Medication for Opioid Use Disorder After Nonfatal Opioid Overdose and Association With Mortality

By Marc R. Larochelle, Dana Bernson, T. Land, T. Stopka, Na Wang, Z. Xuan, S. Bagley, J. Liebschutz, A. Walley. Annals of Internal Medicine, 2018

906 citations recorded by the source corpus.

The United States is in the midst of a crisis of opioid-related harms (1). Some efforts to address this crisis focus on expanding access to effective treatment of opioid use disorders (OUDs) (2). Prior nonfatal opioid overdose is a known risk factor for subsequent nonfatal and fatal overdoses (37), and engaging persons in treatment who survive an overdose may be effective in limiting subsequent fatalities. However, data on the association between treatment of OUD and mortality after a nonfatal overdose are limited to a single retrospective cohort study that analyzed enrollment in methadone maintenance treatment (MMT) at a single time point and found no association (3). The 3 medications for OUD (MOUD) approved by the U.S. Food and Drug Administration are methadone, buprenorphine, and naltrexone. Randomized controlled trials of these medications have shown consistent benefits across many outcomes, including increased treatment retention and suppression of illicit opioid use (810). A recent systematic review and meta-analysis of 19 observational cohort studies identified substantial reductions in all-cause and overdose mortality for methadone and buprenorphine (11). However, the mortality benefit in this analysis was limited to time actively retained in treatment, and the 4-week period after discontinuation was associated with an especially high risk for death. The few studies that examined mortality among patients receiving naltrexone show an unclear effect (1215). Massachusetts has been particularly affected by the opioid crisis: Opioid overdose deaths more than tripled between 2010 and 2016 (16). Through Chapter 55 of the Acts of 2015, the state legislature permitted individual-level linkage of data from 16 Massachusetts government agencies to gain a deeper understanding of the circumstances that influence fatal and nonfatal opioid overdoses (17). For this analysis, we identified a cohort of persons in the Chapter 55 data set who survived an opioid overdose and described any episodes of treatment with MOUD before and after that overdose. Specifically, we sought to determine whether treatment with MOUD, including receipt of MMT, buprenorphine, or naltrexone, was associated with reduced risk for all-cause and opioid-related mortality. Methods Study Design and Data Source We did a retrospective cohort study using the Massachusetts Chapter 55 data set, which includes data between 2011 and 2015 on residents aged 11 years or older with health insurance (as reported in the Massachusetts All-Payer Claims Database [APCD]) and represents more than 98% of Massachusetts residents. Data from APCD were linked at the individual level with records from other data sets using a multistage deterministic linkage technique described elsewhere (18). For this study, we used 7 linked Massachusetts databases: APCD, the Registry of Vital Records and Statistics, the prescription monitoring program, the Acute Hospital Case Mix, the Ambulance Trip Record Information System, the Bureau of Substance Addiction Services licensed treatment encounters, and the cancer registry. This work was mandated by Massachusetts law and conducted by a public health authority that required no institutional board review. The Boston University Medical Campus Institutional Review Board also determined that this study was not human subjects research. Cohort Selection We identified persons who had had a nonfatal opioid overdose between January 2012 and December 2014 to allow 12 months of observation before and after the overdose. We restricted the cohort to persons aged 18 years or older because access to OUD treatment substantially differs in adolescents versus adults (19). We identified opioid overdose in 2 ways. First, we identified emergency department, observation, or inpatient encounters with a medical claim containing a diagnosis code for opioid poisoning from the International Classification of Diseases, Ninth Revision, Clinical Modification (codes 965.00, 965.01, 965.02, 965.09, E850.0, E850.1, and E850.2). A study validated these codes by showing positive predictive values of 81% for identifying fatal or nonfatal opioid overdose and 94% for an opioid overdose or opioid-related adverse event (20). Second, we identified persons with an ambulance encounter for opioid overdose (available in 2013 and 2014 only). In collaboration with the Centers for Disease Control and Prevention, the Massachusetts Department of Public Health created and refined an algorithm to use with emergency medical services data to identify opioid-related overdoses; this algorithm was previously validated against internal emergency medical services data on opioid overdose events (Supplement). Supplement. Supplementary Material We examined the first qualifying event (nonfatal opioid overdose) for each person, hereafter called the index overdose. Of 20155 persons with an event, we excluded 1203 who died within 30 days after the overdose using dates of death from the Registry of Vital Records and Statistics. We excluded 1338 persons with evidence of cancer at any time in the 5 years of Chapter 55 data because of high competing risk for death. Cancer was identified using International Classification of Diseases, Ninth Revision, diagnosis codes in APCD (Supplement) or entry in the state-based cancer registry. We also excluded 46 persons whose age or sex was unknown, yielding a final cohort of 17568 persons. Key Variables We identified exposure to MOUD in monthly intervals. Exposure to MMT was identified in 2 ways: a medical claim from APCD for methadone administration via Healthcare Common Procedure Coding System code H0020 or a record of treatment with methadone in data from the Bureau of Substance Addiction Services. We used the prescription monitoring program to identify dispensing of buprenorphine or buprenorphine and naloxone combined. Naltrexone was identified via a pharmacy claim for injectable or oral naltrexone in APCD. We examined all-cause and opioid-related mortality as identified in death files. Classification of opioid-related death was based on medical examiner determination or standardized assessment by the Massachusetts Department of Public Health (Supplement). We examined potential confounding variables. We obtained patient sex and age from APCD and categorized age as 18 to 29 years, 30 to 44 years, or 45 years or older. We identified monthly dispensings of opioid analgesics and benzodiazepines from the prescription monitoring program. We identified diagnosis of anxiety or depression using International Classification of Diseases, Ninth and Tenth Revisions, diagnosis codes from APCD (Supplement). We identified OUD treatment services, including inpatient detoxification episodes and postdetoxification treatment in short- and long-term residential facilities, through the Bureau of Substance Addiction Services. Statistical Analysis To compare baseline characteristics by receipt of MOUD, we developed the following 5 categories of MOUD receipt in the 12 months after the index overdose: no MOUD during follow-up, 1 or more months of buprenorphine, 1 or more months of methadone, 1 or more months of naltrexone, and 1 or more months of 2 or 3 MOUDs combined. We compared baseline characteristics among these mutually exclusive treatment groups using 2 tests. We did time-to-event analyses for all-cause and opioid-related mortality using MOUD as a monthly time-varying exposure variable. We used 2 dichotomous classifications for MOUD exposure, with discontinuation and on treatment. Several studies have shown an increased risk for all-cause and opioid-related mortality in the 4 weeks immediately after MOUD discontinuation (11, 21). Thus, we defined a with discontinuation exposure variable, which we considered the primary classification, to attribute any effect of MOUD discontinuation on mortality to the MOUD. For this classification, persons were considered exposed to MOUD in any month in which they received it and in the month after last receipt. We defined an on treatment exposure variable as the secondary classification, in which persons were considered exposed to MOUD only in months in which they received it (Figure 1). Figure 1. MOUD exposure classification. For the primary classification (with discontinuation), MOUD exposure extends through the month after discontinuation (light and dark-green months). For the secondary classification (on treatment), exposure is limited to months in which treatment is received (light-green months only). In the illustrative examples, participant 1 is not exposed to MOUD through follow-up; participant 2 is exposed in months 12 and 712 for the primary classification and months 1 and 712 for the secondary classification. In the month of death, participant 3 would be considered exposed in the primary classification only, participant 4 would be considered exposed in both primary and secondary exposure classifications, and participant 5 would be considered not exposed to MOUD. MOUD= medication for opioid use disorder. We used an extended KaplanMeier estimator allowing for time-varying exposure to MOUD to generate cumulative incidence curves (Supplement) (22). We developed a multivariable Cox regression model of time to all-cause and opioid-related mortality. The predictors of interest were monthly receipt of MMT, buprenorphine, and naltrexone as time-varying exposure variables. Covariates were age; sex; monthly time-varying receipt of prescription opioids, benzodiazepines, and OUD treatment services; baseline characteristics, including mental health diagnoses; and prior receipt of medication or OUD treatment services. We calculated the E-value to identify the minimum strength of association that an unmeasured confounder would need to have with both treatment and outcome, conditional on the measured covariates, to explain away the observed associations between MOUD and mortality (23). We used SAS Studio, version 3.5 (SAS Institute), for analyses (Supplemen

Subfields: Anesthesiology and Pain Medicine, Epidemiology, Public Health, Environmental and Occupational Health

View Paper: https://www.semanticscholar.org/paper/215a9c4763dc72dee389f0957839956cec1586bd