Abstract
INTRODUCTION
Whether prescription opioid exposure, duration, and dose are associated with cognitive function remains inconclusive.
METHODS
A longitudinal cohort among 3097 older adults with chronic pain and without dementia was conducted using Health and Retirement Study (HRS) linked to Medicare data from 2006 to 2020. Prescription opioid exposure, cumulative use for ≥ 90 days, and high‐dose use (≥ 90 morphine milligram equivalents [MME] daily) were assessed biennially. Memory score and dementia probability were derived from HRS cognitive measures and analyzed using linear mixed‐effects models.
RESULTS
Adjusted memory decline and dementia probability were not statistically different between patients with (vs. without) opioid exposure and between patients with cumulative use for ≥ 90 days (vs. < 90 days) but were higher between participants with high‐dose opioid use (vs. low‐dose) at the end of the follow‐up.
DISCUSSION
Prescription opioid exposure and duration were not associated, but high‐dose opioid use was associated with greater memory decline and dementia probability.
Highlights
Opioid use versus no use was not related to memory decline and dementia probability.
Long‐term opioid use was not related to memory decline and dementia probability.
High‐dose opioid use was related to greater memory decline and dementia probability.
Keywords: dementia probability, memory decline, older adults, prescription opioid use
1. INTRODUCTION
Opioids have been considered a major component of comprehensive pain treatment for chronic pain among older individuals given their high vulnerability to the serious adverse effects (e.g., gastrointestinal bleeding) of non‐opioid analgesic and adjuvant analgesic tricyclic antidepressants. 1 , 2 , 3 Since 2012, the American Geriatric Society has recommended that non‐opioid analgesics and tricyclic antidepressants be avoided or used cautiously among older individuals, 4 leaving prescription opioids as one of the few available options for pain control. The use of prescription opioids in older adults increased from 11.0% in 2007 to 19.3% in 2015, 5 , 6 followed by a decline to 12.8% in 2021, 7 likely attributable to concerns for increasing prescription opioid misuse and overdose. 8
In addition to opioid misuse and overdose, concerns have also been raised that opioids may worsen cognitive function, particularly among older adults, who have the highest risk of cognitive decline and dementia of all age groups. 9 These concerns have been built on preclinical studies suggesting that opioids may modulate cognitive and decision‐making processes. 10 , 11 Clinical data also suggest that opioid exposure, particularly at high doses and for long durations, may impair cognition, presumably resulting from opioid‐induced neurotoxicity effects on the central nervous system, including mild confusion, hallucinations, and delirium. 11 , 12 , 13
Several clinical and observational studies have assessed the association between prescription opioid exposure and cognition. Yet the results of these studies are inconclusive, leaving clinicians and patients with uncertainty regarding the cognitive safety of opioid therapy. A systematic review of 10 clinical studies has shown that some studies demonstrated null associations between opioid exposure and cognition, while others reported mixed findings—either improvement or impairment of cognition. 14 These clinical studies have notable limitations, including small sample sizes, single clinical sites, and short follow‐up times for detecting changes in cognitive function. Several population‐based longitudinal studies have been conducted and report either a null association 15 , 16 or a positive association of cognitive decline or dementia risk with prescription opioid exposure, 17 , 18 long‐term opioid use (≥ 180 days), 19 and high doses. 20 , 21 When stratified by age, only patients aged between 75 and 80 years with opioid exposure had a significantly higher risk of incident dementia compared to those without opioid exposure. 17 These population‐based studies, however, did not account for pain and its severity, key factors associated with cognitive decline and use of opioid treatment, thus limiting the ability of the studies to associate cognitive decline or dementia risk solely with opioid use. 18 , 22 , 23
To address the inconclusive evidence and limitations of previous studies investigating opioid use and cognition, we conducted a longitudinal cohort study of US older adults in the Health and Retirement Study (HRS) with linked Medicare claims data between 2006 and 2020. This study assessed the associations of prescription opioid exposure, duration, and dose with memory function and probability of dementia, cognitive outcomes validated and derived from biennial HRS surveys among older adults with chronic pain but without dementia, while accounting for key confounders, notably pain intensity.
2. METHODS
2.1. Study design and data sources
We conducted a longitudinal cohort study of HRS participants who consented to the linkage of Medicare fee‐for‐service claims data from January 1, 2006, to December 31, 2020. The HRS is a nationally representative, longitudinal study that has conducted surveys biennially since 1992 among ≈ 40,000 US community dwellers aged ≥ 50 years. 24 We used the HRS surveys to measure cognitive outcomes and important confounders, such as pain intensity, physical function, and depressive symptoms. Medicare claims data contain fee‐for‐service enrollees’ medical billing records for Parts A (inpatient), B (outpatient), and D (prescription drugs), with the prescription drug program implemented in 2006. 25 We used Medicare Part D data to measure prescription opioid exposures. The Ohio State University's Institutional Review Board approved this study and waived the informed consent requirement because of the use of deidentified data. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for cohort studies.
2.2. Study sample
The study sample included HRS Medicare participants aged ≥ 65 years who entered the cohort in the 2008 or 2010 HRS wave (index wave). Between the index wave and prior wave, patients were required to (1) have no dementia, defined as having an HRS self‐reported cognitive summary score > 11 on a 27‐point scale or an HRS proxy‐reported score < 3 on an 11‐point scale, 26 or having no diagnosis of Alzheimer's disease or related dementias; and (2) have HRS‐reported pain or an inpatient or outpatient diagnosis of chronic pain to reduce confounding by pain indication. The first 2 years after the index HRS were designated as an index interval to measure opioid exposures. Exclusion criteria included (1) no continuous enrollment in Medicare Parts A, B, and D for ≥ 1 years during the 2 years before the index HRS; (2) receipt of hospice or palliative care before the index HRS and during the index 2‐year interval; and (3) discontinuous enrollment in Part D, death, or discontinuation of HRS interviews during the index 2‐year interval.
Eligible participants were followed from the end of the 2‐year index interval until the last observed HRS wave before death or administrative censoring (i.e., Medicare Part D enrollment or HRS discontinuation). Figure 1 shows the study sample selection details. The diagnostic and procedure codes for conditions and services considered in the sample selection are given in Table S1 in supporting information.
FIGURE 1.
Cohort inclusion flowchart for the study sample. ADRD, Alzheimer's disease and related dementias; HRS, Health and Retirement Study.
2.3. Key exposures
The three key exposures were prescription opioid exposure, duration, and dose (Table S2 in supporting information) assessed according to dispensed date, days’ supply, and dose strength of drugs from Medicare Part D prescription data. Because individual opioid exposure likely changes over time, we measured these opioid exposures as time‐varying variables at the index 2‐year interval and then biennially in the follow‐up. During each 2‐year interval, participants were classified according to whether they received ≥ 1 opioid prescription, whether they had a cumulative use for ≥ 90 days, and whether they had a mean daily opioid dose ≥ 90 morphine milligram equivalents (MME; Method S1 in supporting information). The cutoff points of 90 days and 90 MME daily were chosen to align with the definitions of long‐term and high‐dose opioid use, respectively, commonly adopted in the current literature and clinical guidelines. 27 , 28 The percentages of participants who received prescription opioids, had a cumulative use for ≥ 90 days, and had high‐dose opioids decreased over time (Table S3 in supporting information), with few (< 11) participants receiving high‐dose opioids after 8 years after the index wave (corresponding to the year 2016). Thus, analyses for high‐dose opioids were performed up to 8 years in the follow‐up.
RESEARCH IN CONTEXT
Systematic review: In a PubMed literature review, while identifying several clinical and observational studies that assessed the association between prescription opioid use and cognition, the results of these studies are inconclusive. Also, none of the studies had accounted for pain and its severity, key factors associated with cognitive decline and use of opioid treatment.
Interpretation: After adjustment for confounders, notably pain intensity, we found that prescription opioid use and duration were not associated, but high‐dose opioid use was associated with greater memory decline and dementia probability.
Future directions: Our results suggest that for older patients with healthy cognition, emphasizing opioid dose rather than use and duration may be the critical component to attenuating the risk of cognitive decline among older adults with chronic pain.
2.4. Cognitive outcomes
The two primary cognitive outcomes were continuous composite memory score and dementia probability score, estimated from HRS data on cognitive measures and neuropsychological tests using the methods developed by Wu et al. 29 The composite memory score is a summary metric that combines multiple self‐report cognitive assessments and proxy‐report assessments for severely impaired HRS participants, with lower scores indicating worse cognition. The dementia probability score represents the chance that an individual would meet the Diagnostic and Statistical Manual of Mental Disorders (DSM), 3rd Edition Revised or Fourth Edition criteria for dementia, with a possible score ranging from 0 (no chance of dementia) to 1 (certain to have dementia). Both cognitive outcome measures have been validated 29 and commonly used in the literature 23 , 30 , 31 , 32 because of their advantages in reducing attrition bias and misclassification, improving statistical power, and increasing sensitivity to cognitive changes over time. 29
The constructs of composite memory score and dementia probability were detailed in Method S2 in supporting information. In brief, both measures were initially developed in the Aging, Demographics, and Memory Study (ADAMS), a subsample of the HRS population; 29 all ADAMS participants received in‐person comprehensive neuropsychological tests and clinical experts’ assessments of dementia, in addition to HRS core biennial assessments. 33 The memory score was derived from a prediction model with the ADAMS‐assessed composite memory score as the continuous outcome and dementia probability was derived from a prediction model with expert‐confirmed diagnosis of dementia as a dichotomous outcome. In both models, predictors included HRS cognitive measures and other measures (age, sex, and race) as the independent variables. Each prediction model yielded acceptable discrimination (r 2 for the memory score, 0.61; C statistic for the dementia probability, 94%). 29 We used the published beta coefficients derived from the ADAMS‐HRS prediction models and applied them to the HRS core interview respondents from 2008 and 2020 to generate estimates of the composite memory score (range, –1.21 to 2.12) and dementia probability (range, 0–1) for the entire HRS study sample. To compare memory changes over time across individuals, we created a composite memory z score by standardizing each wave's memory score during the follow‐up according to the mean (standard deviation [SD]) of the index HRS wave (0.87 [0.45]).
2.5. Covariates
A total of 36 potential covariates was identified from the HRS survey and Medicare claims data (Table S4 in supporting information). These variables included demographic characteristics, socioeconomic variables, lifestyle factors, health indicators, comorbidities, health care use, pain management, use of central nervous system medications or anticholinergic medications that could affect cognition, and cognitive function at the index HRS. These variables are either associated with cognitive outcomes of interest (potential confounders) or associated with both opioid exposures and cognitive outcomes (true confounders), which were recommended to be included in propensity score models to account for confounding bias. 34 , 35 Self‐reported race and ethnicity were examined given racial differences in cognitive decline 36 and were based on the Medicare–Research Triangle Institute race code. Health indicators included four key confounders associated with both opioid use and cognitive function: (1) pain intensity (classified into no‐to‐mild vs. moderate‐to‐severe pain); (2) depression (8‐item Center for Epidemiologic Studies Depression score, ≥ 4 vs. < 4); 37 (3) physical impairment (difficulty in performing ≥ 3 of 7 activities of daily living) or physical dependence (difficulty in performing ≥ 1 of 5 instrumental activities of daily living); 38 and (4) restless sleep (yes vs. no), all of which were derived from HRS survey data. These four health indicators along with marital status, income status, and total number of comorbidities were treated as time‐varying confounders. A small percentage (2.3%) of participants with missing data for covariates on the index HRS were excluded from analyses. To minimize missingness in time‐varying HRS‐assessed covariates during follow‐up (2.1% of total HRS waves), if data were missing, the value of the prior HRS wave was carried forward. 39
2.6. Statistical analysis
We compared baseline covariates of participants according to the status of prescription opioid exposure, cumulative use of ≥ 90 days, and high‐dose opioid use measured during the index 2‐year interval. Differences in baseline variables between groups were accounted for via inverse probability of treatment weighting (IPTW), calculated as the inverse of the propensity score for the exposed group (i.e., opioid use) and the inverse of 1 minus the propensity score for the non‐exposed group (i.e., no opioid use). Propensity score was estimated using a logistic regression that modeled the probability of being assigned to the opioid exposed versus non‐exposed group as the dependent variable, and baseline covariates as the independent variables. 35
To account for time‐varying confounders that simultaneously acted as confounders and intermediate variables, we used the marginal structural model (MSM) approach. 40 In MSM, we estimated the treatment weights by fitting pooled multivariable logistic regression models at each 2‐year interval, with the opioid exposure of interest as the dependent variable, and baseline and time‐varying variables as independent variables. Weights were truncated at the 1st and 99th percentiles to reduce the influence of outliers on estimates (Table S5 in supporting information).
To examine the associations of opioid exposures with the subsequent memory z score and dementia probability, we fitted a separate linear mixed‐effects model with MSM weights for each outcome that included opioid exposure, time since the index HRS survey, and the interaction of both variables. Time since the index HRS wave was used as the underlying time scale because (1) opioid prescribing declined over the follow‐up period 41 during which HRS surveys were conducted; and (2) both opioid exposures and cognitive outcomes were assessed at 2‐year intervals corresponding to the year of biennial HRS surveys. Because the continuous dementia probability score has a restricted range from 0 to 1, following previous studies, 23 , 30 we modeled the probability score on a log‐odds scale and then transformed it back to a probability scale. In all models, we treated individual‐level intercepts as random effects and applied the baseline HRS survey weights. To facilitate interpretations, we computed adjusted estimates and confidence intervals (CIs) of memory z score and dementia probability score according to opioid exposure status and differences in adjusted cognitive outcomes between groups over time.
2.7. Sensitivity analysis
We performed three sensitivity analyses by (1) additionally adjusting for competing risk of death by the inverse probability of censoring weights (IPCWs) in all models; (2) requiring dispensing of ≥ 2 opioid prescriptions during each 2‐year interval to define participants with versus without opioid exposure; (3) restricting participants to opioid‐naive individuals who had no opioid prescription fill within 1 year before the index HRS survey; and (4) stratifying associations of prescription opioid exposure with cognitive outcomes by sex and race/ethnicity (categorized into Whites versus non‐Whites to ensure a sufficient number of patients within each subgroup to yield reliable results). The last stratification analysis was not conducted on long‐term and high‐dose opioid use given small numbers of male and non‐White patients with these opioid exposures. All analyses were using SAS, version 9.4 (SAS Institute Inc.). Statistical significance was set at P < 0.05, and all tests were two‐sided.
3. RESULTS
We identified 3097 HRS Medicare participants with chronic pain and without dementia (mean [SD] age, 73.2 [5.4] years; 2082 [67.2%] female and 1015 [32.8%] male; Table 1). The median (interquartile range) follow‐up was 6 (4–8) years and was similar between participants based on the opioid exposures of interest (Table S6 in supporting information). Rates of loss to follow‐up were similar for participants with versus without opioid exposure but differed between participants with versus without cumulative use of ≥ 90 days and between those with versus without high‐dose opioid use, which were adjusted via IPCW in a sensitivity analysis.
TABLE 1.
Baseline demographic and clinical characteristics of the HRS Medicare study participants overall and stratified by with versus without prescription opioid receipt, defined during the index 2‐year interval.
| Participants, no. (%) | SDiff b | ||||
|---|---|---|---|---|---|
| Baseline characteristic a | Overall sample (n = 3097) | With opioid receipt (n = 1663) | Without opioid receipt (n = 1434) | Before IPTW | After IPTW |
| Age, years | |||||
| Mean (SD) | 73.2 (6.4) | 73.2 (6.4) | 73.4 (6.5) | 0.048 | 0.012 |
| 65–74 | 1956 (63.2) | 1070 (64.4) | 886 (61.8) | 0.053 | 0.015 |
| 75–84 | 945 (30.5) | 498 (29.9) | 447 (31.2) | 0.027 | 0.002 |
| ≥85 | 196 (6.3) | 95 (5.7) | 101 (7.0) | 0.055 | 0.027 |
| Sex | 0.127 | 0.008 | |||
| Female | 2082 (67.2) | 1164 (70.0) | 918 (64.0) | ||
| Male | 1015 (32.8) | 499 (30.0) | 516 (36.0) | ||
| Race and ethnicity | |||||
| White | 2368 (76.4) | 1251 (75.2) | 1117 (77.9) | 0.063 | 0.074 |
| Black | 367 (11.8) | 212 (12.7) | 155 (10.8) | 0.060 | 0.002 |
| Other c | 362 (11.7) | 200 (12.1) | 162 (11.3) | 0.023 | 0.095 |
| US region | |||||
| Northeast | 406 (13.1) | 187 (11.2) | 219 (15.3) | 0.119 | 0.010 |
| Northcentral | 784 (25.3) | 393 (23.6) | 391 (27.3) | 0.083 | 0.006 |
| West | 603 (19.5) | 327 (19.7) | 276 (19.2) | 0.011 | 0.006 |
| South | 1304 (42.1) | 756 (45.5) | 548 (38.2) | 0.147 | 0.008 |
| Dual eligibility (yes) | 577 (18.7) | 389 (23.4) | 188 (13.1) | 0.268 | 0.010 |
| Education level | |||||
| < High school | 849 (27.4) | 487 (29.3) | 360 (25.1) | 0.094 | 0.008 |
| High school or equivalent | 1664 (53.7) | 895 (53.8) | 769 (53.6) | 0.004 | 0.002 |
| ≥College | 586 (18.9) | 281 (16.9) | 305 (21.3) | 0.111 | 0.006 |
| Marital status | |||||
| Married or partnered | 1733 (56.0) | 889 (53.5) | 844 (58.9) | 0.109 | 0.013 |
| Separated or divorced | 345 (11.1) | 196 (11.8) | 149 (10.4) | 0.045 | 0.013 |
| Widowed or never married | 1019 (32.9) | 578 (34.7) | 441 (30.8) | 0.085 | 0.022 |
| Household income | |||||
| ≤$14,000 | 715 (23.1) | 443 (26.7) | 272 (19.0) | 0.185 | 0.015 |
| $14,001–$26,000 | 787 (25.4) | 420 (25.2) | 367 (25.6) | 0.009 | 0.003 |
| $26,001–$50,000 | 844 (27.2) | 411 (24.7) | 433 (30.2) | 0.123 | 0.002 |
| ≥50,001 | 751 (24.2) | 389 (23.4) | 362 (25.2) | 0.043 | 0.009 |
| Self‐reported health status | |||||
| Poor or fair | 1103 (35.6) | 699 (42.0) | 404 (28.2) | 0.292 | 0.000 |
| Good | 1068 (34.4) | 569 (34.1) | 499 (34.8) | 0.013 | 0.004 |
| Very good or excellent | 926 (29.9) | 395 (23.7) | 531 (37.0) | 0.292 | 0.004 |
| Body mass index | |||||
| Underweight or normal weight | 925 (29.9) | 461 (27.7) | 464 (32.4) | 0.101 | 0.015 |
| Overweight | 1171 (37.8) | 595 (35.8) | 576 (40.1) | 0.092 | 0.001 |
| Obese | 1001 (32.3) | 607 (36.5) | 394 (27.2) | 0.194 | 0.015 |
| Ever smoker | 1679 (54.2) | 929 (55.9) | 750 (52.3) | 0.071 | 0.018 |
| Ever consumed alcohol | 1310 (42.3) | 646 (38.8) | 664 (46.3) | 0.151 | 0.003 |
| Self‐reported depression (CES‐D > 4) | 337 (10.9) | 235 (14.1) | 102 (7.1) | 0.229 | 0.015 |
| Any physical impairment or dependence | 516 (16.7) | 341 (20.5) | 175 (12.2) | 0.224 | 0.003 |
| Restless sleep | 1095 (35.4) | 680 (40.9) | 415 (28.9) | 0.244 | 0.003 |
| Pain intensity | 0.445 | 0.022 | |||
| No or mild | 2046 (66.1) | 941 (56.6) | 1105 (77.1) | ||
| Moderate or severe | 1051 (33.9) | 722 (43.4) | 329 (22.9) | ||
| Chronic pain diagnosis | |||||
| Any pain | 2109 (68.1) | 1134 (68.2) | 975 (68.0) | 0.004 | 0.010 |
| Musculoskeletal pain | 2056 (66.4) | 1119 (67.3) | 937 (65.3) | 0.041 | 0.010 |
| Neuropathic pain | 715 (23.1) | 441 (26.5) | 274 (19.1) | 0.177 | 0.014 |
| HRS‐assessed comorbidity | |||||
| Arthritis | 2298 (74.2) | 1328 (79.8) | 970 (67.6) | 0.280 | 0.002 |
| Cancer | 539 (17.4) | 306 (18.4) | 233 (16.2) | 0.569 | 0.005 |
| Diabetes | 728 (23.5) | 432 (26.0) | 296 (20.6) | 0.128 | 0.009 |
| Heart problems | 924 (29.8) | 544 (32.7) | 380 (26.5) | 0.135 | 0.012 |
| High blood pressure | 2081 (67.2) | 1156 (69.5) | 925 (64.5) | 0.107 | 0.014 |
| Lung disease | 355 (11.5) | 230 (13.8) | 125 (8.7) | 0.162 | 0.014 |
| Psychiatric problem | 498 (16.1) | 333 (20.0) | 165 (11.5) | 0.235 | 0.020 |
| Stroke | 287 (9.3) | 179 (10.8) | 108 (7.5) | 0.112 | 0.007 |
| Incontinence | 971 (31.3) | 573 (34.4) | 398 (27.8) | 0.145 | 0.011 |
| Injury due to fall | 1234 (39.8) | 735 (44.2) | 498 (34.7) | 0.196 | 0.025 |
| Total number of comorbidities, Mean (SD) | 3.2 (1.6) | 3.4 (1.0) | 2.9 (1.5) | 0.400 | 0.017 |
| Health care use | |||||
| Any hospital stay | 1038 (33.5) | 678 (40.7) | 360 (25.1) | 0.336 | 0.009 |
| Any nursing home stay | 123 (4.0) | 85 (5.1) | 38 (2.6) | 0.128 | 0.019 |
| Pain management | |||||
| Use of non‐pharmacological therapy for pain management | 815 (26.3) | 489 (29.4) | 326 (22.7) | 0.152 | 0.004 |
| Use of adjuvant analgesic | 722 (23.3) | 502 (30.2) | 220 (15.3) | 0.360 | 0.007 |
| Use of non‐opioid | 1431 (46.2) | 1002 (60.3) | 429 (29.9) | 0.640 | 0.008 |
| Medication use | |||||
| Polypharmacy | 2299 (74.2) | 1380 (83.0) | 919 (64.0) | 0.440 | 0.004 |
| Use of other CNS drug | 753 (24.3) | 514 (30.9) | 239 (16.7) | 0.341 | 0.002 |
| Use of anticholinergic drug | 672 (21.7) | 458 (27.5) | 214 (14.9) | 0.312 | 0.035 |
| Indexed memory z score, mean (SD) | 0.88 (0.45) | 0.87 (0.44) | 0.88 1(0.46) | 0.000 | 0.001 |
Abbreviations: CES‐D, Center for Epidemiological Studies‐Depression; CNS, central nervous system; HRS, Health and Retirement Study; IPTW, inverse probability of treatment weighting; SD, standard deviation; SDiff, standardized mean difference.
HRS‐assessed characteristics and cognitive function were derived from the index HRS wave; Medicare‐assessed use and medication use were measured during 1 year before the index HRS wave.
Standardized mean difference > 0.10 indicated imbalance between the groups for the measured characteristic.
Included Asian, Pacific Islander, and Native American individuals and Hispanic individuals.
Of 3097 participants, 1664 (53.7%) received at least 1 prescription opioid, 493 (15.9%) had cumulative use for ≥ 90 days, and 238 (7.7%) had high‐dose opioid use (≥ 90 MME daily) during the index 2‐year interval (Table 1; Tables S7 and S8 in supporting information). After IPTW, the distribution of all measured baseline characteristics was well balanced between participants with versus without opioid exposure defined during the index 2‐year interval, with standardize mean differences (SMDs) for all characteristics < 0.100 (Table 1). An imbalance in four characteristics with SMDs > 0.01 was observed between patients with versus without cumulative opioid use of ≥ 90 days, and an imbalance in diabetes status was observed between patients with versus without high‐dose opioid use (Tables S7, S8). These imbalanced baseline covariates were double adjusted as covariates in the outcome models. 42
3.1. Composite memory z score and dementia probability over time
Among older participants with chronic pain and without dementia, the crude composite memory z score decreased, whereas the crude dementia probability increased over time in both groups by opioid exposure of interest defined during the index 2‐year interval. In MSM linear mixed‐effects models that accounted for time‐varying opioid exposures and time‐varying confounders, participants with (vs. without) prescription opioid exposure had no difference in memory decline (−0.04 standardized units [SUs] [95% CI: −0.09 to 0.16]; P = 0.58; Table 2 and Figure 2A) and dementia probability (−0.015 [95% CI: −0.059 to 0.028]; P = 0.49) at the 12‐year follow‐up (Table 3 and Figure 2B). Similarly, we observed no difference in cognitive decline (−0.05 SUs [95% CI: −0.26 to 0.15]; P = 0.61) and dementia probability (−0.014 [95% CI: −0.052 to 0.025]; P = 0.49) for participants with cumulative opioid use for ≥ 90 days (vs. < 90 days or no use) at the 12‐year follow‐up (Tables 2, 3 and Figure 2C,D).
TABLE 2.
Linear mixed‐effects regression analysis for composite memory z score over time by prescription opioid exposure status among older HRS Medicare beneficiaries with chronic pain and without dementia.
| Time since index HRS, years | Crude mean memory z score (95% CI) | Crude between‐group difference in memory z score | Adjusted mean memory z score (95% CI) | Adjusted between‐group difference in memory z score | ||||
|---|---|---|---|---|---|---|---|---|
| Patients with opioid use a | Patients without opioid use a | Absolute difference (95% CI) | p value | Patients with opioid use b | Patients without opioid use b | Absolute difference (95% CI) | p value | |
| 2 | −0.36 (−0.44, −0.29) | −0.36 (−0.43, −0.28) | −0.01 (−0.01, −0.10) | 0.88 | −0.34 (−0.40, −0.27) | −0.38 (−0.45, −0.32) | 0.05 (−0.02, 0.12) | 0.19 |
| 4 | −0.63 (−0.70, −0.55) | −0.59 (−0.67, −0.51) | −0.03 (−0.15, 0.08) | 0.55 | −0.63 (−0.69, −0.56) | −0.60 (−0.66, −0.53) | −0.03 (−0.10, 0.04) | 0.40 |
| 6 | −1.00 (−1.08, −0.92) | −0.85 (−0.93, −0.77) | −0.15 (−0.27, −0.04) | 0.01 | −0.97 (−1.03, −0.90) | −0.88 (−0.95, −0.81) | −0.09 (−0.17, 0.00) | 0.05 |
| 8 | −1.31 (−1.39, −1.23) | −1.18 (−1.27, −1.10) | −0.13 (−0.25, −0.01) | 0.04 | −1.25 (−1.32, −1.18) | −1.24 (−1.31, −1.16) | −0.01 (−0.10, 0.07) | 0.77 |
| 10 | −1.43 (−1.52, −1.34) | −1.37 (−1.46, −1.27) | −0.06 (−0.19, 0.07) | 0.36 | −1.38 (−1.47, −1.30) | −1.37 (−1.45, −1.29) | −0.02 (−0.11, 0.09) | 0.83 |
| 12 | −1.83 (−1.93, −1.73) | −1.73 (−1.83, −1.63) | −0.10 (−0.25, 0.04) | 0.15 | −1.71 (−1.83, −1.60) | −1.75 (−1.83, −1.67) | −0.04 (−0.09, 0.16) | 0.58 |
| Patients with cumulative use ≥ 90 days a | Patients with cumulative use < 90 days a | Absolute difference (95% CI) | P value | Patients with cumulative use ≥ 90 days b | Patients with cumulative use < 90 days b | Absolute difference (95% CI) | p value | |
|---|---|---|---|---|---|---|---|---|
| 2 | −0.42 (−0.55, −0.28) | −0.35 (−0.41, −0.29) | −0.07 (−0.22, 0.08) | 0.38 | −0.22 (−0.33, −0.11) | −0.39 (−0.46, −0.32) | 0.17 (0.07, 0.27) | <0.001 |
| 4 | −0.76 (−0.91, −0.62) | −0.58 (−0.64, −0.52) | −0.18 (−0.34, −0.03) | 0.02 | −0.59 (−0.70, −0.48) | −0.62 (−0.69, −0.55) | 0.03 (−0.07, 0.13) | 0.59 |
| 6 | −1.25 (−1.40, −1.10) | −0.88 (−0.94, −0.81) | −0.37 (−0.54, −0.21) | <0.001 | −1.02 (−1.13, −0.91) | −0.91 (−0.99, −0.84) | −0.11 (−0.21, 0.00) | 0.05 |
| 8 | −1.49 (−1.65, −1.33) | −1.21 (−1.28, −1.14) | −0.28 (−0.45, −0.11) | 0.002 | −1.20 (−1.32, −1.08) | −1.24 (−1.32, −1.17) | 0.05 (−0.07, 0.16) | 0.46 |
| 10 | −1.70 (−1.88, −1.52) | −1.35 (−1.42, −1.28) | −0.35 (−0.54, −0.16) | <0.001 | −1.36 (−1.50, −1.21) | −1.38 (−1.46, −1.30) | 0.02 (−0.12, 0.17) | 0.74 |
| 12 | −2.05 (−2.25, −1.85) | −1.74 (−1.82, −1.66) | −0.31 (−0.52, −0.09) | 0.005 | −1.80 (−2.01, −1.60) | −1.75 (−1.83, −1.66) | −0.05 (−0.26, 0.15) | 0.61 |
| Patients with opioid dose ≥ 90 MME/day a | Patients with opioid dose < 90 MME/day a | Absolute difference (95% CI) | P value | Patients with opioid dose ≥ 90 MME/day b | Patients with opioid dose < 90 MME/day b | Absolute difference (95% CI) | p value | |
|---|---|---|---|---|---|---|---|---|
| 2 | −0.34 (−0.53, −0.14) | −0.36 (−0.42, −0.31) | 0.03 (−0.18, 0.23) | 0.83 | −0.30 (−0.44, −0.17) | −0.43 (−0.49, −0.37) | 0.12 (−0.01, 0.25) | 0.07 |
| 4 | −0.60 (−0.80, −0.40) | −0.61 (−0.67, −0.55) | 0.01 (−0.20, 0.22) | 0.91 | −0.70 (−0.86, −0.53) | −0.67 (−0.73, −0.61) | −0.02 (−0.18, 0.14) | 0.77 |
| 6 | −1.13 (−1.34, −0.91) | −0.91 (−0.98, −0.86) | −0.22 (−0.44, 0.01) | 0.06 | −1.33 (−1.63, −1.03) | −0.99 (−1.05, −0.92) | −0.34 (−0.64, −0.04) | 0.03 |
| 8 | −1.55 (−1.78, −1.32) | −1.23 (−1.29, −1.17) | −0.32 (−0.56, −0.08) | 0.008 | −1.76 (−2.12, −1.40) | −1.30 (−1.37, −1.24) | −0.46 (−0.82, −0.10) | 0.01 |
Abbreviation: CI, confidence interval; HRS, Health and Retirement Study; MME, morphine milligram equivalents.
Prescription opioid exposures were defined during the index 2‐year interval.
Prescription opioid exposures were analyzed as time‐varying variables in the models.
FIGURE 2.
Adjusted composite memory z score and adjusted dementia probability during follow‐up periods by participant prescription opioid use, duration, and dose. MME, morphine milligram equivalents.
TABLE 3.
Linear mixed‐effects regression analysis for dementia probability score over time by prescription opioid exposure status among older HRS Medicare beneficiaries with chronic pain and without dementia.
| Time since index HRS, years | Crude mean dementia probability score (95% CI) | Crude between‐group difference in dementia probability score | Adjusted mean dementia probability score (95% CI) | Adjusted between‐group difference in dementia probability score | ||||
|---|---|---|---|---|---|---|---|---|
| Patients with opioid use a | Patients without opioid use a | Absolute difference (95% CI) | p value | Patients with opioid use b | Patients without opioid use b | Absolute difference (95% CI) | p value | |
| 2 | 0.003 (0.002, 0.004) | 0.002 (0.002, 0.003) | 0.001 (−0.001, 0.002) | 0.38 | 0.002 (0.002, 0.003) | 0.003 (0.002, 0.003) | 0.000 (−0.002, 0.001) | 0.73 |
| 4 | 0.006 (0.005, 0.008) | 0.004 (0.003, 0.006) | 0.002 (−0.001, 0.005) | 0.24 | 0.006 (0.004, 0.007) | 0.005 (0.004, 0.006) | 0.001 (−0.002, 0.004) | 0.72 |
| 6 | 0.018 (0.014, 0.024) | 0.008 (0.006, 0.011) | 0.010 (0.001, 0.037) | 0.04 | 0.014 (0.011, 0.018) | 0.010 (0.008, 0.013) | 0.004 (−0.003, 0.011) | 0.23 |
| 8 | 0.040 (0.031, 0.053) | 0.022 (0.016, 0.029) | 0.019 (0.001, 0.036) | 0.04 | 0.029 (0.023, 0.037) | 0.031 (0.024, 0.040) | −0.002 (−0.019, 0.014) | 0.79 |
| 10 | 0.046 (0.034, 0.062) | 0.034 (0.024, 0.046) | 0.012 (−0.009, 0.033) | 0.25 | 0.037 (0.027, 0.048) | 0.038 (0.029, 0.050) | −0.001 (−0.021, 0.018) | 0.89 |
| 12 | 0.111 (0.081, 0.150) | 0.075 (0.054, 0.105) | 0.036 (−0.013, 0.085) | 0.15 | 0.077 (0.051, 0.113) | 0.092 (0.071, 0.119) | −0.015 (−0.059, 0.028) | 0.49 |
| Patients with cumulative use ≥ 90 days a | Patients with cumulative use < 90 days a | Absolute difference (95% CI) | p value | Patients with cumulative use ≥ 90 days b | Patients with cumulative use < 90 days b | Absolute difference (95% CI) | p value | |
|---|---|---|---|---|---|---|---|---|
| 2 | 0.004 (0.002, 0.006) | 0.002 (0.002, 0.003) | 0.002 (−0.001, 0.002) | 0.06 | 0.002 (0.001, 0.003) | 0.003 (0.002, 0.003) | −0.001 (−0.002, 0.001) | 0.34 |
| 4 | 0.011 (0.007, 0.018) | 0.005 (0.004, 0.006) | 0.007 (−0.001, 0.005) | 0.001 | 0.006 (0.004, 0.009) | 0.005 (0.004, 0.006) | 0.001 (−0.002, 0.004) | 0.46 |
| 6 | 0.036 (0.022, 0.058) | 0.010 (0.008, 0.013) | 0.026 (0.001, 0.037) | <0.001 | 0.018 (0.013, 0.026) | 0.011 (0.009, 0.015) | 0.007 (−0.001, 0.015) | 0.09 |
| 8 | 0.065 (0.039, 0.106) | 0.026 (0.021, 0.032) | 0.039 (0.001, 0.036) | 0.001 | 0.025 (0.016, 0.037) | 0.029 (0.023, 0.037) | −0.004 (−0.018, 0.009) | 0.54 |
| 10 | 0.093 (0.053, 0.159) | 0.034 (0.027, 0.043) | 0.059 (−0.009, 0.033) | <0.001 | 0.027 (0.016, 0.045) | 0.038 (0.029, 0.049) | −0.011 (−0.027, 0.006) | 0.21 |
| 12 | 0.238 (0.133, 0.390) | 0.079 (0.062, 0.101) | 0.159 (−0.013, 0.085) | <0.001 | 0.075 (0.037, 0.145) | 0.088 (0.067, 0.114) | −0.014 (−0.052, 0.025) | 0.49 |
| Patients with opioid dose ≥90 MME/days a | Patients with opioid dose < 90 MME/days a | Absolute difference (95% CI) | p value | Patients with opioid dose ≥90 MME/days b | Patients with opioid dose < 90 MME/days b | Absolute difference (95% CI) | p value | |
|---|---|---|---|---|---|---|---|---|
| 2 | 0.003 (0.001, 0.005) | 0.002 (0.002, 0.003) | 0.001 (−0.001, 0.002) | 0.69 | 0.002 (0.001, 0.004) | 0.003 (0.003, 0.004) | −0.001 (−0.002, 0.001) | 0.31 |
| 4 | 0.006 (0.003, 0.012) | 0.005 (0.004, 0.006) | 0.001 (−0.002, 0.004) | 0.51 | 0.008 (0.004, 0.015) | 0.007 (0.006, 0.008) | 0.001 (−0.002, 0.005) | 0.50 |
| 6 | 0.022 (0.011, 0.044) | 0.012 (0.010, 0.014) | 0.010 (0.003, 0.017) | 0.007 | 0.055 (0.018, 0.154) | 0.016 (0.013, 0.019) | 0.040 (0.028, 0.050) | <0.001 |
| 8 | 0.085 (0.042, 0.165) | 0.038 (0.023, 0.034) | 0.057 (0.033, 0.081) | <0.001 | 0.206 (0.061, 0.350) | 0.038 (0.030, 0.046) | 0.168 (0.135, 0.202) | <0.001 |
Abbreviation: CI, confidence interval; HRS, Health and Retirement Study; MME, morphine milligram equivalents.
Prescription opioid exposures were defined during the index 2‐year interval.
Prescription opioid exposures were analyzed as time‐varying variables in the models.
Conversely, high‐dose opioid use (≥ 90 MME daily) was associated with greater memory decline and dementia probability beginning in the sixth year of follow‐up (Tables 2, 3; Figure 2E,F). By the eighth year of follow‐up, memory decline was 0.46 SUs higher (95% CI: −0.82 to −0.10; P = 0.01), and dementia probability was 16.8% higher (95% CI: 13.5% to 20.2%, P < 0.001) among patients with high‐dose opioid use versus patients with low‐dose or no opioid use.
3.2. Sensitivity analysis
We observed similar results in the sensitivity analysis of using ≥ 2 opioid prescriptions to define opioid users (Table S9 in supporting information) and after accounting for competing risk of death (Tables S10, S11 in supporting information). Among opioid‐naive participants, we observed consistent results—null associations of cognitive outcomes with opioid exposure and cumulative use for ≥ 90 days (Tables S12, S13 in supporting information). We examined high‐dose prescription use in the first 4 years of follow‐up given few (i.e., ≤ 10) opioid‐naive participants receiving high‐dose opioids after that period; we observed null associations of cognitive outcomes with high‐dose opioid use, consistent with the results of the first 4 years of follow‐up in the main analysis (Tables S12, S13). When stratified by age and race/ethnicity, the associations of prescription opioid exposure and cognitive outcomes were similar to that observed in the main analysis (Tables S14, S15 in supporting information).
4. DISCUSSION
In this longitudinal cohort study of older HRS Medicare participants with chronic pain and without dementia at baseline, receipt of prescription opioids and cumulative opioid use of ≥ 90 days were not associated with, but receipt of high‐dose opioids (≥ 90 MME daily) was associated with greater memory decline and dementia probability, after accounting for key time‐varying confounders (e.g., pain intensity). Our findings suggested that receipt of high‐dose prescription opioids, independent of pain intensity and mental and physical function, was associated with impaired cognitive function, whereas prescription opioid exposure and duration were not, among older adults with chronic pain.
While opioids have been speculated as a risk factor for cognitive decline among older adults, the evidence for this association remains inconclusive due to several research limitations. Paramount among the limitations is the inability to account for pain conditions and key confounders (e.g., pain intensity) 15 , 16 , 17 , 18 , 19 , 20 , 21 that affect the use of prescription opioids and are known risk factors of cognitive decline. 18 , 22 , 23 Failure to consider these confounders prevented determination of whether cognitive changes are due to chronic pain or to opioid therapy. These limitations may have led to the spurious positive associations of prescription opioid exposure and duration with cognitive decline observed in prior studies. 17 , 18 , 19 The present study is among the first to address confounding by pain indication and severity by including only patients with pain and accounting for time‐varying pain intensity. This study is further strengthened by considering a wealth of factors (e.g., depressive symptoms, physical function, and socioeconomic factors) associated with prescription opioid exposure and cognitive function.
The present study found that high‐dose opioid use (vs. low‐dose or no use) over 2 years was associated with an increase in subsequent memory decline and dementia probability after 6 years of follow‐up among older adults with chronic pain and without dementia at baseline. This finding echoes the results of two prior studies showing that total standardized doses > 90 (vs. 0–10) across 10 years or cumulative MME > 2940 (vs. no use) across 4 years was associated with higher dementia risk 21 or poor cognitive performance. 20 These two studies were conducted using old data (up to the year 2012). The positive association found in the present and prior studies is likely explained by neurotoxicity induced by use of high‐dose opioids, causing neurological symptoms (e.g., confusion or delirium) in the short term 11 , 12 , 13 , 43 and leading to acceleration of cognitive decline in the long term among older adults with chronic pain.
The present study provides referential data for clinicians and patients regarding the cognitive safety of prescription opioids for older adults with chronic pain. For older patients with healthy cognition, emphasizing opioid dose rather than use and duration may be the critical component to attenuating the risk of cognitive decline among older adults with chronic pain.
The key strength of this study is the use of both Medicare claims and HRS assessment data to longitudinally assess prescription opioid exposure, duration, and dose and their associations with cognitive function over a 12‐year period. The use of HRS linked to Medicare data also optimizes adjustments of potential variables, notably pain intensity, that are associated with prescription opioids and cognitive function. This study also has limitations. First, Medicare prescription drug event data provide information on prescription drugs dispensed but not consumed, a limitation shared in previous studies that rely on prescription dispensing data. 15 , 16 , 17 , 18 , 19 , 20 , 21 Second, while accounting for confounding by pain indication and pain intensity, our data have no information on the history of chronic pain before the index HRS, a potential factor associated with opioid use and cognitive function. Thus, our results remain subject to residual confounding. Third, Medicare prescription dispensing data do not capture self‐paid prescriptions or medications covered by non‐Medicare programs. Fourth, use of HRS‐estimated cognitive measures prevents examining the associations of prescription opioid exposures with specific cognitive domains (e.g., executive function). Fifth, given a high number of covariates (36) relative to a small number of patients exposed to high‐dose opioid use (n = 238), overfitting of pooled multivariable logistic regression models for estimating MSM treatment weights is possible. We have truncated MSM weights to reduce any overfitting issues. Future research using penalized regression approaches in MSM modeling to further reduce overfitting is encouraged to examine the robustness of our findings. Finally, our findings are generalizable only to older US community dwellers with a linkage of Medicare fee‐for‐service claims data. Thus, further studies should be conducted among Medicare Advantage enrollees, which has increased from 25% in 2010 to 54% in 2024 of all eligible Medicare beneficiaries. 44 , 45
5. CONCLUSIONS
This population‐based longitudinal cohort study found a positive association beginning in the sixth year of follow‐up for memory decline and dementia probability with use of high‐dose prescription opioids (≥ 90 MME daily) but found null associations with prescription opioid exposure or cumulative opioid use of ≥ 90 days over the 12‐year follow‐up period. These findings suggest limiting high‐dose prescription opioid exposure to slow cognitive decline among older adults with chronic pain.
CONFLICT OF INTEREST STATEMENT
All authors have no conflicts of interest. Author disclosures are available in the supporting information.
ROLE OF THE FUNDER/SPONSOR
The funder 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.
CONSENT STATEMENT
The Ohio State University's Institutional Review Board approved and waived patient informed consent and HIPAA authorization for this study because of minimal risk and lack of feasibility to contact Medicare subjects.
Supporting information
Supporting Information
Supporting Information
ACKNOWLEDGMENTS
This research was supported by grant R01AG073442 from the National Institute on Aging to Dr. Wei.
Wei Y‐JJ, Winterstein AG, Schmidt S, et al. Prescription opioid use and cognitive function in older adults with chronic pain: A population‐based longitudinal cohort study. Alzheimer's Dement. 2025;21:e70002. 10.1002/alz.70002
DATA AVAILABILITY STATEMENT
Data for these analyses are not available for sharing because the data are licensed to the authors through a data user agreement with the US National Institute on Aging (NIA). Individual researchers can access the data by purchasing a license through the NIA Data LINKAGE program at https://www.nia.nih.gov/research/dbsr/nia‐data‐linkage‐program‐linkage.
REFERENCES
- 1. Machado GC, Maher CG, Ferreira PH, et al. Efficacy and safety of paracetamol for spinal pain and osteoarthritis: systematic review and meta‐analysis of randomised placebo controlled trials. BMJ. 2015;350:h1225. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Wongrakpanich S, Wongrakpanich A, Melhado K, Rangaswami J. A comprehensive review of non‐steroidal anti‐inflammatory drug use in the elderly. Aging Dis. 2018;9:143‐150. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Hubbard R, Farrington P, Smith C, Smeeth L, Tattersfield A. Exposure to tricyclic and selective serotonin reuptake inhibitor antidepressants and the risk of hip fracture. Am J Epidemiol. 2003;158:77‐84. [DOI] [PubMed] [Google Scholar]
- 4. American Geriatrics Society Beers Criteria Update Expert P . American Geriatrics Society updated Beers Criteria for potentially inappropriate medication use in older adults. J Am Geriatr Soc. 2012;60:616‐631. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Jeffery MM, Hooten WM, Henk HJ, et al. Trends in opioid use in commercially insured and medicare advantage populations in 2007‐16: retrospective cohort study. BMJ. 2018;362:k2833. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Moriya AS, Fang Z. Any Use and “Frequent Use” of Opioids among Elderly Adults in 2018‐2019, by Socioeconomic Characteristics. Statistical Brief (Medical Expenditure Panel Survey (US)); 2001. [Google Scholar]
- 7. Moriya AS, Fang Z. Any Use and “Frequent Use” of Opioids among Adults Aged 65 and Older in 2020‐2021, by Socioeconomic Characteristics. Statistical Brief (Medical Expenditure Panel Survey (US)); 2001. [Google Scholar]
- 8. National Institute on Drug Abuse . Drug Overdose Death Rates. Accessed May 22, 2024. https://nida.nih.gov/research-topics/trends-statistics/overdose-death-rates
- 9. Ersek M, Cherrier MM, Overman SS, Irving GA. The cognitive effects of opioids. Pain Manag Nurs. 2004;5:75‐93. [DOI] [PubMed] [Google Scholar]
- 10. van Steenbergen H, Eikemo M, Leknes S. The role of the opioid system in decision making and cognitive control: a review. Cogn Affect Behav Neurosci. 2019;19:435‐458. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Moriarty O, McGuire BE, Finn DP. The effect of pain on cognitive function: a review of clinical and preclinical research. Prog Neurobiol. 2011;93:385‐404. [DOI] [PubMed] [Google Scholar]
- 12. Richards GC, Lluka LJ, Smith MT, et al. Effects of long‐term opioid analgesics on cognitive performance and plasma cytokine concentrations in patients with chronic low back pain: a cross‐sectional pilot study. Pain Rep. 2018;3:e669. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Dhingra L, Ahmed E, Shin J, Scharaga E, Magun M. Cognitive effects and sedation. Pain Med. 2015;16(1):S37‐43. [DOI] [PubMed] [Google Scholar]
- 14. Pask S, Dell'Olio M, Murtagh FEM, Boland JW. The Effects of opioids on cognition in older adults with cancer and chronic noncancer pain: a systematic review. J Pain Symptom Manage. 2020;59:871‐893 e1. [DOI] [PubMed] [Google Scholar]
- 15. Taipale H, Hamina A, Lampela P, et al. Is Alzheimer's disease associated with previous opioid use?. Pain Med. 2018;19:2115‐2121. [DOI] [PubMed] [Google Scholar]
- 16. Dublin S, Walker RL, Gray SL, et al. Use of analgesics (opioids and nonsteroidal anti‐inflammatory drugs) and dementia‐related neuropathology in a community‐based autopsy cohort. J Alzheimers Dis. 2017;58:435‐448. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Levine SZ, Rotstein A, Goldberg Y, Reichenberg A, Kodesh A. Opioid exposure and the risk of dementia: a national cohort study. Am J Geriatr Psychiatry. 2023;31:315‐323. [DOI] [PubMed] [Google Scholar]
- 18. Warner NS, Hanson AC, Schulte PJ, Habermann EB, Warner DO, Mielke MM. Prescription opioids and longitudinal changes in cognitive function in older adults: a population‐based observational study. J Am Geriatr Soc. 2022;70:3526‐3537. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Sun M, Chen WM, Wu SY, Zhang J. Long‐term opioid use and dementia risk in patients with chronic pain. J Am Med Dir Assoc. 2023;24:1420‐1426 e2. [DOI] [PubMed] [Google Scholar]
- 20. Neelamegam M, Zgibor J, Chen H, et al. The effect of opioids on the cognitive function of older adults: results from the personality and total health through life study. Age Ageing. 2021;50:1699‐1708. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Dublin S, Walker RL, Gray SL, et al. Prescription opioids and risk of dementia or cognitive decline: a prospective cohort study. J Am Geriatr Soc. 2015;63:1519‐1526. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Mailis‐Gagnon A, Lakha SF, Furlan A, Nicholson K, Yegneswaran B, Sabatowski R. Systematic review of the quality and generalizability of studies on the effects of opioids on driving and cognitive/psychomotor performance. Clin J Pain. 2012;28:542‐555. [DOI] [PubMed] [Google Scholar]
- 23. Whitlock EL, Diaz‐Ramirez LG, Glymour MM, Boscardin WJ, Covinsky KE, Smith AK. Association between persistent pain and memory decline and dementia in a longitudinal cohort of elders. JAMA Intern Med. 2017;177:1146‐1153. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Sonnega A, Faul JD, Ofstedal MB, Langa KM, Phillips JW, Weir DR. Cohort profile: the Health and Retirement Study (HRS). Int J Epidemiol. 2014;43:576‐585. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. The Centers for Medicare and Medicaid Services . Chronic Condition Data Warehouse Chronic Categories. The Centers for Medicare and Medicaid Services; 2020. Accessed March 19, 2024. https://www2.ccwdata.org/web/guest/condition-categories [Google Scholar]
- 26. Langa KM, Larson EB, Crimmins EM, et al. A Comparison of the prevalence of dementia in the United States in 2000 and 2012. JAMA Intern Med. 2017;177:51‐58. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Chou R, Turner JA, Devine EB, et al. The effectiveness and risks of long‐term opioid therapy for chronic pain: a systematic review for a National Institutes of Health Pathways to Prevention Workshop. Ann Intern Med. 2015;162:276‐286. [DOI] [PubMed] [Google Scholar]
- 28. Dowell D, Haegerich TM, Chou R. CDC guideline for prescribing opioids for chronic pain—United States, 2016. MMWR Recomm Rep. 2016;65:1‐49. [DOI] [PubMed] [Google Scholar]
- 29. Wu Q, Tchetgen Tchetgen EJ, Osypuk TL, White K, Mujahid M, Maria Glymour M. Combining direct and proxy assessments to reduce attrition bias in a longitudinal study. Alzheimer Dis Assoc Disord. 2013;27:207‐212. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Qian H, Khadka A, Martinez SM, et al. Food insecurity, memory, and dementia among US adults aged 50 years and older. JAMA Netw Open. 2023;6:e2344186. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Tang AB, Diaz‐Ramirez LG, Smith AK, Lee SJ, Whitlock EL. Preoperative factors predict memory decline after coronary artery bypass grafting or percutaneous coronary intervention in an epidemiological cohort of older Adults. J Am Heart Assoc. 2023;12:e027849. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Whitlock EL, Diaz‐Ramirez LG, Smith AK, et al. Association of coronary artery bypass grafting vs percutaneous coronary intervention with memory decline in older adults undergoing coronary revascularization. JAMA. 2021;325:1955‐1964. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Langa KM, Plassman BL, Wallace RB, et al. The aging, demographics, and memory study: study design and methods. Neuroepidemiology. 2005;25:181‐191. [DOI] [PubMed] [Google Scholar]
- 34. Brookhart MA, Schneeweiss S, Rothman KJ, Glynn RJ, Avorn J, Sturmer T. Variable selection for propensity score models. Am J Epidemiol. 2006;163:1149‐1156. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. Austin PC. An introduction to propensity score methods for reducing the effects of confounding in observational studies. Multivariate Behav Res. 2011;46:399‐424. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Weuve J, Barnes LL, Mendes de Leon CF, et al. Cognitive aging in Black and White Americans: cognition, cognitive decline, and incidence of Alzheimer disease dementia. Epidemiology. 2018;29:151‐159. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37. Turvey CL, Wallace RB, Herzog R. A revised CES‐D measure of depressive symptoms and a DSM‐based measure of major depressive episodes in the elderly. Int Psychogeriatr. 1999;11:139‐148. [DOI] [PubMed] [Google Scholar]
- 38. Brown RT, Diaz‐Ramirez LG, Boscardin WJ, Lee SJ, Williams BA, Steinman MA. Association of functional impairment in middle age with hospitalization, nursing home admission, and death. JAMA Intern Med. 2019;179:668‐675. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39. Allison PD. Survival Analysis Using SAS: A Practical Guide. 2nd ed. SAS Institute; 2010. [Google Scholar]
- 40. Robins JM, Hernan MA, Brumback B. Marginal structural models and causal inference in epidemiology. Epidemiology. 2000;11:550‐560. [DOI] [PubMed] [Google Scholar]
- 41. Bohnert ASB, Guy GP Jr, Losby JL. Opioid prescribing in the United States before and after the centers for disease control and prevention's 2016 opioid guideline. Ann Intern Med. 2018;169:367‐375. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42. Nguyen TL, Collins GS, Spence J, et al. Double‐adjustment in propensity score matching analysis: choosing a threshold for considering residual imbalance. BMC Med Res Methodol. 2017;17:78. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43. Gallagher R. Opioid‐induced neurotoxicity. Can Fam Physician. 2007;53:426‐427. [PMC free article] [PubMed] [Google Scholar]
- 44. Xu L, Welch WP, Ruhter J, et al. Medicare Advantage Overview: A Primer on Enrollment and Spending. Issue Brief. ASPE; 2023. Accessed May 20, 2024. https://aspe.hhs.gov/reports/medicare-advantage-overview [PubMed] [Google Scholar]
- 45. Freed MBJ, Damico A, Neuman T. Medicare Advantage in 2024: Enrollment Update and Key Trends. KFF; 2024. [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Supporting Information
Supporting Information
Data Availability Statement
Data for these analyses are not available for sharing because the data are licensed to the authors through a data user agreement with the US National Institute on Aging (NIA). Individual researchers can access the data by purchasing a license through the NIA Data LINKAGE program at https://www.nia.nih.gov/research/dbsr/nia‐data‐linkage‐program‐linkage.