Secondary Logo

Journal Logo

ORIGINAL RESEARCH

Racial Differences in Hormone Replacement Therapy Prescriptions

MARSH, JANE V. R. MS; BRETT, KATE M. PhD; MILLER, LISA C. MSPH

Author Information
  • Free

Greater use of hormone replacement therapy (HRT) by white than black women has been reported in several studies in the United States.1–7 Reasons for this difference have not yet been fully explored, but differential health behaviors, health risks, and access to care among black and white women have been posited as potential causes1,2 Recent studies of black women's attitudes toward menopause8,9 found a lack of knowledge about menopause or the health risks and benefits of HRT. Evidence shows that physician recommendations of HRT are the primary factors influencing its use among women.10 Therefore, studying aspects of physician visits that might differ by race is an important next step to understanding racial differences in HRT use.

The goal of this study was to estimate differences in prescriptions of HRT to black and white women in a nationally representative sample of visits to U.S. physicians' offices and outpatient departments. Previous studies typically used patient surveys1–6 to estimate racial differences in the use of HRT, so differences associated with physician practice types or aspects of medical visits could not be addressed. Many studies did not collect data to assess accessibility of health care or used geographically limited samples4–7 that may not be nationally generalizable. Using national data on physician visits rather than self-reports of patients' prescription histories, this study sheds light on potential dissimilarities in physicians' prescription patterns for black and white women's use of HRT.

Materials and Methods

The data for this analysis came from the 1993–1995 National Ambulatory Medical Care Surveys and National Hospital Ambulatory Medical Care Surveys. Both surveys are part of the National Health Care Survey and are annual, national probability surveys that sample visits to nonfederally employed, office-based physicians and outpatient and emergency departments of nonfederal, short-stay hospitals in the United States. Details of the sampling procedure have been published elsewhere.11,12 Because a disproportionate sampling design was used to select patient visits for inclusion in both survey samples, all analyses used sample weights to produce national estimates, and all standard error estimation was performed with SUDAAN software (Version 6.40; Research Triangle Institute, Research Triangle Park, NC).13 Data collected over 3 years were combined to obtain more precise estimates.

Emergency room visits were excluded from all analyses because encounters in that setting tend not to involve HRT prescriptions. To limit confounding due to age and to focus on the age group most likely to use HRT, we limited the analyses to women aged 45–64 years. Visits by women of nonwhite and nonblack races were excluded because of small numbers (n = 936 for all other races combined). These exclusion criteria resulted in a combined data set of 25,203 sampled visits made by black (16.4%) and white (83.6%) women. The primary analysis was performed using the combined data set. Because some potentially important confounding variables were available only in the National Ambulatory Medical Care Survey, a secondary analysis was conducted using only that data set. All logistic regression and rate analyses performed in the primary analysis were repeated using 14,878 sampled visits made by black (9.0%) and white (91.0%) women 45–64 years of age.

In both surveys, any new or continuing medication that was ordered, administered, or provided during a visit was collected and coded using the National Drug Code Directory, 1985 edition (1993–1994 surveys)14 and 1995 edition (1995 surveys).15 Up to three of the patient's stated reasons for the visit and three of the physician's principal diagnoses for the visit were collected. Reasons for visits were coded according to “A Reason for Visit Classification for Ambulatory Care,”16 and diagnoses were coded using the International Classification of Diseases, 9th Revision, Clinical Modification. For this analysis, all estrogen drugs, alone or in combination with progesterone, medroxyprogesterone, or norethindrone, were considered HRT, excluding those prescribed for contraception or infertility. Patients' reasons for visits and physicians' diagnoses were used to identify and exclude visits involving a medication that qualified as HRT but was prescribed for contraception or infertility.

Logistic regression analysis was used to examine whether any of the previously identified racial differences in HRT use could be attributed to known confounders. Potential confounders included the woman's age (45–49, 50–54, 55–59, 60–64), expected source of payment for the visit (private, government-only, self-pay or no charge, other), drug mentions other than HRT (yes, no), whether the physician had previously seen the patient (yes, no), physician or clinic specialty type grouped into three categories (general medicine [family practice, general practice, geriatric medicine, and internal medicine], obstetrics and gynecology [obstetrics, gynecology, obstetrics and gynecology, maternal-fetal medicine, critical care medicine in obstetrics and gynecology, reproductive endocrinology, and gynecologic oncology], and all other types), type of care (office-based, hospital outpatient department), and region of practice (Northeast, Midwest, South, West). Potential confounding variables available only in the National Ambulatory Medical Care Survey data and considered only in secondary analyses included obesity (yes, no), duration of the visit (0–15 minutes, 16 or more minutes), and physician's sex.

For the variables considered in analysis of the combined data, we used imputation using a “hot deck” procedure that randomly assigned a value from another visit record with similar characteristics to complete missing data for race (8.0%), sex (1.5%), age (2.6%), and new or old patient status (0.7%). All other variables contained complete data. For the analyses limited to National Ambulatory Medical Care Survey data, imputation was used to complete missing data for race (7.0%), sex (2.2%), age (3.0%), and duration of visit (6.4%). Physician's sex was missing for 10.0% of observations. All remaining variables contained complete data.

Three-way contingency tables of visits by HRT prescription, patient's race, and all other variables were used to explore potential interactions and confounding. Logistic regression was used to estimate both the crude and adjusted associations of race with HRT prescription. Significance of interactions was tested using likelihood ratio tests at the α = .10 level. Potential confounders were dropped from the model if they had no appreciable effect on the coefficient for race.17 To assess the effect of missing race values on odds ratio (OR) estimates, logistic regression models were rerun after eliminating all observations for which race was imputed.

Differences in rates of physician visits by black and white women in the study age group might explain racial differences in the proportion of visits with HRT prescriptions. To address this issue, we used census data to calculate annual ambulatory medical care visit rates per person and annual rates of ambulatory medical care visits with HRT prescriptions per person, separately for black and white women, along with 95% confidence intervals (CIs). White and black rates were stratified by variables shown to be confounders in logistic regression analysis to identify any differences in HRT prescription rates that might be due to different physician usage patterns.

Results

Based on the combined sample, during the 3-year period from 1993 to 1995, there were approximately 296,361,000 ambulatory medical care visits by black and white women aged 45–64 in the United States, or 98,787,000 visits per year. New or continuing prescriptions for HRT were reported in an estimated 9.2% (95% CI 8.0%, 10.3%) of the total number of visits. Prescriptions for HRT at visits by white women were double those of black women during this time; an HRT prescription was reported in 9.7% (95% CI 8.5%, 11.0%) of approximately 263,314,000 visits by white women compared with 4.5% (95% CI 3.2%, 5.8%) of approximately 33,048,000 visits by black women.

Table 1 presents the proportions of all ambulatory medical care visits by race. A significantly higher percentage of visits by black women than white women were made to outpatient clinics (18% compared with 7%), included a non-HRT prescription (71% compared with 63%), and were made to family practitioners (58% compared with 45%). A higher percentage of visits by white women than black women were to obstetrician-gynecologists (9% compared with 6%). The percentages of visits made for reasons related to menopause and menstrual or gynecologic problems did not differ by race.

Table 1
Table 1:
Ambulatory Care Visits According to Selected Visit Characteristics

The crude OR for HRT prescriptions given at visits by white women compared with black women was 2.27 (95% CI 1.62, 3.19). No significant interactions were found with race and the other covariates. Only type of physician or outpatient department, region of the United States, source of payment, existence of non–HRT prescriptions, and health care provider type modified the OR for race and were retained in the model. Visits by white women were more than twice as likely to include prescriptions for HRT than visits by black women, adjusting for other variables in the model (adjusted OR 2.07; 95% CI 1.48, 2.90) (Table 2). Compared with non–gynecologists and non–family or general practitioners, visits to gynecologists were 19 times more likely to involve HRT prescriptions, and visits to family or general practitioners were more than three times as likely to do so, adjusting for the other variables in the model. No remaining variables in the model had a stronger association with HRT prescriptions than race. Logistic regression analyses of data excluding observations with missing race produced virtually identical results.

Table 2
Table 2:
Factors Associated With Ambulatory Care Visits Involving Prescriptions for Hormone Replacement Therapy

The overall rate of ambulatory medical care visits per year per woman was 3.94 (95% CI 3.69, 4.19) for white women and 3.82 (95% CI 3.31, 4.33) for black women. Despite virtually equal medical care visit rates for black and white women, the rate of medical care visits per year in which HRT was prescribed to white women was more than twice the corresponding rate for black women in this age group: 0.38 (95% CI 0.32, 0.45) for whites and 0.17 (95% CI 0.12, 0.23) for blacks. When stratified by covariates, annual rates of physician use were similar by race, although white women had higher visit rates to obstetrician-gynecologists (0.37; 95% CI 0.31, 0.42 compared with 0.21; 95% CI 0.15, 0.28) and higher rates of visits paid for by private insurance (2.48; 95% CI 2.46, 2.50 compared with 1.74; 95% CI 1.46, 2.02). Despite similar annual visit rates by race, stratified annual rates of visits with HRT prescriptions remained higher for white women than for black women in every category of each covariate except for visits to hospital outpatient departments and government-paid visits.

Analyses of data from the National Ambulatory Medical Care Survey produced virtually identical results to those based on the combined data. In the logistic regression model, physician sex, but not visit duration or patient obesity, confounded the effect of race on HRT prescriptions; however, physician sex was not significantly associated with HRT prescriptions when all other variables were included in the model.

Discussion

This study found that visits to office-based physicians and outpatient clinics by white women were more than twice as likely to involve HRT prescriptions as were visits by black women. This association remained after adjusting for several important characteristics of the physician-patient interaction, including specialty and gender of the physician, the patient's use of other medications, source of payment for the visit, and the type and geographic location of the care setting. The rate of visits with HRT prescriptions per white woman was higher than the corresponding rate per black woman, although rates of all ambulatory care for black and white women were similar. The rate of visits to gynecologists was higher among white women, as was the rate of visits covered by private insurance. Visits in these categories tended to be more likely to include HRT prescriptions, so these racial differences may explain part of the overall racial differences in HRT prescriptions.

Two previous studies examined HRT prescription patterns by interviewing physicians.18,19 In one study, 97% of a sample of obstetrician-gynecologists in Los Angeles reported routine use of HRT for postmenopausal patients, regardless of the percentage of minority women in their practices.18 Although it did not address race, the second study found that 88% of women aged 40–69 received information about HRT, but only 61% received new or continuing prescriptions.19 These findings and ours suggest that differences in HRT use by race might result from differences in the proportions of black and white women who receive information about HRT and request prescriptions for it.

A strength of this study was its use of a large, nationally representative sample of ambulatory medical care visits. Thus, it was possible to obtain estimates of frequencies, rates, and relative risks that are generalizable to the United States as a whole. It was also possible to control for factors associated with the medical practice and the medical encounters that have not been measured previously; however, these variables did not explain much of the difference in HRT prescriptions.

This data source also has limitations. Because both the National Ambulatory Medical Care Survey and the National Hospital Ambulatory Medical Care Survey are surveys of medical visits, not individuals, the breadth of patient characteristics is limited. Information such as a history of oophorectomy and hysterectomy, which may influence the racial difference in HRT usage,4 was not collected; nor were menopause status, education, and income levels. It was also not possible to accurately measure overall health status or identify contraindicators for use. Because the unit of analysis was the medical care visit, the data could not be used to address the prevalence of HRT use because individual women may have had multiple visits involving HRT prescriptions in a year. Ignoring the issue of multiple visits, given that the compliance rate for HRT regimens is only 15–50%,20 estimates of HRT prescriptions probably overstate HRT usage in all women. In addition, source of payment and place of care (private office versus outpatient clinic) were used as proxy measures for socioeconomic status, and these are less adequate than directly measured factors. Race was determined by the physician or hospital staff, so race may have been misclassified more often than if it had been collected directly from the women. Finally, although physicians or staff were asked to record new or continuing HRT prescriptions in each visit, continuing prescriptions may have been underreported because of inaccessibility or inaccuracy of medical records or misunderstanding of instructions. Differential misclassification bias cannot be ruled out if under-reporting were more frequent in outpatient clinics than in physicians' offices because visits to outpatient clinics were more common among black women in the sample.

The finding that visits by black women compared with white women were more likely to include non-HRT prescriptions may indicate that black women were more likely to have medical problems that needed attention during physician visits or that white women were more likely to visit physicians for preventive care. If the former supposition is true, then the lower rate of HRT prescriptions for black women could have been due to more contraindications to HRT. If either supposition is true, physicians may have had more time to discuss and encourage the use of preventive therapies such as HRT with white women. However, prescriptions for non-HRT medications were positively associated with HRT prescriptions, indicating a need for further research into the racial differences in HRT use.

Another possible explanation for our findings is that patients with physicians of the same race are more likely to take HRT. It has been shown that patients of female physicians are more likely to take HRT than patients of male physicians,21 although in our study, physician sex was not significantly related to the outcome. This hypothesis could not be tested because physician race was not ascertained by the surveys used in our analysis. Another explanation for the findings is the possibility that black women have fewer menopausal symptoms than white women.

A study of attitudes among a group of low-income, perimenopausal black women8 found a relatively stoic attitude toward menopausal symptoms and a lack of knowledge about the health risks and benefits associated with menopause and HRT. That study also suggested that black women tend to obtain medical information from health care providers and other women. However, another study found that black women were uncomfortable talking to physicians about menopause, fearing that they would sound unintelligent or mentally impaired, and were dissatisfied with the discussions when they did raise the subject with their doctors.9 Given that women using HRT must discuss this medication with their physicians,22 it seems important for medical providers to be aware of differences in the ease with which women of different cultures and races talk about HRT and menopause and to help break down any barriers to proper treatment.

References

1. Brett KM, Madans JH. Use of postmenopausal hormone replacement therapy: Estimates from a nationally representative cohort study. Am J Epidemiol 1997;145:536–45.
2. Brett KM, Madans JH. Differences in use of postmenopausal hormone replacement therapy by black and white women. Menopause 1997;4:66–70.
3. Dawson DA, Thompson GB. Breast cancer risk factors and screening: United States, 1987. Vital Health Stat 10 1990;172:1–60.
4. Egeland GM, Matthews KA, Kuller LH, Kelsey SF. Characteristics of noncontraceptive hormone users. Prev Med 1988;17:403–11.
5. Matthews KA, Kuller LH, Wing RR, Meilahn EN, Plantinga P. Prior to use of estrogen replacement therapy, are users healthier than nonusers? Am J Epidemiol 1996;143:971–8.
6. Hill HA, Coates RJ, Austin H, Correa P, Robboy SJ, Chen V, et al. Racial differences in tumor grade among women with endometrial cancer. Gynecol Oncol 1995;56:154–63.
7. Liu JR, Conaway M, Rodriguez GC, Soper JT, Clarke-Pearson DL, Berchuck A. Relationship between race and interval to treatment in endometrial cancer. Obstet Gynecol 1995;86:486–90.
8. Homes-Rovner M, Padonu G, Kroll J, Breer L, Rovner DR, Talarczyk G, et al. African-American women's attitudes and expectations of menopause. Am J Prev Med 1996;12:420–3.
9. Padonu G, Homes-Rovner M, Rothert M, Schmitt N, Kroll J, Rovner D, et al. African-American women's perception of menopause. Am J Health Behav 1996;20:242–51.
10. Ferguson KJ, Hoegh C, Johnson S. Estrogen replacement therapy: A survey of women's knowledge and attitudes. Arch Intern Med 1989;149:133–6.
11. Woodwell DA. National Ambulatory Medical Care Survey: 1995 summary. Advance Data from Vital and Health Statistics no. 286. Hyattsville, Maryland: National Center for Health Statistics, 1997.
12. McCaig LF. National Hospital Ambulatory Medical Care Survey: 1995 outpatient department summary. Advance Data from Vital and Health Statistics no. 284. Hyattsville, Maryland: National Center for Health Statistics, 1997.
13. Shah BV, Barnwell BG, Bieler GS. SUDAAN user's manual: Software for analysis of correlated data, release 6.40. Research Triangle Park, North Carolina: Research Triangle Institute, 1995.
14. Food and Drug Administration. National drug code directory, 1985 edition. Washington, DC: Public Health Service, 1985.
15. Food and Drug Administration. National drug code directory, 1995 edition. Washington, DC: Public Health Service, 1995.
16. Schneider D, Appleton L, McLemore T. A reason for visit classification for ambulatory care. National Center for Health Statistics. Vital Health Stat 2 1979;78:1–63.
17. Kleinbaum DG, Kupper LL, Morgenstern H. Epidemiologic research. 1st ed. New York: Van Nostrand Reinhold, 1982.
18. Ross RK, Paganini-Hill A, Roy S, Chao A, Henderson BE. Past and present preferred prescribing practices of hormone replacement therapy among Los Angeles gynecologists: Possible implications for public health. Am J Public Health 1988;78:516–9.
19. Barlow DH, Brockie JA, Rees CM. Study of general practice consultations and menopausal problems. BMJ 1991;302:274–6.
20. Scalley EK, Henrich JB. An overview of estrogen replacement therapy in postmenopausal women. J Womens Health 1993;2:289–94.
21. Seto TB, Taira DA, Davis RB, Safran C, Phillips RS. Effect of physician gender on the prescription of estrogen replacement therapy. J Gen Intern med 1996;11:197–203.
22. McNagny SE, Jacobson TA. Use of postmenopausal hormone replacement therapy by African American women. The importance of physician discussion. Arch Intern Med 1997;157:1337–42.

Cited By

This article has been cited 2 time(s).

Obstetrics & Gynecology
Changes in U.S. Prescribing Patterns of Menopausal Hormone Therapy, 2001–2003
Hing, E; Brett, KM
Obstetrics & Gynecology, 108(1): 33-40.
10.1097/01.AOG.0000220502.77153.5a
PDF (250) | CrossRef
Obstetrics & Gynecology
Factors Associated With Undergoing Bilateral Salpingo-oophorectomy at the Time of Hysterectomy for Benign Conditions
Jacoby, VL; Vittinghoff, E; Nakagawa, S; Jackson, R; Richter, HE; Chan, J; Kuppermann, M
Obstetrics & Gynecology, 113(6): 1259-1267.
10.1097/AOG.0b013e3181a66c42
PDF (343) | CrossRef
© 1999 The American College of Obstetricians and Gynecologists