Analytical approach:
A decision-analytic model and data from published studies were used to simulate a hypothetical cohort of comatose patients who had experienced the return of spontaneous circulation after a witnessed ventricular fibrillation OHCA. The time horizon was the lifetime of the patient and the authors stated that the study perspective was societal.
Effectiveness data:
The effectiveness data were from published literature. The authors reported that most of the estimates used in the model were from a single randomised controlled trial (Hypothermia after Cardiac Arrest Study Group. 2002, see 'Other Publications of Related Interest' below for bibliographic details). The main clinical effectiveness estimate was the neurological outcomes of the patients, which were based on Cerebral Performance Category scores.
Monetary benefit and utility valuations:
Quality of life data for cardiac arrest survivors was obtained from the published literature.
Measure of benefit:
Quality-adjusted life-years were the benefit measure.
Cost data:
The costs included those of the cooling devices, blankets, and pads; nurse time; staff training; tubing and intravenous fluids; emergency visits; intensive care unit stay; and care following hospital discharge. The costs were from the manufacturers of the cooling equipment, the authors of RCTs, and hospital administrators. The costs of standard equipment for intensive care were from purchasing administrators at two large academic institutions. The estimates of equipment depreciation were from device manufacturers and those for equipment usage were from hospital equipment administrators. Care costs were from published Markov models. All costs were reported in 2008 prices, and the currency was US dollars ($).
Analysis of uncertainty:
: A series of one- and two-way sensitivity analyses was undertaken by varying each model parameter across a wide range of values. All parameters were assigned a distribution, and 10,000 Monte Carlo simulations were performed to assess the overall variability in the costs and outcomes, and the proportion of simulations below a $100,000 per QALY threshold. The results of the one-way sensitivity analyses were presented in a tornado diagram, while those of the probabilistic sensitivity analysis were presented in a scatter plot.