Clinician use of the Statin Choice Shared Decision-making Encounter Tool in a Major Health System.

BACKGROUND: Effective shared decision-making (SDM) tools for use during clinical encounters are available, but, outside of study settings, little is known about clinician use of these tools in practice. OBJECTIVE: To describe real-world use of an SDM encounter tool for statin prescribing, Statin Choice, embedded into the workflow of an electronic health record. DESIGN: Cross-sectional study. PARTICIPANTS: Clinicians and their statin-eligible patients who had outpatient encounters between January 2020 and June 2021 in Cleveland Clinic Health System. MAIN MEASURES: Clinician use of Statin Choice was recorded within the Epic record system. We categorized each patient's 10-year atherosclerotic cardiovascular disease risk into low (< 5%), borderline (5-7.5%), intermediate (7.5-20%), and high (≥ 20%). Other patient factors included age, sex, insurance, and race. We used mixed effects logistic regression to assess the odds of using Statin Choice for statin-eligible patients, accounting for clustering by clinician and site. We generated a residual intraclass correlation coefficient (ICC) to characterize the impact of the clinician on Statin Choice use. KEY RESULTS: Statin Choice was used in 7% of 68,505 eligible patients. Of 1047 clinicians, 48% used Statin Choice with ≥ 1 patient, and these clinicians used it with a median 9% of their patients (interquartile range: 3-22%). In the mixed effects logistic regression model, patient age (adjusted OR per year: 1.04; 95%CI 1.03-1.04) and 10-year ASVCD risk (aOR for 5-7.5% versus < 5% risk: 1.28; 95%CI: 1.14-1.44) were associated with use of Statin Choice. Black versus White race was associated with a lower odds of Statin Choice use (aOR: 0.83; 95%CI: 0.73-0.95), as was female versus male sex (aOR: 0.83; 95%CI: 0.76-0.90). The model ICC demonstrated that 53% of the variation in use of Statin Choice was clinician-driven. CONCLUSIONS: Patient factors, including race and sex, were associated with clinician use of Statin Choice; half the variation in use was attributable to individual clinicians.

Antibiotics on Demand: Advances in Asynchronous Telemedicine Call for Increased Antibiotic Surveillance.

The rapid growth of telehealth services has brought about direct-to-consumer telemedicine platforms, enabling patients to request antibiotics online without a virtual or face-to-face consultation. While telemedicine aims to enhance accessibility, this trend raises significant concerns regarding appropriate antimicrobial use and patient safety. In this viewpoint, we share our first-hand experience with 2 direct-to-consumer platforms, where we intentionally sought inappropriate antibiotic prescriptions for nonspecific symptoms strongly indicative of a viral upper respiratory infection. Despite the lack of clear necessity, requested antibiotic prescriptions were readily transmitted to our local pharmacy following a simple monetary transaction. The effortless acquisition of patient-selected antibiotics online, devoid of personal interactions or consultations, underscores the urgent imperative for intensified antimicrobial stewardship initiatives led by state and national public health organizations in telehealth settings. By augmenting oversight and regulation, we can ensure the responsible and judicious use of antibiotics, safeguard patient well-being, and preserve the efficacy of these vital medications.

Patient Portal Message Volume and Time Spent on the EHR: an Observational Study of Primary Care Clinicians.

BACKGROUND: As patient-initiated messaging rises, identifying variation in message volume and its relationship to clinician workload is essential. OBJECTIVE: To describe the association between variation in message volume over time and time spent on the electronic health record (EHR) outside of scheduled hours. DESIGN: Retrospective cohort study. PARTICIPANTS: Primary care clinicians at Cleveland Clinic Health System. MAIN MEASURES: We categorized clinicians according to their number of quarterly incoming medical advice messages (i.e., message volume) between January 2019 and December 2021 using group-based trajectory modeling. We assessed change in quarterly messages and outpatient visits between October-December 2019 (Q4) and October-December 2021 (Q12). The primary outcome was time outside of scheduled hours spent on the EHR. We used mixed effects logistic regression to describe the association between incoming portal messages and time spent on the EHR by clinician messaging group and at the clinician level. KEY RESULTS: Among the 150 clinicians, 31% were in the low-volume group (206 messages per quarter per clinician), 47% were in the moderate-volume group (505 messages), and 22% were in the high-volume group (840 messages). Mean quarterly messages increased from 340 to 695 (p < 0.001) between Q4 and Q12; mean quarterly outpatient visits fell from 711 to 575 (p = 0.005). While time spent on the EHR outside of scheduled hours increased modestly for all clinicians, this did not significantly differ by message group. Across all clinicians, each additional 10 messages was associated with an average of 12 min per quarter of additional time spent on the EHR (p < 0.001). CONCLUSIONS: Message volume increased substantially over the study period and varied by group. While messages were associated with additional time spent on the EHR outside of scheduled hours, there was no significant difference in time spent on the EHR between the high and low message volume groups.

Trends, Distribution, and Impact of Pediatric Messages in a Large Health System From 2019 to 2021.

This cohort study assesses changes in the volume of medical advice messages between 2019 and 2021, variation among pediatricians, and the association of volume with time spent working on the electronic health record outside clinical hours.

Informed Decision Making for Anticoagulation Therapy for Atrial Fibrillation.

BACKGROUND: Patients with atrial fibrillation (AF) must decide between warfarin and direct oral anticoagulants (DOACs), a decision involving important tradeoffs. Our objective was to understand whether physicians engage patients in informed decision making for anticoagulants. DESIGN: We performed an analysis of recorded conversations between physicians and anticoagulation-naïve patients in the Verilogue Point-of-Practice database. We assessed the presence of 7 elements of informed decision making, as well as a discussion of financial costs. RESULTS: Of 37 encounters with 21 physicians, 92% resulted in a DOAC prescription and 8% resulted in a warfarin prescription. Seventy percent met criteria for discussion of pros and cons, 70% for discussion of the alternatives, 43% presented the decision, 30% included an assessment of patient understanding, 22% included an explanation of the patient's role in decision making, 22% included an assessment of patient preferences, and 19% included a discussion of uncertainty. Two encounters (5%) included all 7 elements and 9 (24%) included none. Physicians discussed treatment costs with patients in 43% of encounters. LIMITATIONS: We assessed informed decision making in a single encounter. Physicians and patients may have had other discussions that were not captured in our data. CONCLUSIONS: Physicians often presented the alternatives but did not generally engage patients in informed decision making. The high rate of DOAC prescriptions is likely the result of physician preferences, as patient preferences were rarely assessed. IMPLICATIONS: Strategies to support physicians in engaging patients in informed decision making for anticoagulation are needed. HIGHLIGHTS: While physicians discussed the alternatives and presented pros and cons with patients, they rarely assessed patient preferences, explained the patient's role in decision making, or addressed uncertainty.The cost of treatment with DOACs versus warfarin was discussed by physicians in less than half of encounters, limiting patients' ability to make informed decisions for anticoagulation.Only 2 encounters (5%) fulfilled all 7 elements of informed decision making.

Physician Gender and Its Association With Patient Satisfaction and Visit Length: An Observational Study in Telemedicine.

Introduction Female physicians conduct longer visits than male physicians, with negative implications for their compensation. Yet patients often report higher satisfaction with female physicians. It is unknown whether satisfaction scores for female physicians are associated with their visit lengths. Our objective was to characterize the role of the physician and patient gender with respect to visit length and patient satisfaction. Methods We conducted an observational cohort study with patients and physicians of a nationwide telemedicine service between 2016 and 2018. Visit length was measured by the telemedicine system. Patients rated physicians on scales of one to five stars, with five considered "top box" satisfaction. We used logistic regression to estimate differences in patient satisfaction and linear regression to estimate differences in visit length by the physician and patient gender. We tested interactions between physician and patient gender and accounted for clustering by the physician. Results Among 102,664 visits with 405 physicians, the mean visit length was 5.8 minutes. Visits with male physicians were 1.11 minutes shorter than those with female physicians (95% CI:-1.58, -0.65). Controlling for visit length, male physicians were less likely than female physicians to receive top-box satisfaction scores (OR: 0.72; 95% CI: 0.61, 0.85). Visits between female physicians and male patients were the longest and visits between male physicians and female patients were the shortest. Female physicians had longer visits than male physicians but this did not explain their higher satisfaction scores. Conclusions To reduce inequity in compensation resulting from differences in visit length, female physicians could shorten their visits without negative consequences for their satisfaction ratings.

Qualitative Analysis of Patient-Physician Discussions Regarding Anticoagulation for Atrial Fibrillation.

Importance: For patients with atrial fibrillation (AF), the decision to initiate anticoagulation involves the choice between warfarin or a direct oral anticoagulant (DOAC). How physicians engage patients in this decision is unknown. Objective: To describe the content of discussions between patients with AF and physicians regarding choice of anticoagulation. Design, Setting, and Participants: This qualitative content analysis included clinical encounters between physicians and anticoagulation-naive patients discussing anticoagulation initiation between 2014 and 2020. Main Outcomes and Measures: Themes identified through content analysis. Results: Of 37 encounters, almost all (34 [92%]) resulted in a prescription for a DOAC. Most (25 [68%]) patients were White; 15 (41%) were female and 22 (59%) were male; and 24 (65%) were aged 65 to 84 years. Twenty-one physicians conducted the included encounters, the majority of whom were cardiologists (14 [67%]) and male (19 [90%]). The analysis revealed 4 major categories and associated subcategories of themes associated with physician discussion of anticoagulation with anticoagulation-naive patients: (1) benefit vs risk of taking anticoagulation-in many cases, this involved an imbalance in completeness of discussion of stroke vs bleeding risk, and physicians often used emotional language; (2) tradeoffs between warfarin and DOACs-physicians typically discussed pros and cons, used persuasive language, and provided mixed signals, telling patients that warfarin and DOACs were basically equivalent, while simultaneously saying warfarin is rat poison; (3) medication costs-physicians often attempted to address patients' questions about out-of-pocket costs but were unable to provide concrete answers, and they often provided free samples or coupons; and (4) DOACs in television commercials-physicians used direct-to-consumer pharmaceutical advertising about DOACs to orient patients to the issue of anticoagulation as well as the advantages of DOACs over warfarin. Patients and physicians also discussed class action lawsuits for DOACs that patients had seen on television. Conclusions and Relevance: This qualitative analysis of anticoagulation discussions between physicians and patients during clinical encounters found that physicians engaged in persuasive communication to convince patients to accept anticoagulation with a DOAC, yet they were unable to address questions regarding medication costs. For patients who are ultimately unable to afford DOACs, this may lead to unnecessary financial burden or abandoning prescriptions at the pharmacy, placing them at continued risk of stroke.

Change in Physician Burnout between 2013 and 2020 in a Major Health System.

OBJECTIVE: Assessing changes in physician burnout over time can offer insight into the causes of burnout. Existing studies are limited by using different samples of physicians at each time point. Our objective was to assess changes in burnout between 2013-2014 and 2019-2020 overall and within a cohort of physicians who took the survey twice. METHODS: This is a longitudinal cohort and cross-sectional study of physicians in a major health system. They were administered the Maslach Burnout Inventory in 2013-2014 and 2019-2020. We separately assessed differences in odds of burnout and its subscales in 2013-2014 and 2019-2020 by physician characteristics and clinical time. We then assessed differences in the odds of reporting burnout and its subscales in 2019-2020 compared with 2013-2014 overall and by physician sex, race, and change in clinical full-time employment. RESULTS: There were 1220 respondents in 2013-2014, 503 in 2019-2020, and 149 who responded at both time points. Burnout increased from 35% to 56%. Compared with 2013-2014, physicians had 2.39 higher odds (95% confidence interval [CI] 1.92-2.98) of burnout in 2019-2020, and this change in burnout was significantly more pronounced for female versus male physicians (odds ratio 1.80; 95% CI 1.57-1.80). Compared with White physicians, non-White physicians had significantly lower odds of burnout at both time points, but their odds increased significantly more over time (odds ratio 1.36; 95% CI 1.05-1.57). CONCLUSIONS: We found a substantial increase in burnout over time, which was particularly pronounced for non-White and female physicians. Assessment over time is essential for understanding problematic trajectories of burnout that may be obscured by cross-sectional studies.

Primary care physician referral practices regarding BRCA1/2 genetic counseling in a major health system.

PURPOSE: The United States Preventive Services Task Force recommends primary care physicians refer patients at high risk for BRCA1/2 mutations to genetic testing when appropriate. The objective of our study was to describe referrals for BRCA1/2 testing in a large integrated health system and to assess factors associated with referral. METHODS: This retrospective cohort study includes female patients between 18 and 50 years who had a primary care visit in the Cleveland Clinic Health System between 2010 and 2019. We used multivariable logistic regression to estimate differences in the odds of a woman being referred for BRCA1/2 testing by patient factors and referring physician specialty. We also assessed variation in referrals by physicians. RESULTS: Among 279,568 women, 5% were high risk. Of those, 22% were referred for testing. Black patients were significantly less likely to be referred than white patients (aOR 0.87; 95% CI 0.77, 0.98) and Jewish patients were more likely to be referred than non-Jewish patients (aOR 2.13; 95% CI 1.68, 2.70). Patients primarily managed by OB/GYN were significantly more likely to be referred than those cared for via Internal/Family Medicine (aOR 1.45; 95% CI 1.30, 1.61). Less than a quarter of primary care physicians ever referred a patient for testing. CONCLUSION: The majority of primary care patients at high risk for a BRCA1/2 mutation were not referred for testing, and over a decade, most physicians never referred a single patient. Internal/Family Medicine physicians, in particular, need support in identifying and referring women who could benefit from testing.

Prescribing of anticoagulation for atrial fibrillation in primary care.

Atrial fibrillation (AF) is common in primary care patients. Many patients who could benefit from anticoagulation do not receive it. The objective of this study was to describe anticoagulation prescribing by primary care physicians. We conducted an observational study in the Cleveland Clinic Health System among patients with AF and ≥ 1 primary care appointment between 2015 and 2018 and their physicians. We estimated differences in the odds of an eligible patient receiving anticoagulation versus not and a DOAC versus warfarin using two mixed effects logistic regression models, adjusted for patient sociodemographic factors, history of falls or dementia, and CHA2DS2-VASc and HAS-BLED scores. We categorized physicians into prescribing tertiles, based on their adjusted prescribing rate, which we included as predictors in the models. Among 5253 patients, 47% received anticoagulation. Of those, 56% received a DOAC. CHA2DS2-VASc and HAS-BLED scores were not associated with anticoagulation prescription. Black race was negatively associated with receiving anticoagulation overall (aOR:0.71; 95%CI:0.56-0.89) and with prescription for a DOAC (aOR:0.65; 95%CI:0.45-0.93). Among 195 physicians, the anticoagulation prescribing rate ranged from 27% to 57% and DOAC rates ranged from 34% to 69%. Physician prescribing tertile was associated with odds of a patient receiving anticoagulation overall (aOR:1.51; 95%CI: 1.13-2.01 for the highest versus lowest tertile), but not DOAC prescriptions. When prescribing anticoagulation, physicians appear not to consider risk of stroke or bleeding but patient race is an important determinant. Seeing a physician with a high anticoagulation prescribing rate was strongly associated with a patient receiving it, suggesting a lack of individualization.

Use of Direct-to-Consumer Telemedicine to Access Mental Health Services.

BACKGROUND: The coronavirus disease 2019 pandemic has contributed to growing demand for mental health services, but patients face significant barriers to accessing care. Direct-to-consumer(DTC) telemedicine has been proposed as one way to increase access, yet little is known about its pre-pandemic use for mental healthcare. OBJECTIVE: To characterize patients, providers, and their use of a large nationwide DTC telemedicine platform for mental healthcare. DESIGN: Retrospective cross-sectional study. SETTING: Mental health encounters conducted on the American Well DTC telemedicine platform from 2016 to 2018. PARTICIPANTS: Patients and physicians. MAIN MEASURES: Patient measures included demographics, insurance report, and number of visits. Provider characteristics included specialty, region, and number of encounters. Encounter measures included wait time, visit length and timing, out-of-pocket payment, coupon use, prescription outcome, referral receipt, where care otherwise would have been sought, and patient satisfaction. Factors associated with five-star physician ratings and prescription receipt were assessed using logistic regression. KEY RESULTS: We analyzed 19,270 mental health encounters between 6708 patients and 1045 providers. Visits were most frequently for anxiety (39.1%) or depression (32.5%), with high satisfaction (4.9/5) across conditions. Patients had a median 2.0 visits for psychiatry (IQR 1.0-3.0) and therapy (IQR 1.0-5.0), compared to 1.0 visit (IQR 1.0-1.0) for urgent care. High satisfaction was positively correlated with prescription receipt (OR 1.89, 95% CI 1.54-2.32) and after-hours timing (aOR 1.18, 95% CI 1.02-1.36). Prescription rates ranged from 79.6% for depression to 32.2% for substance use disorders. Prescription receipt was associated with increased visit frequency (aOR 1.95, 95% CI 1.57-2.42 for ≥ 3 visits). CONCLUSIONS: As the burden of psychiatric disease grows, DTC telemedicine offers one solution for extending access to mental healthcare. While most encounters were one-off, evidence of some continuity in psychiatry and therapy visits-as well as overall high patient satisfaction-suggests potential for broader DTC telemental health use.

Physicians' Views on Utilization of an Electronic Health Record-Embedded Calculator to Assess Risk for Venous Thromboembolism among Medical Inpatients: A Qualitative Study.

Background Venous thromboembolism (VTE) causes preventable in-hospital morbidity. Pharmacologic prophylaxis reduces VTE in at-risk patients but also increases bleeding. To increase appropriate prescribing, a risk calculator to guide prophylaxis decisions was developed. Despite efforts to promote its use, providers accessed it infrequently. Objective This study aimed to understand provider perspectives on VTE prophylaxis and facilitators and barriers to using the risk calculator. Design This is a qualitative study exploring provider perspectives on VTE prophylaxis and the VTE risk calculator. Participants We interviewed attending physicians and advanced practice providers who used the calculator, and site champions who promoted calculator use. Providers were categorized by real-world usage over a 3-month period: low (<20% of the time), moderate (20-50%), or high (>50%). Approach During semistructured interviews, we asked about experiences with VTE, calculator use, perspectives on its implementation, and experiences with other risk assessment tools. Once thematic saturation was reached, transcripts were analyzed using content analysis to identify themes. Results Fourteen providers participated. Five were high utilizers, three were moderate utilizers, and six were low utilizers. Three site champions participated. Eight major themes were identified as follows: (1) ease of use, (2) perception of VTE risk, (3) harms of thromboprophylaxis, (4) overestimation of calculator use, (5) confidence in own ability, (6) underestimation of risk by calculator, (7) variability of trust in calculator, and (8) validation to withhold prophylaxis from low-risk patients. Conclusions While providers found the calculator is easy to use, routine use may be hindered by distrust of its recommendations. Inaccurate perception of VTE and bleeding risk may prevent calculator use.

Patient and physician factors contributing to polypharmacy among older patients.

OBJECTIVE: Polypharmacy, or use of multiple medications, is associated with patient factors. Less is known regarding variation in polypharmacy by individual physicians. The objective of this study was to assess patient and physician factors associated with polypharmacy among older patients. METHODS: This is a cross-sectional study of patients aged ≥65 years with a primary care visit at Cleveland Clinic Health System in 2015 and their physicians. We collected patient demographics, comorbidities and current medications from the electronic health record, including potentially inappropriate medications (PIMs). We used mixed effects linear regression to estimate adjusted differences in the number of medications by patient factors. We generated adjusted prescribing rates for individual physicians and assessed differences in physician performance on quality measures by their prescribing rate. RESULTS: Our study included 44,570 patients who were prescribed an average of 6.8 medications (standard deviation: 4.0) by 701 physicians. Female sex, higher BMI, having Medicaid insurance, current or former smoking status, comorbidities and seeing a specialist were associated with number of medications. Age was not. Among 267 physicians who saw ≥20 study-eligible patients, the adjusted mean number of medications per patient ranged from 5.2 to 9.6. Compared to physicians who prescribed above the mean, lower prescribing physicians performed significantly better on medication reconciliation (p = .007) and hypertension control (p < .001) and prescribed fewer PIMs (p < .001). CONCLUSIONS: Individual physicians varied in their prescribing practices, even after adjusting for patient demographic and clinical characteristics. Interventions to reduce polypharmacy in older adults should target high prescribing physicians, as physician behavior is more actionable than patient factors.

Change in individual physicians' screening mammography completion rates following the updated USPSTF guideline supporting shared decision making: An observational cohort study.

OBJECTIVE: To understand changes in physician screening practices in response to the 2009 U.S. Preventive Services Task Force recommendation supporting shared decision making (SDM) for mammography in women aged 40-49 years. METHODS: We assessed screening completion rates for physicians in the Cleveland Clinic Health System pre-2009 (2006-2008) and post-2009 (2010-2015), and rates for physicians new to the system post-2009. We used mixed effects logistic regression to estimate the odds of a woman receiving screening post-2009. If physicians practiced SDM, we hypothesized their screening rates would change after 2009. To test this, we included each physician's pre-2009 screening rate as a predictor in the model. RESULTS: Among 125 physicians, the screening rate increased from 40% to 45% from pre-2009 to post-2009. For physicians new to the health system post-2009 the rate was 32%. In the mixed effects model (N = 17,007), the strongest predictor of mammography receipt among patients post-2009 was their physician's pre-2009 screening rate (aOR:3.57 per 10% increase in pre-2009 rate; 95%CI:1.69-7.50). CONCLUSIONS: Whether a woman received a mammogram post-2009 was highly associated with her physicians' pre-2009 screening rate, suggesting physicians are not individualizing screening decisions via SDM. PRACTICE IMPLICATIONS: Physicians may need support to effectively practice SDM.

Shared Decision Making in Cardiac Electrophysiology Procedures and Arrhythmia Management.

Shared decision making (SDM) has been advocated to improve patient care, patient decision acceptance, patient-provider communication, patient motivation, adherence, and patient reported outcomes. Documentation of SDM is endorsed in several society guidelines and is a condition of reimbursement for selected cardiovascular and cardiac arrhythmia procedures. However, many clinicians argue that SDM already occurs with clinical encounter discussions or the process of obtaining informed consent and note the additional imposed workload of using and documenting decision aids without validated tools or evidence that they improve clinical outcomes. In reality, SDM is a process and can be done without decision tools, although the process may be variable. Also, SDM advocates counter that the low-risk process of SDM need not be held to the high bar of demonstrating clinical benefit and that increasing the quality of decision making should be sufficient. Our review leverages a multidisciplinary group of experts in cardiology, cardiac electrophysiology, epidemiology, and SDM, as well as a patient advocate. Our goal is to examine and assess SDM methodology, tools, and available evidence on outcomes in patients with heart rhythm disorders to help determine the value of SDM, assess its possible impact on electrophysiological procedures and cardiac arrhythmia management, better inform regulatory requirements, and identify gaps in knowledge and future needs.

COVID-19 Is Associated With an Increase in Visits for Anxiety but Not Depression.

Background The coronavirus disease 2019 (COVID-19) pandemic has increased concerns about mental health. We conducted a time-series analysis to determine whether the percentage of primary care visits for anxiety and depression changed after COVID-19. Methodology We assessed the adjusted weekly change in the percentage of primary care visits for anxiety and depression between August 2019 and October 2020 at a large integrated health system. To account for changes in overall visit behavior during the pandemic, we created three periods: pre-period (August 1, 2019 to March 8, 2020), initial period (March 9, 2020 to June 31, 2020), and return period (July 1, 2020 to October 31, 2020). We used hierarchical linear regression models (clustered by month) to identify the association between the time period and the adjusted mean weekly percentage of visits for depression or anxiety. We conducted the analysis in 2020 and 2021. Results There were 1,691,071 encounters among 605,105 unique adults. The median age was 55 years (interquartile range = 39-68), 57% were female, 78% were white, and 59% had private insurance. Most visits were office-based (versus virtual), of which 99% were in the pre-COVID-19 period and 75% in the return period. There was a significant increase in the percentage of visits associated with anxiety after July compared to before COVID-19 (10.4% versus 9.2%; p = 0.006), and there was no difference in the percentage of visits for depression (p > 0.05). Conclusions Outreach to individuals with depression who have not sought care may be necessary.

Association between Opioids Prescribed to Medical Inpatients with Pain and Long-Term Opioid Use.

OBJECTIVES: Opioid receipt during medical hospitalizations may be associated with subsequent long-term use. Studies, however, have not accounted for pain, which may explain chronic use. The objective of this study was to identify the association between opioid exposure during a medical hospitalization and use 6 to 12 months later. METHODS: This was an observational cohort study using electronic health record data from 10 hospitals in the Cleveland Clinic Health System in 2016. Eligible patients were opioid-naïve adults with pain age 18 years and older, admitted to a medical service. Outcomes were opioid receipt during hospitalization and on discharge, and long-term opioid use, defined as ≥2 prescriptions for at least 30 pills 6 to 12 months posthospitalization. We estimated the odds of long-term opioid use by opioid exposure during the hospitalization. Models controlled for patient demographic and clinical characteristics, including patient-reported pain. RESULTS: Among the 2971 patients in the sample, 64% received opioids during their hospitalization and 28% were discharged with opioids. Overall, 3% of patients had long-term use. Higher pain score was associated with greater odds of long-term use (adjusted odds ratio [aOR] per point increase 1.11; 95% confidence interval [CI] 1.03-1.19). No patient factors were associated with long-term use. Receipt of an opioid during a hospitalization only was not associated with long-term use (aOR 1.44, 95% CI 0.81-2.57), but receipt at discharge was (aOR 1.96, 95% CI 1.08-3.56). CONCLUSIONS: Although opioid receipt at discharge was associated with long-term use, the number of patients this applied to was small. Pain severity was an important predictor of long-term use and should be accounted for in future studies.