CEDRN: Opioid Registry

Every day 21 Canadians die from opioid poisoning. Opioid poisoning deaths are predominantly driven by high potency, synthetic opioids (e.g. fentanyl, carfentanil) and other substances, which are highly addictive. Seven years after the toxic unregulated drug crisis was first declared a public health emergency, the human impact is staggering. More than 40,000 Canadians have lost their lives from opioid poisoning since 2016, most commonly people who are between 20 and 59 years old. Toxic illicit drugs have become the largest driver of potential years-of-life lost in our country. Emergency departments are frequented by people who use drugs. They are accessible 24-hours per day, 365-days per year without the need to schedule an appointment. Emergency department visits for substance use have increased, and more than half of patients who ultimately die from an opioid poisoning visit an emergency department in the year before they die, yet 20% of them leave against medical advice before their evaluation or treatment is complete. Among patients who leave against medical advice, 9% experience another episode of opioid poisoning within one week. Emergency department visits are opportunities to initiate treatment and connect people to harm reduction services, and may be the last chance to save a life. The main objectives are to:1) Harmonize data collection on patient and presentation characteristics, emergency department and in-hospital treatments and services, and the clinical course of patients who present to emergency departments with opioid poisoning and/or dependence across participating sites to create the CEDRN Opioid Registry; 2) Describe variation in and associations between presentations for opioid poisoning and/or dependence, clinical practice and patient outcomes including the safety and effectiveness of treatments and services geographically and over time; a) Determine the association between naloxone administration and precipitated opioid withdrawal (safety outcome) and time to opioid toxicity reversal (effectiveness outcome).b) Determine the association between buprenorphine/naloxone induction method and leaving against medical advice (safety outcome), and 30- and 60-day, and 3- and 12-month continued use of opioid agonist therapy and mortality (effectiveness outcomes).3) Derive a risk stratification tool to predict the risk of 60-day mortality in patients presenting to emergency departments with opioid poisoning and/or dependence to identify the highest risk patients for harm reduction services and treatments in our resource constrained environment. The investigators hypothesize that patient and/or visit characteristics will predict patient outcomes among patients presenting to emergency departments with opioid poisoning and/or dependence. Derivation of clinical decision rules to identify the highest risk patients will allow physicians to prioritize them for potentially life-saving treatments and services. The investigators also hypothesize that certain treatments or other variations in clinical practice will be associated with improved outcomes among patients presenting to emergency departments with opioid poisoning and/or dependence. These will be used to guide clinical care, decision-making and policy. The investigators will leverage existing infrastructure and governance to create a registry of consecutive eligible patients presenting to emergency departments with opioid poisoning and/or dependence. The investigators will create a single nationally-harmonized curated accessible dataset. It will capture the patient journey from ambulance arrival to hospital discharge filling critical gaps in administrative data, including individual-level sociodemographic variables, clinical signs and symptoms, investigations, pre-hospital, ED and hospital medications, interventions and clinical course. The investigators will include all patients who arrive at the emergency department and screen positive for opioid poisoning and/or dependence using the following screening criteria:1) Medication orders in the present or past (last 2 years) ED encounters: Buprenorphine, Buprenorphine-Naloxone, Methadone, Kadian, or Naloxone2) The following ED discharge diagnosis: Opioid overdose or intoxication, opioid withdrawal, overdoses of any of the following: Percocet, Oxycontin, Oxycodone, Overdose, Methadone, Heroin, Fentanyl, Codeine, Carfentanil3) Inpatient discharge diagnosis: Opioid use disorder (mild/moderate/severe, abuse/dependence), or opioid-related disorders. Adverse effect, and poisoning by polysubstance use. Opioid intoxication or withdrawal. Intoxication or withdrawal (from any substance, including unspecified ones). Adverse effect, and poisoning by opioids, benzodiazepines, other antiepileptic and sedative-hypnotic drugs, and unspecified drugs. Any psychoactive substance abuse resulting in intoxication, withdrawal, or any adverse effects. The investigators will include patients in cardiac arrest being resuscitated by paramedics on arrival and patients presenting without valid health care cards (e.g., comatose without an ID) using institution-assigned temporary IDs which are then assigned valid provincial health numbers. The investigators will track multiple visits by the same patient by assigning them the same Study ID using temporary (if needed) and provincial health numbers. The investigators will include patients with no fixed address and those who report staying in a shelter. The investigators will identify patient presentations as being due to opioid poisoning, opioid dependence, or polysubstance use to subgroup the cohort for our assessment of patients' risk profile in pre-planned subgroup analyses. The investigators will apply the following standardized definitions and refine them, if needed, to achieve high between-rater agreement between research assistants and physicians:1) Opioid poisoning. The investigators define opioid poisoning as: (1) historical accounts by EMS personnel, bystanders or the patient of presumed opioid use; and (2) a respiratory rate of <10 breaths/minute documented in EMS or emergency department records or reported by bystanders in the context of decreased level of consciousness.30,35,36 The diagnosis is supported by a known exposure to opioids, but consumption may have been unwitnessed or unintentional (e.g., through contamination of another substance with fentanyl). Miosis (small pupils) and hypoxia (low blood oxygen levels) are often but not always present. The diagnosis is supported by known exposure to opioids, but consumption may have been unwitnessed or unintentional (e.g., contamination of another drug). Miosis (small pupils) and hypoxemia (low blood oxygen) are often but not always present. Breathing improves after naloxone administration or a period of observation. If treatment was delayed, the patient may suffer cardiac arrest and/or brain injury from low blood oxygen. There was no obvious alternative cause (e.g., head trauma) to explain the patient's presentation. Research assistants will document whether clinicians implied that the patient experienced opioid poisoning or intoxication in the medical record, in the absence of the definition being met. 2) Opioid dependence. Patients with opioid dependence experience withdrawal symptoms (see below) after opioid use. Patients with withdrawal symptoms, a documented diagnosis of opioid use disorder in the list of ED discharge diagnoses or as a comorbidity in clinician notes, and/or an active prescription for OAT at the index visit would be classified as a patient with opioid dependence. Research assistants will document whether clinicians implied that the patient has opioid dependence in the medical record, in the absence of the definition being met.3) Opioid withdrawal. The investigators define opioid withdrawal as the documentation of any of the following in the medical record: new or worsening nausea requiring treatment, new or worsening agitation, aggressive behavior towards staff, restlessness, heart rate >100 beats per minute, diarrhea, tremor, flushing, sweating, piloerection (gooseflesh skin), bone or joint aches, rhinorrhea, lacrimation, or yawning, or documentation of a score >5 on the Clinical Opioid Withdrawal Score scale, which has been incorporated into medical records at many sites and validated for use by emergency nurses. Precipitated withdrawal occurs within 1h of naloxone or buprenorphine/naloxone administration. Research assistants will document whether clinicians implied the presence of opioid withdrawal in the medical record, in the absence of the definition being met. Outcomes will include: (a) emergency department discharge, (b) leaving against medical advice or without being seen, (c) hospitalization, (d) naloxone administration, (e) provision of a take-home naloxone kit, (f) administration of opioid agonist therapy (methadone, slow release buprenorphine, buprenorphine/naloxone, slow-release oral morphine), (g) consultations, (h) opioid poisoning reversal, (i) opioid withdrawal, (j) in-hospital mortality, and (k) death determined by neurologic criteria. The investigators have planned linkages of registry with administrative data to ascertain (l) retention on opioid agonist therapy, and (m) mortality at 30 and 60-days, and 3 and 12-months. Analysis: The investigators will use summary statistics to describe the patient population, including patient demographics and proxies for Indigeneity, prehospital and emergency department presentation characteristics, and patient outcomes. The investigators will use descriptive statistics to understand the treatments and services provided by EMS and in emergency departments, and to understand the dosing data distribution and the routes of administration of naloxone, as well as the method of buprenorphine/induction used. The investigators will stratify analyses by site characteristics (province, urban/rural, academic/non-academic) and over time to describe practice variation. The investigators will stratify our analyses by sex and gender, as biological sex is a known risk factor for opioid-related mortality, and non-conforming gender (e.g., transgender identity) is associated with health access barriers and poor health outcomes. The investigators will also stratify our analyses using proxies for Indigeneity as the Indigenous population is overrepresented in the patient population affected by the toxic drug crisis and may be treated differently in the ED due to systemic racism. While the proxies for Indigeneity available in medical records are not comprehensive, the VCHRI Indigenous Health Research Unit encouraged us to collect what is available. They or the Indigenous Advisory will guide us in the use and interpretation of these variables. The investigators will assign patients who report staying in a shelter to the shelter's forward sortation area, and those with no fixed address to the forward sortation area of the hospital they presented to.The investigators will derive the clinical decision rule using state-of-the-art methodology. The investigators will pre-define candidate predictor variables based on current knowledge. Sex-at-birth, gender identity, and proxies for Indigeneity will be considered potential predictor variables. The investigators will use multiple imputation for variables with incomplete data (the investigators anticipate <5% missing data based on prior studies) so all cases can be included. The investigators will assess potential predictor variables for multi-collinearity. The investigators will investigate sex and gender differences, given reported sex and gender-based differences in outcomes through potential interactions with other candidate variables. The investigators will fit logistic regression models and use spline functions for continuous variables. The investigators will then assess the strength of associations using an ANOVA plot to inform how many degrees of freedom to allocate to each variable in the model. The investigators will use a fast stepdown procedure to create a more concise model, then internal bootstrap validation to assess for over-fitting and to calculate optimism. The final reduced model will have points allocated to each variable based on their predictive strength. The investigators will calculate the sensitivity and specificity of the clinical decision rule, along with discrimination and calibration of regression-based risk scores. The investigators will present the performance of the derived rule using each patient's total point score and predicted risk category (low, medium, high) as follows: (a) discrimination between cases that experience the outcome of interest or not, using the area under the receiver operating curve with 95% CIs; (b) calibration by graphically comparing the observed and expected probabilities of outcomes for each level of the score, and the Hosmer-Lemeshow χ2 goodness-of-fit statistic; and (c) classification performance of the scoring system and risk categories using 95% CIs for sensitivity, specificity, and predictive values. A Knowledge Translation committee will ensure our work remains relevant and impactful to knowledge users and patients by employing effective strategies refined to support the dissemination of findings from the study. Patients with lived/living experience and Indigenous Advisory will continue to help refine research questions, variables, screening procedures and outcomes. The investigators plan to present our results back to them, to assist with interpretation of our findings. Knowledge users include CanadiEM (generate infographics), Emergency Care BC (provincial health improvement network), the Canadian Association of Emergency Physicians (national rounds), and Public Health and Poison Centre collaborators who will facilitate translation of our work into practice. Others will engage with us via emails, briefing notes, and in presentations and webinars. The investigators will seek guidance from the Indigenous Advisory Committee on who should receive findings to ensure they reach Indigenous Peoples, including non-academic healthcare providers. The investigators will develop simple, accessible knowledge translation materials to support patients in emergency settings.