The impact of structural assumptions on decision uncertainty: a case study on the prevention of neonatal infections in preterm babies

Research output: Working paper


Early modelling coupled with value of information (VOI) typically reflects any parameter uncertainty; however there is little guidance on how to include the uncertainty from structural assumptions in this process. This study shows the impact of failure to consider structural uncertainty in the VOI results.
This study was conducted as part of the PREVAIL project, which evaluates the effectiveness and cost-effectiveness of Anti-Microbial Impregnated Peripherally Inserted Central Catheters (AM-PICCs) to prevent neonatal infections in preterm babies. In the model, the two key structural assumptions are: (i) infection increases the risk of disability and death later in life, based on observational studies; (ii) AM-PICC reduce infection risk, based on a trial in older children. Sensitivity and VOI analyses with bias adjustment explore the impact of uncertainty and the value of future research.
Avoiding impairment in babies is associated with a lifetime gain of up to 10 QALYs and £19,009 in savings per infant for the NHS. If neonatal infections cause NDI and death as suggested by the observational evidence, AM-PICCs have a probability of being cost-effective of over 90% at all thresholds above £2000/QALY even for small reductions in the risk of infection. The value of additional research is £234,452 per year and it is mostly related to the uncertainty around the effectiveness of AM-PICC. If the uncertainty on the effect of infection on impairment and death is increased to reflect the structural uncertainty, AM-PICCs are still cost-effective but the value of additional research increases to £560,795.
The cost-effectiveness of AM-PICCs depends not only on its effectiveness but also on the link between infection and NDI. Failure to consider structural uncertainty in the VOI analysis risks underestimating decision uncertainty and the value of future research. More research is needed on methods to include structural uncertainty in VOI analysis.
Original languageEnglish
Publication statusUnpublished - Jun 2018

Cite this