Table 3.

Comparison of costs and outcomes between programme and non-programme ICUs

	Mean	Median	2.5th–97.5th centile
Non-programme ICU
CLABSIs	52	52	39–66
Deaths	8	8	2–14
Costs*	$987 000	$937 000	$488 000–$1 760 000
Programme ICU
CLABSIs	10	9	3–29
Deaths	2	1	0–5
Costs$*	$738 000	$710 000	$453 000–$1 190 000
Benefit of programme†
CLABSIs prevented	42	42	23–58
Deaths averted	6	6	2–12
Net costs$	−$249 000	−$221 000	−$976 000 to $300 000
Incremental cost-effectiveness ratio (probability)
Cost per CLABSI prevented	Strongly dominant (0.80)‡
Cost per death prevented	Strongly dominant (0.80)‡

Mean, median, 2.5% and 97.5% centile estimates for outputs from probabilistic sensitivity analysis of 10 000 model runs representing uncertainty in epidemiological and economic parameters are reported.

All mean, median, and centile values are expressed per 1000 patients to make the scale easier to interpret. Values have been rounded to three significant digits at most.

*Costs are not presented separately for each outcome (CLABSI and death) because no additional cost was assumed to occur for death; discounted at 3%.

†Benefit of programme determined by subtracting programme ICU estimates from non-programme ICU estimates within the model.

‡Probability that the programme is more effective and less costly than current practice.

CLABSI, central line-associated bloodstream infection; ICU, intensive care unit.