Opioid drug utilization and comparative safety: a population-based cohort study

Study type

• Study type:
  Non-interventional study
• Scope of the study:
  Assessment of risk minimisation measure implementation or effectiveness

  Drug utilisation

Non-interventional study

• Non-interventional study design:
  Cohort

Study drug and medical condition

• Anatomical Therapeutic Chemical (ATC) code:
  (N02AA01) morphine

  morphine

  (N02AX02) tramadol

  tramadol

  (N01AH01) fentanyl

  fentanyl

  (R05DA04) codeine

  codeine

  (A01AD05) acetylsalicylic acid

  acetylsalicylic acid

  (N02BE01) paracetamol

  paracetamol

  (N02BB02) metamizole sodium

  metamizole sodium

  (M01AE01) ibuprofen

  ibuprofen

  (M01AC01) piroxicam

  piroxicam
• Medical condition to be studied:
  Cardiovascular event prophylaxis

  Spinal fracture

  Femoral neck fracture

  Sleep apnoea syndrome

  Constipation

  Drug dependence

  Drug abuse

  Death
• Additional medical condition(s):
  Delirium, wrist and humerus fracture. co-morbid conditions and prescriptions that act as confounders

Population studied

• Age groups:
  Adults (18 to < 46 years)

  Adults (46 to < 65 years)

  Adults (65 to < 75 years)

  Adults (75 to < 85 years)

  Adults (85 years and over)
• Estimated number of subjects:
  1186887

Study design details

• Main study objective:
  To characterize the population that consume opioids and to compare the incidence of pre-defined adverse events between incident users of opioids and compared to non-opioid pain killers in subjects of at least 18 years old in Catalonia (Spain).
• Data analysis plan:
  Drug Utilisation: Descriptive analysis for categorical and continuous variables. Kaplan-Meier for opioid treatment persistence.Drug Safety: Unadjusted IR (and 95%CIs) of the events of interest stratified by drug exposure cohort will be calculated. We will use a propensity score-matched survival analysis (PMS) to compare time to event between users of opioids, opioids vs anti-inflammatory drug users and between opioids vs non-anti-inflammatory drug users. PMS will be calculated by fitting multivariable logistic regression models (including confounders). On PMS, drug users will be matched to active comparison drug users, using a caliper width of 0.2 SD. Both “intention to treatment” and “on treatment” analyses will be adopted when estimate association between drugs. We will account for competing risk of death. Adjusted multivariable models will be done if there is remaining imbalance after PMS. Cox regression models will be used to estimate relative risk according to drug use.