Health care costs, long-term survival, and quality of life following intensive care unit admission after cardiac arrest

The annual incidence of sudden cardiac arrest in central Europe is approximately nine arrests per 10,000 inhabitants [1]. Thus, more than 600,000 people in Europe may be affected each year. Since the 1960s, immediate cardiopulmonary resuscitation (CPR) has been considered life-saving for sudden cardiac arrest [2], and following successful CPR, patients are routinely admitted to intensive care units (ICUs) to manage both the causes and acute sequelae. ICUs consume a large proportion of hospital budgets yet care for a minority of patients [3]. Longstanding economic constraints, present within all health care systems, create pressures to ration ICU care ethically [4]. Restricting the demands for futile medical services by limiting access to the ICU [5], at least for those patients likely to die anyway [6], has been proposed as a theoretical model to lower expenditures. In patients with sudden cardiac arrest, ICU and hospital lengths of stay are often protracted and incurred health care costs are high. Despite high short-term mortality and significant morbidity [7], long-term functional capacity for those surviving the initial hospitalization remains good [8]. Objective cost-outcome studies, integrating costs and quality-adjusted life years (QALYs) gained, are required to determine whether ICU admission constitutes a reasonable use of constrained resources in this patient population. To investigate the costs and long-term health status outcomes after CPR for out-of-hospital or in-hospital cardiac arrest, we conducted an individual patient-level assessment of health status at 5 years post-ICU discharge and combined these outcomes with a fully costed economic evaluation. All consecutive patients admitted to a single tertiary ICU were eligible for follow-up. We calculated the costs per survivor, costs per life year gained, and costs per QALY gained. A sensitivity analysis was conducted to model the impact of changes in utility (life years gained and health status index [HSI]) on the development of cost-outcome indices.

Patients who survive cardiac arrest are often considered to have a grim prognosis and discussion ensues as to whether they should be universally welcomed to the ICU: costs are expected to be high and benefits are expected to be moderate at best. Herein, we present information from a cohort of patients with HRQL obtained at 5 years following cardiac arrest and subsequent CPR. We found that HRQL 5 years after hospital discharge was only slightly lower than age- and gender-matched apparently healthy German controls. In addition, both the reported survival (ICU and hospital stays) [24] and HRQL of our cohort did not differ significantly when compared with publications based on similar patient populations (in-hospital and out-of-hospital cardiac arrest) [8], other medical ICU patients [25, 26], or ICU patients with sepsis [27]. We followed our patients for 5 years to allow sufficient recovery time before the assessment of health status. This time frame was selected because the slope of the survival curve can be expected to proceed in parallel with that of a control population and it is unlikely that patients' HRQL attributable to the index hospitalization will improve [25, 26, 28, 29]. These assumptions permit the prediction of the cohort's remaining life span based on data from the German Census Bureau and the calculation of a valid HSI by close approximation. Mean patient costs per ICU day were twice those obtained for average ICU patients admitted to the same institution (2,693 € compared with 1,334 € in less severely ill patients [30]). Consequently, costs per 5-year survivor were also considerably higher than the average ICU patient (68,116 € versus 14,130 € [26]). This is attributable to both higher costs per ICU stay and higher short-term (that is, ICU and hospital) mortality in the cohort of patients with cardiac arrest; however, the health status outcomes and costs per life year saved and per QALY of our patients compared favorably with general cardiovascular and pulmonary ICU patient populations from the same ICU [26]. They also compare favorably with cost-outcome profiles of a variety of other interventions routinely undertaken in the critically ill as well as non-ICU patients (Table 4). Our findings are robust under a wide range of sensitivity analyses adjusting for patient mortality rates and HSI projected over the estimated remaining life expectancy. In simulation, significant cost increases per life year gained or per QALY gained were observed only after the mortality was increased to over 50% (Figure 3) or by decreasing the HSI below 0.58.

There are several unique aspects of our study which should be considered further. First, health care expenditures do not usually end with hospital discharge, especially for critically ill patients. In our cohort, the estimated costs incurred for post-hospital discharge health care services of all 5-year survivors surpassed their initial ICU and hospital costs by more than twofold. It is important to point out that, in the absence of specific data for patients following cardiac arrest, we based these estimates on average age- and gender-adjusted health care utilization costs provided by the German Bureau of Census. Since the majority of the patients in our cohort were readmitted to hospital at least once during the 5-year follow-up, true long-term health care costs of this patient group may be above the expected averages reported by the Bureau of Census; however, there is no reason to expect that they are above the average that could be expected for other ICU survivors. It is important to note that we did not consider the potential overall costs for the society as a whole (earlier retirement, higher employees' sickness rates, and so on). The compilation of such data is largely based on estimations and was beyond the scope of this study. Second, the expected remaining life span of our 5-year survivors was also based on census data. Since the best estimates suggest that the hazard of death for survivors of in-hospital arrest parallels the appropriately age- and gender-matched general population after 2 years [31], it is reasonable to assume that the hazard of death for our patient cohort parallels the appropriate age- and gender-matched population. Thus, the application of standard life table (actuarial) methods, using population-based life expectancy tables, is likely a reasonable approach for estimating remaining life expectancy. Third, there is a dilemma with patients surviving hospital discharge with severe neurological deficits (as defined here with a GCS score of less than 6 points). What is the quality of life or HSI of such a patient? To the best of our knowledge, we cannot judge. On a utility scale from zero (representing death) to one (representing perfect health), such patients definitely do not represent one but are most likely not zero either. We therefore calculated costs of these patients separately for life years gained but did not include any utility measures. In these patients, we did not discount life expectancy although one may expect shorter overall survival in this group [32]. Thus, costs incurred for this subgroup as well as for the whole population are most likely overestimated. Fourth, both socioeconomic status and occupational class may affect patients' perception of quality of life, with patients belonging to lower status groups reporting a quicker decline in self-reported health [33]. However, we did not assess socioeconomic status or occupational class of our patients and thus could not adjust our data. Finally, we should point out that the Utstein style protocol [34] of basic and advanced life support was not available in our patient population. Although this would have been desirable for auditing resuscitation efforts, it was deemed of secondary interest since our study focused on outcomes and costs, not quality of care.

Despite some restrictions that emerge owing to the methodologic complexities inherent in any cost-outcome description conducted in ICU patients [20, 35], we found that the costs per life year and per QALY gained for patients with cardiac arrest who require ICU admission are reasonable (approximately 9,930 € and 13,000 €, respectively). Moreover, our data highlight the somewhat skewed notion that extreme expenses result from the care of patients who have undergone basic life support following cardiac arrest. Although it may be true that patients with cardiac arrest incur considerable costs and resource consumption, the trade-off between input and output, costs and outcome, justifies such resource allocation, at least in comparison with other ICU patient groups. We believe our study is the first to demonstrate that patients who leave the hospital following cardiac arrest without severe neurological disabilities may expect fair long-term survival and quality of life for reasonable expenses to the health care system.