Systematic Review of Decision-Analytic Models Estimating the Economic Impact of Prescription Opioid Use and Associated Adverse Events
Author(s)
Bowen Yang, MSc1, Rachel Ann Elliott, PhD2, Darren Ashcroft, PhD3, Joe Hilton, PhD4.
1University of Manchester, Manchester, United Kingdom, 2University of Manchester, Chorley, United Kingdom, 3NIHR Greater Manchester Patient Safety Research Collaboration (GM PSRC), Manchester, United Kingdom, 4The University of Manchester, Manchester, United Kingdom.
1University of Manchester, Manchester, United Kingdom, 2University of Manchester, Chorley, United Kingdom, 3NIHR Greater Manchester Patient Safety Research Collaboration (GM PSRC), Manchester, United Kingdom, 4The University of Manchester, Manchester, United Kingdom.
OBJECTIVES: Prescription opioid use is common and can lead to opioid use disorder and overdose. Other mental and physical adverse events (AEs) (e.g. cognitive impairment, fractures, cardiovascular events) may be overlooked, resulting in underestimation of the economic impact of harm. This review aimed to identify and appraise models estimating economic impact of prescription opioid use and identify research gaps to inform future model development.
METHODS: Medline, Embase, PsycINFO, EconLit, CINAHL and Web of Science were searched systematically (01/01/1990-10/01/2025). Title/abstract and full text of records were screened. Data on model structure, input sources, outcomes, and uncertainty analysis were extracted and descriptively synthesised. The AdViSHE checklist and an evidence hierarchy were used to assess model validity and input data quality. Model open-access availability was checked.
RESULTS: Fifty-nine studies and 45 distinct model structures were identified. Models examined the long-term impact of opioid use by simulating AEs during prolonged use (n=8), treatment and outcomes after AEs (n=22), or both (n=15). Gastrointestinal events (n=16) were the most frequently modelled AEs with short-term impacts, and opioid dependence (n=35) was the most frequently modelled AE with long-term impacts. Forty-four studies used cohort-based Markov or decision tree models, and 15 studies adopted microsimulation or system dynamics models to model long-term consequences or opioid use trajectories. Most studies used low-quality to moderate-quality data and conducted 0-2 of the validation steps listed in AdViSHE. Five studies provided source code for their model.
CONCLUSIONS: Markov models and decision trees are the predominant approaches used to model the economic impact of prescription opioid use. However, these methods provide limited capacity to model time-dependency and interactions between AEs. Few models incorporated multiple AEs, and the downstream outcomes of serious AEs beyond dependence were rarely modelled. A transparent and validated model incorporating all clinically important AEs is needed to estimate the economic impact of prescription opioid use.
METHODS: Medline, Embase, PsycINFO, EconLit, CINAHL and Web of Science were searched systematically (01/01/1990-10/01/2025). Title/abstract and full text of records were screened. Data on model structure, input sources, outcomes, and uncertainty analysis were extracted and descriptively synthesised. The AdViSHE checklist and an evidence hierarchy were used to assess model validity and input data quality. Model open-access availability was checked.
RESULTS: Fifty-nine studies and 45 distinct model structures were identified. Models examined the long-term impact of opioid use by simulating AEs during prolonged use (n=8), treatment and outcomes after AEs (n=22), or both (n=15). Gastrointestinal events (n=16) were the most frequently modelled AEs with short-term impacts, and opioid dependence (n=35) was the most frequently modelled AE with long-term impacts. Forty-four studies used cohort-based Markov or decision tree models, and 15 studies adopted microsimulation or system dynamics models to model long-term consequences or opioid use trajectories. Most studies used low-quality to moderate-quality data and conducted 0-2 of the validation steps listed in AdViSHE. Five studies provided source code for their model.
CONCLUSIONS: Markov models and decision trees are the predominant approaches used to model the economic impact of prescription opioid use. However, these methods provide limited capacity to model time-dependency and interactions between AEs. Few models incorporated multiple AEs, and the downstream outcomes of serious AEs beyond dependence were rarely modelled. A transparent and validated model incorporating all clinically important AEs is needed to estimate the economic impact of prescription opioid use.
Conference/Value in Health Info
2025-11, ISPOR Europe 2025, Glasgow, Scotland
Value in Health, Volume 28, Issue S2
Code
EE678
Topic
Economic Evaluation, Methodological & Statistical Research, Study Approaches
Topic Subcategory
Cost/Cost of Illness/Resource Use Studies
Disease
Injury & Trauma, Musculoskeletal Disorders (Arthritis, Bone Disorders, Osteoporosis, Other Musculoskeletal), No Additional Disease & Conditions/Specialized Treatment Areas, Surgery, Systemic Disorders/Conditions (Anesthesia, Auto-Immune Disorders (n.e.c.), Hematological Disorders (non-oncologic), Pain)