Cargando…
How real can we get in generating real world evidence? Exploring the opportunities of routinely collected administrative data for evaluation of medical devices
Real‐world data are considered a potentially valuable source of evidence for assessing medical technologies in clinical practice, but their widespread use is hampered by numerous challenges. Using the case of coronary stents in Italy, we investigate the potential of administrative databases for esti...
Autores principales: | , |
---|---|
Formato: | Online Artículo Texto |
Lenguaje: | English |
Publicado: |
John Wiley and Sons Inc.
2022
|
Materias: | |
Acceso en línea: | https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9796733/ https://www.ncbi.nlm.nih.gov/pubmed/35762465 http://dx.doi.org/10.1002/hec.4562 |
Sumario: | Real‐world data are considered a potentially valuable source of evidence for assessing medical technologies in clinical practice, but their widespread use is hampered by numerous challenges. Using the case of coronary stents in Italy, we investigate the potential of administrative databases for estimating costs and health outcomes associated with the use of medical devices in real world conditions. An administrative dataset was created ad hoc by merging hospital records from patients admitted between 2013 and 2019 for stent implantations with ambulatory records, pharmaceutical use data and vital statistics. Health outcomes were multifold: all‐cause and cardiac mortality and myocardial infarction, within 30 days, 1, 2, 5 years. Costs were estimated from the National Health System perspective. We used multivariable Cox models and propensity score (PS) methods (PS matching; stratification on PS; inverse probability of treatment weighting using PS; PS adjustment). 257,907 coronary stents were implanted in 113,912 patients. For all health outcomes and follow‐up times, and across all methods, patients receiving drug‐eluting stents (DES) presented lower risk. For all‐cause mortality, the DES patient advantage over bare‐metal stent (BMS) patients declined over time but remained significant even at 5 years. For myocardial infarction, results remained quite stable. The DES group presented lower cumulative total costs (ranging from 3264 to 2363 Euros less depending on methods). Our results confirm the consolidated evidence of the benefits of DES compared to BMS. The consistency of results across methods suggests internal validity of the study, while highlighting strengths and limitations of each depending on research context. Administrative data yield great potential to perform comparative effectiveness and cost‐effectiveness analysis of medical devices provided certain conditions are met. |
---|