Prescribing patterns before and after a non-fatal drug overdose using Tennessee's controlled substance monitoring database linked to hospital discharge data

Person-centered hospital discharge data: Essential existing infrastructure to enhance public health surveillance of maternal substance use disorders in the midst of a national maternal overdose crisis

OBJECTIVES

As crises of drug-related maternal harms escalate, US public health surveillance capacity remains suboptimal for drug-related maternal morbidities. Most state hospital discharge databases (HDDs) are encounter-based, and thus limit ascertainment of morbidities to delivery visits and ignoring those occurring during the 21 months spanning pregnancy and postpartum year. This study analyzes data from a state that curates person-centered HDD to compare patterns of substance use disorder (SUD) diagnoses at delivery vs. the full 21 pregnancy/postpartum months, overall and by maternal social position.

METHODS

Among people who experienced an in-hospital birth in New York State between 9/1/2016 and 1/1/2018 (N = 330,872), we estimated SUD diagnosis (e.g., opioids, stimulants, benzodiazepines, cannabis) prevalence at delivery; across the full 9 months of pregnancy and 12 postpartum months; and by trimester and postpartum quarter. Risk ratio and risk difference estimated disparities by race/ethnicity, age, rurality, and payor.

RESULTS

The 21-month SUD prevalence rate per 100,000 was 2671 (95% CI 2616-2726), with 31% (29.5%-31.5%) missing SUD indication when ascertained at delivery only (1866; 95% CI 1820-1912). Quarterly rates followed a roughly J-shaped trajectory. Structurally marginalized individuals suffered the highest 21-month SUD prevalence (e.g., Black:White risk ratio=1.80 [CI:1.73-1.88]).

CONCLUSION

By spanning the full 21 months of pregnancy/postpartum, person-centered HDD reveal than the maternal SUD crisis is far greater than encounter-based delivery estimates had revealed. Generating person-centered HDD will improve efforts to tailor interventions to help people who use drugs survive while pregnant and postpartum, and eliminate inequities.

Ensemble learning to predict opioid-related overdose using statewide prescription drug monitoring program and hospital discharge data in the state of Tennessee

Objective

To develop and validate algorithms for predicting 30-day fatal and nonfatal opioid-related overdose using statewide data sources including prescription drug monitoring program data, Hospital Discharge Data System data, and Tennessee (TN) vital records. Current overdose prevention efforts in TN rely on descriptive and retrospective analyses without prognostication.

Materials and Methods

Study data included 3 041 668 TN patients with 71 479 191 controlled substance prescriptions from 2012 to 2017. Statewide data and socioeconomic indicators were used to train, ensemble, and calibrate 10 nonparametric “weak learner” models. Validation was performed using area under the receiver operating curve (AUROC), area under the precision recall curve, risk concentration, and Spiegelhalter z-test statistic.

Results

Within 30 days, 2574 fatal overdoses occurred after 4912 prescriptions (0.0069%) and 8455 nonfatal overdoses occurred after 19 460 prescriptions (0.027%). Discrimination and calibration improved after ensembling (AUROC: 0.79–0.83; Spiegelhalter P value: 0–.12). Risk concentration captured 47–52% of cases in the top quantiles of predicted probabilities.

Discussion

Partitioning and ensembling enabled all study data to be used given computational limits and helped mediate case imbalance. Predicting risk at the prescription level can aggregate risk to the patient, provider, pharmacy, county, and regional levels. Implementing these models into Tennessee Department of Health systems might enable more granular risk quantification. Prospective validation with more recent data is needed.

Conclusion

Predicting opioid-related overdose risk at statewide scales remains difficult and models like these, which required a partnership between an academic institution and state health agency to develop, may complement traditional epidemiological methods of risk identification and inform public health decisions.

Prescription Opioid Characteristics and Nonfatal Overdose Among Patients Discharged from Tennessee Emergency Departments

BACKGROUND

Despite increasing trends of nonfatal opioid overdoses in emergency departments (EDs), population-based studies comparing prescription opioid dosing patterns before and after nonfatal opioid overdoses are limited.

OBJECTIVES

To evaluate characteristics of prescribing behaviors before and after nonfatal overdoses, with a focus on opioid dosage.

METHODS

Included were 5,395 adult residents of Tennessee discharged from hospital EDs after a first nonfatal opioid overdose (2016-2017). Patients were linked to eligible prescription records in the Tennessee Controlled Substance Monitoring Database. We estimated odds ratios (OR) and 95% confidence intervals (CI) to evaluate characteristics associated with filling opioid prescriptions 90 days before overdose and with high daily dose (≥ 90 morphine milligram equivalents) 90 days after overdose.

RESULTS

Among patients who filled a prescription both before and after an overdose, the percentage filling a low, medium, and high dose was 33.7%, 31.9%, and 34.4%, respectively, after an opioid overdose (n = 1,516). Most high-dose users before an overdose (>70%) remained high-dose users with the same prescriber after the overdose. Male gender, ages ≥ 35 years, and medium metro residence were associated with increased odds of high-dose filling after an opioid overdose. Patients filling overlapping opioid-benzodiazepine prescriptions and with > 7 days' supply had increased odds of filling high dose after an opioid overdose (OR 1.4, 95% CI 1.08-1.70 and OR 3.7, 95% CI 2.28-5.84, respectively).

CONCLUSIONS

In Tennessee, many patients treated in the ED for an overdose are still prescribed high-dose opioid analgesics after an overdose, highlighting a missed opportunity for intervention and coordination of care between ED and non-ED providers.

Impact of benzodiazepines and polysubstance status on repeat non-fatal drug overdoses

Research has shown that benzodiazepines and mental health disorders can increase the likelihood of repeat overdose, but researchers have not explored this association in Tennessee (TN). We examined benzodiazepines, polysubstance overdose status with/without benzodiazepines, and mental health comorbidities with repeat overdose using statewide data in TN. This study analyzed TN hospital discharge data on nonfatal overdoses for patients ages 18-64 from 2012 to 2016 for 21,066 patients with an initial inpatient visit and 36,244 patients with an initial outpatient visit. The study assessed each patient at one year after initial overdose to determine likelihood of repeat overdose. We used a Cox proportional hazards model to compute hazard ratios (HRs) and 95% confidence intervals (CIs) to determine the factors associated with repeat nonfatal overdose. Repeat overdose rates, by one year after index overdose, were 12.9% of the sample for inpatients and 13.9% of the sample for outpatients. The visit factors (overdose characteristics and comorbidities determined from the initial visit) that the study found to be independently associated with repeat overdoses among inpatients were polysubstance status (HR: 0.88, 95% CI 0.78-0.99), benzodiazepine/polysubstance interaction (HR: 1.29, 95% CI 1.02-1.64), and presence of any mental health disorder (HR: 1.28, 95% CI: 1.18-1.39). For outpatients, the benzodiazepine/polysubstance interaction (HR: 1.21, 95% CI 1.01-1.44) was significant without adjusting for demographic factors. We found evidence that benzodiazepine/polysubstance status and mental health disorders were associated with repeat overdose for inpatients, and that benzodiazepine/polysubstance status was associated with repeat overdose for outpatients. Findings support the need to include polysubstance status and mental health in overdose prevention efforts.