Abstract
This study uses data from the National Ambulatory Medical Care Survey to describe the central nervous system polypharmacy practices among outpatient older adults.
With each new revision of the Beers Criteria, the list of psychotropic medications considered potentially inappropriate in the elderly has grown. Opioids have recently been included in a Beers measure of central nervous system (CNS) polypharmacy. Prescribing related drug combinations also received increased regulatory attention when the US Food and Drug Administration recently ordered a black-box warning to alert patients of serious risks, including death, caused by opioids coprescribed with CNS depressants. While evidence builds concerning harms of CNS polypharmacy, little is known about the trends in relevant prescribing practices.
Methods
This analysis used data from the 2004 through 2013 National Ambulatory Medical Care Survey (NAMCS), an annual survey of office-based physicians. We limited our analysis to patients aged 65 years or older (n = 97 910). Because this research is on publicly available, deidentified data, the University of Michigan Medical School institutional review board did not require approval. An outpatient visit met Beers CNS polypharmacy criteria if 3 or more of the following medications were initiated or continued: antipsychotics, benzodiazepines, nonbenzodiazepine benzodiazepine receptor agonists, tricyclic antidepressants, selective serotonin reuptake inhibitors, and opioids. We recorded up to 3 visit diagnoses and included NAMCS-collected information such as chronic medical conditions, whether psychotherapy was provided or ordered, whether stress management or other mental health counseling services were provided or ordered, and time spent with physician.
We used logistic regression to assess time trends in polypharmacy. For the most recent period (2011-2013), difference-in-proportion tests were used to compare visit characteristics by polypharmacy type. Analyses were conducted in Stata, version 13.1 (StataCorp LLC), using 2-sided tests (α = .05).
Data were collected from January 2004 to December 2013. Data analysis took place from October 7, 2016, to October 28, 2016.
Results
Between 2004 and 2013, annual polypharmacy visits by adults 65 years or older increased from 1.50 million (95% CI, 1.12-1.87 million) to 3.68 million (95% CI, 3.23-4.13 million), rising from 0.6% to 1.4% of visits (adjusted odds ratio [AOR], 3.12; 95% CI, 2.28-4.28; P < .001) (Table 1). Among demographic groups, the largest increase in CNS polypharmacy was among rural visits (AOR, 4.99; 95% CI, 2.67-9.33; P < .001). Additionally, CNS polypharmacy increased among visits with no mental health or pain diagnoses (AOR, 2.65; 95% CI, 1.65-4.27; P < .001).
Table 1. Central Nervous System Polypharmacy at Office-Based Physician Visits by Older Adults Overall and by Demographic and Clinical Subpopulation, 2004-2013a.
| Characteristic | CNS Polypharmacy Visits per 100 Visitsb | AOR (95% CI)c | ||
|---|---|---|---|---|
| 2004-2006 | 2007-2010 | 2011-2013 | ||
| Overall (n = 97 910) | 0.6 | 1.0 | 1.4 | 3.12 (2.28-4.28) |
| Demographic Characteristics | ||||
| Age, y | ||||
| 65-74 | 0.8 | 1.0 | 1.4 | 2.40 (1.59-3.63) |
| 75-84 | 0.5 | 0.8 | 1.5 | 4.28 (2.44-7.51) |
| ≥85 | 0.4 | 1.1 | 1.5 | 4.15 (2.04-8.43) |
| Sex | ||||
| Male | 0.4 | 0.7 | 1.1 | 3.10 (1.73-5.57) |
| Female | 0.8 | 1.1 | 1.7 | 3.15 (2.25-4.40) |
| Race/ethnicity | ||||
| Non-Hispanic white | 0.6 | 1.0 | 1.5 | 3.23 (2.33-4.46) |
| Non-Hispanic African American | 0.5 | 0.7 | 0.9 | 1.74 (0.51-5.99) |
| Hispanic | 0.7 | 1.0 | 1.7 | 3.28 (0.93-11.59) |
| Dual eligible | ||||
| Yes | 0.8 | 0.5 | 0.7 | 0.46 (0.03-6.63) |
| No | 0.6 | 1.0 | 1.5 | 3.21 (2.37-4.34) |
| Geography | ||||
| Urban | 0.6 | 0.9 | 1.4 | 2.91 (2.03-4.17) |
| Rural | 0.7 | 1.0 | 2.2 | 4.99 (2.67-9.33) |
| Clinical Characteristicsd | ||||
| Anxiety | 8.9 | 4.4 | 5.4 | 0.65 (0.21-1.96) |
| No anxiety | 0.6 | 0.9 | 1.4 | 3.31 (2.39-4.60) |
| Insomnia | 7.6 | 3.9 | 5.3 | 0.78 (0.16-3.93) |
| No insomnia | 0.6 | 0.9 | 1.4 | 3.18 (2.31-4.37) |
| Depression | 6.7 | 6.7 | 10.4 | 1.31 (0.66-2.60) |
| No depression | 0.5 | 0.8 | 1.3 | 3.24 (2.28-4.60) |
| Dementia | 1.2 | 2.2 | 2.4 | 2.14 (0.55-8.27) |
| No dementia | 0.6 | 0.9 | 1.4 | 3.11 (2.26-4.29) |
| Substance use disorder | 0.5 | 3.3 | 2.3 | 2.63 (0.43-16.01) |
| No substance use disorder | 0.6 | 0.9 | 1.4 | 3.13 (2.28-4.30) |
| Pain | 0.9 | 1.8 | 2.8 | 4.54 (2.59-7.97) |
| No pain | 0.6 | 0.8 | 1.2 | 2.64 (1.82-3.83) |
| Any mental health or pain dx | 1.5 | 2.3 | 3.6 | 3.34 (2.25-4.94) |
| No mental health or pain dx | 0.4 | 0.6 | 0.9 | 2.65 (1.65-4.27) |
| Visit Characteristics | ||||
| Physician type | ||||
| Family practice | 0.9 | 0.9 | 2.3 | 3.92 (2.04-7.55) |
| Internal medicine | 0.6 | 1.3 | 1.4 | 2.48 (1.31-4.72) |
| Psychiatry | 7.7 | 8.9 | 8.2 | 0.92 (0.40-2.14) |
| Other medical specialty | 0.4 | 0.7 | 1.0 | 3.72 (2.14-6.46) |
Abbreviations: AOR, adjusted odds ratio; CNS, central nervous system; dx, diagnosis.
Analyses were completed using survey design elements for visit weight, clustering, and stratification to allow national inferences.
The term CNS polypharmacy encompasses visits where patients are prescribed 3 or more medications from any of the following medication groups included in the Beers CNS polypharmacy measure: antipsychotics, benzodiazepines, nonbenzodiazepine benzodiazepine receptor agonist hypnotics, tricyclic antidepressants, selective serotonin reuptake inhibitors, and opioids. The 2012 and 2013 surveys included up to 10 medications, but this analysis used the first 8 listed to be consistent across all years.
The AOR is associated with the transformed survey year variable ([survey year – 2004] / 9) and therefore estimates the change in odds of a visit with CNS polypharmacy relative to a visit where none of the CNS-active medications were prescribed over the entire 2004-2013 study period. Logistic regression models were adjusted for age, gender, and race/ethnicity. For example, CNS polypharmacy occurred at 0.8% of encounters by women from 2004-2006 and 1.7% from 2011-2013, with an AOR of 3.15 for CNS polypharmacy visits by women in 2013 compared with 2004.
A given visit could include more than 1 diagnosis of interest; categories were not mutually exclusive.
Women and rural patients accounted for 68.1% and 16.6% of CNS polypharmacy visits, compared with 55% and 10.5% of nonpolypharmacy visits (P < .001 for both comparisons) (Table 2). While mental health or pain diagnoses were more common among the polypharmacy group, 45.9% of the polypharmacy visits included neither mental health nor pain diagnoses. A small minority of polypharmacy visits included psychotherapy (5.3%) or stress management (6.7%). There were no significant demographic differences between polypharmacy visits with and without opioids. Polypharmacy encounters without opioids were more likely to include an anxiety or a depression diagnosis but were less likely to include a pain diagnosis (15.3% vs 38.5%; P = .01).
Table 2. Demographic and Clinical Characteristics of Office-Based Physician Visits by Older Adults With and Without CNS Polypharmacy, With and Without Opioids, 2011-2013a.
| Characteristic | Visits, % | CNS Polypharmacy Visits, % | ||||
|---|---|---|---|---|---|---|
| CNS Polypharmacy (n = 618) |
No CNS Polypharmacyb (n = 36 225) | P Value | With Opioid (n = 465) | Without Opioid (n = 153) | P Value | |
| Demographic Characteristics | ||||||
| Age, y | ||||||
| 65-74 | 51.3 | 52.1 | .93 | 49.6 | 56.7 | .60 |
| 75-84 | 35.0 | 34.8 | 35.5 | 33.3 | ||
| ≥85 | 13.8 | 13.2 | 15.0 | 10.0 | ||
| Sex | ||||||
| Male | 31.9 | 45.1 | <.001 | 32.7 | 29.3 | .57 |
| Female | 68.1 | 55.0 | 67.3 | 70.7 | ||
| Ethnicity | ||||||
| Non-Hispanic white | 86.1 | 84.9 | .19 | 85.6 | 87.6 | .40 |
| Non-Hispanic African American | 4.6 | 7.4 | 5.4 | 1.8 | ||
| Hispanic | 9.4 | 7.7 | 9.0 | 10.6 | ||
| Medicaid | 0.8 | 1.5 | .13 | 0.5 | 1.7 | .13 |
| Rural | 16.6 | 10.5 | <.001 | 16.6 | 16.4 | .98 |
| Clinical Characteristicsc | ||||||
| Diagnosis | ||||||
| Anxiety | 5.8 | 0.6 | <.001 | 10.8 | 30.6 | <.001 |
| Insomnia | 2.7 | 0.4 | <.001 | 4.0 | 11.3 | .03 |
| Depression | 15.5 | 0.9 | <.001 | 0.6 | 9.6 | <.001 |
| Dementia | 1.8 | 0.8 | .03 | 1.9 | 1.5 | .79 |
| Substance use disorder | 0.5 | 0.3 | .29 | 0.6 | 0.0 | .35 |
| Pain | 33.0 | 14.7 | <.001 | 38.5 | 15.3 | .01 |
| No mental health or pain | 45.9 | 82.6 | <.001 | 48.4 | 37.7 | .12 |
| Other medical condition | ||||||
| Asthma/COPD | 10.8 | 11.7 | .64 | 11.3 | 9.4 | .65 |
| CAD/CVD | 10.3 | 13.0 | .15 | 11.3 | 7.0 | .26 |
| HTN/HL | 63.4 | 59.9 | .22 | 68.2 | 47.9 | <.01 |
| CHF | 6.6 | 4.8 | .18 | 7.9 | 2.4 | .17 |
| Obesity | 8.9 | 6.4 | .07 | 10.0 | 5.6 | .19 |
| Osteoporosis | 13.3 | 7.2 | .002 | 12.3 | 16.5 | .50 |
| Total chronic conditions, mean (SE), n | 2.41 (0.13) | 1.72 (0.05) | <.001 | 2.50 (0.15) | 2.13 (0.24) | .18 |
| Total medications, mean (SE), n | 6.90 (0.10) | 3.13 (0.07) | <.001 | 7.08 (0.10) | 6.30 (0.26) | <.001 |
| Visit Characteristics | ||||||
| Physician type | ||||||
| Family practice | 33.7 | 19.4 | <.001 | 37.9 | 20.0 | <.001 |
| Internal medicine | 18.9 | 18.3 | 20.0 | 15.1 | ||
| Psychiatry | 8.8 | 0.6 | 0.9 | 34.2 | ||
| Other medical specialty | 38.7 | 61.7 | 41.2 | 30.7 | ||
| Psychotherapy | 5.3 | 0.3 | <.001 | 0.9 | 19.4 | <.001 |
| Stress management or health education | 6.7 | 1.0 | <.001 | 3.3 | 17.6 | <.001 |
| Time with physician, mean (SE), min | 23.09 (0.71) | 21.94 (0.24) | .12 | 22.61 (0.73) | 24.64 (1.74) | .28 |
Abbreviations: CAD, coronary artery disease; CHF, congestive heart failure; CNS, central nervous system; COPD, chronic obstructive pulmonary disease; CVD, cerebrovascular disease; HL, hyperlipidemia; HTN, hypertension.
Analyses were completed using survey design elements for visit weight, clustering, and stratification to allow national inferences.
The term No CNS polypharmacy encompasses visits where patients are prescribed no medications from any of the following medication groups included in the Beers CNS polypharmacy measure: antipsychotics, benzodiazepines, nonbenzodiazepine benzodiazepine receptor agonist hypnotics, tricyclic antidepressants, selective serotonin reuptake inhibitors, or opioids.
A given visit could include more than 1 diagnosis of interest; categories were not mutually exclusive.
Discussion
From 2004 to 2013, CNS polypharmacy more than doubled. While it was most common at visits with anxiety, insomnia, or depression, there was not a significant increase at such visits. By contrast, polypharmacy significantly increased for patients with a pain diagnosis, occurring in the context of the overall growth in opioid prescribing. Visits without pain, insomnia, or other mental health diagnoses accounted for nearly half (45.9%) of CNS polypharmacy visits and grew significantly from 2004 to 2013.
Older adults have become more open to mental health treatment. Because of limited access to specialty care and a preference to receive treatment in primary care settings, it is unsurprising that mental health treatment has expanded in nonpsychiatric settings. The growth in polypharmacy in rural settings, where access to specialty mental health or pain care is particularly limited, is part of this broader trend.
This study has several limitations. First, NAMCS does not account for whether a medication is prescribed as needed, so regular use may have been overestimated. Second, NAMCS does not capture outcomes nor include nonphysicians. Nonresponse might introduce bias, but survey weights account for nonresponse to produce unbiased estimates. Finally, our visit diagnoses were limited to 3, so prescribing without a diagnosis may have been overestimated. However, psychotropic use without a diagnosis has been described in other studies and thus is unlikely an artifact of NAMCS.
References
- 1.American Geriatrics Society 2015 Beers Criteria Update Expert Panel American Geriatrics Society 2015 updated Beers Criteria for potentially inappropriate medication use in older adults. J Am Geriatr Soc. 2015;63(11):2227-2246. [DOI] [PubMed] [Google Scholar]
- 2.National Center for Health Statistics, Centers for Disease Control and Prevention National Ambulatory Medical Care Survey: 2013 NAMCS micro-data file documentation. ftp://ftp.cdc.gov/pub/Health_Statistics/NCHS/Dataset_Documentation/NAMCS/doc2013.pdf. Accessed January 17, 2017.
- 3.Currin JB, Hayslip B Jr, Schneider LJ, Kooken RA. Cohort differences in attitudes toward mental health services among older persons. Psychotherapy. 1998;35(4):506-518. [Google Scholar]
- 4.Areán PA, Hegel MT, Reynolds CF III. Treating depression in older medical patients with psychotherapy. J Clin Geropsychology. 2001;7(2):93-104. [Google Scholar]
- 5.Bishop TF, Seirup JK, Pincus HA, Ross JS. Population of US practicing psychiatrists declined, 2003-13, which may help explain poor access to mental health care. Health Aff (Millwood). 2016;35(7):1271-1277. [DOI] [PubMed] [Google Scholar]
- 6.Takayanagi Y, Spira AP, Bienvenu OJ, et al. Antidepressant use and lifetime history of mental disorders in a community sample: results from the Baltimore Epidemiologic Catchment Area Study. J Clin Psychiatry. 2015;76(1):40-44. [DOI] [PMC free article] [PubMed] [Google Scholar]