Parameters of adapted Markov cohort cost-utility model*
| Parameter | Base case | Distribution† |
|---|---|---|
| Patient cohort demographic characteristics | ||
| Age, mean ± SD, yr | 67 ± 1215 | Normal |
| Female sex (SE), % | 39.7 (0.13)33 | β |
| Patient outcomes in acute phase of mechanical ventilation‡ | ||
| Patients with clusters of ineffective efforts, mean (estimated 95% CI), % | 38 (29.1 to 47.3)33 | β |
| Patient asynchrony on entering model‡ | ||
| Asynchrony > 10% at initiation of mechanical ventilation if clusters of ineffective efforts, % ± SD | 8.5 ± 1.833 | β |
| Asynchrony > 10% at initiation of mechanical ventilation if no clusters of ineffective efforts, % ± SD | 1.5 ± 1.033 | β |
| Reference efficacy standard of care (PSV), mean (95% CI) | ||
| Duration of mechanical ventilation, d | 8.1 (4.5 to 28.3)34 | Normal |
| Time in intensive care unit, d | 12.6 (7.4 to 33.3)34§ | Normal |
| Time in hospital, d | 43.5 (18.6 to 68.4)34§ | Normal |
| Spontaneous breathing trial success, % | 77.9 (73.8 to 82.1)23 | β |
| Liberation success, % (95% CI) | 85.3 (85.1 to 85.6)35 | β |
| Adverse event rates, mean (95% CI), % | ||
| Tracheostomy | 26.0 (8.1 to 44.0)15 | β |
| Ventilator-associated pneumonia | 8.8 (5.7 to 11.9)34 | β |
| Nosocomial infection | 0.85 (0.66 to 1.04)36 | β |
| Intensive care unit death | 25.4 (20.7 to 30.1)34 | β |
| Hospital death | 30.3 (25.3 to 35.3)34 | β |
| Postdischarge death | ||
| Year 1 | 12.5 (12.4 to 12.6)37 | β |
| Year 2 | 19.3 (19.2 to 19.5)37¶ | β |
| Year 3 | 27.5 (27.3 to 27.7)37¶ | β |
| Year 4 | onward Life tables¶ | β |
| Comparative effectiveness, PAV+ mode v. PSV** | ||
| Total duration of mechanical ventilation, mean (95% CI), d | −1.53 (−2.24 to −0.83) | Normal |
| Intensive care unit length of stay, mean (95% CI), d | −1.54 (−2.19 to −0.90) | Normal |
| Hospital length of stay, mean (95% CI), d | −1.83 (−2.51 to −1.16) | Normal |
| Successful weaning/liberation, OR (95% CI) | 1.49 (0.59 to 3.79) | Log-normal |
| Intensive care unit death, OR (95% CI) | 0.70 (0.41 to 1.20) | Log-normal |
| Hospital death, OR (95% CI) | 0.70 (0.40 to 1.22) | Log-normal |
| Tracheostomy, OR (95% CI) | 0.76 (0.44 to 1.31) | Log-normal |
| Extubation failure/reintubation, OR (95% CI) | 0.52 (0.25 to 1.08) | Log-normal |
| Asynchrony index ≥ 10, OR (95% CI) | 0.13 (0.07 to 0.23) | Log-normal |
| Costs†† | ||
| Intensive care unit, cost per day, mean (range of reported means), $ | 2765 (2354–3690)38 | γ |
| General ward, cost per day, mean (range of reported means), $ | 019 (717–1400)38 | γ |
| Mechanical ventilation initiation, cost per event, mean (95% CI), $ | 139 (125 to 153)39 | γ |
| Mechanical ventilation maintenance, cost per day, mean (95% CI), $ | 851 (766 to 936)39 | γ |
| Tracheostomy, cost per event, mean (95% CI), $ | 4193 (3908 to 4477)40 | γ |
| Ventilator-associated pneumonia, cost per day, mean (95% CI), $ | 58 (30 to 73)41,42 | γ |
| Other nosocomial infection, cost per event, mean (± 10%), $ | 870 (783 to 956)43 | γ |
| PSV, purchase cost, $ | 0‡‡ | γ |
| PAV+ mode, 1-time purchase cost, $ | 27 00022§§ | γ |
| After discharge, annual cost, mean (95% CI), $ | ||
| Year 1–2 | 13 707 (6241 to 37 631)44¶ | γ |
| Year 3 onward | 10 032 (5835 to 17 169)44¶ | γ |
| Ventilator-associated pneumonia, additional length of stay, median (range), d | 9.5 (8.8–10.1)27 | Normal |
| Health state utility, mean (95% CI) | ||
| Baseline | 0.776 (0.677 to 0.899)22 | Normal |
| Mechanical ventilation | −0.390 (−0.590 to 0.090)22 | Normal |
| Intensive care unit | 0.402 (0.362 to 0.442)22 | Normal |
| Hospital | 0.520 (0.450 to 0.590)22 | Normal |
| After discharge to 1 yr | 0.550 (0.480 to 0.610)22 | Normal |
| Adverse event disutility | ||
| Tracheostomy | 0¶¶ | Normal |
| Ventilator-associated pneumonia | 0*** | Normal |
| Extubation failure | 0¶¶ | Normal |
Note: CI = confidence interval, OR = odds ratio, PAV+ mode = proportional-assist ventilation with load-adjustable gain factors, PSV = pressure-support ventilation, SD = standard deviation, SE = standard error.
↵* Canadian data in italics.
↵† We made the choice of distribution to reflect the uncertainty of each parameter from the perspective of population-level uncertainty as opposed to uncertainty at the individual patient level.
↵‡ See Figure 1.
↵§ In the analysis by Sinuff and colleagues,34 no upper bound was presented owing to the patient’s remaining in hospital. For our calculations, we assumed that the upper bound is given by: mean + (mean – lower bound).
↵¶ Used for scenario analyses only.
↵** Seven clinical studies comparing PAV+ mode to PSV15–17,23–26 were identified by K.J.B. and in systematic reviews.45,46 As these systematic reviews did not report on all required outcomes, and no single study presented robust clinical data on the required model inputs, we determined the comparative efficacy of PAV+ mode versus PSV by means of a pragmatic meta-analysis (Appendix 1).
↵†† 2017 Canadian dollars.
↵‡‡ Conservative assumption.
↵§§ Assumed to be $24.64 per day of use, assuming a 5-year life cycle and that the ventilator is in use on 60% of days. Probabilistic model inputs (used for the probabilistic sensitivity analysis) were based on input variance, calculated from reported CIs.
↵¶¶ Assumed none in addition to mechanical ventilation.
↵*** Additional duration of mechanical ventilation is assumed to cover the disutility.