Abstract
Purpose:
Because of toxicities, benzodiazepines are not usually recommended in older adults. We therefore sought to describe the trends in benzodiazepine use in long-term care and examine the variation in benzodiazepine use among nursing homes.
Methods:
In this retrospective repeated cross-sectional analysis of Medicare Parts A, B and D claims data linked to the Minimum Data Set from 2013 to 2018, we included long-term residents who stayed in a nursing home for at least one entire quarter of a calendar year in 2013-2018. The outcome was whether residents were prescribed a benzodiazepine drug for at least 30 days during each quarter stay. We use mixed effects logistic regression models to assess the variation in benzodiazepine use among nursing homes, adjusting for patient and nursing home characteristics.
Results:
The cohort for the time trend analysis included 270,566 unique residents and 1,843,580 quarter stays for 2013-2018. Prescribing rates for short-acting benzodiazepines were stable over 2013-2016, then declined from 12.1% in 2016 to 10.4% in 2018. The rate of long-acting benzodiazepine use remained relatively steady at around 4% over 2013-2018. During 2017-2018, the variation among nursing homes in benzodiazepine use was 7.2% for short-acting vs. 9.3% for long-acting benzodiazepines, after controlling for resident characteristics.
Conclusion:
Prescribing for short-acting benzodiazepines in long-term care declined after 2016, while long-acting benzodiazepine use did not change. The variation in benzodiazepine use among nursing homes is substantial. Identifying factors that explain this variation may help in developing strategies for deprescribing benzodiazepines in nursing home residents.
Keywords: Nursing home, benzodiazepine, long term care, Medicare
1. INTRODUCTION
Benzodiazepine drugs are widely used for insomnia and anxiety. Benzodiazepine use is associated with falls [1], fractures [2], and cognitive impairment [3] in older adults. Most benzodiazepine prescriptions are for relatively short-acting medications, drugs with half-lives of 24 hours or less [4]. Earlier findings [5] and recommendations from Beers criteria [6,7] suggested that long-acting benzodiazepine use was associated with increased risk of hip fracture relative to short- and intermediate-acting benzodiazepines. However, subsequent studies have found that the risk is similarly elevated for short-acting benzodiazepines [2, 8]. Therefore, the American Geriatric Society’s Beers Criteria considers benzodiazepines as potentially inappropriate medication for older adults, regardless of half-life [9]. While benzodiazepines are widely prescribed to older adults [10], Medicare coverage for prescription drugs (Part D) did not initially cover benzodiazepine prescriptions. In response to concerns that the exclusion caused a financial burden to beneficiaries [11, 12], Medicare Part D added benzodiazepine coverage in 2013 [13].
The prescribing of benzodiazepines during ambulatory care visits increased from 3.8% in 2003 to 7.5% in 2015 [14]. Among long-term care nursing home residents, benzodiazepine use was around 20% in 2013 to 2014 [15]. However, recent data on the use of benzodiazepine among nursing home residents are lacking.
Nursing home residents are vulnerable to adverse effects of benzodiazepines. The rate of falls- and fractures-related injury is higher among nursing home residents compared to community-dwelling older adults, presumable because of higher risk factors such as poor cognition and physical function [16]. Several studies have found that prescribing patterns for many medications vary considerably among nursing homes [17–19]. Whether such variation exists in benzodiazepine drug use and what factors explain such variation is unknown.
We describe the trends of short- and long-acting benzodiazepine use in long-term care and examine the variation in benzodiazepine use among nursing homes. We also examine the association of resident and nursing home characteristics associated with short- and long-acting benzodiazepine use. We used Medicare Parts A, B and D claims data linked to the Minimum Data Set (MDS) 2013–2018 to describe trends, and 2016-2018 data for variation analysis.
2. METHODS
2.1. Study Design and Data Sources
This cross-sectional analysis used Medicare beneficiary summary, Medicare Provider Analysis and Review, Outpatient, and Carrier files for 2013-2018. Data from the claims files were linked to the MDS v3.0 for 2013-2018. Data originated from a 20% random sample of fee-for-service beneficiaries enrolled in Medicare Parts A, B and D. Medicare is the primary federal health insurance program in the US for people aged 65 years or older. Medicare Part A covers hospital and skilled nursing facility stays, Part B covers physician’s services and Part D covers prescription drugs. The MDS was used to identify residents of nursing homes. Medicare Part D prescription data provided the National Drug Code (NDC) identifier, drug name, date of dispensing and estimated days’ supply for all prescriptions covered by Part D. Medicare Provider Analysis and Review, outpatient and carrier claim files were used to identify diagnoses. We used the Centers for Medicare and Medicaid Services (CMS) Provider of Services file to capture nursing home characteristics.
2.2. Cohort Derivation
We used one cohort from 2013-2018 to derive the time trends of benzodiazepine use in long-term care and a cohort from recent years (2017-2018) to examine variation and identify resident and facility factors associated with receiving at least a 30-day supply of benzodiazepines.
For trend analysis, the study cohort included residents who stayed in a nursing home for at least one entire quarter of a calendar year. During 2013-18, residents were included if 1) aged 65 years or older, 2) eligible for Medicare Parts A and B and without health maintenance organization (HMO) insurance 6 months prior to and during their quarter stay to ensure they are fully enrolled in Medicare fee-for-service plan and 3) eligible for Medicare Part D in the 3 months prior to and during their quarter stay to capture their prescription drug use. Residents could contribute information in more than one quarter but each resident is included only once in each quarter.
For variation analysis, the study cohort included long-term care nursing home residents with 90 to 365 consecutive days of a nursing home stay in 2017-2018. We generated a single cohort for both 2017 and 2018 calendar years and a resident’s stay may span both years (Table 1). We used a validated algorithm and excluded skilled nursing home stays to identify long-term care nursing home stays [20]. Long-term care stays overlapping with a resident’s last six months of life were excluded because drug use may be very different during a person’s end of life. If a stay was longer than 1 year, we considered in the analysis only the first 365 days. If a resident had multiple stays longer than 90 days during this period, we selected the longest one. First, we included residents aged 65 years or older. Second, we included residents with continuous enrollment in Medicare Parts A and B, and without HMO coverage for at least six months prior to and during their stay to ensure they are fully covered by Medicare and to capture their comorbidities. Third, we included patients with continuous enrollment in Medicare Part D for at least 3 months prior to and during their stay to capture their prescription drug use. Fourth, we included residents who resided in nursing homes which had no missing data for nursing home characteristics for 2017-2018.
Table 1.
Cohort Derivation for years 2017-2018.
| Step | Description | Residents (%) |
|---|---|---|
| 1 | Long-term nursing residents 2017-2018 (excluding SNF stays) | 597,110 (100) |
| 2 | > 90 consecutive days stay excluding the last six months of life | 231,504 (38.8) |
| 3 | Resident’s age equal or over 65 years | 205,366 (88.7) |
| 4 | Medicare Parts A and B (6 months prior to and during stays) | 198,146 (96.5) |
| 5 | Without HMO insurance (6 months prior to and during stay) | 130,107 (65.7) |
| 6 | Medicare Part D (3 months prior to and during stay) | 113,159 (87.0) |
| 7 | Residents in nursing homes without missing data | 112,541 (99.5) |
Abbreviations: HMO, health maintenance organization; SNF, skilled nursing facility.
2.3. Independent Variables
Both resident and nursing home characteristics were used in the variation analysis. Resident age, sex and race were extracted from the Medicare enrollment files. Resident length of stay ranged from 90 to 365 days. Resident characteristics from the MDS datasets were extracted from the first MDS assessment during their stay. We defined marital status (married, unmarried), dementia severity (none, mild, moderate, severe) [21], aggressive behavior (none, moderate, severe), depression (none, mild, moderate/severe), hallucination/delusions and activities of daily living (ADL) (6 items each with 0-4 score, summed to produce a score of 0-24, with higher scores indicating greater dependence). The Medicare Provider Analysis and Review, Outpatient, and Carrier files during the 6 months prior to each resident’s stay were used to identify diagnoses used in the calculation of the weighted Elixhauser comorbidity score [22] and diagnoses related to indications for benzodiazepine use, including anxiety, seizures, sleep disorder and substance abuse.
Nursing home characteristics included ownership type (for-profit, non-profit, government), location (urban, rural), bed number and geographic region (Northeast, Midwest, West, South). The nursing home quality ratings were retrieved from the Nursing Home Compare web site. Every Medicare- and Medicaid-certified nursing home in the US is evaluated monthly. For each resident stay (90 to 365 days) we defined quality rating as the facility’s median rating during the calendar months during which the stay occurred.
2.4. Outcome Variable
The outcome was a binary variable indicating whether residents were prescribed at least a 30-day supply of benzodiazepines during their stay. We used Medicare Part D prescription data to obtain date of dispensing and estimated the days’ supply for all prescriptions. We considered use of short-acting and long-acting benzodiazepines separately.
Micromedex RED BOOK provided all NDC drug codes corresponding to the generic names of short- and intermediate-acting (alprazolam, estazolam, lorazepam, oxazepam, temazepam, triazolam) and long-acting (clorazepate, chlordiazepoxide, clonazepam, diazepam, flurazepam, quazepam) benzodiazepines reported in the American Geriatrics Society 2015 updated Beers Criteria Beers criteria [23] were matched to the NDCs in the Part D claims data.
2.5. Statistical Analysis
Descriptive statistics, including counts and percentages, were used to describe the study cohorts. Bivariate analysis was performed to assess the effect of the independent variables and estimate the unadjusted rates and time-trends of short- and long-acting benzodiazepine use. For the variation analysis, the intraclass correlation coefficient (ICC), obtained from mixed effects logistic regression models, was used to quantify the proportion of total variance in the outcome attributed to differences in prescribing patterns among nursing homes. Initially, we assessed the variation among facilities without including any predictors in the model. Subsequently, we assessed how patient- and facility-level characteristics affected the variation among nursing homes and estimated the adjusted odd ratios of these characteristics. All analyses were conducted separately for short- and long-acting benzodiazepines. SAS version 6.2 (SAS, Inc., Cary, NC) was used for all analyses.
3. RESULTS
Figure 1 shows the rates of short- and long-acting benzodiazepine use among nursing home residents for each quarter from 2013 to 2018. Short-acting benzodiazepine rates did not noticeably change between 2013 and 2016, and then declined from 12.1% in 2016 to 10.4% in 2018. The rate of long-acting benzodiazepine use remained relatively constant throughout this period, at approximately 4%.
Fig. 1. Unadjusted rates of short- and long-acting benzodiazepine prescriptions in nursing homes for the period 2013 – 2018, by quarter.
Benzodiazepine use was defined as receipt of a prescription for at least a 30-day supply during a quarter. Only residents who stayed in a nursing home for the entire quarter were included. Residents may contribute information in more than one quarter (quarter stays n = 1,843,580; unique residents n = 270,566).
Variation analyses focused on years 2017 and 2018. The demographic, clinical and facility characteristics of 112,541 unique residents residing in 14,641 nursing homes are presented in Table 2. Overall, 45.9% of nursing home residents were 85 years or older, 71.5% were female and 79.2% were White. Most residents were diagnosed with mild (24.2%), moderate (34.6%) and severe dementia (10.5%). The overall prescribing rate was 15% for short-acting and 4.7% for long-acting benzodiazepines.
Table 2.
Unadjusted rates and adjusted odd ratios for receipt of a prescription for short- and long-acting benzodiazepine use by patient and nursing home characteristics in 2017- 2018.
| Characteristic | Number of Residents | Short acting | Long acting | ||
|---|---|---|---|---|---|
|
| |||||
| N (%) | Rate (%) | Adjusted OR (95% CI) | Rate (%) | Adjusted OR (95% CI) | |
| Total | 112,541 (100) | 15.0 | 4.7 | ||
|
| |||||
| Age | |||||
| 65-74 | 24,702 (21.9) | 16.3 a | REF | 8.3 a | REF |
| 75-84 | 36,166 (32.1) | 15.8 | 0.98 (0.93, 1.03) | 4.8 | 0.60 (0.56, 0.64) |
| 85+ | 51,673 (45.9) | 13.9 | 0.89 (0.85, 0.93) | 2.9 | 0.38 (0.35, 0.41) |
|
| |||||
| Sex | |||||
| Male | 32,104 (28.5) | 11.5 a | REF | 4.3 a | REF |
| Female | 80,437 (71.5) | 16.5 | 1.37 (1.32, 1.44) | 4.9 | 1.21 (1.13, 1.29) |
|
| |||||
| Race | |||||
| White | 89,091 (79.2) | 16.2 a | REF | 5.1 a | REF |
| Black | 13,596 (12.1) | 9.9 | 0.67 (0.63, 0.72) | 3.1 | 0.60 (0.54, 0.67) |
| Hispanic | 6,077 ( 5.4) | 13.8 | 0.92 (0.84, 1.00) | 4.2 | 0.86 (0.75, 0.99) |
| Other | 3,777 ( 3.4) | 8.6 | 0.64 (0.57, 0.73) | 3.0 | 0.75 (0.62, 0.92) |
|
| |||||
| Marital Status | |||||
| Married | 20,560 (18.3) | 15.0 | REF | 4.7 | REF |
| Unmarried | 91,981 (81.7) | 15.1 | 0.93 (0.89, 0.98) | 4.7 | 0.98 (0.91, 1.06) |
|
| |||||
| Dementia | |||||
| None | 34,543 (30.7) | 15.5 a | REF | 6.0 a | REF |
| Mild | 27,246 (24.2) | 13.8 | 0.89 (0.84, 0.93) | 5.0 | 0.91 (0.85, 0.98) |
| Moderate | 38,955 (34.6) | 15.4 | 1.01 (0.96, 1.05) | 3.7 | 0.69 (0.64, 0.75) |
| Severe | 11,797 (10.5) | 15.4 | 1.03 (0.96, 1.11) | 3.8 | 0.65 (0.58, 0.73) |
|
| |||||
| Aggressive Behavior | |||||
| None | 94,778 (84.2) | 13.8 a | REF | 4.4 a | REF |
| Moderate | 12,908 (11.5) | 20.4 | 1.43 (1.36, 1.51) | 6.0 | 1.23 (1.13, 1.34) |
| Severe | 4,855 ( 4.3) | 25.6 | 1.75 (1.62, 1.89) | 7.3 | 1.44 (1.27, 1.63) |
|
| |||||
| Mood disorder | |||||
| None | 57,169 (50.8) | 13.1 a | REF | 4.1 a | REF |
| Mild | 50,036 (44.5) | 16.8 | 1.20 (1.16, 1.25) | 5.2 | 1.19 (1.12, 1.26) |
| Mod/Severe | 5,336 ( 4.7) | 19.8 | 1.37 (1.26, 1.48) | 6.9 | 1.46 (1.29, 1.65) |
|
| |||||
| Hallucinations / Delusions | |||||
| no | 106,098 (94.3) | 14.4 a | REF | 4.5 a | REF |
| yes | 6,443 ( 5.7) | 25.3 | 1.47 (1.38, 1.58) | 7.5 | 1.36 (1.22, 1.51) |
|
| |||||
| Anxiety diagnosis | |||||
| no | 74,393 (66.1) | 8.6 a | REF | 2.4 a | REF |
| yes | 38,148 (33.9) | 27.7 | 3.81 (3.67, 3.95) | 9.1 | 3.52 (3.31, 3.74) |
|
| |||||
| Seizure diagnosis | |||||
| no | 103,436 (91.9) | 15.0 | REF | 4.6 a | REF |
| yes | 9,105 ( 8.1) | 15.2 | 0.98 (0.91, 1.04) | 6.4 | 1.15 (1.05, 1.27) |
|
| |||||
| Sleep disorder diagnosis | |||||
| no | 109,504 (97.3) | 14.8 a | REF | 4.6 a | REF |
| yes | 3,037 ( 2.7) | 24.9 | 1.37 (1.25, 1.50) | 8.0 | 1.20 (1.04, 1.38) |
|
| |||||
| Substance abuse diagnosis | |||||
| no | 104,050 (92.5) | 14.8 a | REF | 4.5 a | REF |
| Yes | 8,491 ( 7.5) | 17.7 | 1.10 (1.03, 1.17) | 7.3 | 1.07 (0.97, 1.18) |
|
| |||||
| Elixhauser comorbidity weighted score | |||||
| <4 | 29,789 (26.5) | 16.0 a | REF | 4.9 b | REF |
| 4-9 | 28,070 (24.9) | 15.1 | 0.95 (0.91, 1.00) | 4.8 | 0.96 (0.88, 1.03) |
| 10-16 | 25,541 (22.7) | 14.9 | 0.93 (0.88, 0.98) | 4.7 | 0.89 (0.82, 0.97) |
| >16 | 29,141 (25.9) | 14.1 | 0.86 (0.82, 0.90) | 4.3 | 0.73 (0.67, 0.79) |
|
| |||||
| Activities of Daily Living | |||||
| 0-6 | 12,556 (11.2) | 16.9 a | REF | 5.9 a | REF |
| 7-12 | 19,035 (16.9) | 16.2 | 0.93 (0.87, 1.00) | 5.2 | 0.96 (0.86, 1.06) |
| 13-18 | 56,007 (49.8) | 14.6 | 0.85 (0.80, 0.90) | 4.4 | 0.91 (0.83, 1.00) |
| 19-24 | 24,943 (22.2) | 14.2 | 0.82 (0.77, 0.88) | 4.4 | 1.02 (0.92, 1.14) |
|
| |||||
| Length of stay (months) | |||||
| 3-4 | 12,015 (10.7) | 11.8 a | REF | 4.3 b | REF |
| 4-8 | 31,670 (28.1) | 14.5 | 1.30 (1.21, 1.39) | 4.5 | 1.07 (0.97, 1.19) |
| 8-12 | 68,856 (61.2) | 15.8 | 1.53 (1.44, 1.63) | 4.8 | 1.22 (1.10, 1.34) |
|
| |||||
| Region | |||||
| Northeast | 25,244 (22.4) | 12.4 a | REF | 4.6 a | REF |
| Midwest | 30,705 (27.3) | 14.6 | 1.13 (1.06, 1.20) | 4 | 0.78 (0.71, 0.86) |
| West | 13,301 (11.8) | 11.7 | 1.16 (1.07, 1.25) | 3.4 | 0.81 (0.72, 0.92) |
| South | 43,291 (38.5) | 17.9 | 1.54 (1.46, 1.63) | 5.7 | 1.17 (1.08, 1.28) |
|
| |||||
| Ownership | |||||
| Profit | 79,640 (70.8) | 15.2 b | REF | 4.9 a | REF |
| Non-profit | 24,741 (22.0) | 14.6 | 1.02 (0.97, 1.07) | 4 | 0.96 (0.89, 1.04) |
| Government | 8,160 ( 7.3) | 14.7 | 1.00 (0.92, 1.08) | 4.5 | 1.02 (0.90, 1.16) |
|
| |||||
| Location | |||||
| Urban | 79,849 (71.0) | 14.2 a | REF | 4.6 b | REF |
| Rural | 32,692 (29.0) | 17.0 | 1.15 (1.10, 1.21) | 5.1 | 1.10 (1.03, 1.18) |
|
| |||||
| Quality rating | |||||
| 1-2 | 37,276 (33.1) | 15.5 b | REF | 5.1 a | REF |
| 3 | 22,428 (19.9) | 15.4 | 1.01 (0.96, 1.07) | 4.9 | 1.02 (0.94, 1.10) |
| 4-5 | 52,837 (46.9) | 14.6 | 0.99 (0.95, 1.03) | 4.3 | 0.98 (0.92, 1.06) |
|
| |||||
| Bed number | |||||
| 200− | 98,925 (87.9) | 15.3 a | REF | 4.7 | REF |
| 200+ | 13,616 (12.1) | 13.2 | 1.01 (0.94, 1.09) | 4.5 | 1.05 (0.94, 1.16) |
Benzodiazepine use was defined as drug use for at least 30 days during stay. A stay ranges from 90 to 365 days.
Rates of benzodiazepine use are unadjusted.
Adjusted odds ratios were derived from two-level mixed logistic regression model for each outcome (short- and long-acting benzodiazepine use for at least 30 days during stay). In each of the two models, we included resident characteristics age, sex, race, marital stays, length of stay, Elixhauser comorbidity weighted score, dementia, Activities of Daily Living, aggressive behavior, mood disorder, hallucinations or delusions, anxiety, seizures, sleep disorder, substance abuse, and the facility characteristics region, ownership, location, bed number, and quality rating. Nursing home facility was used as random effect. 736 residents (0.7% of the sample) who used both short- and long-acting benzodiazepines for at least 30 days each were included in analyses for both short- and long-acting benzodiazepine use.
p < .0001
p < .05
Table 2 presents unadjusted rates and adjusted odds ratios (OR) of short- and long-acting benzodiazepine prescriptions. For both types of benzodiazepines, residents older than 85 years were less likely to be prescribed compared to those aged 65-74 years [short-acting: OR: 0.89, 95% Confidence Interval (CI): 0.85-0.93; long-acting: OR: 0.38, 95% CI: 0.35-0.41]. Those aged 75-84 years were significantly less likely to receive long-acting drugs (OR: 0.60, 95% CI: 0.56-0.64). Females were more likely than males to receive either type of benzodiazepine (OR: 1.37, 95% CI: 1.32-1.44 for short-acting and OR: 1.21, 95% CI: 1.13-1.29 for long-acting). The odds of benzodiazepine use were higher for those with any aggressive behavior, hallucinations, mood disorders or a diagnosis of anxiety and sleep disorders. Residents who were White, had fewer comorbidities, a lower ADL score or a longer nursing home stay were significantly more likely to use benzodiazepines.
Increased severity of dementia was associated with progressively lower odds of long-acting benzodiazepine use compared to residents without dementia while, for short-acting benzodiazepine use, only those with mild dementia were significantly less likely than those without dementia to use benzodiazepines (OR: 0.89, 95% CI: 0.84-0.93). Seizure diagnosis was associated with increased odds of long-acting benzodiazepine use (OR: 1.15, 95% CI: 1.05-1.27), whereas substance abuse diagnosis was associated with short-acting benzodiazepine use only (OR: 1.10, 95% CI: 1.03-1.17). Short-acting benzodiazepine use was significantly lower in the Northeast and higher in the South compared to the other regions. For long-acting drugs, the Midwest and West had lower odds of use than the Northeast, whereas the South had higher.
Residents of rural nursing homes were more likely to receive prescriptions for benzodiazepines (OR: 1.15, 95% CI: 1.10-1.21 for short-acting and OR: 1.10, 95% CI: 1.03-1.18 for long-acting). There was no significant association of short- or long-acting benzodiazepine prescribing with the CMS nursing home quality ratings.
Table 3 reports variation among nursing homes expressed as the ICCs for short- and long-acting benzodiazepine prescribing. Before adjusting for resident and facility characteristics, the percent of variation attributed to nursing homes was estimated at 7.9% for short- and 10.9% for long-acting benzodiazepines. After adjustment for both resident and facility characteristics, the percent of nursing home variation was reduced to 6.3% for short-acting and 8.4% for long-acting benzodiazepines.
Table 3.
Variation among nursing homes in benzodiazepine use of at least 30 days during residents’ stay in 2017- 2018.
| Covariate group | Additional covariates | ICC (95% CI) |
|
|---|---|---|---|
| Short acting | Long acting | ||
| Null model | - | 7.9 (7.2, 8.6) | 10.9 (9.2, 12.5) |
|
| |||
| Patient | Age, sex, race, marital status, dementia, aggressive behavior, mood disorder, hallucinations/delusions, anxiety, seizures, sleep disorders, substance abuse, Elixhauser comorbidity score, ADL, LOS | 7.2 (6.5, 8.0) | 9.3 (7.7, 10.9) |
|
| |||
| Facility | Region, ownership, location, bed number, quality rating | 6.3 (5.6, 7.1) | 8.4 (6.8, 10.0) |
Abbreviations: ADL, activities of daily living; ICC, intraclass correlation coefficient; LOS, length of nursing stay in months.
Nursing home facility was used as random effect.
The effect of facility characteristics on the ICC were assessed while adjusting for patient characteristics.
4. DISCUSSION
In this national sample of long-term nursing home residents from 2013 to 2018, the rates of short-acting benzodiazepine use were stable at approximately 12% between 2013 and 2016, then decreased to 10.5% in 2018, while long-acting benzodiazepine use was relatively stable at around 4% throughout the 2013-2018 period. We also found that an appreciable portion of the variance in benzodiazepine prescribing, 7% to 9%, is attributed to differences in use patterns among nursing homes.
Our study expands the findings of a previous study which found that benzodiazepine use was nearly 20% from 2013 to 2014 among Medicare beneficiaries in nursing homes [15]. We report use for short- and long-acting benzodiazepine separately from 2013 to 2018. We also used a stricter definition of benzodiazepine use, i.e., at least 30 days of supply in each quarter.
We found a decrease in the use of short-acting benzodiazepines at the beginning of 2016. Emerging evidence on the harms associated with benzodiazepine use and periodic negative press reports may have contributed to the decline of benzodiazepine use. However, we did not find any policy or consensus statement that may have resulted in significant decline in use around 2016.
After adjusting for resident characteristics, the nursing home contribution to the variance in benzodiazepine use in 2017-2018 was 7.2% for short-acting benzodiazepines and 9.3% for long-acting benzodiazepines. As a comparison, hospitals account for 0.1% to 2.6% of the variation in readmission rates and are considered targets to reduce variation [24–26]. Including nursing home characteristics into the model resulted in a modest reduction of the ICCs. We had anticipated finding a relationship between the CMS facility quality rating and benzodiazepine prescribing rates as we have found in the case of anticholinergic drugs [19]. However, we found no such relationship.
We identified several resident characteristics associated with benzodiazepine use. The finding of decreasing prescription rates with increasing age and also with dementia severity is consistent with evidence that benzodiazepine use may promote cognitive decline in older adults [3]. Benzodiazepine treatment is used for aggressive behavior, hallucinations or mood disorders which may explain higher use among individuals with such conditions. Both short- and long-acting benzodiazepines are commonly prescribed to treat residents with anxiety, and long-acting benzodiazepines are occasionally used to treat seizure disorders. Residents with greater comorbid burden may use other medications to manage their diseases. The lower use of benzodiazepine among those with greater comorbid burden may reflect more careful selection of drugs to reduce potential for drug-drug interactions and reduce the overall medication burden. Greater use in rural area maybe due to lower access to high-quality care and geriatric services [27]. Deprescribing guidelines for benzodiazepine may help clinicians and patients make shared decisions on how to reduce dosage or deprescribe drugs and use alternative non-drug approaches [28].
Our study has several limitations. First, the percent of Medicare recipients with Part D coverage grew over the study period. Thus, the sample we used in the trend analysis may have changed in characteristics over time. Second, we have no way of determining at an individual resident level whether the prescription of a short- or long-acting benzodiazepine might be clinically appropriate. Finally, we assessed prescriptions for benzodiazepines with at least a 30-day supply. We were not able to assess whether the drugs were administered. Despite these limitations, our study has significant strengths, such as the large representative cohort size and the accurate estimation of quarterly benzodiazepine prescription rates by including in the cohort only residents who stayed in a nursing homes for the whole quarter. Finally, this is the first study that investigates trends and patterns of benzodiazepine use in nursing homes after Medicare Part D coverage expansion in 2013.
5. CONCLUSIONS AND IMPLICATIONS
There was a decline in short-acting benzodiazepine use at the beginning of 2016 while the use of long-acting benzodiazepines did not change from 2013 to 2018. The variation in benzodiazepine use among nursing homes is substantial. Identifying the factors that explain this variation may help in developing strategies for deprescribing benzodiazepines in nursing home residents.
Funding:
This study was funded by The National Institutes of Health (grants number R01 DA039192, 1 K01 AG070329-01 and P30 AG024832).
Footnotes
Conflicts of interest / Competing interests: None
Code availability: N/A
Ethics Approval: This research study was conducted retrospectively from de-identified data. The UTMB Institutional Review Board approved this study.
Consent to Participate / for Publication: N/A. The UTMB Institutional Review Board waived the requirement for consent to participate/ consent to publish because the data was de-identified prior to use.
Availability of data and material:
The datasets generated during and/or analysed during the current study are available from the Research Data Assistance Center, The Centers for Medicare and Medicaid Services, https://www.cms.gov/Research-Statistics-Data-and-Systems/Research/ResearchGenInfo/ResearchDataAssistanceCenter
References
- 1.Woolcott JC, Richardson KJ, Wiens MO, et al. (2009) Meta-analysis of the impact of 9 medication classes on falls in elderly persons. Arch Intern Med 169:1952–1960. 10.1001/archinternmed.2009.357 [DOI] [PubMed] [Google Scholar]
- 2.Wang PS, Bohn RL, Glynn RJ, Mogun H, Avorn J (2001) Hazardous benzodiazepine regimens in the elderly: Effects of half-life, dosage, and duration on risk of hip fracture. Am J Psychiatry 158:892–898. 10.1176/appi.ajp.158.6.892 [DOI] [PubMed] [Google Scholar]
- 3.Tannenbaum C, Paquette A, Hilmer S, Holroyd-Leduc J, Carnahan R (2012) A systematic review of amnestic and non-amnestic mild cognitive impairment induced by anticholinergic, antihistamine, GABAergic and opioid drugs. Drugs Aging 29:639–658. 10.1007/bf03262280 [DOI] [PubMed] [Google Scholar]
- 4.Maust DT, Blow FC, Wiechers IR, Kales HC, Marcus SC (2017) National trends in antidepressant, benzodiazepine, and other sedative-hypnotic treatment of older adults in psychiatric and primary care. J Clin Psychiatry 78:e363–71. 10.4088/JCP.16m10713 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Ray WA (1989) Benzodiazepines of long and short elimination half-life and the risk of hip fracture. JAMA 262(23):3303. 10.1001/jama.1989.03430230088031 [DOI] [PubMed] [Google Scholar]
- 6.Beers MH (1997) Explicit criteria for determining potentially inappropriate medication use by the elderly. Arch Intern Med 157:1531. 10.1001/archinte.1997.00440350031003 [DOI] [PubMed] [Google Scholar]
- 7.Fick DM, Cooper JW, Wade WE, Waller JL, Maclean JR, Beers MH (2003) Updating the Beers Criteria for potentially inappropriate medication use in older adults. Arch Intern Med 163:2716. 10.1001/archinte.163.22.2716 [DOI] [PubMed] [Google Scholar]
- 8.Wagner AK, Zhang F, Soumerai SB, et al. (2004) Benzodiazepine use and hip fractures in the elderly. Arch Intern Med 164:1567. 10.1001/archinte.164.14.1567 [DOI] [PubMed] [Google Scholar]
- 9.Fick DM, Semla TP, Steinman M, et al. (2019) American Geriatrics Society 2019 Updated AGS Beers Criteria® for Potentially Inappropriate Medication Use in Older Adults. J Am Geriatr Soc 67:674–694. 10.1111/jgs.15767 [DOI] [PubMed] [Google Scholar]
- 10.Bykov K, He M, Gagne JJ (2020) Trends in utilization of prescribed controlled substances in US commercially insured adults, 2004-2019. JAMA Intern Med 180:1006. 10.1001/jamainternmed.2020.0989 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Chen H, Nwangwu A, Aparasu R, Essien E, Sun S, Lee K (2008) The impact of Medicare Part D on psychotropic utilization and financial burden for community-based seniors. Psychiatr Serv 59:1191–1197. 10.1176/ps.2008.59.10.1191 [DOI] [PubMed] [Google Scholar]
- 12.Ong MK, Xu H, Zhang L, Azocar F, Ettner SL (2012) Effect of Medicare Part D benzodiazepine exclusion on psychotropic use in benzodiazepine users. J Am Geriatr Soc 60:1292–1297. 10.1111/j.1532-5415.2012.04031.x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Centers for Medicare & Medicaid Services (CMS) (2012) Transition to Part D coverage of benzodiazepines and barbiturates beginning in 2013. CMS web page. 1927:21244. https://www.cms.gov/Medicare/Prescription-Drug-Coverage/PrescriptionDrugCovContra/Downloads/BenzoandBarbituratesin2013.pdf [Google Scholar]
- 14.Agarwal SD, Landon BE (2019) Patterns in outpatient benzodiazepine prescribing in the United States. JAMA Netw Open 2:e187399. 10.1001/jamanetworkopen.2018.7399 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Maust DT, Kim HM, Chiang C, Kales HC (2018) Association of the Centers for Medicare & Medicaid Services’ national partnership to improve dementia care with the use of antipsychotics and other psychotropics in long-term care in the United States from 2009 to 2014. JAMA Intern Med 178:640–647. 10.1001/jamainternmed.2018.0379 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Vu MQ, Weintraub N, Rubenstein LZ (2006) Falls in the nursing home: Are they preventable? J Am Med Dir Assoc 7:53–58. 10.1016/j.jamda.2005.12.016 [DOI] [PubMed] [Google Scholar]
- 17.Cioltan H, Alshehri S, Howe C, et al. (2017) Variation in use of antipsychotic medications in nursing homes in the United States: A systematic review. BMC Geriatr 17:32. 10.1186/s12877-017-0428-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Fog AF, Mdala I, Engedal K, Straand J (2020) Variation between nursing homes in drug use and in drug-related problems. BMC Geriatr 20:336. 10.1186/s12877-020-01745-y [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Malagaris I, Mehta HB, Li S, Goodwin JS (2020) Decrease of anticholinergic drug use in nursing home residents in the United States, 2009 to 2017. J Am Geriatr Soc 68:2797–2804. 10.1111/jgs.16776 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Goodwin JS, Li S, Zhou J, Graham JE, Karmarkar A, Ottenbacher K (2017) Comparison of methods to identify long term care nursing home residence with administrative data. BMC Health Serv Res 17:1–8. 10.1186/s12913-017-2318-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Downer B, Thomas KS, Mor V, Goodwin JS, Ottenbacher KJ (2017) Cognitive status of older adults on admission to a skilled nursing facility according to a hospital discharge diagnosis of dementia. J Am Med Dir Assoc 18:726–728. 10.1016/j.jamda.2017.04.021 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Van Walraven C, Austin PC, Jennings A, Quan H, Forster AJ (2009) A modification of the Elixhauser comorbidity measures into a point system for hospital death using administrative data. Med Care 47:626–633. 10.1097/MLR.0b013e31819432e5 [DOI] [PubMed] [Google Scholar]
- 23.Samuel MJ (2015) American Geriatrics Society 2015 updated beers criteria for potentially inappropriate medication use in older adults. J Am Geriatr Soc 63:2227–2246. 10.1111/jgs.13702 [DOI] [PubMed] [Google Scholar]
- 24.Singh S, Lin Y-L, Kuo Y-F, Nattinger AB, Goodwin JS (2014) Variation in the risk of readmission among hospitals: the relative contribution of patient, hospital and inpatient provider characteristics. J Gen Intern Med 2:572–578. 10.1007/s11606-013-2723-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Korda RJ, Du W, Day C, Page K, Macdonald PS, Banks E (2017) Variation in readmission and mortality following hospitalisation with a diagnosis of heart failure: prospective cohort study using linked data. BMC Health Serv Res 17:220. 10.1186/s12913-017-2152-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Hekkert K, Kool RB, Rake E, et al. (2018) To what degree can variations in readmission rates be explained on the level of the hospital? a multilevel study using a large Dutch database. BMC Health Serv Res 18:999. 10.1186/s12913-018-3761-y [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Mattos MK, Sereika SM, Naples JG, Albert SM. Differences in Benzodiazepine Receptor Agonist Use in Rural and Urban Older Adults. Drugs Real World Outcomes. 2016;3(3):289–296. doi: 10.1007/s40801-016-0080-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Pottie K, Thompson W, Davies S, Grenier J, Sadowski CA, Welch V, Holbrook A, Boyd C, Swenson R, Ma A, Farrell B (2018) Deprescribing benzodiazepine receptor agonists: Evidence-based clinical practice guideline. Can Fam Physician. 64(5):339–351. [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The datasets generated during and/or analysed during the current study are available from the Research Data Assistance Center, The Centers for Medicare and Medicaid Services, https://www.cms.gov/Research-Statistics-Data-and-Systems/Research/ResearchGenInfo/ResearchDataAssistanceCenter