This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.

Objectives: Clinical studies of mechanical ventilation (MV) are often small with large uncertainty in outcomes. Meta-analysis provides a method to combine data into a single estimate of efficacy. Meta-analysis of proportional assist ventilation+ (PAV+) versus pressure support ventilation (PSV) was recently undertaken but did not report outcomes relevant to our cost-effectiveness model. A pragmatic meta-analysis was undertaken to provide estimates of efficacy and explore how data sources used impact on outcomes. Methods: A Markov-model of patient care from MV in the intensive care unit (ICU) through to discharge home or death was developed for the Canadian setting. Structured searches identified studies of PAV+ versus PSV that were then subject to meta-analysis. Outcomes of interest were MV/ICU/hospital time, tracheostomy, and ICU/hospital mortality. The model was populated with efficacy inputs from either Canadian trials or meta-analysis estimates. Outcomes were over 20 years, with costs in 2017 CAD with quality-adjusted life years (QALYs) using EQ-5D. Performed sensitivity analyses (n=2,000) used a willingness-to-pay (WTP) threshold of CAD 50,000 per QALY gained. Results: Seven studies comparing PAV+ with PSV were identified, totalling 271 PAV+ patients and 253 PSV patients and meta-analysis included at least 4 studies for each outcome. The cost of care with PSV was CAD 141,003 and 6.07 QALYs were accrued. Using Canadian data, the cost of care with PAV+ was CAD 129,333 and 6.29 QALYs were accrued, making PAV+ dominant. With meta-analysis data, PAV+ cost CAD 147,276 and accrued 6.98 QALYs over the 20 years, meaning it was cost effective at CAD 6,875 per QALY gained. In the Canadian scenario 80% of simulations were under the WTP threshold, compared with 100% when using meta-analysis. Conclusions: In our test, efficacy data from individual trials or meta-analysis substantially changed the numerical results but the interpretation remained largely intact. PAV+ is expected to be cost-effective for MV in Canada.

Background