Introduction
Older adults with dementia are commonly prescribed central nervous system (CNS)–active medications, including opioids, sedatives, or antipsychotics.1 These medications are known to cause adverse events and are associated with increased all-cause mortality.2 Community-based prescribing has been characterized previously.1 We hypothesized that hospitalization is also a major contributor to initiation of these medications, as they are often used to treat acute pain or delirium. When continued on discharge, these medications can be particularly problematic owing to the already heightened risk of adverse events during this vulnerable transition.3 Thus, we aimed to estimate the rate of initiation of CNS-active medications and associated predictors of prescribing among older adults with dementia following hospital discharge home.
Methods
We conducted a retrospective cohort study using the 20% random sample of US Medicare fee-for-service beneficiaries with Part D coverage age 65 or older with a short-stay hospitalization in 2016 and a diagnosis of dementia, defined using the Medicare Chronic Condition (MCC) of “Alzheimer’s Disease, Related Disorders, or Senile Dementia.” We excluded beneficiaries with an MCC of “Schizophrenia and Other Psychotic Disorders,” and those admitted from or discharged to another facility, since only outpatient prescriptions are captured in Part D.
To estimate the incidence of initiation of any CNS-active medication, we identified beneficiaries with claims for opioids, sedatives, or antipsychotic medications1 within 7 days of hospital discharge among beneficiaries without claims for these medications during the 90 days prior to admission. We used a generalized estimating equation model with a binomial distribution and log link to estimate the relative risk of initiation, adjusting for patient, hospitalization, and hospital characteristics.
Analyses were conducted using SAS v9.4 (SAS Institute Inc., Cary, NC); tests were 2-sided with α = .05. This study was approved by the Beth Israel Deaconess Medical Center institutional review board; informed consent was waived.
Results
There were 1,419,688 adults age 65 and over with continuous traditional Medicare and Part D coverage, discharged alive in 2016. After excluding beneficiaries admitted from or discharged to a facility (n=942,247), those with a psychotic disorder (n=19,697), and those without a dementia diagnosis (n=385,810), a total of 71,934 hospitalizations met inclusion criteria; the mean age was 81.3 years, 57.8% were female, and 14.1% were non-Hispanic Black. Among the 38,222 hospitalizations without a prior claim for any CNS-active medication, 5902 (15.4%) were initiated on any CNS-active medication on hospital discharge (Table 1). Initiation of opioids was most common (8.7%), followed by sedatives (2.0%), and antipsychotics (1.3%) (Table 2). Patient characteristics associated with initiation of any CNS-active medication included age between 65 and 74, female sex, non-Hispanic White race/ethnicity, and presence of delirium; hospitalization factors included not having an intensive care unit stay and being admitted for a surgical diagnosis; hospital characteristics included non-teaching status, rurality, and geographic location. Results were similar across medication classes, except that delirium was associated with reduced likelihood of opioid initiation and markedly increased likelihood of antipsychotic initiation, and hospitalizations for a medical diagnosis had higher rates of sedative and antipsychotic initiation while hospitalizations for a surgical diagnosis had higher rates of opioid initiation.
Table 1.
Incidence and Adjusted Relative Risk of Initiation of CNS-Active Medications Within Seven Days of Hospital Discharge Among Those Without Prior Exposure to CNS-Active Medications (n=38,222)
| Any new CNS-active medication | ||
|---|---|---|
| n=38,222 adults without prior CNS-active medication | ||
| n (row %) | ARR (95% CI) | |
| Overall | 5902 (15.4) | |
| Patient characteristics | ||
| Age | ||
| 65–74 | 1579 (23.2) | 1 [reference] |
| 75–84 | 2558 (17.0) | 0.78 (0.74–0.82) |
| ≥ 85 | 1765 (10.8) | 0.56 (0.53–0.60) |
| Sex | ||
| Male | 2419 (15.0) | 1 [reference] |
| Female | 3483 (15.8) | 1.23 (1.17–1.29) |
| Race/ethnicity | ||
| Non-Hispanic White | 4755 (16.4) | 1 [reference] |
| Non-Hispanic Black | 623 (11.6) | 0.74 (0.68–0.80) |
| Hispanic | 220 (14.7) | 0.99 (0.88–1.12) |
| Other | 304 (12.8) | 0.83 (0.74–0.92) |
| Low-income subsidy eligible | ||
| No | 4234 (16.2) | 1 [reference] |
| Yes | 1668 (13.8) | 0.95 (0.90–1.01) |
| Presence of delirium* | ||
| No | 5337 (15.4) | 1 [reference] |
| Yes | 565 (15.5) | 1.26 (1.16–1.36) |
| Hospitalization factors | ||
| ICU stay | ||
| No | 4877 (16.0) | 1 [reference] |
| Yes | 1025 (13.3) | 0.83 (0.78–0.88) |
| DRG | ||
| Surgical | 2444 (39.1) | 1 [reference] |
| Medical | 3458 (10.8) | 0.29 (0.28–0.31) |
| Hospital characteristics | ||
| Teaching status | ||
| Non-teaching | 2984 (16.1) | 1 [reference] |
| Teaching | 2918 (14.8) | 0.91 (0.86–0.95) |
| Hospital ownership† | ||
| For profit | 4203 (15.1) | 1 [reference] |
| Nonprofit | 1075 (16.1) | 1.01 (0.95–1.07) |
| Government | 623 (16.5) | 1.01 (0.94–1.09) |
| Rurality† | ||
| Urban | 5286 (15.3) | 1 [reference] |
| Rural | 616 (16.5) | 1.13 (1.05–1.22) |
| US region† | ||
| Northeast | 1141 (12.0) | 1 [reference] |
| Midwest | 1178 (15.7) | 1.18 (1.10–1.27) |
| South | 2579 (17.1) | 1.27 (1.19–1.36) |
| West | 992 (16.5) | 1.29 (1.19–1.39) |
| Other | 12 (12.4) | 0.89 (0.55–1.46) |
Abbreviations: ARR, adjusted relative risk; ICU, intensive care unit; DRG, diagnosis-related group
*Delirium defined using diagnosis codes specified by Dae et al.6
†Numbers of missing data: hospital ownership (n=3 hospitals), rurality (n=4 hospitals), and US region (n=1 hospital)
Table 2.
Incidence and Adjusted Relative Risk of Initiation of Opioids, Sedatives, and Antipsychotics Among Those Without Prior Exposure to the Same Medication Class
| Opioid Initiation† | Sedative Initiation‡ | Antipsychotic Initiation§ | ||||
|---|---|---|---|---|---|---|
| n=47,922 opioid-naive adults | n=54,358 sedative-naive adults | n=66,217 antipsychotic-naive adults | ||||
| n (row %) | ARR (95% CI) | n (row %) | ARR (95% CI) | n (row %) | ARR (95% CI) | |
| Overall* | 4148 (8.7) | 1096 (2.0) | 877 (1.3) | |||
| Patient characteristics | ||||||
| Age | ||||||
| 65–74 | 1195 (17.6) | 1 [reference] | 269 (4.0) | 1 [reference] | 182 (2.7) | 1 [reference] |
| 75–84 | 1825 (12.2) | 0.78 (0.74–0.83) | 458 (3.0) | 0.72 (0.62–0.83) | 367 (2.4) | 0.85 (0.71–1.01) |
| ≥ 85 | 1128 (6.9) | 0.55 (0.51–0.59) | 369 (2.2) | 0.50 (0.42–0.58) | 328 (2.0) | 0.63 (0.52–0.76) |
| Sex | ||||||
| Male | 1746 (10.8) | 1 [reference] | 390 (2.4) | 1 [reference] | 372 (2.3) | 1 [reference] |
| Female | 2402 (10.9) | 1.22 (1.16–1.29) | 706 (3.2) | 1.40 (1.24–1.59) | 505 (2.3) | 0.98 (0.86–1.12) |
| Race/ethnicity | ||||||
| Non-Hispanic White | 3321 (11.5) | 1 [reference] | 894 (3.1) | 1 [reference] | 718 (2.5) | 1 [reference] |
| Non-Hispanic Black | 442 (8.2) | 0.80 (0.73–0.88) | 103 (1.9) | 0.58 (0.47–0.72) | 98 (1.8) | 0.67 (0.54–0.83) |
| Hispanic | 163 (10.9) | 1.07 (0.93–1.23) | 46 (3.1) | 1.05 (0.77–1.42) | 20 (1.3) | 0.59 (0.38–0.94) |
| Other | 222 (9.3) | 0.89 (0.78–1.003) | 53 (2.2) | 0.72 (0.54–0.96) | 41 (1.7) | 0.72 (0.52–1.003) |
| Low-income subsidy eligible | ||||||
| No | 2996 (11.4) | 1 [reference] | 770 (2.9) | 1 [reference] | 618 (2.4) | 1 [reference] |
| Yes | 1152 (9.6) | 0.94 (0.88–1.004) | 326 (2.7) | 0.91 (0.79–1.05) | 259 (2.2) | 0.98 (0.84–1.15) |
| Presence of delirium† | ||||||
| No | 3946 (11.4) | 1 [reference] | 973 (2.8) | 1 [reference] | 596 (1.7) | 1 [reference] |
| Yes | 202 (5.6) | 0.68 (0.6–0.78) | 123 (3.4) | 1.21 (1.01–1.46) | 281 (7.7) | 4.41 (3.84–5.07) |
| Hospitalization factors | ||||||
| ICU stay | ||||||
| No | 3425 (11.2) | 1 [reference] | 875 (2.9) | 1 [reference] | 750 (2.5) | 1 [reference] |
| Yes | 723 (9.4) | 0.83 (0.77–0.89) | 221 (2.9) | 0.99 (0.85–1.15) | 127 (1.6) | 0.68 (0.57–0.83) |
| DRG | ||||||
| Surgical | 2314 (37.0) | 1 [reference] | 138 (2.2) | 1 [reference] | 55 (0.9) | 1 [reference] |
| Medical | 1834 (5.7) | 0.17 (0.16–0.18) | 958 (3.0) | 1.45 (1.21–1.73) | 822 (2.6) | 2.61 (1.98–3.45) |
| Hospital characteristics | ||||||
| Teaching status | ||||||
| Non-teaching | 1982 (10.7) | 1 [reference] | 629 (3.4) | 1 [reference] | 500 (2.7) | 1 [reference] |
| Teaching | 2166 (11.0) | 0.96 (0.91–1.01) | 467 (2.4) | 0.79 (0.69–0.89) | 377 (1.9) | 0.75 (0.65–0.86) |
| Hospital ownership‖ | ||||||
| For profit | 2972 (10.7) | 1 [reference] | 755 (2.7) | 1 [reference] | 620 (2.2) | 1 [reference] |
| Nonprofit | 745 (11.2) | 0.98 (0.91–1.05) | 221 (3.3) | 1.11 (0.95–1.30) | 157 (2.4) | 1.11 (0.93–1.32) |
| Government | 430 (11.4) | 1.01 (0.93–1.10) | 120 (3.2) | 1.04 (0.85–1.26) | 100 (2.7) | 1.11 (0.89–1.37) |
| Rurality‖ | ||||||
| Urban | 3771 (10.9) | 1 [reference] | 951 (2.8) | 1 [reference] | 752 (2.2) | 1 [reference] |
| Rural | 377 (10.1) | 1.06 (0.97–1.16) | 145 (3.9) | 1.20 (1.003–1.44) | 125 (3.4) | 1.24 (1.03–1.51) |
| US region‖ | ||||||
| Northeast | 762 (8.0) | 1 [reference] | 214 (2.3) | 1 [reference] | 210 (2.2) | 1 [reference] |
| Midwest | 842 (11.2) | 1.22 (1.12–1.33) | 205 (2.7) | 1.13 (0.93–1.36) | 171 (2.3) | 0.97 (0.79–1.19) |
| South | 1807 (12.0) | 1.28 (1.18–1.39) | 482 (3.2) | 1.24 (1.05–1.48) | 373 (2.5) | 1.00 (0.84–1.20) |
| West | 729 (12.2) | 1.36 (1.25–1.49) | 191 (3.2) | 1.30 (1.06–1.59) | 122 (2.0) | 0.83 (0.65–1.04) |
| Other | -¶ | 0.89 (0.47–1.65) | -¶ | 1.53 (0.58–4.03) | -¶ | 0.44 (0.06–3.14) |
Abbreviations: ARR adjusted relative risk, ICU intensive care unit, DRG diagnosis-related group
*Incidence of initiation of individual drug classes sum to greater than the incidence of initiation of any CNS-active medication owing to different sample sizes in each analysis, based on exclusion of patients with pre-admission exposure to the corresponding medication class
†Based on Centers for Disease Control and Prevention definition
‡Includes the following medication classes: barbiturates, benzodiazepines, non-benzodiazepine hypnotics, and first generation anti-histamines
§Includes the following medication classes: atypical antipsychotics and first generation antipsychotics including butyrophenones, phenothiazines, and thioxanthenes
‖Numbers of missing data: hospital ownership (n=3 hospitals), rurality (n=4 hospitals), and US region (n=1 hospital)
¶Cell suppressed owing to small cell size, in accordance with CMS policy
Discussion
Prescription of new CNS-active medications occurred in one in six hospitalized older adults with dementia being discharged to home. Our finding of higher rates of prescribing among younger, non-Hispanic White, female patients is consistent with prior studies of outpatient use.1 Delirium was associated with a fourfold higher rate of new antipsychotic prescribing, suggesting a need to examine why these medications are continued on discharge for a generally acute, reversible inpatient diagnosis.
Our findings may not be generalizable to older adults with non-Medicare insurance, or those admitted from or discharged to a facility. Since patients with more severe dementia/delirium may be more likely to be prescribed antipsychotic medications and/or be discharged to a facility, this may have resulted in underestimation of the rate of initiation of antipsychotics. Additionally, our data is from 2016 and prescribing patterns may have subsequently shifted.
This study highlights the substantial contribution of hospitalization to new initiation of CNS-active medication in older adults with dementia. This is concerning since patients with dementia are often hospitalized, and these medications contribute to adverse events in this vulnerable population.2,4,5 Further studies are necessary to determine duration and appropriateness of prescribing and to understand factors driving the observed prescribing.
Funding
The study was funded by grant number R01HS026216 from the Agency for Healthcare Research and Quality.
Declarations
Conflict of Interest
Dr. Timothy Anderson is a paid consultant of Alosa Health, unrelated to the present work.
Footnotes
Prior Presentations
None
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Maust DT, Strominger J, Bynam JP. Prevalence of psychotropic and opioid prescription fills among community-dwelling older adults with dementia in the US. JAMA. 2020;324(7):706–708. doi: 10.1001/jama.2020.8519. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.2019 American Geriatrics Society Beers Criteria Update Expert Panel American Geriatrics Society 2019 updated AGS Beers Criteria for potentially inappropriate medication use in older adults. J Am Geriatr Soc. 2019;67(4):674–694. doi: 10.1111/jgs.15767. [DOI] [PubMed] [Google Scholar]
- 3.Forster AJ, Murff HJ, Peterson JF, Gandhi TK, Bates DW. The incidence and severity of adverse events affecting patients after discharge from the hospital. Ann Intern Med. 2003;138(3):161–167. doi: 10.7326/0003-4819-138-3-200302040-00007. [DOI] [PubMed] [Google Scholar]
- 4.Phelan EA, Borson S, Grothaus L, Balch S, Larson EB. Association between incident dementia and risk of hospitalization. JAMA. 2012;307(2):165–172. doi: 10.1001/jama.2011.1964. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Growdon ME, Gan S, Yaffe K, Steinman M. Polypharmacy among older adults with dementia compared with those without dementia in the United States. J Am Geriatr Soc. 2021;69(9):2464–2475. doi: 10.1111/jgs.17291. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Kim DH, Lee J, Kim CA, Huybrechts KF, Bateman BT, Patorno E, et al. Evaluation of algorithms to identify delirium in administrative claims and drug utilization database. Pharmacoepidemiol Drug Saf. 2017;26(8):945–953. doi: 10.1002/pds.4226. [DOI] [PMC free article] [PubMed] [Google Scholar]