Perioperative Gabapentin Use and In-Hospital Adverse Clinical Events Among Older Adults After Major Surgery

Key Points

Question  Is perioperative gabapentin use associated with in-hospital adverse clinical events among older adults after major surgery?

Findings  In this cohort study of 237 872 propensity score–matched adults aged 65 years or older, perioperative gabapentin users had significantly increased risk of delirium, new antipsychotic use, and pneumonia compared with nonusers after major surgery.

Meaning  This study suggests that careful risk-benefit assessment is needed before prescribing gabapentin for perioperative pain management to older patients.

Importance  Gabapentin has been increasingly used as part of a multimodal analgesia regimen to reduce opioid use in perioperative pain management. However, the safety of perioperative gabapentin use among older patients remains uncertain.

Objective  To examine in-hospital adverse clinical events associated with perioperative gabapentin use among older patients undergoing major surgery.

Design, Setting, and Participants  This retrospective cohort study using data from the Premier Healthcare Database included patients aged 65 years or older who underwent major surgery at US hospitals within 7 days of hospital admission from January 1, 2009, to March 31, 2018, and did not use gabapentin before surgery. Data were analyzed from June 14, 2021, to May 23, 2022.

Exposures  Gabapentin use within 2 days after surgery.

Main Outcomes and Measures  The primary outcome was delirium, identified using diagnosis codes, and secondary outcomes were new antipsychotic use, pneumonia, and in-hospital death between postoperative day 3 and hospital discharge. To reduce confounding, 1:1 propensity score matching was performed. Risk ratios (RRs) and risk differences (RDs) with 95% CIs were estimated.

Results  Among 967 547 patients before propensity score matching (mean [SD] age, 76.2 [7.4] years; 59.6% female), the rate of perioperative gabapentin use was 12.3% (119 087 patients). After propensity score matching, 237 872 (118 936 pairs) gabapentin users and nonusers (mean [SD] age, 74.5 [6.7] years; 62.7% female) were identified. Compared with nonusers, gabapentin users had increased risk of delirium (4040 [3.4%] vs 3148 [2.6%]; RR, 1.28 [95% CI, 1.23-1.34]; RD, 0.75 [95% CI, 0.75 [0.61-0.89] per 100 persons), new antipsychotic use (944 [0.8%] vs 805 [0.7%]; RR, 1.17 [95% CI, 1.07-1.29]; RD, 0.12 [95% CI, 0.05-0.19] per 100 persons), and pneumonia (1521 [1.3%] vs 1368 [1.2%]; RR, 1.11 [95% CI, 1.03-1.20]; RD, 0.13 [95% CI, 0.04-0.22] per 100 persons), but there was no difference in in-hospital death (362 [0.3%] vs 354 [0.2%]; RR, 1.02 [95% CI, 0.88-1.18]; RD, 0.00 [95% CI, –0.04 to 0.05] per 100 persons). Risk of delirium among gabapentin users was greater in subgroups with high comorbidity burden than in those with low comorbidity burden (combined comorbidity index <4 vs ≥4: RR, 1.20 [95% CI, 1.13-1.27] vs 1.40 [95% CI, 1.30-1.51]; RD, 0.41 [95% CI, 0.28-0.53] vs 2.66 [95% CI, 2.08-3.24] per 100 persons) and chronic kidney disease (absence vs presence: RR, 1.26 [95% CI, 1.19-1.33] vs 1.38 [95% CI, 1.27-1.49]; RD, 0.56 [95% CI, 0.42-0.69] vs 1.97 [95% CI, 1.49-2.46] per 100 persons).

Conclusion and Relevance  In this cohort study, perioperative gabapentin use was associated with increased risk of delirium, new antipsychotic use, and pneumonia among older patients after major surgery. These results suggest careful risk-benefit assessment before prescribing gabapentin for perioperative pain management.

Multimodal analgesia has been an increasingly adopted strategy in the Enhanced Recovery After Surgery pathway,¹ which aims to reduce opioid use by using nonopioid analgesia, such as regional or epidural analgesia, nonsteroidal anti-inflammatory drugs, acetaminophen, and gabapentinoids.² Randomized clinical trials (RCTs) have demonstrated that multimodal analgesia improves pain control^3-5 and decreases opioid use and its adverse effects.^2-7 The American Pain Society recommends gabapentin as a component of multimodal analgesia.⁸ Although the contribution of perioperative gabapentin use to the overall population trend remains unknown, gabapentin use tripled from 2002 to 2015 and has been the 10th most prescribed drug in the US since 2016.^9,10

Despite the widespread use of gabapentin, recent studies raised concerns about the marginal benefit and immediate harms of gabapentin use for perioperative pain management.¹¹ Studies,^12-14 including meta-analyses of RCTs, concluded that the evidence on gabapentin’s effectiveness is low quality owing to inconsistent and imprecise results and that its analgesic and opioid-sparing effects may be clinically insignificant. Moreover, perioperative gabapentin use was associated with dizziness and visual disturbances.¹³ Although there were no statistically significant differences in perioperative delirium, respiratory failure, ataxia, or falls,¹³ the studies did not exclude the possibility of clinically meaningful adverse events owing to small sample sizes, underrepresentation of older patients, and heterogenous surgical procedures. The American Geriatrics Society Beers Criteria¹⁵ lists gabapentin as a potentially inappropriate medication owing to its risk of sedation and respiratory depression, especially when used with opioids. Because older surgical patients are vulnerable to these adverse effects and are at increased risk of perioperative delirium, pneumonia, and death, there is a need to examine the safety of perioperative gabapentin use using a health care database of the general population that includes a large number of older adults undergoing different types of surgical procedures.

We conducted a retrospective cohort study to investigate the association of perioperative gabapentin use with in-hospital adverse clinical events using a nationwide administrative inpatient database of older adults undergoing major surgery. We hypothesized that gabapentin use would be associated with increased risk of delirium, pneumonia, and in-hospital death.

This retrospective cohort study using the Premier Healthcare Database was approved by the institutional review board of Brigham and Women’s Hospital, and a waiver of informed consent was obtained because the data were deidentified. The Premier Healthcare Database is a deidentified, hospital-based, service-level, all-payer database containing more than 900 small-sized to medium-sized hospitals that covers approximately 25% of annual inpatient admissions in the US. The database contains information on demographic characteristics, admission status, diagnosis codes, discharge status, and date-stamped records of drugs and procedures. This nationwide database has been used to investigate the safety of medical interventions in the inpatient care setting.^16-18 This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.

The analytic sample included adults aged 65 years or older who underwent major surgical procedures within 7 days of hospital admission from January 1, 2009, to March 31, 2018. Major surgical procedures included cardiac, gastrointestinal, genitourinary, orthopedic, neurological (excluding procedures involving the brain), thoracic, and vascular surgery, as defined by the Agency for Healthcare Research and Quality procedure classification.^19,20 We excluded patients who died or were discharged before or on postoperative day 2 because we defined gabapentin use based on the exposure on the day of surgery or on postoperative days 1 or 2 (exposure-defining period). Because gabapentin is typically started before surgery and continued postoperatively, we refer to this regimen as perioperative use. We also excluded patients who had diagnosis codes for psychosis, received antipsychotics before or on postoperative day 2, received gabapentin before the day of surgery, or had diagnosis codes for other potential indications for gabapentin (alcohol use disorder, alcohol withdrawal, fibromyalgia, neuropathic pain, postherpetic neuralgia, restless legs syndrome, seizure, and social anxiety disorder) or contraindications to gabapentin (myasthenia gravis). Moreover, patients who received critical care, mechanical ventilation, or a feeding tube before or on postoperative day 2 were excluded because they had a higher acuity of illness and were unlikely to receive gabapentin orally.

We defined perioperative gabapentin use based on charge codes, which identify each service item (eg, medications and procedures) for billing and reimbursement purposes, on the day of surgery or postoperative day 1 or 2. We calculated total daily gabapentin dose (in milligrams) given during the exposure-defining period. To avoid immortal time bias,²¹ the group that used gabapentin and the group that did not use gabapentin were required to have survived the exposure-defining period.

The following patient-level characteristics were assessed: age, sex, race and ethnicity (Black, White, or other [American Indian, Asian/Pacific Islander, Hispanic, and other]), insurance type (commercial, Medicare, Medicaid, or other), admission type (elective, urgent, emergent, or other) and source (outpatient, emergency department, transfer, or other), surgery type (cardiac, gastrointestinal, genitourinary, orthopedic, neurological, thoracic, or vascular), combined comorbidity index (scores range from −2 to 26, with higher scores indicating greater risk of death),²² and comorbidity diagnoses. Race and ethnicity were obtained from Uniform Billing Code of 1992 billing forms. Inpatient medication use, including analgesics (acetaminophen, cyclooxygenase-2 inhibitors, nonsteroidal anti-inflammatory drugs, and opioid use and dose in morphine milligram equivalent [MME] per day), and receipt of cardiopulmonary resuscitation, dialysis, or blood transfusion before surgery and during the first 2 postoperative days were measured. We also obtained hospital bed capacity, teaching status, location (urban or rural), and geographic region. To account for changes in gabapentin-prescribing patterns and surgical outcomes over time, we recorded calendar year of the hospital admission.

The primary outcome was delirium, identified using a validated claims-based algorithm.^23,24 This algorithm, which consists of explicit (ie, delirium is directly mentioned) and implicit (eg, encephalopathy) diagnosis codes of delirium (eTable 1 in the Supplement), has a positive predictive value of 80% against the Confusion Assessment Method, as validated in a previous study of 184 patients.²³ As secondary outcomes, we assessed (1) new antipsychotic use, which has a positive predictive value of 92% for delirium²³; (2) pneumonia, which was defined based on diagnosis codes plus intravenous antibiotic use or computed tomography of the chest; and (3) in-hospital death. Analogous to the intention-to-treat analysis in an RCT, patients were followed up from postoperative day 3 until the occurrence of the outcomes or hospital discharge regardless of the presence or duration of gabapentin therapy.

Data were analyzed from June 14, 2021, to May 23, 2022. To reduce confounding, we performed propensity score matching.^25,26 A propensity score was estimated as the probability of receiving gabapentin from a logistic regression model that included all patient- and hospital-level characteristics. We conducted 1:1 nearest-neighbor matching and assessed covariate balance based on the standardized mean difference (<0.1 was considered adequate).²⁵ We estimated risk ratios (RRs), risk differences (RDs), and 95% CIs for delirium, new antipsychotic use, pneumonia, and in-hospital death associated with gabapentin use. We also examined the association in subgroups defined by age (<80 or ≥80 years), sex, comorbidity burden (combined comorbidity index <4 or ≥4), chronic kidney disease status (presence or absence), opioid dose (MME<15 mg per day or ≥15 mg per day), and surgery type (cardiac, gastrointestinal, genitourinary, orthopedic, neurological, thoracic, or vascular). Within each subgroup, we reestimated the propensity score, and performed 1:1 matching; we tested heterogeneity across subgroups.²⁷ Two-sided P < .05 for heterogeneity was considered significant. We performed 3 sensitivity analyses regarding (1) gabapentin exposure present on the day of surgery, (2) dose-response relationship, and (3) unmeasured confounding. First, we repeated the analysis by defining gabapentin exposure on the day of surgery without requiring a 2-day minimum length of stay after surgery. Then, we explored a dose-response relationship using a 4-dose category of gabapentin (no use, 1 mg to <600 mg, 600 mg to <1200 mg, or ≥1200 mg). Because 4-group propensity score matching was not feasible, we used multivariable logistic regression models to adjust for covariates selected by a stepwise algorithm with a 2-sided P value threshold of 0.10 for entry and removal of a covariate from the model. In addition, because pain intensity may be an unmeasured confounder of the association between gabapentin use and delirium, we examined how the RR estimate would change under various scenarios: (1) the prevalence difference in severe pain between gabapentin users and nonusers and (2) the relative risk between severe pain and delirium from the literature (severe perioperative pain was associated with a 1.2-fold increase in delirium).²⁸ Analyses were performed using SAS, version 9.4 (SAS Institute Inc).

After applying inclusion and exclusion criteria, 967 547 patients (mean [SD] age, 76.2 [7.4] years; 59.6% female) were eligible to initiate gabapentin for perioperative pain management (Figure 1). Of these patients, 119 087 (12.3%; mean [SD] age, 74.5 [6.7] years; 62.7% female) received gabapentin between the day of surgery (108 190 [90.8%]) and 2 days after surgery. The most common surgery type among gabapentin users was orthopedic (91 014 [76.4%]). Compared with those who did not use gabapentin, gabapentin users were younger (mean [SD] age, 74.5 [6.7] vs 76.4 [7.5] years), were more likely to be female (74 627 [62.7%] vs 501 934 [59.2%]), underwent more elective surgical procedures (93 674 [78.7%] vs 422 262 [49.8%]), and had a lower comorbidity index (mean [SD], 1.2 [2.2] vs 1.8 [2.6]). However, gabapentin users were more frequently treated with analgesics, including opioids (109 389 [91.9%] vs 686 293 [80.9%]), a higher daily opioid dose (mean [SD], 7.6 [6.1] MME vs 5.8 [5.6] MME), and psychoactive drugs, such as antidepressants (29 772 [25.0%] vs 141 942 [16.7%]) and anxiolytics (84 529 [71.0%] vs 525 736 [62.0%]) (Table 1; the complete covariate list is given in eTable 2 in the Supplement). After 1:1 propensity score matching, we identified 237 872 (118 936 pairs) gabapentin user and nonusers. Baseline characteristics were well balanced (standardized mean difference, <0.1) in the propensity score–matched cohort (Table 1).

Before propensity score matching, gabapentin users had lower risk of adverse clinical events than nonusers (Table 2). After propensity score matching, gabapentin users had increased risk (RR and RD) of delirium (4040 [3.4%] vs 3148 [2.6%]; RR, 1.28 [95% CI, 1.23-1.34]; RD, 0.75 [95% CI, 0.61-0.89] per 100 persons), new antipsychotic use (944 [0.8%] vs 805 [0.7%]; RR, 1.17 [95% CI, 1.07-1.29]; RD, 0.12 [95% CI, 0.05-0.19] per 100 persons), and pneumonia (1521 [1.3%] vs 1368 [1.2%]; RR, 1.11 [95% CI, 1.03-1.20]; RD, 0.13 [95% CI, 0.04-0.22] per 100 persons) compared with nonusers. In-hospital death was similar between gabapentin users and nonusers (362 [0.3%] vs 354 [0.2%]; RR, 1.02 [95% CI, 0.88-1.18]; RD, 0.00 [95% CI, –0.04 to 0.05] per 100 persons).

Across the subgroups defined by age, sex, comorbidity burden, chronic kidney disease status, and opioid dose, the associations of gabapentin use with delirium, new antipsychotic use, and pneumonia were consistent (Figure 2 and Figure 3). The RR of delirium was greater among gabapentin users younger than 80 years than among those aged 80 years or older (1.34 [95% CI, 1.27-1.43] vs 1.21 [95% CI, 1.12-1.30]; P = .02 for heterogeneity), but the RD was similar (0.71 [95% CI, 0.57-0.86] vs 0.91 [95% CI, 0.56-1.27] per 100 persons; P = .31 for heterogeneity). The RR and RD were greater among patients with high comorbidity burden (≥4) than among those with low comorbidity burden (<4) (RR, 1.40 [95% CI, 1.30-1.51] vs 1.20 [95% CI, 1.13-1.27]; P = .001 for heterogeneity; RD, 2.66 [95% CI, 2.08-3.24] vs 0.41 [95% CI, 0.28-0.53] per 100 persons; P < .001 for heterogeneity) and among patients with chronic kidney disease (absence vs presence: RR, 1.26 [95% CI, 1.19-1.33] vs 1.38 [95% CI, 1.27-1.49]). Subgroup estimates differed only on the RD scale for risk of delirium by chronic kidney disease status (absence vs presence: RD, 0.56 [95% CI, 0.42-0.69] vs 1.97 [95% CI, 1.49-2.46] per 100 persons; P < .001 for heterogeneity), risk of new antipsychotic use by comorbidity burden (<4 vs ≥4: 0.09 [95% CI, 0.03-0.15] vs 0.65 [0.33-0.97] per 100 persons; P = .001 for heterogeneity), and risk of pneumonia by comorbidity burden (<4 vs ≥4: 0.15 [95% CI, 0.07-0.22] vs 0.66 [0.25-1.07] per 100 persons; P = .02 for heterogeneity) and chronic kidney disease status (absence vs presence: 0.14 [95% CI, 0.06-0.23] vs 0.47 [95% CI, 0.17-0.77] per 100 persons; P = .04 for heterogeneity). There was no statistically significant evidence for heterogeneity by sex or opioid dose on the RR or RD scale. Because of the small number of clinical events, certain surgery-specific estimates were imprecise, but in general, the associations were consistent with increased risk of outcomes with gabapentin use (eTable 3 in the Supplement). Perioperative gabapentin use was not associated with in-hospital death in all subgroups.

When gabapentin exposure was defined on the day of surgery without requiring a minimum 2-day length of stay after surgery, the results were unchanged from our primary analysis (eTable 4 in the Supplement). On multivariable logistic regression analysis (eTable 5 in the Supplement), increasing gabapentin dose was associated with progressively increased risk of delirium diagnosis (1 mg to <600 mg: adjusted odds ratio [AOR], 1.25 [95% CI, 1.18-1.32]; 600 mg to <1200 mg: AOR, 1.30 [95% CI, 1.24-1.36]; ≥1200 mg: AOR, 1.43 [95% CI, 1.28-1.60]) and pneumonia (1 mg to <600 mg: AOR, 1.05 [95% CI, 0.96-1.14]; 600 mg to <1200 mg: AOR, 1.16 [95% CI, 1.07-1.25]; ≥1200 mg: AOR, 1.34 [95% CI, 1.13-1.59]) but not with new antipsychotic use and in-hospital death. In a sensitivity analysis for unmeasured confounding (eFigure in the Supplement), we found that the RR of 1.28 in our study would become null if (1) the prevalence difference of severe pain was at least 30% between the treatment groups or (2) the RR between severe pain and delirium was at least 2.0, which was greater than previously reported in the literature (ie, >2.0).²⁸ These results suggest that unmeasured, severe pain alone is unlikely to explain our results.

In this cohort study, we found that perioperative gabapentin use was associated with modestly increased risk of delirium, new antipsychotic use, and pneumonia but not with in-hospital death among adults aged 65 years or older after major surgery. Considering the increasing number of major surgeries performed in older adults^29,30 and the negative consequences of perioperative delirium, our findings raise concern about an increasingly adopted clinical practice that involves routine use of gabapentin as part of multimodal analgesia. Our study provides evidence on the safety of perioperative gabapentin use in a representative population of older patients undergoing surgery as part of routine care.

To our knowledge, more than 200 RCTs have been conducted to evaluate the effect of gabapentin on perioperative pain control, reduction in opioid use, and adverse events. The trials varied in terms of sample size (20-697 patients),^31,32 gabapentin regimen (single dose vs continued treatment with daily dose ranging from 300 mg to 1200 mg), surgery type (orthopedic, abdominal, and vascular), and study quality (low to high risk of bias). Several meta-analyses^{3-5,12,13,33,34} concluded that reductions in pain intensity 24 hours after surgery and opioid-related adverse events associated with gabapentin and placebo were inconsistent and not clinically meaningful. An RCT by Hah et al³⁵ showed that perioperative gabapentin use had no effect on time to cessation of perioperative pain but reduced the median time to opioid cessation after surgery (25 days vs 32 days) compared with lorazepam. The rate of drug discontinuation owing to sedation or dizziness (25% in the gabapentin group and 20% in the lorazepam group) was not statistically significant, possibly because of the use of lorazepam as an active comparator. Some experts caution that the immediate harm of gabapentin may outweigh the long-term benefits of opioid cessation among older adults.¹¹ A recent meta-analysis¹³ found higher rates of dizziness and visual disturbances after use of gabapentin, with no statistically significant associations with respiratory failure, ataxia, falls, or delirium. However, owing to the small sample size and underrepresentation of older adults in RCTs, the safety of perioperative gabapentin use in this population remains uncertain.

The main site of action of gabapentin is on the α2-δ subunit of calcium channels that are found in the peripheral and central nervous system,^36,37 which may explain its adverse effects, such as dizziness, visual disturbance, sedation, and confusion.³⁷ The same mechanism may also explain the increased risk of delirium, new antipsychotic use, and aspiration pneumonia. In addition, because gabapentin is solely metabolized through the kidneys, adverse effects may lead to more severe clinical consequences (eg, delirium and respiratory depression)^3,38,39 among older patients with a higher prevalence of multimorbidity and chronic kidney disease. It was hypothesized that gabapentin may prevent delirium by improving pain control and reducing opioid dose, whereas gabapentin-related adverse effects may increase delirium.⁴⁰ To date, several studies investigated whether gabapentin could reduce perioperative delirium. In a post hoc analysis of an RCT of 161 patients (mean age, 63 years) undergoing total knee replacement, Dighe et al⁴¹ reported that 12.0% of patients in the gabapentin group and 9% of patients in the placebo group developed delirium (P = .53). In another RCT of 697 patients (mean age, 73 years) undergoing orthopedic surgery, Leung et al³² showed that 24.0% in the gabapentin group and 20.8% in the placebo group had delirium (P = .30). However, the difference in the delirium incidence was not statistically significant. The RR estimates of delirium from these studies were consistent with our findings.

Our study has limitations. The findings should be interpreted within the limitations of an administrative database study. First, delirium incidence in the present study population (3.4%) was lower compared with a previously reported incidence of 15% to 25% after major surgery⁴² owing to low sensitivity (18%) and high specificity (98%) of the present study’s delirium identification algorithm.²³ This delirium identification algorithm was better at identifying hyperactive delirium, which was associated with poorer prognosis compared with hypoactive or normoactive delirium.⁴³ For these reasons, our RD estimates may have been underestimated. Moreover, the diagnosis codes for delirium and pneumonia did not have an exact onset date in our data sets; thus, these outcomes may have been present before surgery in some patients. Second, confounding was possible. Gabapentin users were healthier and more likely to have an elective surgery compared with nonusers. After propensity score matching, the RR estimate increased from 0.84 to 1.28. If residual confounding was present in the same direction, the true RR would have been higher than the value after propensity score matching. Our sensitivity analysis suggests that unmeasured pain severity alone was unlikely to explain our results. Although the association between gabapentin and delirium did not differ by opioid dose, patients who received a higher dose had lower absolute risk of outcomes, which suggests that those patients may have been healthier than those who received a lower dose. Therefore, this subgroup analysis should be interpreted with caution. Third, outpatient medication use was unavailable in the Premier Healthcare Database. To identify patients eligible to newly receive gabapentin for perioperative pain management, we excluded those who received gabapentin before surgery, had other indications or contraindications for gabapentin, or received critical care, mechanical ventilation, or a feeding tube in the immediate perioperative period. In addition, we required at least a 2-day length of stay after surgery to define gabapentin exposure status, which may have limited the generalizability of our findings. However, the results did not change when we defined the exposure on the day of surgery without requiring a minimum length of stay after surgery.

In this cohort study, perioperative gabapentin use was associated with increased risk of delirium, new antipsychotic use, and pneumonia among older patients after major surgery. On the basis of these findings and those of meta-analyses of RCTs^12-14 showing a weak opioid-sparing effect of gabapentin, clinicians should reconsider routine use of gabapentin for perioperative pain management among older adults and individualize the treatment decision after assessing the risk of immediate harms vs opioid-sparing benefits of perioperative gabapentin use. For older patients who receive gabapentin as part of multimodal analgesia, daily assessment of the appropriateness of gabapentin use may be necessary to avoid unintended harm.

Accepted for Publication: July 3, 2022.

Published Online: September 19, 2022. doi:10.1001/jamainternmed.2022.3680

Correction: This article was corrected on January 9, 2023, to replace eTable 4 in the Supplement.

Corresponding Author: Dae Hyun Kim, MD, ScD, Harvard Medical School, 1620 Tremont St, Boston, MA 02120 (dkim12@bwh.harvard.edu).

Author Contributions: Dr. Kim had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Park, Inouye, Bateman, Kim.

Acquisition, analysis, or interpretation of data: Park, Marcantonio, Metzger, Bateman, Lie, Lee, Levin, Kim.

Drafting of the manuscript: Park, Lie, Kim.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Park, Levin, Kim.

Obtained funding: Kim.

Administrative, technical, or material support: Inouye, Lee, Kim.

Supervision: Inouye, Marcantonio, Kim.

Conflict of Interest Disclosures: Dr Marcantonio reported receiving grants from the National Institute on Aging (NIA). Dr Bateman reported receiving grants from the National Institutes of Health (NIH) during the conduct of the study. Dr Kim reported receiving personal fees from Alosa Health and VillageMD and receiving grants from the NIA, NIH, outside the submitted work and during the conduct of the study. No other disclosures were reported.

Funding/Support: This study was supported by grant R01AG056368 (Dr Kim) and grant R33AG071744 (Dr Inouye) from the NIA.

Role of the Funder/Sponsor: The NIA 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.