Periods of altered risk for non-fatal drug overdose: a self-controlled case series

Abstract

Background Being recently released from prison or discharged from hospital, or being dispensed opioids, benzodiazepines, or antipsychotics have been associated with an increased risk of fatal drug overdose. This study aimed to examine the association between these periods and non-fatal drug overdose using a within-person design.Methods In this self-controlled case series, we used data from the provincial health insurance client roster to identify a 20% random sample of residents (aged ≥10 years) in British Columbia, Canada between Jan 1, 2015, and Dec 31, 2017 (n=921 346). Individuals aged younger than 10 years as of Jan 1, 2015, or who did not have their sex recorded in the client roster were excluded. We used linked provincial health and correctional records to identify a cohort of individuals who had a non-fatal overdose resulting in medical care during this time period, and key exposures, including periods of incarceration, admission to hospital, emergency department care, and supply of medications for opioid use disorder (MOUD), opioids for pain (unrelated to MOUD), benzodiazepines, and antipsychotics. Using a self-controlled case series, we examined the association between the time periods during and after each of these exposures and the incidence of non-fatal overdose with case-only, conditional Poisson regression analysis. Sensitivity analyses included recurrent overdoses and pre-exposure risk periods. FindingsWe identified 4149 individuals who had a non-fatal overdose in 2015-17. Compared with unexposed periods (ie, all follow-up time that was not part of a designated risk period for each exposure), the incidence of non-fatal overdose was higher on the day of admission to prison (adjusted incidence rate ratio [aIRR] 2•76 [95% CI 1•51-5•04]), at 1-2 weeks (2•92 [2•37-3•61]), and 3-4 weeks (1•34 [1•01-1•78]) after release from prison, 1-2 weeks after discharge from hospital (1•35 [1•11-1•63]), when being dispensed opioids for pain (after ≥4 weeks) or benzodiazepines (entire use period), and from 3 weeks after discontinuing antipsychotics. The incidence of non-fatal overdose was reduced during use of MOUD (aIRRs ranging from 0•33 [0•26-0•42] to 0•41 [0•25-0•67]) and when in prison (0•12 [0•08-0•19]).Interpretation Expanding access to and increasing support for stable and long-term medication for the management of opioid use disorder, improving continuity of care when transitioning between service systems, and ensuring safe prescribing and medication monitoring processes for medications that reduce respiratory function (eg, benzodiazepines) could decrease the incidence of non-fatal overdose.

Non-fatal overdose risk during and after opioid agonist treatment: A primary care cohort study with linked hospitalisation and mortality records

“…Negative binomial regression models allowed to study repeated overdoses without censoring the observation period on the occurrence of the first event as applied previously. 21 , 26 This expands our understanding pertaining to overdoses, since censoring restricts the denominator when calculating rate ratios, possibly leading to under-estimation of overdose risk. Although we accounted to some extent for residual confounding, differences pertaining to treatment modality and overdose risk might not be exclusively attributable to treatment modality but instead, mediated by treatment retention, illicit drug use, or affected by patients’ varying degrees of dependence.…”

“…A recently conducted self-controlled case-series study by Keen et al 21 examined non-fatal overdoses among patients who received OAT and, revealing that patients who were exposed to OAT had a lower non-fatal overdose risk compared to unexposed patients. That study highlighted the time-dependent risk of non-fatal overdose for people who received OAT with 1.24 decrease in non-fatal overdose incidence risk from weeks 1-2 to weeks 3-4 during treatment initiation, a stabilisation for the remaining period in treatment and a possible increase for the weeks 1-4 after treatment cessation.…”

“…First, propensity scores were generated on the probability of being prescribed methadone versus buprenorphine and then the inverse probability was calculated to estimate stabilised IPWs. Age at baseline and gender, socioeconomic position 18 (using patient neighbourhood and, if missing, practice locality IMD quintiles), ethnicity, region, history of overdose or self-harm, 2 severe mental illness diagnosis, 19 alcohol misuse, 20 use of antipsychotics or benzodiazepines, 21 in the year before the index date all may act as potential confounders. These variables were used to balance baseline characteristics and generate a pseudo-population that was weighted using IPWs.…”

Overdose deaths and the COVID‐19 pandemic in British Columbia, Canada

“…This analysis used data collected by the BC Coroners Service as part of their ongoing surveillance of illicit drug toxicity deaths. Future studies may consider linking data to investigate other factors that are known to influence overdose risk such as discontinuation from opioid agonist treatment or release from prison [15]. These data could help to identify factors associated with increased vulnerability to overdose death during COVID‐19 that are not fully captured in this analysis.…”

Association between contact with mental health and substance use services and reincarceration after release from prison

“…First, we divided the exposures into early versus later service initiation, using separate indicator variables if the first contact occurred early (<30 days after release) versus later (≥30 days after release). This one-month period is widely regarded as a critical time window during which people released from prison are especially vulnerable to substance-related harm [ 34 ], and there is evidence that contact with health services during the first month post-release is important to ensuring ongoing engagement with care [ 35 ]. Second, we divided the exposures into single service use versus repeat service use, with separate indicator variables representing the period between first and second service contact (if any), and the period following the second contact.…”