Effectiveness and efficiency of intensive care medicine: variable costs in different diagnosis groups.

Background: To establish the effectiveness of ICU treatment and the efficiency in the use of resources in patients stratified according to 10 diagnosis and two levels-of-care. To propose strategies aimed at reducing costs and improving efficiency in each patient group.

Methods: Multicentre prospective observational study. ICUs enrolled two cohorts of up to 10 consecutive patients with ICU stay >/= 48 h. Each with one of these diagnoses: trauma, brain-trauma, brain-hemorrhage, stroke, acute-on-chronic-obstructive-pulmonary disease, lung-injury/acute respiratory distress syndrome, heart failure, and scheduled/unscheduled abdominal surgery. The presence of active-life support divides high from low level-of-care treatments. Variable ICU costs were collected daily (bottom-up) for 21 days. We evaluated effectiveness (hospital survival) and efficiency (hospital-survivors variable-cost as a ratio of overall cost).

Results: Forty-two Italian general ICUs recruited 529 patients in 5 months. Mean ICU variable-costs significantly differed with diagnosis and level-of-care. Costs were positively affected by ICU length-of-stay, by duration of active-treatment. Outcome variably influenced costs. Medians of variable-costs per patient (1715 Euro) and patient-groups efficiencies (60.7%) identified four possible combinations between (low and high) cost and (low and high) efficiency groups. Moreover, efficiency was better than effectiveness in stroke, brain-hemorrhage and trauma, while it was worse in heart failure, acute-on-COPD or acute-lung injury. Overall ICU cost attributed only to survivors ranged from 699 (scheduled surgical) to 5906 (unscheduled surgical) Euro. Cost of non-survivors distributed to all patient was between 95 (scheduled-surgical) to 1633 (unscheduled-surgical) Euro.

Conclusions: Analysis of variable patient-specific cost was used as a tool to assess intensive care performance in patient subgroups with different diagnosis and levels-of-care.