Can accurate demographic information about people who use prescription medications nonmedically be derived from Twitter?

Traditional substance use (SU) surveillance methods, such as surveys, incur substantial lags. Due to the continuously evolving trends in SU, insights obtained via such methods are often outdated. Social media-based sources have been proposed for obtaining timely insights, but methods leveraging such data cannot typically provide fine-grained statistics about subpopulations, unlike traditional approaches. We address this gap by developing methods for automatically characterizing a large Twitter nonmedical prescription medication use (NPMU) cohort (n = 288,562) in terms of age-group, race, and gender. Our natural language processing and machine learning methods for automated cohort characterization achieved 0.88 precision (95% CI:0.84 to 0.92) for age-group, 0.90 (95% CI: 0.85 to 0.95) for race, and 94% accuracy (95% CI: 92 to 97) for gender, when evaluated against manually annotated gold-standard data. We compared automatically derived statistics for NPMU of tranquilizers, stimulants, and opioids from Twitter with statistics reported in the National Survey on Drug Use and Health (NSDUH) and the National Emergency Department Sample (NEDS). Distributions automatically estimated from Twitter were mostly consistent with the NSDUH [Spearman r: race: 0.98 (P < 0.005); age-group: 0.67 (P < 0.005); gender: 0.66 (P = 0.27)] and NEDS, with 34/65 (52.3%) of the Twitter-based estimates lying within 95% CIs of estimates from the traditional sources. Explainable differences (e.g., overrepresentation of younger people) were found for age-group-related statistics. Our study demonstrates that accurate subpopulation-specific estimates about SU, particularly NPMU, may be automatically derived from Twitter to obtain earlier insights about targeted subpopulations compared to traditional surveillance approaches.

National Institute on Drug Abuse Clinical Trials Network Meeting Report: Managing Patients Exposed to Xylazine-Adulterated Opioids in Emergency, Hospital and Addiction Care Settings.

Used as a veterinary sedative and not approved for human use, xylazine has been increasingly linked with opioid overdose deaths in the United States. A growing number of people have been exposed to xylazine in the illicit opioid supply (especially fentanyl) or in other drugs, particularly in some areas of the Northeast. Xylazine is an α-2 adrenergic agonist that decreases sympathetic nervous system activity. When combined with fentanyl or heroin, it is purported to extend the duration of the opioid's sedative effect and to cause dependence and an associated withdrawal syndrome; however, data to support these concerns are limited. Despite the escalating frequency of detection of xylazine in people with nonfatal and fatal opioid overdose, direct links to these outcomes have not been identified. Because the strongest causal link is to fentanyl coexposure, ventilatory support and naloxone remain the cornerstones of overdose management. Xylazine is also associated with severe tissue injury, including skin ulcers and tissue loss, but little is known about the underlying mechanisms. Nonetheless, strategies for prevention and treatment are emerging. The significance and clinical effects of xylazine as an adulterant is focused on 4 domains that merit further evaluation: fentanyl-xylazine overdose, xylazine dependence and withdrawal, xylazine-associated dermal manifestations, and xylazine surveillance and detection in clinical and nonclinical settings. This report reflects the Proceedings of the National Institute on Drug Abuse Center for the Clinical Trials Network convening of clinical and scientific experts, federal staff, and other stakeholders to describe emerging best practices for treating people exposed to xylazine-adulterated opioids. Participants identified scientific gaps and opportunities for research to inform clinical practice in emergency departments, hospitals, and addiction medicine settings.

Leveraging a Learning Collaborative Model to Develop and Pilot Quality Measures to Improve Opioid Prescribing in the Emergency Department.

The American College of Emergency Physicians (ACEP) Emergency Medicine Quality Network (E-QUAL) Opioid Initiative was launched in 2018 to advance the dissemination of evidence-based resources to promote the care of emergency department (ED) patients with opioid use disorder. This virtual platform-based national learning collaborative includes a low-burden, structured quality improvement project, data benchmarking, tailored educational content, and resources designed to support a nationwide network of EDs with limited administrative and research infrastructure. As a part of this collaboration, we convened a group of experts to identify and design a set of measures to improve opioid prescribing practices to provide safe analgesia while reducing opioid-related harms. We present those measures here, alongside initial performance data on those measures from a sample of 370 nationwide community EDs participating in the 2019 E-QUAL collaborative. Measures include proportion of opioid administration in the ED, proportion of alternatives to opioids as first-line treatment, proportion of opioid prescription, opioid pill count per prescription, and patient medication safety education among ED visits for atraumatic back pain, dental pain, or headache. The proportion of benzodiazepine and opioid coprescribing for ED visits for atraumatic back pain was also evaluated. This project developed and effectively implemented a collection of 6 potential measures to evaluate opioid analgesic prescribing across a national sample of community EDs, representing the first feasibility assessment of opioid prescribing-related measures from rural and community EDs.

Xylazine Adulteration of the Heroin-Fentanyl Drug Supply : A Narrative Review.

Xylazine is an animal sedative, approved by the U.S. Food and Drug Administration, that is commonly used in veterinary medicine and is not approved for human use. Since 2016, xylazine has consistently appeared in the illicitly manufactured fentanyl supply and has significantly increased in prevalence, likely due to its low cost, easy availability, and presumed synergistic psychoactive effect. Clinical experience along with the available pertinent research were used to review xylazine adulteration of the drug supply and provide guidance on the care of patients exposed to xylazine. This review discusses xylazine pharmacology, animal and human clinical effects, and what is known to date about care of patients experiencing acute overdose, xylazine-fentanyl withdrawal, and xylazine-associated wounds.

National Institute on Drug Abuse Clinical Trials Network Meeting Report: Advancing Emergency Department Initiation of Buprenorphine for Opioid Use Disorder.

Opioid use disorder and opioid overdose deaths are a major public health crisis, yet highly effective evidence-based treatments are available that reduce morbidity and mortality. One such treatment, buprenorphine, can be initiated in the emergency department (ED). Despite evidence of efficacy and effectiveness for ED-initiated buprenorphine, universal uptake remains elusive. On November 15 and 16, 2021, the National Institute on Drug Abuse Clinical Trials Network convened a meeting of partners, experts, and federal officers to identify research priorities and knowledge gaps for ED-initiated buprenorphine. Meeting participants identified research and knowledge gaps in 8 categories, including ED staff and peer-based interventions; out-of-hospital buprenorphine initiation; buprenorphine dosing and formulations; linkage to care; strategies for scaling ED-initiated buprenorphine; the effect of ancillary technology-based interventions; quality measures; and economic considerations. Additional research and implementation strategies are needed to enhance adoption into standard emergency care and improve patient outcomes.

Impact of Universal Screening and Automated Clinical Decision Support for the Treatment of Opioid Use Disorder in Emergency Departments: A Difference-in-Differences Analysis.

STUDY OBJECTIVE: Emergency department (ED)-initiated buprenorphine improves outcomes in patients with opioid use disorder; however, adoption varies widely. To reduce variability, we implemented a nurse-driven triage screening question in the electronic health record to identify patients with opioid use disorder, followed by targeted electronic health record prompts to measure withdrawal and guide next steps in management, including initiation of treatment. Our objective was to assess the impact of screening implementation in 3 urban, academic EDs. METHODS: We conducted a quasiexperimental study of opioid use disorder-related ED visits using electronic health record data from January 2020 to June 2022. The triage protocol was implemented in 3 EDs between March and July 2021, and 2 other EDs in the health system served as controls. We evaluated changes in treatment measures over time and used a difference-in-differences analysis to compare outcomes in the 3 intervention EDs with those in the 2 controls. RESULTS: There were 2,462 visits in the intervention hospitals (1,258 in the preperiod and 1,204 in the postperiod) and 731 in the control hospitals (459 in the preperiod and 272 in the postperiod). Patient characteristics within the intervention and control EDs were similar across the time periods. Compared with the control hospitals, the triage protocol was associated with a 17% greater increase in withdrawal assessment, using the Clinical Opioid Withdrawal Scale (COWS) (95% CI 7 to 27). Buprenorphine prescriptions at discharge also increased by 5% (95% CI 0% to 10%), and naloxone prescriptions increased by 12% points (95% CI 1% to 22%) in the intervention EDs relative to controls. CONCLUSION: An ED triage screening and treatment protocol led to increased assessment and treatment of opioid use disorder. Protocols designed to make screening and treatment the default practice have promise in increasing the implementation of evidence-based treatment ED opioid use disorder care.

Variability in opioid use disorder clinical presentations and treatment in the emergency department: A mixed-methods study.

BACKGROUND: There is strong evidence for emergency department (ED)-initiated treatment of opioid use disorder (OUD). However, implementation is variable, and ED management of OUD may differ by clinical presentation. Our aim was to use mixed methods to explore variation in ED-based OUD care by patient clinical presentation and understand barriers and facilitators to ED implementation of OUD treatment across scenarios. METHODS: We analyzed treatment outcomes in OUD-related visits within three urban, academic EDs from 12/2018 to 7/2020 following the implementation of interventions to increase ED-initiated OUD treatment. We assessed differences in treatment with medications for OUD (MOUDs) by clinical presentation (overdose, withdrawal, others). These data were integrated with results from 5 focus groups conducted with 28 ED physicians and nurses January to April 2020 to provide a richer understanding of clinician perspectives on caring for ED patients with OUD. RESULTS: Of the 1339 total opioid-related visits, there were 265 (20%) visits for overdose, 123 (9%) for withdrawal, and 951 (71%) for other OUD-related conditions. 23% of patients received MOUDs during their visit or at discharge. Treatment with MOUDs was least common in overdose presentations (6%) and most common in withdrawal presentations (69%, p < 0.001). Buprenorphine was prescribed at discharge in 15% of visits, including 42% of withdrawal visits, 14% of other OUD-related visits, and 5% of overdose visits (p < 0.001). In focus groups, clinicians highlighted variation in ED presentations among patients with OUD. Clinicians also highlighted key aspects necessary for successful treatment initiation including perceived patient receptivity, provider confidence, and patient clinical readiness. CONCLUSIONS: ED-based treatment of OUD differed by clinical presentation. Clinician focus groups identified several areas where targeted guidance or novel approaches may improve current practices. These results highlight the need for tailored clinical guidance and can inform health system and policy interventions seeking to increase ED-initiated treatment for OUD.

URINE TOXICOLOGY PROFILES OF EMERGENCY DEPARTMENT PATIENTS WITH UNTREATED OPIOID USE DISORDER: A MULTI-SITE VIEW.

BACKGROUND: Opioid overdose deaths in 2021 were the highest ever, driven by fentanyl and polysubstance use. OBJECTIVE: The aim of the study was to characterize drug use, assessed by urine drug screens (UDSs), in patients with untreated opioid use disorder (OUD) presenting to 28 emergency departments (EDs) nationally and by region. METHODS: We analyzed UDSs from patients enrolled in the CTN-0099 ED-INNOVATION (Emergency Department-Initiated Buprenorphine Validation) trial between July 12, 2020 and March 9, 2022. Participants were adult ED patients with OUD not engaged in addiction treatment with a UDS positive for an opioid, but negative for methadone. Sites were divided into "East" and "West" regions. RESULTS: A UDS was available for all 925 enrolled participants, 543 from East and 382 from West. Fentanyl was in 702 specimens (76%) (n = 485 [89%] East vs. n = 217 [57%] West; p < 0.01) and was the only opioid in 269 (29%). After fentanyl, the most common opioids were morphine (presumably heroin; n = 411 [44%]; n = 192 [35%] East vs. n = 219 [57%] West; p < 0.01) and buprenorphine (n = 329 [36%]; n = 186 [35%] East vs. n = 143 [37%] West; p = 0.32). The most common drugs found with opioids were stimulants (n = 545 [59%]), tetrahydrocannabinol (n = 417 [45%]), and benzodiazepines (n = 151 [16%]). Amphetamine-type stimulants were more common in West (n = 209 [55%] vs. East (n = 125 [23%]). Cocaine was more common in East (n = 223 [41%]) vs. West (n = 82 [21%]). The presence of multiple drugs was common (n = 759 [82%]). CONCLUSIONS: Most participants had UDS specimens containing multiple substances; a high proportion had fentanyl, stimulants, and buprenorphine. Regional differences were noted. Given the increased risk of death with fentanyl and polysubstance use, ED providers should be providing risk reduction counseling, treatment, and referral.

A Multicentered Randomized Controlled Trial Comparing the Effectiveness of Pain Treatment Communication Tools in Emergency Department Patients With Back or Kidney Stone Pain.

Objectives. To compare the effectiveness of 3 approaches for communicating opioid risk during an emergency department visit for a common painful condition. Methods. This parallel, multicenter randomized controlled trial was conducted at 6 geographically disparate emergency department sites in the United States. Participants included adult patients between 18 and 70 years of age presenting with kidney stone or musculoskeletal back pain. Participants were randomly assigned to 1 of 3 risk communication strategies: (1) a personalized probabilistic risk visual aid, (2) a visual aid and a video narrative, or 3) general risk information. The primary outcomes were accuracy of risk recall, reported opioid use, and treatment preference at time of discharge. Results. A total of 1301 participants were enrolled between June 2017 and August 2019. There was no difference in risk recall at 14 days between the narrative and probabilistic groups (43.7% vs 38.8%; absolute risk reduction = 4.9%; 95% confidence interval [CI] = -2.98, 12.75). The narrative group had lower rates of preference for opioids at discharge than the general risk information group (25.9% vs 33.0%; difference = 7.1%; 95% CI = 0.64, 0.97). There were no differences in reported opioid use at 14 days between the narrative, probabilistic, and general risk groups (10.5%, 10.3%, and 13.3%, respectively; P = .44). Conclusions. An emergency medicine communication tool incorporating probabilistic risk and patient narratives was more effective than general information in mitigating preferences for opioids in the treatment of pain but was not more effective with respect to opioid use or risk recall. Trial Registration. Clinical Trials.gov identifier: NCT03134092. (Am J Public Health. 2022;112(S1):S45-S55. https://doi.org/10.2105/AJPH.2021.306511).

A Neuropharmacological Model to Explain Buprenorphine Induction Challenges.

Buprenorphine induction for treating opioid use disorder is being implemented in emergency care. During this era of high-potency synthetic opioid use, novel and divergent algorithms for buprenorphine induction are emerging to optimize induction experience, facilitating continued treatment. Specifically, in patients with chronic fentanyl or other drug exposures, some clinicians are using alternative buprenorphine induction strategies, such as quickly maximizing buprenorphine agonist effects (eg, macrodosing) or, conversely, giving smaller initial doses and slowing the rate of buprenorphine dosing to avoid antagonist/withdrawal effects (eg, microdosing). However, there is a lack of foundational theory and empirical data to guide clinicians in evaluating such novel induction strategies. We present data from clinical studies of buprenorphine induction and propose a neuropharmacologic working model, which posits that acute clinical success of buprenorphine induction (achieving a positive agonist-to-withdrawal balance) is a nonlinear outcome of the opioid balance at the time of initial buprenorphine dose and mu-opioid-receptor affinity, lipophilicity, and mu-opioid-receptor intrinsic efficacy (the "ALE value") of the prior opioid. We discuss the rationale for administering smaller or larger doses of buprenorphine to optimize the patient induction experience during common clinical situations.

Sustained Implementation of a Multicomponent Strategy to Increase Emergency Department-Initiated Interventions for Opioid Use Disorder.

STUDY OBJECTIVE: There is strong evidence supporting emergency department (ED)-initiated buprenorphine for opioid use disorder, but less is known about how to implement this practice. Our aim was to describe implementation, maintenance, and provider adoption of a multicomponent strategy for opioid use disorder treatment in 3 urban, academic EDs. METHODS: We conducted a retrospective analysis of electronic health record data for adult patients with opioid use disorder-related visits before (March 2017 to November 2018) and after (December 2018 to July 2020) implementation. We describe patient characteristics, clinical treatment, and process measures over time and conducted an interrupted time series analysis using a patient-level multivariable logistic regression model to assess the association of the interventions with buprenorphine use and other outcomes. Finally, we report provider-level variation in prescribing after implementation. RESULTS: There were 2,665 opioid use disorder-related visits during the study period: 28% for overdose, 8% for withdrawal, and 64% for other conditions. Thirteen percent of patients received medications for opioid use disorder during or after their ED visit overall. Following intervention implementation, there were sustained increases in treatment and process measures, with a net increase in total buprenorphine of 20% in the postperiod (95% confidence interval 16% to 23%). In the adjusted patient-level model, there was an immediate increase in the probability of buprenorphine treatment of 24.5% (95% confidence interval 12.1% to 37.0%) with intervention implementation. Seventy percent of providers wrote at least 1 buprenorphine prescription, but provider-level buprenorphine prescribing ranged from 0% to 61% of opioid use disorder-related encounters. CONCLUSION: A combination of strategies to increase ED-initiated opioid use disorder treatment was associated with sustained increases in treatment and process measures. However, adoption varied widely among providers, suggesting that additional strategies are needed for broader uptake.

Buprenorphine precipitated opioid withdrawal: Prevention and management in the ED setting.

INTRODUCTION: Buprenorphine precipitated opioid withdrawal (BPOW) is an uncommon complication of buprenorphine initiation in the emergency department (ED), but it can produce significant discomfort and be distressing to patients. As EDs continue to care for those with opioid use disorder (OUD), clinicians should be aware of how to prevent and treat BPOW. OBJECTIVE: This narrative review provides an evidence-based update of the epidemiology, prevention strategies, and management of BPOW for the emergency clinician. DISCUSSION: BPOW is a rapid worsening of opioid withdrawal symptoms upon initiating buprenorphine. BPOW can be prevented by waiting for the onset of moderate Clinical Opioid Withdrawal Scale (COWS) > 13 opioid withdrawal symptoms and a sufficient amount of time since last full opioid agonist use before buprenorphine administration. Risk factors for BPOW include chronic fentanyl use, methadone use, and concurrent benzodiazepine use. Alternative dosing strategies such as low-dose or "microdosing" and high-dose or "macrodosing" are options for buprenorphine that may impact the development of BPOW. The strategy of treating BPOW with more buprenorphine has a pharmacological basis and has been effective in case reports. Additional management is symptom-based and supportive. Although most cases have a benign course, patients may be significantly less likely to use buprenorphine for OUD in the future or seek care for substance use disorder. CONCLUSIONS: Appropriate initiation of buprenorphine is important to prevent BPOW. Dosing buprenorphine should be based on the patient's patterns of opioid use and response to therapy. Management of BPOW should be symptom-based but include additional buprenorphine and adjunctive medications.

Variation in emergency department visit rates for opioid use disorder: Implications for quality improvement initiatives.

STUDY OBJECTIVE: Emergency departments (ED) are critical touchpoints for encounters among patients with opioid use disorder (OUD), but implementation of ED initiated treatment and harm reduction programs has lagged. We describe national patient, visit and hospital-level characteristics of ED OUD visits and characterize EDs with high rates of OUD visits in order to inform policies to optimize ED OUD care. METHODS: We conducted a descriptive, cross-sectional study with the 2017 Nationwide Emergency Department Sample (NEDS) from the Healthcare Cost and Utilization Project, using diagnostic and mechanism of injury codes from ICD-10 to identify OUD related visits. NEDS weights were applied to generate national estimates. We evaluated ED visit and clinical characteristics of all OUD encounters. We categorized hospitals into quartiles by rate of visits for OUD per 1000 ED visits and described the visit, clinical, and hospital characteristics across the four quartiles. RESULTS: In 2017, the weighted national estimate for OUD visits was 1,507,550. Overdoses accounted for 295,954. (19.6%) of visits. OUD visit rates were over 8× times higher among EDs in the highest quartile of OUD visit rate (22.9 per 1000 total ED visits) compared with EDs in the lowest quartile of OUD visit rate (2.7 per 1000 ED visits). Over three fifths (64.2%) of all OUD visits nationwide were seen by the hospitals in the highest quartile of OUD visit rate. These hospitals were predominantly in metropolitan areas (86.2%), over half were teaching hospitals (51.7%), and less than a quarter (23.3%) were Level 1 or Level 2 trauma centers. CONCLUSION: Targeting initial efforts of OUD care programs to high OUD visit rate EDs could improve care for a large portion of OUD patients utilizing emergency care.

Patient perspectives on naloxone receipt in the emergency department: a qualitative exploration.

BACKGROUND: Emergency departments (EDs) are important venues for the distribution of naloxone to patients at high risk of opioid overdose, but less is known about patient perceptions on naloxone or best practices for patient education and communication. Our aim was to conduct an in-depth exploration of knowledge and attitudes toward ED naloxone distribution among patients who received a naloxone prescription. METHODS: We conducted semi-structured telephone interviews with 25 adult participants seen and discharged from three urban, academic EDs in Philadelphia, PA, with a naloxone prescription between November 2020 and February 2021. Interviews focused on awareness of naloxone as well as attitudes and experiences receiving naloxone in the ED. We used thematic content analysis to identify key themes reflecting patient attitudes and experiences. RESULTS: Of the 25 participants, 72% had previously witnessed an overdose and 48% had personally experienced a non-fatal overdose. Nineteen participants (76%) self-disclosed a history of substance use or overdose, and one reported receiving an opioid prescription during their ED visit and no history of substance use. In interviews, we identified wide variability in participant levels of knowledge about overdose risk, the role of naloxone in reducing risk, and naloxone access. A subset of participants was highly engaged with community harm reduction resources and well versed in naloxone access and use. A second subset was familiar with naloxone, but largely obtained it through healthcare settings such as the ED, while a final group was largely unfamiliar with naloxone. While most participants expressed positive attitudes about receiving naloxone from the ED, the quality of discussions with ED providers was variable, with some participants not even aware they were receiving a naloxone prescription until discharge. CONCLUSIONS: Naloxone prescribing in the ED was acceptable and valued by most participants, but there are missed opportunities for communication and education. These findings underscore the critical role that EDs play in mitigating risks for patients who are not engaged with other healthcare or community health providers and can inform future work about the effective implementation of harm reduction strategies in ED settings.

Tramadol Use in United States Emergency Departments 2007-2018.

BACKGROUND: Amidst the opioid epidemic, there has been an increasing focus on opioid utilization in U.S. emergency departments (EDs). Compared with other opioids, little is known about the use of tramadol over the past decade. Tramadol has uncertain efficacy and a concerning adverse effect profile compared with traditional opioids. OBJECTIVE: Our aim was to describe trends in tramadol use in U.S. EDs between 2007 and 2018. METHODS: We analyzed the National Hospital Ambulatory Medical Care Survey from 2007 to 2018 to examine ED visits by patients 18 years or older in which tramadol was administered or prescribed. We examined trends in demographics and resource utilization and compared these trends with those of traditional opioids. Survey-weighted analyses were conducted to provide national-level estimates. RESULTS: Between 2007 and 2018, ED visits in which tramadol was used increased 70.6%, from 1.7% of all ED visits in 2007 to 2.9% in 2018. The largest increases were noted among patients aged 55 through 64 years and 65 years and older. Diagnostic resource utilization increased across the study period. Overall opioid utilization during the study period decreased from 28.4% in 2007 to 17.9% in 2018 (p < 0.001). The use of other specific opioids declined or remained stable between 2007 and 2018. CONCLUSIONS: Although the use of traditional opioids decreased from 2007 to 2018, the use of tramadol increased. Increases were largest among older patients, who may be more susceptible to the adverse effects associated with this medication. Further research in the appropriate use of tramadol in the ED setting is warranted.

Naloxone prescriptions following emergency department encounters for opioid use disorder, overdose, or withdrawal.

OBJECTIVE: To determine the rate at which commercially-insured patients fill prescriptions for naloxone after an opioid-related ED encounter as well as patient characteristics associated with obtaining naloxone. METHODS: This is a retrospective cohort study of adult patients discharged from the ED following treatment for an opioid-related condition from 2016 to 2018 using a commercial insurance claims database (Optum Clinformatics® Data Mart). The primary outcome was a pharmacy claim for naloxone in the 30 days following the ED encounter. A multivariable logistic regression model examined the association of patient characteristics with filled naloxone prescriptions, and predictive margins were used to report adjusted probabilities with 95% confidence intervals. RESULTS: 21,700 patients had opioid-related ED encounters during the study period, of which 1743 (8.0%) had encounters for heroin overdose, 8825 (40.7%) for overdose due to other opioids, 5400 (24.9%) for withdrawal, and 5732 (26.4%) for other opioid use disorder conditions. 230 patients (1.1%) filled a prescription for naloxone within 30 days. Patients with heroin overdose (2.6%; 95%CI 1.7 to 3.4), recent prescriptions for opioid analgesics (1.4%; 95%CI 1.1 to 1.7), recent prescriptions for buprenorphine (1.9%; 95%CI 1.0 to 2.9), and naloxone prescriptions in the prior year (3.3%; 95%CI 1.8 to 4.8) were more likely to obtain naloxone. The rate was significantly higher in 2018 [1.9% (95%CI 1.5 to 2.2)] as compared to 0.4% (95%CI 0.3 to 0.6) in 2016. CONCLUSIONS: Few patients use insurance to obtain naloxone by prescription following opioid-related ED encounters. Clinical and policy interventions should expand distribution of this life-saving medication in the ED.

Text classification models for the automatic detection of nonmedical prescription medication use from social media.

BACKGROUND: Prescription medication (PM) misuse/abuse has emerged as a national crisis in the United States, and social media has been suggested as a potential resource for performing active monitoring. However, automating a social media-based monitoring system is challenging-requiring advanced natural language processing (NLP) and machine learning methods. In this paper, we describe the development and evaluation of automatic text classification models for detecting self-reports of PM abuse from Twitter. METHODS: We experimented with state-of-the-art bi-directional transformer-based language models, which utilize tweet-level representations that enable transfer learning (e.g., BERT, RoBERTa, XLNet, AlBERT, and DistilBERT), proposed fusion-based approaches, and compared the developed models with several traditional machine learning, including deep learning, approaches. Using a public dataset, we evaluated the performances of the classifiers on their abilities to classify the non-majority "abuse/misuse" class. RESULTS: Our proposed fusion-based model performs significantly better than the best traditional model (F1-score [95% CI]: 0.67 [0.64-0.69] vs. 0.45 [0.42-0.48]). We illustrate, via experimentation using varying training set sizes, that the transformer-based models are more stable and require less annotated data compared to the other models. The significant improvements achieved by our best-performing classification model over past approaches makes it suitable for automated continuous monitoring of nonmedical PM use from Twitter. CONCLUSIONS: BERT, BERT-like and fusion-based models outperform traditional machine learning and deep learning models, achieving substantial improvements over many years of past research on the topic of prescription medication misuse/abuse classification from social media, which had been shown to be a complex task due to the unique ways in which information about nonmedical use is presented. Several challenges associated with the lack of context and the nature of social media language need to be overcome to further improve BERT and BERT-like models. These experimental driven challenges are represented as potential future research directions.

Impact of a State Opioid Prescribing Limit and Electronic Medical Record Alert on Opioid Prescriptions: a Difference-in-Differences Analysis.

BACKGROUND: Prescribing limits are one policy strategy to reduce short-term opioid prescribing, but there is limited evidence of their impact. OBJECTIVE: Evaluate implementation of a state prescribing limit law and health system electronic medical record (EMR) alert on characteristics of new opioid prescriptions, refill rates, and clinical encounters. DESIGN: Difference-in-differences study comparing new opioid prescriptions from ambulatory practices in New Jersey (NJ) to controls in Pennsylvania (PA) from 1 year prior to the implementation of a NJ state prescribing limit (May 2016-May 2017) to 10 months after (May 2017-March 2018). PARTICIPANTS: Adults with new opioid prescriptions in an academic health system with practices in PA and NJ. INTERVENTIONS: State 5-day opioid prescribing limit plus health system and health system EMR alert. MAIN MEASURES: Changes in morphine milligram equivalents (MME) and tablet quantity per prescription, refills, and encounters, adjusted for patient and prescriber characteristics. KEY RESULTS: There were a total of 678 new prescriptions in NJ and 4638 in PA. Prior to the intervention, median MME/prescription was 225 mg in NJ and 150 mg in PA, and median quantity was 30 tablets in both. After implementation, median MME/prescription was 150 mg in both states, and median quantity was 20 in NJ and 30 in PA. In the adjusted model, there was a greater decrease in mean MME and tablet quantity in NJ relative to PA after implementation of the policy plus alert (- 82.99 MME/prescription, 95% CI - 148.15 to - 17.84 and - 10.41 tabs/prescription, 95% CI - 19.70 to - 1.13). There were no significant differences in rates of refills or encounters at 30 days based on exposure to the interventions. CONCLUSIONS: Implementation of a prescribing limit and EMR alert was associated with an approximately 22% greater decrease in opioid dose per new prescription in NJ compared with controls in PA. The combination of prescribing limits and alerts may be an effective strategy to influence prescriber behavior.

Providing Incentive for Emergency Physician X-Waiver Training: An Evaluation of Program Success and Postintervention Buprenorphine Prescribing.

STUDY OBJECTIVE: Emergency department (ED) initiation of buprenorphine for patients with opioid use disorder increases treatment engagement but remains an uncommon practice. One important barrier to ED-initiated buprenorphine is the additional training requirement (X waiver). Our objective is to evaluate the influence of a financial incentive program on emergency physician completion of X-waiver training. Secondary objectives are to evaluate the program's effect on buprenorphine prescribing and to explore physician attitudes toward the incentive. METHODS: We conducted a prospective, observational cohort study set in 3 urban academic EDs before and after implementation of a financial incentive program providing $750 for completion of X-waiver training. We describe program participation as well as rates of buprenorphine prescribing per opioid use disorder-related encounter before and after the intervention period, using electronic health record data. We also completed a postintervention physician survey assessing attitudes about the incentive program. RESULTS: Overall, 89% of eligible emergency physicians (56/63) completed the X-waiver training during the 6-week incentive period. In the 5 months after the incentive, buprenorphine prescribing per opioid use disorder-related encounter increased from 0.5% to 16% (Δ 15%; 95% confidence interval 10.6% to 19.9%), with substantial variability across sites (range 8% to 22% of opioid use disorder-related encounters). In a postintervention survey, 67% of participating physicians indicated that they would have completed the training for a lower amount. CONCLUSION: A financial incentive paying approximately half the clinical rate was effective in promoting emergency physician X-waiver training. The effect on ED-based buprenorphine prescribing was positive but variable across sites, and likely dependent on the availability of additional supports.

Promoting Reproducible Research for Characterizing Nonmedical Use of Medications Through Data Annotation: Description of a Twitter Corpus and Guidelines.

BACKGROUND: Social media data are being increasingly used for population-level health research because it provides near real-time access to large volumes of consumer-generated data. Recently, a number of studies have explored the possibility of using social media data, such as from Twitter, for monitoring prescription medication abuse. However, there is a paucity of annotated data or guidelines for data characterization that discuss how information related to abuse-prone medications is presented on Twitter. OBJECTIVE: This study discusses the creation of an annotated corpus suitable for training supervised classification algorithms for the automatic classification of medication abuse-related chatter. The annotation strategies used for improving interannotator agreement (IAA), a detailed annotation guideline, and machine learning experiments that illustrate the utility of the annotated corpus are also described. METHODS: We employed an iterative annotation strategy, with interannotator discussions held and updates made to the annotation guidelines at each iteration to improve IAA for the manual annotation task. Using the grounded theory approach, we first characterized tweets into fine-grained categories and then grouped them into 4 broad classes-abuse or misuse, personal consumption, mention, and unrelated. After the completion of manual annotations, we experimented with several machine learning algorithms to illustrate the utility of the corpus and generate baseline performance metrics for automatic classification on these data. RESULTS: Our final annotated set consisted of 16,443 tweets mentioning at least 20 abuse-prone medications including opioids, benzodiazepines, atypical antipsychotics, central nervous system stimulants, and gamma-aminobutyric acid analogs. Our final overall IAA was 0.86 (Cohen kappa), which represents high agreement. The manual annotation process revealed the variety of ways in which prescription medication misuse or abuse is discussed on Twitter, including expressions indicating coingestion, nonmedical use, nonstandard route of intake, and consumption above the prescribed doses. Among machine learning classifiers, support vector machines obtained the highest automatic classification accuracy of 73.00% (95% CI 71.4-74.5) over the test set (n=3271). CONCLUSIONS: Our manual analysis and annotations of a large number of tweets have revealed types of information posted on Twitter about a set of abuse-prone prescription medications and their distributions. In the interests of reproducible and community-driven research, we have made our detailed annotation guidelines and the training data for the classification experiments publicly available, and the test data will be used in future shared tasks.