicians, general medicine clinics), or any other providers (P < .05). Chi–square tests of association of visits involving HT, ET, and no hormone therapy prescriptions by selected characteristics are shown in Table 2.
Results of a multinomial logistic regression for use of HT and ET relative to no hormone therapy used as the dependent variable are presented in Table 3. Relative to visits with no hormone prescriptions, visits with HT prescriptions declined slightly more than ET visits, controlling for all other characteristics. This difference, however, was not statistically significant. Most covariates were similarly related to HT visits and ET visits. The exceptions were age (older age was associated with decreased odds of only HT visits) and source of payment, which was more strongly associated with HT visits.
Although not shown, we also ran the analysis including interaction terms for year by age, race, and expected payment in the model. None of these interactions terms were significant at the P = .05 level in this model.
Because the Women’s Health Initiative health findings were targeted primarily to use of hormone therapy for preventive reasons, we ran our logistic model stratified by presence of menopausal symptoms. The results were virtually the same for those without menopausal symptoms; there was a significant decrease in visits with hormone prescriptions over the period. There was no significant trend over time in visits with either type of hormone prescription among women with menopausal symptoms.
These nationally representative data indicate that the clinical response to the Women’s Health Initiative clinical trial and subsequent guidelines was a significant decline in HT and ET prescriptions written in 2002 and 2003 in both physician offices and hospital outpatient departments. This study found that visit rates with HT and with ET prescriptions declined during this period for women in the age groups 50–59 years and 60 years and older. The largest decline in hormone therapy occurred among women being prescribed HT. However, a significant decline of ET use was also noted during the study period. These findings are consistent with a previous study that found sales of hormone replacement therapy prescriptions dropped, as well as the percentage of visits involving estrogen–progestin.11
The model used in this study replicates models estimated in earlier studies of hormone replacement therapy using the 1993–1995 NAMCS and NHAMCS data.15,16 Therefore, it is not unexpected that some of the patient demographic characteristics that were associated with hormone therapy use in these previous studies were also found using the 2001–2003 data. During 2001–2003, ET was more frequently prescribed than HT during visits to physician offices and hospital outpatient departments, as has been found previously.1 Although our data cannot differentiate menopause type among the sampled patients, it is generally known that women with surgical menopause are more likely to use hormone therapy than women with natural menopause and that women with surgical menopause are most often treated with ET.1 A previous study found that women who experience a bilateral oophorectomy were 1.9 times more likely to use hormone therapy for 5 years or more than women with natural menopause.17 Another household-based survey found that 58.7% of women who underwent a hysterectomy were currently using hormone therapy, compared with 19.6% of women with intact uteri.18
A major limitation of the data are that it is cross-sectional. Information on issues that occurred in the past cannot be connected to the current visit. Several pieces of data that would have been very useful for this analysis were missing due to this limitation: hysterectomy status and past menopausal symptoms. For example, although visits with ET prescriptions went down over time, we cannot say whether this is true among all women or was concentrated among women either with or without intact uteri. Because the Women’s Health Initiative results that had been announced in 2002 were for women with intact uteri, this information would have been very useful as a stratifying variable. Another example of a way this limited our analysis is that although we would like to know if use of hormone therapy has gone down among women whose primary reason for use was menopausal symptoms, the only information available to us is on current symptoms. This group would by definition be women whose use of hormone therapy was for alleviation of symptoms and for whom hormone therapy was either newly being prescribed or had been ineffective for that indication, not the total proportion of women for whom symptoms were the reason for hormone therapy use.
Another limitation of this study is the fact that the data being analyzed sampled ambulatory medical care visits rather than patients. Since 2001, a new methodology has become available to estimate patients from encounter data such as found in the NAMCS and NHAMCS surveys. Specifically, beginning in 2001, the 2 surveys started collecting an item on the multiplicity of visits made by the sample patient to the same provider during the last 12 months. This information, in turn, was used to reweight the visit estimates to reflect patients by reducing the influence of multiple visits made by the same patient to the same provider (Burt CW, Hing E. Making patient-level estimates from medical encounter records with a multiplicity estimator. Stat Med. In press.). Using this methodology for the 2003 data year, we found that the percentage of women aged 40 years and older with an HT prescription was statistically similar to the percent of visits with an HT prescription. Likewise, the percentages of female patients with an ET prescription and visits with an ET prescription were statistically similar. Thus, the fact that this study used data that was sampled by visit and not by person is probably not a major limitation, because estimates of HT or ET use were similar.
Prescribing patterns for hormone replacement therapy have fluctuated as findings of its potential health risks and benefits appear in the medical literature. The NAMCS and NHAMCS surveys, as well as a previous study of pharmacy sales have documented 3 successive patterns in hormone therapy since 1995: increases in 1995 to 2000, primarily due to the adoption of new estrogen–progestin formulations and findings of beneficial effects of hormone therapy in observational studies, stabilization between 2000 and 2001, and a rapid drop starting in 2002 to 2003 to 1995 levels, with the largest decline among oral estrogen and estrogen–progestin.11,12,15,19 This study has shown that ambulatory visits involving HT prescriptions declined after release of HT findings from the Women’s Health Initiative. Visits involving ET prescriptions also declined. The decline occurred among women aged 50 years and older for both therapies.
The results of this investigation are noteworthy for 2 reasons. First, all types of hormone therapy use declined at the time the Women’s Health Initiative results for HT were announced, although no results were revealed regarding the risks and benefits of ET use. The subsequent termination of the Women’s Health Initiative ET trial due to greater risk than benefit of ET occurred after the study period of this investigation.20 Although the HT research results were heavily covered by the news media, our results suggest that the effects on prescribing patterns went beyond HT. Some research has found the public lacked understanding21,22 or were worried23 about the details of the Women’s Health Initiative findings during the period of the current study. Further research into the consequences of the release of Women’s Health Initiative findings in the future could illuminate whether the passage of time, and possibly additional interactions with primary care physicians, will alter the trends in menopausal hormone therapy use.
A second issue this research raises is the rapid response to clinical findings. In the case of hormone therapy use, some women stopped use without counseling by a physician about the menopausal symptoms that return with medication termination.24,25 If menopausal symptoms resume and continue for extended lengths of time, some women may resume use of hormone therapy, because it is still considered the best treatment for those menopausal symptoms.26 It will be important to continue to investigate use of hormone therapy over time to study the effect of additional findings regarding ET10, specifically, the Women’s Health Initiative finding that there is no association between hormone therapy and cognition on the trends in use.27
1. Brett KM, Reuben CA. Prevalence of estrogen or estrogen–progestin hormone therapy use. Obstet Gynecol 2003;102:1240–9.
2. Brett KM, Burt CW. Utilization of ambulatory medical care by women: United States, 1997-98. Vital Health Stat 2001;13:1–46.
3. MacLennan AH, Broadbent JL, Lester S, Moore V. Oral oestrogen and combined oestrogen/progestogen therapy versus placebo for hot flushes. Cochrane Database Syst Rev 2004; CD002978.
4. Stampfer MJ, Colditz GA. Estrogen replacement therapy and coronary heart disease: a quantitative assessment of the epidemiologic evidence. Prev Med 1991;20:47–63.
5. Barrett-Connor E, Stuenkel CA. Hormone replacement therapy (HRT)–risks and benefits. Int J Epidemiol 2001;30:423–6.
6. Grady D, Rubin SM, Petitti DB, Fox CS, Black D, Ettinger B, et al. Hormone therapy to prevent disease and prolong life in postmenopausal women. Ann Intern Med 1992;117:1016–37.
7. LeBlanc ES, Janowsky J, Chan BK, Nelson HD. Hormone replacement therapy and cognition: systematic review and meta-analysis. JAMA 2001;285:1489–99.
8. Grady D, Herrington D, Bittner V, Blumenthal R, Davidson M, Hlatky M, et al. Cardiovascular disease outcomes during 6.8 years of hormone therapy: Heart and Estrogen/progestin Replacement Study follow-up (HERS II). JAMA 2002;288:49–57.
9. Rossouw JE, Anderson GL, Prentice RL, LaCroix AZ, Kooperberg C, Stefanick ML, et al. 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.
10. Majumdar SR, Almasi EA, Stafford RS. Promotion and prescribing of hormone therapy after report of harm by the Women’s Health Initiative. JAMA 2004;292:1983–8.
11. Hersh AL, Stefanick ML, Stafford RS. National use of postmenopausal hormone therapy: annual trends and response to recent evidence. JAMA 2004;291:47–53.
12. Hing E, Cherry DK, Woodwell DA. National Ambulatory Medical Care Survey: 2003 summary; No. 365. Hyattsville, MD: National Center for Health Statistics; 2005.
13. Sirken MG, Shimizu I, French DK, Brock DB. Manual on standards and procedures for reviewing statistical reports. Hyattsville, MD: National Center for Health Statistics; 1990.
14. Research Triangle Institute. SUDAAN User’s Manual, Release 9.0. Research Triangle Park (NC): Research Triangle Institute; 2004.
15. Marsh JV, Brett KM, Miller LC. Racial differences in hormone replacement therapy prescriptions. Obstet Gynecol 1999;93:999–1003.
16. Stafford RS, Saglam D, Causino N, Blumenthal D. Low rates of hormone replacement in visits to United States primary care physicians. Am J Obstet Gynecol 1997;177:381–7.
17. Brett KM, Madans JH. Use of postmenopausal hormone replacement therapy: estimates from a nationally representative cohort study. Am J Epidemiol 1997;145:536–45.
18. Keating NL, Cleary PD, Rossi AS, Zaslavsky AM, Ayanian JZ. Use of hormone replacement therapy by postmenopausal women in the United States. Ann Intern Med 1999;130:545–53.
19. Stafford RS, Saglam D, Causino N, Blumenthal D. The declining impact of race and insurance status on hormone replacement therapy. Menopause 1998;5:140–4.
20. Anderson GL, Limacher M, Assaf AR, Bassford T, Beresford SA, Black H, et al. Effects of conjugated equine estrogen in postmenopausal women with hysterectomy: the Women’s Health Initiative randomized controlled trial. JAMA 2004;291:1701–12.
21. Sawka AM, Huh A, Dolovich L, Papaioannou A, Eva K, Thabane L, et al. Attitudes of women who are currently using or recently stopped estrogen replacement therapy with or without progestins: results of the AWARE survey. J Obstet Gynaecol Can 2004;26:967–73.
22. Ettinger B, Grady D, Tosteson AN, Pressman A, Macer JL. Effect of the Women’s Health Initiative on women’s decisions to discontinue postmenopausal hormone therapy. Obstet Gynecol 2003;102:1225–32.
23. Barber CA, Margolis K, Luepker RV, Arnett DK. The impact of the Women’s Health Initiative on discontinuation of postmenopausal hormone therapy: the Minnesota Heart Survey (2000-2002). J Womens Health (Larchmt) 2004;13:975–85.
24. Schonberg MA, Wee CC. Menopausal symptom management and prevention counseling after the Women’s Health Initiative among women seen in an internal medicine practice. J Womens Health (Larchmt) 2005;14:507–14.
25. Ness J, Aronow WS, Beck G. Menopausal symptoms after cessation of hormone replacement therapy. Maturitas 2006;53:356–61.
26. Hickey M, Davis SR, Sturdee DW. Treatment of menopausal symptoms: what shall we do now? Lancet 2005;366:409–21.
27. Rapp SR, Espeland MA, Shumaker SA, Henderson VW, Brunner RL, Manson JE, et al. Effect of estrogen plus progestin on global cognitive function in postmenopausal women: the Women’s Health Initiative Memory Study: a randomized controlled trial. JAMA 2003;289:2663–72.