Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2026 Mar 1.
Published in final edited form as: J Am Geriatr Soc. 2024 Dec 1;73(3):737–749. doi: 10.1111/jgs.19290

Rates and Predictors of Opioid Deprescribing after Fracture: A Retrospective Study of Medicare Fee-for-Service Claims

Kevin T Pritchard 1, Chun-Ting Yang 1, Qiaoxi Chen 1, Yichi Zhang 1, James M Wilkins 2, Dae Hyun Kim 3,4,*, Kueiyu Joshua Lin 1,5,*
PMCID: PMC11908922  NIHMSID: NIHMS2037490  PMID: 39618093

Abstract

Background:

Adults with Alzheimer’s Disease and Alzheimer’s Disease Related Dementias (ADRD) or frailty are susceptible to fractures. Opioid analgesics are frequently prescribed after fractures. Documenting post-fracture opioid discontinuation rates and predictors of discontinuation among adults with ADRD or frailty can inform clinical practice, identify potential disparities, and improve pain management guidelines. The objective of this paper was to investigate opioid discontinuation in opioid-naïve older adults who used opioids after an acute fracture.

Methods:

This retrospective cohort study included opioid-naïve Medicare fee-for-service beneficiaries (N=33,027) ≥ 65 years of age who filled an opioid prescription within 30 days of a vertebral, lower extremity, or upper extremity fracture from 2013–2018. Beneficiaries were classified according to ADRD (yes/no) and frailty (yes/no) status using validated claims-based algorithms. The primary outcome was opioid discontinuation, defined as a 30-day supply gap. We estimated discontinuation rates with the Kaplan-Meier method and identified predictors of opioid discontinuation using Cox proportional hazards regression.

Results:

The 30-day opioid discontinuation rate was similar among non-frail beneficiaries without ADRD (81% [95% CI, 80–81%]) and those who were non-frail with ADRD (83% [81–84%]). Comparatively, 30-day discontinuation rates were lower among those with frailty and ADRD (76% [75–77%]) and those with frailty alone (77% [75–78%]). After adjusting for sociodemographic characteristics, health status, healthcare utilization, and calendar year, beneficiaries with both ADRD and frailty (HR, 0.90 [0.87–0.93]) and those with frailty alone (HR, 0.85 [0.82–0.89]), but not those with ADRD alone (HR, 1.06 [1.01–1.10]), were less likely to discontinue opioids compared with those without ADRD or frailty.

Conclusions and Relevance:

Our findings suggest that frailty, but not ADRD, was associated with a lower likelihood of opioid discontinuation among older adults who initiated opioids after an acute fracture. Further research is needed to understand how opioid deprescribing practices depend on patient and provider preferences.

Keywords: (MeSH), Dementia, Fracture, Medicare, Opioid, Pain

INTRODUCTION

Among Americans aged 65 years and older, an estimated 1,458 women and 757 men per 100,000 were hospitalized with an extremity fracture in 2017.1 An estimated 30% of this population receive opioid treatment.2,3 Opioid treatment is more common when the fracture is surgically repaired or proximal instead of distal2 and when higher levels of care or rehabilitation are needed.3 Nonetheless, acute pain treatment guidelines encourage discontinuing prescription opioids within 30 days.4–6 Early deprescribing can prevent opioid-related adverse events including overdose,7 diversion,8 prolonged use,9 and falls with injury.10 Most patients undergo guideline-concordant opioid discontinuation within 30 days of elective orthopedic surgeries.11–14 Less is known about the discontinuation rate after acute fractures, which present a different pain experience due to their emergent nature. While some evidence indicates that 92% to 96% of patients discontinue opioids within 180 days after fracture,2,15 discontinuation rates have not been studied at earlier timepoints (e.g., 30 days) that align with current guidelines.

Acute pain control is important for older adults living with Alzheimer’s Disease and Alzheimer’s Disease Related Dementias (ADRD) or frailty. Most fractures among adults 65 years and older (78.1% in women and 69.1% in men) occur from low energy injuries, predominantly falls, and disproportionately affect individuals with more comorbidities and worse physical or cognitive function.1 Best practices are uncertain16 for these high-risk populations because opioid analgesia may increase their risk for adverse events. Although the severity of ADRD is associated with a lower likelihood of initiating opioids,17,18 no studies have assessed opioid discontinuation in people with ADRD or frailty.

The purpose of this research was to describe the rates of opioid discontinuation among opioid-naïve patients who were hospitalized for acute fractures. Based on prior research2,17 and practice guidelines,4,16 we hypothesized that discontinuation rates would be highest in the first 30-days and that adults with ADRD or frailty would be more likely to discontinue than those without ADRD or frailty. This information may help healthcare providers identify risks early, potentially reducing the prevalence of prolonged opioid use and adverse events.

METHODS

Data Sources and Study Design

In this retrospective cohort study, we examined fee-for-service Medicare beneficiaries from January 1, 2013, through December 31, 2018. To aid our comparison of discontinuation trends by ADRD status, we oversampled 100% of beneficiaries likely to have ADRD using the Chronic Conditions Warehouse algorithm19 and combined them with a random sample of beneficiaries without ADRD that represented 16% of fee-for-service beneficiaries. The combined oversample and random sample aided our study of high-risk subgroups based on ADRD or frailty status. We analyzed the Master Beneficiary Summary Files, the Master Beneficiary Summary File, Inpatient, Outpatient, Carrier, Home Health Agency, Skilled Nursing Facility, Hospice Base Claims, and the Prescription Claim Event Files. Our University’s Institutional Review Board determined that this study is exempt human subject research. A waiver of informed consent was obtained. A graphical depiction of the study design (Supplementary Figure S1) designates key study time points.20

Study Cohort

Beneficiaries entered the study cohort on the date (index date) they initiated opioids according to their first prescription fill in Part D data. We defined opioids using the Food and Drug Administration’s National Drug Code Directory Data Files,21 excluding formulations with decongestants or antitussive agents (Supplementary Table S1). Our cohort selection, depicted in Table 1, included Medicare beneficiaries with continuous Medicare Part A, B, and D enrollment during the year preceding the index date (baseline assessment period), which was necessary for covariate assessment. We permitted an enrollment gap of 31 days or fewer to define continuous enrollment.

Table 1.

Cohort Flowchart

Less Excluded Patients Remaining Patients
Initiated an opioid prescription 7,620,847
Excluded due to insufficient Part A, B, or D enrollment -2,197,031 5,423,816
Excluded due to incident opioid use in the 180 days preceding fracture -3,003,443 2,420,373
Excluded beneficiaries who were not part of the random sample or dementia oversample -433,633 1,986,740
Excluded due to age < 65 years old -204,271 1,782,469
Excluded due to no fracture in the 30 days preceding the initial opioid prescription -1,729,889 52,580
Excluded due to prevalent opioid use in the 180 days preceding fracture -288 52,292
Excluded based on initiating long-acting opioids -554 51,738
Excluded due to fibromyalgia and chronic pain -3,674 48,064
Excluded due to cancer (without non-melanoma skin) -3,283 44,781
Excluded due to cirrhosis -387 44,394
Excluded due to hospice / palliative care -7,268 37,126
Excluded due to alcohol use disorder -354 36,772
Excluded due to substance use disorder ** **
Excluded due to age missing ** **
Did not begin follow-up -3,558 33,076
Final cohort 33,076
Open in a new tab
**

To maintain confidentiality, cells were suppressed in accordance with the Centers for Medicare and Medicaid Services standards for minimum cells sizes.

We excluded beneficiaries younger than 65 years old and those without a fracture in the 30 days preceding the index date. Fractures were defined by inpatient claims using a validated algorithm with positive predictive values (PPVs) exceeding 90% for hip, pelvis, tibia/fibula or ankle, radius/ulna, humerus or scapula, and vertebral fractures.22 We used a crosswalk file23 to convert International Classification of Diseases, Ninth Revision (ICD-9) to International Classification of Diseases, Tenth Revision (ICD-10) diagnosis and procedure codes.24 To identify opioid-naïve patients, we excluded anyone with an opioid prescription in the 180 days preceding the index date. We excluded long-acting opioid initiators to avoid exposure misclassification, because this may indicate chronic pain. We employed the Chronic Conditions Warehouse criteria to exclude individuals with chronic pain or fibromyalgia, cancer, liver disease, alcohol use disorder, or substance use disorder during the 1-year baseline assessment period because these conditions are legitimate indications for long-term, chronic use or may expedite discontinuation (Supplementary Table S2).19 Similarly, we excluded those with a hospice (Q5001 through Q5010) or palliative (G9992, G9994, or G9996) claim in any setting and beneficiaries with missing demographic data. Since we could not measure prescriptions from Part A data, our cohort excluded those who remained in a hospital or post-acute care facility for skilled rehabilitation 30 days after fracture.

Outcomes

Our outcome was the number of days until discontinuing opioids. Discontinuation was defined as a gap exceeding 30 days after the prior prescription ended. For example, if a patient filled a 30-day prescription on the index date and refilled before or on day 60, it was considered continuous use. If no refill occurred before day 61, opioid use was deemed discontinued on day 30 when the initial prescription ended.13 We censored patients for the following reasons: death, end of patient data, end of study period (December 31, 2018), and Medicare disenrollment. We also censored for the occurrence of events that could affect opioid use or our ability to accurately measure opioid use via Part D data. These alternative events included hospital admissions, hospice or palliative care, nursing facility admissions, new diagnoses of our exclusion criteria including cancer, liver disease, alcohol use disorder, or substance use disorder.

Covariates

Baseline covariates were assessed during the 365 days preceding the index date. We recorded age, sex, dual eligibility, race and ethnicity (Asian, Black, Hispanic, White, or North American Native/Unknown/Missing), and Rural-Urban Continuum Codes (RUCC) (metro, urban, or rural, defined in Table 2‘s footnote).9,25 Health conditions included the Gagne Comorbidity Score, which combines health conditions from the Charlson and Elixhauser comorbidity measures to predict 1-year mortality with 79% discriminative ability,26 depression or anxiety,19 and fall history.27 We defined ADRD (yes/no) according to diagnoses from ≥ 2 health care settings (PPV 77%).28 We defined frailty (yes/no) using the Kim Claims-Based Frailty Index ≥ 0.25, which measures the accumulated number of deficits with diagnosis and procedure codes to predict mortality, osteoporotic fracture, institutionalization, and disability in mobility, activities of daily living, and instrumental activities of daily living.29–31 We measured the most common fracture types, hip, pelvis, or femur (PPVs: 95%, 93%, and 91%, respectively) and vertebral (PPV 99%) according to a validated claims algorithm.22 Measures of healthcare utilization included the total number of hospital days (quartiles), surgical treatment for fracture (Supplementary Table S3),32 whether a nursing home admission occurred between the fracture and opioid initiation, and total Medicare expenditures in U.S. dollars (quartiles). Pharmacologic pain treatments were measured using Part D claims and included opioids, gabapentinoids, non-steroidal anti-inflammatory drugs (NSAIDs), serotonin-norepinephrine reuptake inhibitors (SNRIs) and duloxetine separately, and muscle relaxants. We classified opioids as weak or strong according to their strength and prior research (Supplementary Table S4).33 Nonpharmacologic pain treatments included occupational therapy and physical therapy (OT/PT), defined by Current Procedural Terminology codes.34 To account for temporal changes in pain management policy and practice,35 we adjusted for cohort entry year.

Table 2.

Characteristics of Medicare beneficiaries who initiated opioids after hospitalization for acute fracture in 2013 through 2018

ADRD and Frailty Status
N = 33,076 Frail with ADRD (N = 12,112) Frail without ADRD (N = 4,721) Non-Frail with ADRD (N = 2,978) Non-Frail without ADRD (N = 13,265) P Value
Discontinued Opioids 30 Day Gap (Yes) 9400 (77.6%) 4052 (85.8%) 2557 (85.9%) 11902 (89.7%) <0.001***
Days to Discontinuation Mean (SD) 22.65 (49.0) 22.73 (42.0) 17.78 (37.8) 18.98 (34.2) <0.001***
Censoring Reason 30 Day Gap
 Outcome 9400 (77.6%) 4052 (85.8%) 2557 (85.9%) 11902 (89.7%) <0.001***
 Death 638 (05.3%) 38 (00.8%) 70 (02.4%) 22 (00.2%)
 Max Follow-up 128 (01.1%) 24 (00.5%) 14 (00.5%) 42 (00.3%)
 Disenrollment 35 (00.3%) 12 (00.3%) N/A (<1.00%) 28 (00.2%)
 End of Data 113 (00.9%) 25 (00.5%) 21 (00.7%) 79 (00.6%)
 Alternative 1798 (14.8%) 570 (12.1%) 313 (10.5%) 1192 (09.0%)
Age
 Mean (SD) 85.25 (7.5) 81.13 (8.3) 84.71 (7.8) 78.91 (8.2) <0.001***
 65–78 years 2284 (18.9%) 1799 (38.1%) 664 (22.3%) 6604 (49.8%) <0.001***
 79–87 years 4764 (39.3%) 1752 (37.1%) 1132 (38.0%) 4388 (33.1%)
 88+ years 5064 (41.8%) 1170 (24.8%) 1182 (39.7%) 2273 (17.1%)
Sex (Female) 9881 (81.6%) 3659 (77.5%) 2433 (81.7%) 10071 (75.9%) <0.001***
Race
 White 10745 (88.7%) 4302 (91.1%) 2622 (88.1%) 12094 (91.2%) <0.001***
 Black 695 (05.7%) 181 (03.8%) 145 (04.9%) 361 (02.7%)
 Asian 187 (01.5%) 64 (01.4%) 87 (02.9%) 273 (02.1%)
 Hispanic 285 (02.4%) 98 (02.1%) 55 (01.9%) 224 (01.7%)
 Unknown, Missing & Other 200 (01.7%) 76 (01.6%) 69 (02.3%) 313 (02.4%)
Disability Entitlement (Yes) 1101 (09.1%) 601 (12.7%) 243 (08.2%) 953 (07.2%) <0.001***
Rural-Urban Continuum Code
 Metro 9106 (75.2%) 3525 (74.7%) 2139 (71.8%) 9922 (74.8%) 0.004**
 Urban 1943 (16.0%) 798 (16.9%) 525 (17.6%) 2131 (16.1%)
 Rural 1063 (08.8%) 398 (08.4%) 314 (10.5%) 1212 (09.1%)
Index Year
 2014 2214 (18.3%) 1211 (25.7%) 496 (16.7%) 3181 (24.0%) <0.001***
 2015 2290 (18.9%) 1168 (24.7%) 523 (17.6%) 2818 (21.2%)
 2016 2517 (20.8%) 823 (17.4%) 620 (20.8%) 2598 (19.6%)
 2017 2540 (21.0%) 801 (17.0%) 673 (22.6%) 2410 (18.2%)
 2018 2551 (21.1%) 718 (15.2%) 666 (22.4%) 2258 (17.0%)
Gagne Comorbidity Index Mean (SD) 5.11 (2.5) 3.91 (2.6) 2.93 (2.0) 1.56 (2.0) <0.001***
Kim Frailty Index Mean (SD) 0.32 (0.1) 0.30 (0.04) 0.22 (0.02) 0.19 (0.04) <0.001***
Spine Fracture (Yes) 1856 (15.3%) 827 (17.5%) 592 (19.9%) 2399 (18.1%) <0.001***
Pelvis, Hip, or Femur Fracture (Yes) 9158 (75.6%) 3159 (66.9%) 1936 (65.0%) 7712 (58.1%) <0.001***
Depression (Yes) 5675 (46.9%) 1738 (36.8%) 644 (21.6%) 2284 (17.2%) <0.001***
Anxiety (Yes) 4168 (34.4%) 1429 (30.3%) 508 (17.1%) 2210 (16.7%) <0.001***
Baseline Falls (Yes) 7059 (58.3%) 2651 (56.2%) 1683 (56.5%) 7519 (56.7%) 0.019**
Baseline Days Hospitalized Mean (SD) 34.97 (49.2) 31.98 (50.0) 18.94 (34.8) 16.97 (31.2) <0.001***
 1st quartile, < 7 days 1781 (14.7%) 545 (11.5%) 983 (33.0%) 4299 (32.4%) <0.001***
 2nd quartile, < 13 days 2750 (22.7%) 921 (19.5%) 862 (29.0%) 3717 (28.0%)
 3rd quartile, < 25 days 2998 (24.8%) 1574 (33.3%) 654 (22.0%) 3422 (25.8%)
 4th quartile, ≥ 25 days 4583 (37.8%) 1681 (35.6%) 479 (16.1%) 1827 (13.8%)
Any Nursing Facility Admission (Yes) 5839 (48.2%) 1779 (37.7%) 828 (27.8%) 2990 (22.5%) <0.001***
Post-Fracture Nursing Home Stay (Yes) 1671 (13.8%) 884 (18.7%) 377 (12.7%) 1781 (13.4%) <0.001***
Total Medicare Cost Mean (SD) 40745.73 (28382.9) 45551.07 (31089.2) 27646.17 (17691.5) 29547.21 (21961.4) <0.001***
 1st quartile, < $19,669 2114 (17.5%) 456 (09.7%) 1108 (37.2%) 4589 (34.6%) <0.001***
 2nd quartile, < $30,423 2985 (24.6%) 856 (18.1%) 880 (29.6%) 3550 (26.8%)
 3rd quartile, < $43,900 3204 (26.5%) 1471 (31.2%) 600 (20.2%) 2994 (22.6%)
 4th quartile, ≥ $43,900 3809 (31.5%) 1938 (41.1%) 390 (13.1%) 2132 (16.1%)
Fracture Surgically Repaired (Yes) 2178 (18.0%) 519 (11.0%) 794 (26.7%) 4141 (31.2%) <0.001***
Strong Opioid (Versus Weak) 8313 (68.6%) 3449 (73.1%) 2101 (70.6%) 10418 (78.5%) <0.001***
Gabapentinoid Use (Yes) 1110 (09.2%) 645 (13.7%) 188 (06.3%) 1151 (08.7%) <0.001***
NSAID Use (Yes) 1890 (15.6%) 988 (20.9%) 451 (15.1%) 2792 (21.1%) <0.001***
Duloxetine Use (Yes) 387 (03.2%) 153 (03.2%) 53 (01.8%) 261 (02.0%) <0.001***
SNRI Use (Yes) 746 (06.2%) 295 (06.3%) 108 (03.6%) 507 (03.8%) <0.001***
Muscle Relaxant Use (Yes) 385 (03.2%) 309 (06.6%) 69 (02.3%) 808 (06.1%) <0.001***
OT/PT (Yes) 4904 (40.5%) 1389 (29.4%) 720 (24.2%) 2609 (19.7%) <0.001***
Open in a new tab

Note: Standard Deviation (SD), Alzheimer’s Disease and Alzheimer’s Disease Related Dementias (ADRD), Non-Steroidal Anti-Inflammatory Drugs (NSAID), Serotonin-Norepinephrine Reuptake Inhibitors (SNRIs), Occupational Therapy and/or Physical Therapy (OT/PT). Urbanicity is measured by the U.S. Department of Agricultural Rural-Urban Continuum Codes. Metropolitan counties included all counties in metropolitan areas (collapsing 1, Counties in metro areas of 1 million population or more; 2, Counties in metro areas of 250,000 to 1 million population; and 3, Counties in metro areas of fewer than 250,000 population). Urban counties included all urban populations adjacent to a metro area (collapsing 4, Urban population of 20,000 or more, adjacent to a metro area; 6, Urban population of 5,000 to 20,000, adjacent to a metro area; and 8, Urban population of fewer than 5,000, adjacent to a metro area). Rural counties included all counties non-adjacent to metro areas (collapsing 5, 7, 9).25 To maintain confidentiality, cells labeled “N/A” were suppressed in accordance with the Centers for Medicare and Medicaid Services standards for minimum cells sizes.

Analysis

We described the study cohort, based on ADRD and frailty status, using proportions (%) for categorical data and means (SD) for continuous data. To compare the distribution of baseline covariates by ADRD and frailty status, we used χ2 tests or ANOVA. To visualize the time until opioid discontinuation during the first 90 days of follow-up, while accounting for censoring, we used the Kaplan-Meier method. We derived the opioid discontinuation rate at follow-up days 30, 60, 90, 180 for each high-risk subgroup and compared them using the log-rank test. We estimated adjusted hazard ratios (HR) for opioid discontinuation with Cox proportional hazards regression. We used Fine and Gray’s method36 to address the competing risk of death, and inverse probability of censoring weights (IPCW)37 to handle informative censoring.

To select predictors in the Cox proportional hazards models, we used least absolute shrinkage and selection operator regression with the Schwarz Bayesian Criterion for model selection. We visually and statistically assessed model fit, influence of outliers, and the proportional hazards assumption. The proportional hazard assumption was violated late in the follow-up period by baseline measures of falls and surgical fracture repair according to their interactions with follow-up time and Schoenfeld residuals. We included time-interaction terms with these variables, but this approach did not improve our model or change our main estimates. Therefore, we retained the original time-invariant measures in our final model for a simpler interpretation. In sensitivity analyses, we considered a 15- and 60-day opioid discontinuation gap definition and restricted our cohort to those with lower extremity fractures. All analyses were two-sided with the significance level set at p < .05. Analyses were conducted using SAS statistical software v.9.4 (SAS Institute, Cary, NC).

RESULTS

Cohort Characteristics

Our study cohort (N=33,076) included 13,265 (40.1%) individuals who were non-frail without ADRD. They were on average 78.9 (SD: 8.2) years old, 75.9% female, 2.1% Asian, 2.7% Black, 1.7% Hispanic, 2.4% North American Native/Unknown/Missing, and 91.2% White. The 4,721 (9.0%) non-frail patients with ADRD were on average 84.7 (SD: 7.8) years old, 81.7% female, 2.9% Asian, 4.9% Black, 1.9% Hispanic, 2.3% North American Native/Unknown/Missing, and 88.1% White. The 4,721 (14.2%) patients with frailty and no ADRD, were on average 81.1 (SD: 8.3) years old, 77.5% female, 1.4% Asian, 3.8% Black, 2.1% Hispanic, 1.6% North American Native/Unknown/Missing, and 91.1% White. The 12,112 (36.6%) with frailty and ADRD were on average 85.3 (SD: 7.5) years old, 81.6% female, 1.5% Asian, 5.7% Black, 2.4% Hispanic, 1.7% North American Native/Unknown/Missing, and 88.7% White.

Compared to those without frailty or ADRD, patients with frailty or ADRD tended to be older, female, and either Black or Hispanic. Regarding health status, they also tended to have a disability entitlement, a higher comorbidity index, and more mental health diagnoses. Healthcare utilization was higher according to the average number of hospitalized days and Medicare costs, but they were less likely to undergo surgery after fracture. Prescription opioids were more commonly weak instead of strong and those with frailty or ADRD also tended to use more SNRIs or OT/PT (Table 2). The average discontinuation time, discontinuation proportions, and censoring reasons are shown in Table 2 for the discontinuation definition using a 30-day gap (primary analysis) and in Supplementary Table S5 for the 15- and 60-day gap definitions.

Opioid Discontinuation

Kaplan-Meier failure curves adjusted using IPCW showed rapid opioid discontinuation during the first ninety days of follow-up in all subgroups (Figure 1). A log-rank test indicated that opioid discontinuation varied by ADRD and frailty status (p < 0.001). Table 3 presents 30-, 60-, 90-, and 180-day discontinuation rates for each subgroup, comparing unadjusted rates with those adjusted using IPCW and Fine and Gray methods. The IPCW-adjusted discontinuation rates showed that, among non-frail individuals without ADRD, 81% (95% Confidence Interval [CI], 80%−81%) discontinued by day 30 and 99% (95% CI, 98%−99%) discontinued by day 180. Among non-frail individuals with ADRD, 83% (95% CI, 81%−84%) discontinued by day 30 and 97% (95% CI, 97–98%) discontinued by day 180.

Figure 1. The Probability of Discontinuing Prescription Opioids among Medicare Beneficiaries with a Fracture (2013–2018): Inverse Probability of Censoring Weight Adjusted Failure Rates.

Figure 1.

Open in a new tab

Note: Alzheimer’s Disease and Alzheimer’s Disease Related Dementias (ADRD), Discontinuation (D/C). The number at risk was identified at days 30, 60, and 90 for individuals who were frail with ADRD (N=12,112), frail without ADRD (N=4,721), non-frail with ADRD (N=2,978), and non-frail without ADRD (N=13,265). Log-Rank test statistics for equality across strata indicated that discontinuation rates varied by ADRD and frailty status (p < 0.001).

Table 3.

Prescription Opioid Discontinuation Rates after Fracture: Medicare beneficiaries (2013–2018)

High-Risk Subgroup Discontinuation Rate, % (95% Confidence Interval) N = 33,076
Days of Follow-up Unadjusted IPCW Adjusted Fine & Gray Adjusted
Frail with ADRD N=12,112
30 76 (76 to 77) 76 (75 to 77) 74 (73 to 75)
60 87 (86 to 88) 87 (86 to 88) 83 (83 to 84)
90 92 (91 to 92) 92 (91 to 92) 87 (87 to 88)
180 95 (94 to 95) 95 (94 to 95) 90 (89 to 90)
Frail without ADRD N=4,721
30 77 (75 to 78) 77 (75 to 78) 76 (75 to 77)
60 89 (88 to 90) 89 (88 to 90) 88 (87 to 89)
90 94 (93 to 94) 94 (93 to 94) 93 (92 to 94)
180 97 (97 to 98) 97 (97 to 98) 96 (96 to 97)
Non-Frail with ADRD N=2,978
30 83 (81 to 84) 83 (81 to 84) 82 (80 to 83)
60 92 (91 to 93) 92 (91 to 93) 90 (89 to 92)
90 96 (95 to 97) 96 (95 to 97) 94 (93 to 95)
180 99 (98 to 99) 97 (97 to 98) 95 (94 to 96)
Non-Frail without ADRD N=13,265
30 81 (80 to 81) 81 (80 to 81) 80 (80 to 81)
60 92 (92 to 93) 92 (92 to 93) 92 (92 to 93)
90 96 (96 to 96) 96 (96 to 96) 96 (96 to 96)
180 99 (99 to 99) 99 (98 to 99) 99 (98 to 99)
Open in a new tab

Note: Alzheimer’s Disease and Alzheimer’s Disease Related Dementias (ADRD), Inverse Probability of Censoring Weight (IPCW) indicates the discontinuation rates were adjusted for competing risk and informative censoring.

Adults with frailty discontinued at lower rates during the early follow-up period. For example, 76% (95% CI, 75%−77%) of beneficiaries with frailty and ADRD discontinued by day 30 and 95% (95% CI, 94%−95%) by day 180. Similarly, 77% (95% CI, 75%−78%) of frail individuals without ADRD discontinued by day 30 and 97% (95% CI, 97%−98%) discontinued by day 180. Our sensitivity analyses also indicated that discontinuation rates were slightly lower among those with frailty (Supplementary Table S6).

Predictors of Opioid Discontinuation

The IPCW model adjusted for informative censoring and all variables listed in Table 4 including age, sex, race, urbanicity, calendar year, comorbidity score, anxiety, fall history, days hospitalized in the baseline year, fracture type, and post-fracture nursing facility admission. Compared to non-frail individuals without ADRD, opioid discontinuation was less likely among those with frailty and ADRD (HR, 0.90, [95% CI, 0.87–0.93]) and among those with frailty but without ADRD (HR, 0.85, [95% CI, 0.82–0.89]). These findings were robust in the analyses using Fine and Gray (Table 4), 15- and 60-day discontinuation gaps, and in the subgroup analysis restricting to those with a lower extremity fracture (Supplementary Table S7).

Table 4.

Adjusted Hazard Ratios for Opioid Discontinuation after Fracture: Medicare beneficiaries (2013–2018)

Adjusted Hazard Ratio (95% Confidence Interval)
Characteristic Cox Multivariable Regression IPCW Adjusted Fine & Gray Adjusted
Non-Frail without ADRD Reference Reference Reference
 Frail with ADRD 0.90 (0.87 to 0.93) 0.90 (0.87 to 0.93) 0.86 (0.83 to 0.88)
 Frail without ADRD 0.85 (0.82 to 0.89) 0.85 (0.82 to 0.89) 0.83 (0.79 to 0.87)
 Non-Frail with ADRD 1.06 (1.01 to 1.11) 1.06 (1.01 to 1.10) 1.03 (0.99 to 1.07)
Age 65-78 Reference Reference Reference
 Age 79-87 years 1.08 (1.04 to 1.11) 1.07 (1.04 to 1.11) 1.05 (1.03 to 1.08)
 Age 88+ years 1.04 (1.01 to 1.07) 1.04 (1.01 to 1.07) 0.97 (0.94 to 1.00)
Sex Male Reference Reference Reference
 Sex Female 0.93 (0.90 to 0.95) 0.93 (0.90 to 0.95) 0.95 (0.93 to 0.98)
Race Non-Hispanic White Reference Reference Reference
 Race Asian 1.06 (0.97 to 1.16) 1.06 (0.97 to 1.16) 1.06 (0.98 to 1.16)
 Race Black 1.00 (0.94 to 1.06) 1.00 (0.94 to 1.06) 1.03 (0.98 to 1.09)
 Race Hispanic 1.04 (0.96 to 1.13) 1.04 (0.96 to 1.13) 1.08 (1.00 to 1.16)
 Race Unknown, Missing &Other 1.10 (1.02 to 1.20) 1.10 (1.02 to 1.20) 1.12 (1.04 to 1.21)
Rural-Urban Continuum Code Metro Reference Reference Reference
 Rural-Urban Continuum Code Rural 0.93 (0.89 to 0.97) 0.93 (0.89 to 0.97) 0.91 (0.88 to 0.95)
 Rural-Urban Continuum Code Urban 0.96 (0.93 to 1.00) 0.96 (0.93 to 0.99) 0.95 (0.92 to 0.98)
Index Year 2014 Reference Reference Reference
 Index Year 2015 1.00 (0.96 to 1.04) 1.00 (0.96 to 1.03) 0.99 (0.96 to 1.02)
 Index Year 2016 1.04 (1.00 to 1.08) 1.04 (1.00 to 1.08) 1.03 (0.99 to 1.06)
 Index Year 2017 1.10 (1.06 to 1.14) 1.10 (1.06 to 1.14) 1.08 (1.04 to 1.12)
 Index Year 2018 1.27 (1.22 to 1.32) 1.27 (1.22 to 1.32) 1.22 (1.18 to 1.27)
Gagne Comorbidity Index, score 0.99 (0.98 to 0.99) 0.99 (0.98 to 0.99) 0.98 (0.98 to 0.99)
No Anxiety Reference Reference Reference
 Anxiety 0.90 (0.88 to 0.93) 0.90 (0.88 to 0.93) 0.92 (0.89 to 0.94)
No Fall History Reference Reference Reference
 Fall history 0.99 (0.97 to 1.02) 1.00 (0.97 to 1.02) 1.00 (0.98 to 1.02)
Hospital days 1st quartile Reference Reference Reference
 Hospital days 2nd quartile 0.92 (0.89 to 0.96) 0.92 (0.89 to 0.96) 0.93 (0.90 to 0.96)
 Hospital days 3rd quartile 0.97 (0.94 to 1.00) 0.97 (0.94 to 1.01) 0.98 (0.95 to 1.02)
 Hospital days 4th quartile 0.90 (0.87 to 0.94) 0.90 (0.87 to 0.94) 0.90 (0.87 to 0.93)
No Surgical Repair of Fracture Reference Reference Reference
 Fracture Surgically Repaired 1.15 (1.11 to 1.18) 1.15 (1.12 to 1.19) 1.13 (1.09 to 1.16)
No Pelvis, Hip, or Femur Fracture Reference Reference Reference
 Fractured Pelvis, Hip, or Femur 1.01 (0.98 to 1.04) 1.01 (0.98 to 1.04) 0.99 (0.97 to 1.02)
No Spine Fracture Reference Reference Reference
 Fractured Spine 0.92 (0.89 to 0.96) 0.92 (0.89 to 0.96) 0.93 (0.89 to 0.96)
No Post-Fracture Nursing Facility Admission Reference Reference Reference
 Nursing Facility Admission After Fracture 1.04 (1.00 to 1.07) 1.04 (1.00 to 1.08) 1.06 (1.03 to 1.10)
Open in a new tab

Note: Alzheimer’s Disease and Alzheimer’s Disease Related Dementias (ADRD), Inverse Probability of Censoring Weight (IPCW). All hazard ratios were calculated over the 365-day at-risk period.

Compared to non-frail individuals without ADRD, those with ADRD alone had a marginally higher likelihood to discontinue opioids (HR, 1.06, [95% CI, 1.01–1.10]) as shown in Table 4. Although we found similar results using a 60-day gap definition for discontinuation (HR, 1.05, [95% CI, 1.01–1.10]), this association was not robust in our Fine and Gray analysis (HR, 1.03, [95% CI, 0.99–1.07]) or when restricting to lower extremity fractures (HR 1.04, [95% CI, 0.99–1.10]). After controlling for ADRD status, frailty, and all baseline characteristics in Table 4, older age, recent calendar years, and surgical treatment for fracture were positively associated with opioid discontinuation. These findings were robust in sensitivity analyses (Supplementary Table S7). Similarly, women, anxiety, spine fracture, and residing outside of a metropolitan area were associated with a decreased likelihood of discontinuation (Table 4 and Supplementary Table S7).

DISCUSSION

In this retrospective cohort study of U.S. Medicare beneficiaries with an acute fracture, we found the 30-day opioid discontinuation rates were 83% among non-frail individuals with ADRD, 81% among non-frail individuals without ADRD, 77% for those with frailty alone, and 78% for those with frailty and ADRD. After controlling for sociodemographic characteristics, health status including fracture type, healthcare utilization, and calendar year, our findings indicate that the likelihood to discontinue prescription opioids after fracture depended on frailty status, but not ADRD.

Previous research demonstrated that most Medicare beneficiaries discontinue opioids within one month of surgery, which may be explained by pre-operative education that prepares patients for their upcoming pain experience.11–14 Compared to those elective surgery populations, we found that most post-fracture patients also discontinued opioids within one month. These findings suggest that opioid prescribing practices are similar for acute pain events resulting from elective surgery and fracture, but more work is needed to elucidate whether patients’ pain experiences are comparable.

In our analysis of 2013–2018 Medicare fee-for-service claims, we observed a 60-day opioid discontinuation rate comparable to the 130-day discontinuation rate reported by Torchia et al2 using 2007–2010 Medicare data. Notably, our more rapid discontinuation rate was likely influenced by 2014 policy38 and 2016 practice guidelines39 that instigated national decreases in opioid use9,35,40 and deprescribing.41 Our finding that recent cohort entry years had a higher likelihood of opioid discontinuation relative to 2014 further supports this conclusion. An encouraging new finding from our analysis of discontinuation rates at earlier timepoints (e.g., 30 days) suggests that modern prescribing practices align with recent guidelines.4,6

Frailty, but not ADRD, was associated with a longer duration of opioid use after fracture. Compared to non-frail adults without ADRD, frailty reduced the likelihood of discontinuing opioids by 10% among those with ADRD and 15% for those without ADRD. Healthcare providers may have limited pain treatment options for frail adults with high comorbidity burdens. For example, NSAIDs and gabapentinoids may be contraindicated for patients taking anticoagulants or with chronic kidney disease. Alternatively, patients with frailty may prefer ongoing prescription opioids as a palliative treatment. Regardless of frailty and ADRD status, clinical guidelines should consider nonpharmacologic treatments, patient and caregiver goals, and the impact of pain on daily activities.

Individuals with ADRD were 6% more likely to discontinue opioids compared to non-frail adults without ADRD, although this small association was not robust across sensitivity analyses. Previous opioid prescribing research among individuals with ADRD categorized opioid use based on prevalent18 and incident17 use. Mehta et al17 studied Medicare beneficiaries residing in long-term care facilities and reported that the odds of initiating opioids in the year after a fracture were 33% lower among those with mild ADRD, 38% lower in those with moderate ADRD, and 54% lower in those with severe ADRD compared to those without ADRD. Together, these findings indicate that people with ADRD may be less likely to receive opioids because of barriers expressing their pain.42 However, once they begin opioids, the duration of use appears to depend more on frailty than ADRD status.

Our findings have implications for clinical practice and health policy. When pharmacologic pain treatments are contraindicated in adults with frailty, healthcare providers should consider referring to OT/PT because this nonpharmacologic treatment predicted opioid discontinuation better than non-opioid pain medications. Patient and caregiver preferences may also favor nonpharmacologic treatment. Caregivers, both formal in long-term care and informal in the community, often manage multiple medications for those with ADRD.43 Informal caregivers face challenges due to limited time and training to manage pain and related behaviors. To support caregivers, the Center for Medicare and Medicaid Innovation Center implemented the Guiding an Improved Dementia Experience Model in 2024, which reimburses OT/PT for caregiver training.44 Although results are not yet available, healthcare providers should be prepared to expand their caregiver training services. For example, including OT/PT in geriatric clinics for caregiver training, pain evaluation and treatment, and improving daily functioning among people with ADRD could reduce caregiver burden.45

Randomized trials support the use of skilled rehabilitation to improve daily function among frail adults; however, there is limited evidence on how such interventions improve pain in ADRD populations.46,47 Among Medicare beneficiaries, participating in home health OT/PT upon opioid initiation is associated with an earlier opioid discontinuation.13 Similarly, higher doses of OT/PT have been associated with improved pain control regardless of daily opioid dosages.34 Despite this evidence, prescription opioids are more commonly used for acute pain than nonpharmacologic alternatives.35

The restrained use of nonpharmacologic alternatives may be a result of limited access to skilled rehabilitation services, particularly in rural areas. Our observation that less populated urban and rural areas had lower rates of opioid discontinuation supports previous research linking a shortage of nonpharmacologic pain treatment providers per capita in these regions to increased opioid use.48 Since adults with frailty need more time to rehabilitate compared to their non-frail counterparts, individuals with frailty are less likely to achieve their functional goals because they exhaust their insurance benefits sooner.49 Policy and payment incentives may be necessary to equip rural areas with adequate healthcare resources, as individuals who are frail, in pain, and possibly disabled face challenges in traveling to rehabilitation appointments.

Limitations

Our study has several limitations. First, due to data limitations, we were unable to measure the severity of pain, the invasiveness of post-fracture surgical treatment, or treatment preferences. To mitigate this residual confounding, we controlled for post-fracture orthopedic surgery, fracture type, post-fracture nursing facility admission, alternative pain treatments such as OT/PT, SNRIs, and muscle relaxants, and urbanicity because opioid deprescribing patterns vary geographically.41 Since NSAIDs are often purchased over-the-counter, their use was likely underreported and increased the risk for differential misclassification. An E Value analysis50 indicated an unmeasured confounder would need a hazard ratio confidence interval with a lower bound of 1.36 or 1.11 to explain away the frailty or ADRD association, respectively. It is unlikely that residual confounding would explain the significant association for frailty, as none of our covariates had such a strong association (HR 1.36). The association between ADRD and opioid discontinuation is less robust. Second, despite using validated claims-based algorithms for ADRD and frailty, we lacked performance-based measures of disability and cognition to determine ADRD stage. Measuring ADRD severity using post-acute care data could reduce exposure misclassification but would limit generalizability to community-dwelling adults. Outcome misclassification is a limitation of the gap-based discontinuation definition in survival analysis. To account for time when discontinuation could not occur, we truncated follow-up and censoring to day 335, and our results remained unchanged since ~99% had already discontinued. Third, our results may not generalize to managed care beneficiaries. Additionally, our reliance on Part D prescription fill data may preclude individuals with the most severe frailty, ADRD, or fractures if they remained in a hospital or post-acute care facility more than 30 days post-fracture. Nonetheless, our findings remain clinically relevant to older adults who experienced a fracture and were discharged to the community, long-term care, or assisted living within 30 days.

Conclusions

In accordance with existing pain management guidelines, approximately 8-in-every-10 Medicare beneficiaries discontinued opioids within 30 days of filling their first prescription. After controlling for sociodemographic factors, health status, healthcare utilization, and calendar year, we found that frailty, but not ADRD status, was associated with a lower likelihood of discontinuing prescription opioids. Initiating opioids in a more recent calendar year increased the likelihood of opioid discontinuation, which was likely a result of 2014 policy and 2016 practice guidelines that encouraged non-opioid pain management. Although individuals with frailty are vulnerable to opioid-related adverse events, they appear less likely to discontinue these high-risk medications compared to their non-frail counterparts. Our findings should encourage healthcare providers to consider low-risk alternatives to pain management to expedite opioid deprescribing when caring for people with frailty.

Supplementary Material

Supinfo

Key Points:

  • In accordance with recent opioid prescribing guidelines, about 8 in every 10 older adults discontinued opioids in the first month after a fracture.

  • Frailty status, but not Alzheimer’s Disease and Alzheimer’s Disease Related Dementias, delayed the opioid deprescribing process among older adults with acute pain.

Why Does this Paper Matter?

  • Opioid discontinuation became more common in recent years; however, some sociodemographic factors, mental health diagnoses, rurality, and frailty delayed discontinuation. Although non-opioid pain treatments may facilitate discontinuation, access to frequent physical and occupational therapy services can be challenging.

ACKNOWLEDGEMENTS

Sponsor’s Role:

This study was funded by the National Institute on Aging (R01AG081268 and R01AG081412) to Dr. Lin. Dr. Kim is supported by K24AG073527. Dr. Kim has been supported by the grants from the National Institute on Aging of the National Institutes of Health for unrelated work. He received personal fee from Alosa Health (ended on 12/31/2022) and VillageMD (ended on 12/13/2022) for unrelated work. The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Abbreviations:

ADRD

Alzheimer’s Disease and Alzheimer’s Disease Related Dementias

ICD-9

International Classification of Diseases, Ninth Revision

ICD-10

International Classification of Diseases, Tenth Revision

IPCW

Inverse Probability of Censoring Weights

NSAIDs

Non-steroidal anti-inflammatory drugs

OT/PT

Occupational Therapy and/or Physical Therapy

PPV

Positive Predictive Value

SNRIs

Serotonin-Norepinephrine Reuptake Inhibitors

Footnotes

Disclosures and financial benefits to the authors: Financial disclosure statements have been obtained, and no conflicts of interest have been reported by the authors or by any individuals in control of the content of this article.

Conflict of Interest: None declared.

REFERENCES

  • 1.Reider L, Pollak A, Wolff JL, Magaziner J, Levy JF. National trends in extremity fracture hospitalizations among older adults between 2003 and 2017. Journal of the American Geriatrics Society. 2021;69(9):2556–2565. [DOI] [PubMed] [Google Scholar]
  • 2.Torchia MT, Munson J, Tosteson TD, et al. Patterns of opioid use in the 12 months following geriatric fragility fractures: a population-based cohort study. Journal of the American Medical Directors Association. 2019;20(3):298–304. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Cupp MA, Beaudoin FL, Hayes KN, et al. Post-acute care setting after hip fracture hospitalization and subsequent opioid use in older adults. Journal of the American Medical Directors Association. 2023;24(7):971–977. e4. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Bhashyam AR, Basilico M, Weaver MJ, Harris MB, Heng M. Using historical variation in opioid prescribing immediately after fracture surgery to guide maximum initial prescriptions. Journal of orthopaedic trauma. 2019;33(4):e131. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Stepan JG, Lovecchio FC, Premkumar A, et al. Development of an institutional opioid prescriber education program and opioid-prescribing guidelines: impact on prescribing practices. JBJS. 2019;101(1):5–13. [DOI] [PubMed] [Google Scholar]
  • 6.OPEN: Opioid Prescribing Engagement Network. OPEN Prescribing Recommendations. Accessed August, 2021. 10.56137/OPEN.000054 [DOI]
  • 7.Kuo Y-F, Raji MA, Goodwin JS. Association of disability with mortality from opioid overdose among US Medicare adults. JAMA network open. 2019;2(11):e1915638-e1915638. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Seamans MJ, Carey TS, Westreich DJ, et al. Association of household opioid availability and prescription opioid initiation among household members. JAMA internal medicine. 2018;178(1):102–109. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Kuo YF, Raji MA, Liaw V, Baillargeon J, Goodwin JS. Opioid prescriptions in older medicare beneficiaries after the 2014 federal rescheduling of hydrocodone products. Journal of the American Geriatrics Society. May 2018;66(5):945–953. doi: 10.1111/jgs.15332 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Shah R, Raji MA, Westra J, Kuo Y-F. Association of co-prescribing of opioid and benzodiazepine substitutes with incident falls and fractures among older adults: a cohort study. BMJ Open. 2021;11(12):e052057. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Nguyen AV, Ross E, Westra J, et al. Opioid Utilization in Geriatric Patients After Operation for Degenerative Spine Disease. Journal of Neurosurgical Anesthesiology. 2020; [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Cook DJ, Kaskovich S, Pirkle S, et al. Benchmarks of duration and magnitude of opioid consumption after common spinal procedures: a database analysis of 47,823 patients. Spine. 2019;44(23):1668–1675. [DOI] [PubMed] [Google Scholar]
  • 13.Pritchard KT, Baillargeon J, Raji MA, Chou L-N, Downer B, Kuo Y-F. Association of occupational and physical therapy with duration of prescription opioid use after hip or knee arthroplasty: A retrospective cohort study of Medicare enrollees. Archives of Physical Medicine and Rehabilitation. 2021;102(7):1257–1266. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Cook DJ, Kaskovich SW, Pirkle SC, Mica MAC, Shi LL, Lee MJ. Benchmarks of duration and magnitude of opioid consumption after total hip and knee arthroplasty: a database analysis of 69,368 patients. J Arthroplasty. 2019;34(4):638–644. [DOI] [PubMed] [Google Scholar]
  • 15.Okike K, Chang RN, Chan PH, Paxton EW, Prentice HA. Prolonged Opioid Usage Following Hip Fracture Surgery in Opioid-Naïve Older Patients. J Arthroplasty. 2023; [DOI] [PubMed] [Google Scholar]
  • 16.U.S. Department of Health and Human Services. Pain Management Best Practices Inter-Agency Task Force Report: Updates, Gaps, Inconsistencies, and Recommendations. Accessed February, 26, 2024. https://www.hhs.gov/sites/default/files/pain-mgmt-best-practices-draft-final-report-05062019.pdf
  • 17.Mehta HB, Kuo Y-F, Raji M, Li S, Westra J, Goodwin JS. Time trends in opioid use by dementia severity in long-term care nursing home residents. Journal of the American Medical Directors Association. 2021;22(1):124–131. e1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Shen C, Zhao X, Dwibedi N, Wiener RC, Findley PA, Sambamoorthi U. Opioid use and the presence of Alzheimer’s disease and related dementias among elderly Medicare beneficiaries diagnosed with chronic pain conditions. Alzheimer’s & Dementia: Translational Research & Clinical Interventions. 2018;4:661–668. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Chronic Conditions Warehouse. Medicare Administrative Data User Guide Version 3.9. Accessed February, 2024. https://www2.ccwdata.org/documents/10280/19002246/ccw-medicare-data-user-guide.pdf
  • 20.Schneeweiss S, Rassen JA, Brown JS, et al. Graphical depiction of longitudinal study designs in health care databases. Annals of internal medicine. 2019;170(6):398–406. [DOI] [PubMed] [Google Scholar]
  • 21.U.S Department of Health and Human Services and Food and Drug Administration. National Drug Code Directory. Accessed October 17, 2023. https://www.accessdata.fda.gov/scripts/cder/ndc/index.cfm
  • 22.Wright NC, Daigle SG, Melton ME, Delzell ES, Balasubramanian A, Curtis JR. The design and validation of a new algorithm to identify incident fractures in administrative claims data. Journal of Bone and Mineral Research. 2019;34(10):1798–1807. [DOI] [PubMed] [Google Scholar]
  • 23.National Bureau of Economic Research. ICD-9-CM to and from ICD-10-CM and ICD-10-PCS crosswalk or general equivalence mappings. Accessed October 17, 2023, 2023. https://www.nber.org/research/data/icd-9-cm-and-icd-10-cm-and-icd-10-pcs-crosswalk-or-general-equivalence-mappings
  • 24.Rees-Punia E, Newton CC, Parsons HM, et al. Fracture risk among older cancer survivors compared with older adults without a history of cancer. JAMA oncology. 2023;9(1):79–87. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.U.S. Department of Agriculture. Measuring Rurality: Rural-Urban Continuum Codes. Accessed July 26, 2022. https://www.ers.usda.gov/data-products/rural-urban-continuum-codes/documentation/
  • 26.Gagne JJ, Glynn RJ, Avorn J, Levin R, Schneeweiss S. A combined comorbidity score predicted mortality in elderly patients better than existing scores. Journal of clinical epidemiology. 2011;64(7):749–759. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Min L, Tinetti M, Langa KM, Ha J, Alexander N, Hoffman GJ. Measurement of fall injury with health care system data and assessment of inclusiveness and validity of measurement models. JAMA network open. 2019;2(8):e199679-e199679. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Festa N, Price M, Moura LM, et al. Evaluation of Claims-Based ascertainment of Alzheimer disease and related dementias across health care settings. American Medical Association; 2022:e220653-e220653. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Kim DH, Schneeweiss S, Glynn RJ, Lipsitz LA, Rockwood K, Avorn J. Measuring frailty in Medicare data: development and validation of a claims-based frailty index. The Journals of Gerontology: Series A. 2018;73(7):980–987. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Kim DH, Glynn RJ, Avorn J, et al. Validation of a claims-based frailty index against physical performance and adverse health outcomes in the health and retirement study. The Journals of Gerontology: Series A. 2019;74(8):1271–1276. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Sison SDM, Shi SM, Oh G, Jeong S, McCarthy EP, Kim DH. Claims-based frailty index and its relationship with commonly used clinical frailty measures. The Journals of Gerontology: Series A. 2024;79(7) [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Agency for Healthcare Research and Quality. Clinical Classifications Software (CCS). Accessed October, 2023. https://hcup-us.ahrq.gov/toolssoftware/ccs/ccs.jsp#overview
  • 33.Nahin RL, Sayer B, Stussman BJ, Feinberg TM. Eighteen-Year Trends in the Prevalence of, and Health Care Use for, Noncancer Pain in the United States: Data from the Medical Expenditure Panel Survey. J Pain. Jul 2019;20(7):796–809. doi: 10.1016/j.jpain.2019.01.003 [DOI] [PubMed] [Google Scholar]
  • 34.Pritchard KT, Baillargeon J, Westra J, et al. The impact of high-versus low-dose home rehabilitation for functional independence after hip or knee replacement. Journal of the American Medical Directors Association. 2023; [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Pritchard KT, Baillargeon J, Lee W-C, Raji MA, Kuo Y-F. Trends in the Use of Opioids vs Nonpharmacologic Treatments in Adults With Pain, 2011–2019. JAMA Network Open. 2022;5(11):e2240612-e2240612. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Fine JP, Gray RJ. A proportional hazards model for the subdistribution of a competing risk. Journal of the American statistical association. 1999;94(446):496–509. [Google Scholar]
  • 37.Robins JM, Finkelstein DM. Correcting for noncompliance and dependent censoring in an AIDS clinical trial with inverse probability of censoring weighted (IPCW) log-rank tests. Biometrics. 2000;56(3):779–788. [DOI] [PubMed] [Google Scholar]
  • 38.Drug Enforcement Administration DoJ. Schedules of controlled substances: rescheduling of hydrocodone combination products from schedule III to schedule II. Final rule. Federal register. 2014;79(163):49661. [PubMed] [Google Scholar]
  • 39.Dowell D, Haegerich TM, Chou R. CDC guideline for prescribing opioids for chronic pain—United States, 2016. Jama. 2016;315(15):1624–1645. doi:doi: 10.1001/jama.2016.1464 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Liaw V, Kuo Y-F, Raji MA, Baillargeon J. Opioid Prescribing Among Adults With Disabilities in the United States After the 2014 Federal Hydrocodone Rescheduling Regulation. Public Health Reports. 2020;135(1):114–123. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41.Westra J, Raji M, Kuo Y-f. National patterns of cessation of prescription opioids among Medicare beneficiaries, 2013–2018. Medicine. 2022;101(34) [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42.Dunbar MS, Edelen MO, McMullen T, et al. Development and testing of a standardized pain interview assessment for use in post-acute care. Journal of the American Geriatrics Society. 2022;70(4):1035–1046. [DOI] [PubMed] [Google Scholar]
  • 43.O’Conor R, Russell AM, Pack A, et al. Managing medications among individuals with mild cognitive impairment and dementia: Patient-caregiver perspectives. Journal of the American Geriatrics Society. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44.Centers for Medicare and Medicaid Services. Guiding an Improved Dementia Experience (GUIDE) Model. Accessed August 1, 2024. https://www.cms.gov/priorities/innovation/innovation-models/guide
  • 45.Bennett S, Laver K, Voigt-Radloff S, et al. Occupational therapy for people with dementia and their family carers provided at home: a systematic review and meta-analysis. BMJ open. 2019;9(11):e026308. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46.Szanton SL, Thorpe RJ, Boyd C, et al. Community aging in place, advancing better living for elders: A bio-behavioral-environmental intervention to improve function and health-related quality of life in disabled older adults. Journal of the American Geriatrics Society. 2011;59(12):2314–2320. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47.Gill TM, Baker DI, Gottschalk M, Peduzzi PN, Allore H, Byers A. A program to prevent functional decline in physically frail, elderly persons who live at home. New England Journal of Medicine. 2002;347(14):1068–1074. [DOI] [PubMed] [Google Scholar]
  • 48.Karmali RN, Skinner AC, Trogdon JG, Weinberger M, George SZ, Hassmiller Lich K. The association between the supply of select nonpharmacologic providers for pain and use of nonpharmacologic pain management services and initial opioid prescribing patterns for Medicare beneficiaries with persistent musculoskeletal pain. Health services research. 2021;56(2):275–288. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 49.Falvey JR, Ye JZ, Parker EA, Beamer BA, Addison O. Rehabilitation outcomes among frail older adults in the United States. International journal of environmental research and public health. 2022;19(17):11021. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 50.VanderWeele TJ, Ding P. Sensitivity analysis in observational research: introducing the E-value. Annals of internal medicine. 2017;167(4):268–274. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supinfo
NIHMS2037490-supplement-Supinfo.pdf (294.6KB, pdf)

RESOURCES