Analytical approach:

The analysis was based on one study with a maximum follow-up of three months, which was projected to a lifetime horizon. The authors did not explicitly state the perspective adopted.

Effectiveness data:

The clinical analysis was part of the Elderly in European Intensive Care Units (ELDICUS) project, a prospective multicentre cohort study that enrolled consecutive adults, in 11 hospitals in seven countries, between September 2003 and March 2005. There were 7,449 patients enrolled, with 6,312 in the ICU group and 1,137 in the ward group. The mean age was 59.3 years for ICU care and 65.7 years for ward care. Patients were referred to the ICU by doctors in the emergency department, on the medical ward, on the surgical ward, or in the operating room. Possible confounders, such as age, chronic health, and the indication for referral, were taken into account using statistical analyses. The primary outcome was the 28-day mortality, while three-month mortality was a secondary outcome.

Monetary benefit and utility valuations:

Not considered.

Measure of benefit:

Lives saved and life-years saved were the summary benefit measures. Life expectancy was based on official statistics.

Cost data:

The economic analysis included the costs of the ICU stay and the ward stay. A top-down approach was used to calculate the daily hospital costs based on the cost-block method, which considered consumables (drugs, nutritional products, blood products, and disposables), clinical support services (laboratory, radiology, and physiotherapy), and staff (junior and senior doctors, and nurses). Overheads were considered. The calculation was made for each country in the study. The costs were determined in the country's currency and then converted into US dollars ($) and Euros (EUR). Statistical analyses were carried out to consider the impact of potential confounders, as in the clinical analysis.

Analysis of uncertainty:

: Three alternative scenarios were considered: the exclusion of centres with extreme results for effectiveness, the exclusion of patients with ICU admissions due to terminal cancer or admitted for observation, and ICU daily costs based on the level of care.

The relative risk of death at 28 days was 0.7 (95% CI 0.5 to 0.9) in ICU patients compared with ward patients. It was 1.5 in patients with less than 5% predicted mortality, 0.7 (p<0.05) in patients with 5% to 40% predicted mortality, and 0.6 (p<0.01) in patients with over 40% predicted mortality.

After three months, the relative risk of death was 0.7 (95% CI 0.5 to 1.0). It was 1.2 for less than 5%, 0.8 for 5% to 40%, and 0.5 (p<0.01) for over 40% predicted mortality.

The incremental cost per life saved with ICU triage over control was $103,771 (95% CI 56,855 to 150,687) or EUR 82,358 (95% CI 45,123 to 119,593).

The incremental cost per life-year saved was $7,065 (95% CI 3,871 to 10,259) or EUR 5,607 (95% CI 3,072 to 8,142).

The cost-effectiveness of ICU admission increased with increasing predicted mortality. The results were similar in all sensitivity analyses and when using the three-month mortality data.

Interventions:

The selection of the comparators was appropriate as the proposed intervention (ICU triage) was compared against the conventional approach for these patients.

Effectiveness/benefits:

The authors stated that a randomised trial was not feasible because it would not have been ethical to allocate patients to treatment that was expected to be less effective than the usual care. Thus, the observational design appears to have been valid. The main limitation of the clinical data was the baseline differences between the groups in their demographic and clinical factors. The authors used statistical tools to consider the impact of these confounding factors on the clinical endpoints. The multicentre enrolment and the large number of patients should have ensured a representative sample. The benefit measure appears to have been appropriate as short-term mortality was the main outcome for critically ill patients admitted to the ICU.

Costs:

The perspective was not explicitly stated, but it appears to have been that of the third-party payer. The authors stated that a top-down approach, with a cost-block method, was used to estimate the daily cost for the ICU and the ward; this is a common approach for ICU treatment. It was acknowledged that a bottom-up approach would have been more detailed, but was too expensive to carry out. The resource use came directly from the large observational trial, but little information on the sources for the unit costs was provided. Appropriate statistical analyses were conducted to account for possible confounding. There was no need for discounting, given the short-term analysis. The price year was not reported.

Analysis and results:

The results were presented selectively, without the expected costs, and only the incremental cost-effectiveness ratios were given. The uncertainty was partly investigated in alternative scenarios. Appropriate statistical tests were carried out to consider ranges of likely values for the model outcomes. The authors acknowledged that the main limitation of their study was the observational nature of the clinical analysis, which produced baseline differences between the groups, but this was necessary on ethical grounds. The results were valid for several countries and can be extrapolated to similar settings; a subgroup analysis by country would have been interesting.

Concluding remarks:

The analytic framework was valid and, despite some methodological limitations, the authors’ conclusions appear to be robust.

Edbrooke DL, Minelli C, Mills GH, Iapichino G, Pezzi A, Corbella D, Jacobs P, Lippert A, Wiis J, Pesenti A, Patroniti N, Pirracchio R, Payen D, Gurman G, Bakker J, Kesecioglu J, Hargreaves C, Cohen SL, Baras M, Artigas A, Sprung CL. Implications of ICU triage decisions on patient mortality: a cost-effectiveness analysis. Critical Care 2011; 15(1):R56

Subject indexing assigned by NLM

Adult; Aged; Cost-Benefit Analysis; Europe /epidemiology; Female; Health Resources /utilization; Hospital Mortality; Humans; Intensive Care Units /economics /statistics & Male; Middle Aged; Patient Admission /statistics & Patients' Rooms /economics /statistics & Risk Assessment; Treatment Outcome; Triage; numerical data; numerical data; numerical data

22011000566

29/06/2011