Aims: To examine the rates of suicidal self-harm hospital presentations and death by intentional self-harm in two chronic pain populations, cancer and chronic non-cancer pain (CNCP) patients, who have been prescribed pharmaceutical opioids. To investigate sociodemographic and clinical risk factors associated with increased self-harm rates in these two populations and apply machine learning techniques to develop risk-prediction models. We hypothesise that self-harm rates will be common and will be associated with a number of clinical factors such as comorbid mental health problems and chronic, high dose opioid use.
Sample: We will leverage off an already established and funded population-based cohort, of all people in the state of New South Wales, Australia, who have initiated a pharmaceutical opioid from 2002-2019. It is estimated that more than two million people will have commenced a new opioid treatment episode in this time and just under one-million would have been prescribed an opioid for CNCP conditions. We will extend the study to examine suicidal outcomes.
Measures: Data will be linked to seven Commonwealth and NSW collections. We will examine two main outcomes. We will use the NSW Admitted Patient Data Collection and the NSW Emergency Department Data Collection to identify self-harm hospital presentations and the National Deaths Index to identify deaths related to intentional self-harm. Pharmaceutical opioid use will be operationalized to oral morphine equivalents, duration of use, i.e. acute, episodic and chronic, and concomitant opioid and other concomitant medicines use. We will utilize several datasets to identify patient groups, specifically cancer and CNCP patients, and sociodemographic (i.e. age, gender, area of residence, including relative socio-economic advantage and disadvantage) and clinical characteristics (i.e. mental health history, other prescription medication use and substance use history).
Procedures: Funding and ethics for the established cohort have been obtained and data linkage is underway and expected to be completed in March 2020. If funded, we will recruit a biostatistician with experience in data linkage.
Potential impact: Unlike the United States, Australia has universal health care that is available to everyone, irrespective of private health insurance. This means we will have a population-based sample with highly generalisable findings. The findings from this study will be crucial in identifying areas for suicide prevention and intervention, i.e. frequent self-harm hospital presentations and characteristics associated with suicidal behaviours. High risk patterns of opioid use, such as dose and duration, and associations with patient characteristics will also be identified and will assist in developing appropriate opioid prescribing guidelines.
