Prescription of estrogen therapy and sedating medications in older women in a national ambulatory care sample
Year: 2024
Type: article
Study design: Observational Study
Abstract: Since 2003, systemic estrogen has been included in the American Geriatric Society's Beers Criteria as a potentially inappropriate medication (PIM) for women aged 65 and older due to concerns about its potential adverse effects,1 including increased risk for cerebrovascular disease, cognitive impairment, and dementia.2, 3 However, clinicians and patients who are inclined toward long-term estrogen therapy (ET) may have less appreciation for the impact of medication use in older age on cognitive risk. Hypothesizing that ET might be a marker for the use of other PIMs associated with increased dementia risk in older women,4, 5 we examined trends in ET use from a national sample of ambulatory care visits to evaluate co-prescription of ET and sedating PIMs and also to assess whether ET was a marker for polypharmacy in women aged 65 and older. The National Ambulatory Medical Care Survey (NAMCS) incorporates multistage probability sampling to obtain a representative sample of ambulatory care data across the United States.6 We merged NAMCS datasets from 2006 to 2016, which included all comparably coded medication data available as of this report, to examine ET prescribing in visits of women 65 and older. Visits were categorized as involving either (a) estrogen-plus-progestin therapy, (b) estrogen-only therapy, or (c) no ET. Medication codes specific to vaginal-only estrogen were excluded. We examined the proportion of visits involving either form of ET and compared patient demographic characteristics as well as selected chronic conditions associated with increased dementia risk among visits stratified by ET prescription status. We inspected co-prescribing of sedating PIMs, including first-generation antihistamines, antispasmodics, benzodiazepines, non-benzodiazepine receptor agonists, opioids, and skeletal muscle relaxants.4, 5 Multivariable logistic regression models assessed whether ET (estrogen only, estrogen plus progestin, or either) was associated with co-prescription of (a) any sedating PIM, and (b) two or more sedating PIMs, compared to no ET as the reference group. To assess whether ET was associated with high levels of co-prescribing beyond sedating PIMs, additional models assessed whether ET was associated with general polypharmacy defined as five or more medications, and with super-polypharmacy defined as 10 or more medications. All models adjusted for patient demographic characteristics including age, race, Hispanic ethnicity, geographic region, and health insurance status. Statistical weights and survey design features were used in analyses using SAS version 9.4. From 2006 to 2016, ET was documented in 2084 of 62,322 visit records representative of an estimated 55.6 million of 1.6 billion visits involving women aged 65 and older. Prevalence of ET prescribing in visits involving older women varied across study years from 2.6% to 5.2% (p < 0.001 for heterogeneity). The prevalence of estrogen prescribing in ambulatory visits varied modestly by patient age, ethnicity, and geographic region (Table 1). Average age was lower in visits involving estrogen, (70.9 years for estrogen plus progestin vs. 75.6 for no estrogen, p < 0.01), and estrogen prescribing was slightly less prevalent in visits with Hispanic patients (3.7% in non-Hispanic patients vs. 1.6% in Hispanic patients, p < 0.01) and more prevalent in those living in western U.S. regions (4.7% in western regions vs. 1.7% in northeastern regions, p < 0.01). Sedating PIMs were co-prescribed in 23.4% of visits involving no ET, 22.8% involving estrogen plus progestin, and 36.6% involving estrogen-only therapy. In multivariable models, ET was positively associated with the use of any sedating PIM (adjusted odds ratio [AOR] 1.6, 95% confidence interval [CI] 1.4–1.9) as well as the use of two or more sedating PIMs (AOR 2.1, 95% CI 1.6–2.7, Figure 1). Associations were primarily driven by estrogen-only therapy rather than estrogen-plus-progestin therapy. Polypharmacy (5 or more medications) was detected in 39.8% of visits not involving ET, 66.5% of visits involving estrogen plus progestin, and 76.9% involving estrogen-only therapy; super-polypharmacy (assessed in years 2012–2016 only) was detected in 23.6% of visits not involving ET, 30.8% of visits involving estrogen-plus-progestin, and 52.2% involving estrogen-only therapy. ET was strongly associated with polypharmacy (AOR 4.6, 95% CI 3.9–5.5) and with super-polypharmacy (AOR 3.1, 95% CI 2.4–4.0). Among ambulatory visits in U.S. women aged 65 and older, ET was associated with over two-fold greater odds of prescription of two or more sedating PIMs. These findings from a nationally representative database suggest ET of any type comprises a marker of high-risk prescribing in older women, particularly of sedating PIMs associated with increased risk of cognitive impairment. Future research should investigate potential shared factors underlying systemic ET and sedating PIM utilization in older women, including symptom clustering, differences in access to healthcare, or preferences for overmedicalization. Reducing prescription rates of these medications could reduce harm to individual women and lower the associated burden on healthcare systems and society. Mary M. Hunter, Nadra E. Lisha, and Alison J. Huang contributed to the conception and design of the study, acquisition of data, analysis, interpretation of data, drafting and revising manuscript, and final approval. This research was supported by National Institute on Aging grant K24AG068601. The authors report no conflicts of interest. None.
DOI: 10.1111/jgs.18844
Language: English
Subfield: Geriatrics and Gerontology
Field: Medicine
Domain: Health Sciences
Open Access status: hybrid
Estimated APC paid: $4,760
Funder: National Institute on Aging
OpenAlex ID: W4392166342