Table 2

Predictors of risk for a defined adverse outcome following opioid use in AI research, reported in 34 included studies

Adverse outcomeNumber of studies (%)Variable predictive of risk
Prolonged opioid use after surgery7 (21)Age; marital status; preoperative opioid use and duration; preoperative medications (antidepressants, benzodiazepines, nonsteroidal anti-inflammatory drugs, gabapentin and beta-2-agonists); medications commonly used to treat anxiety and insomnia; preoperative haemoglobin; tobacco use; comorbidity of depression or diabetes; instrumentation; Medicaid insurance; and particular pharmacy ordering sites.
Opioid use disorder4 (12)Mean annual amount of opioid use days; number of overlaps in opioid prescriptions per year; mean annual opioid prescriptions; annual benzodiazepine and muscle relaxant prescriptions; initiation of marijuana before 18 years; pain; mental health issues; traumatic brain injury; and male gender. Dynamics through time-in-treatment of decision-making parameters and symptom intensity (craving, anxiety and withdrawal symptoms).
Opioid dependence3 (9)Psychopathy; higher WBC and respiratory disturbances; malnutrition, and reduced sensitivity to loss.
Opioid poisoning and overdose3 (9)Sedative, hypnotic or anxiolytic dependence; arrest history; the number of overdoses in a person’s social network; early refills; total days’ supply; concomitant use of antidepressants; concomitant use of antipsychotics; total opioid claims; and high-dose opioid-benzodiazepine use.
Opioid abuse and misuse2 (6)Violation of opioid agreements; release from prison; and an indicator for an arrest.
Chronic opioid therapy1 (3)More than 10 mg of morphine equivalent/per day during hospitalisation; two or more opioid prescriptions filled in the year preceding the index hospitalisation; past year receipt of non-analgesic pain medications; and past year receipt of benzodiazepines.
Emergency department opioid prescription1 (3)CT scan ordered, abdominal pain and back pain.
  • AI, artificial intelligence.