Counting the cost of ICU survivorship after acute lung injury
Nazir I. Lone
University of Edinburgh, Centre for Population Health Sciences
Corresponding Author
Senior Clinical Lecturer in Critical Care
Centre for Population Health Sciences
University of Edinburgh, Teviot Place
Edinburgh, EH8 9AG
+44 (0)131 651 1340
Source of support: None.
Running head: Counting the cost of ICU survivorship
MeSH: Critical care; Costs of illness; Acute Lung Injury
Word Count: 1066
Treating critically ill patients is expensive and costs are predicted to rise.[1, 2] These costs are likely to extend beyond the acute episode of critical illness. Detailed outcomes reported in an increasing body of literature relating to ICU patients, with and without acute lung injury, indicate that many patients survive a critical illness with significant ongoing impairment.[3, 4] However, uncertainty remains as to the cost associated with this ‘post-ICU syndrome’ as well as factors which might help identify those at risk of accruing the highest costs.
In this issue of AnnalsATS, Ruhl and colleagues provide a clear and detailed report of the costs accrued by a cohort of ICU survivors mechanically ventilated for acute lung injury in the two years after discharge from four hospitals in Baltimore, USA.[5] Using robust epidemiological and statistical methods, they found that the vast majority (80%) of their cohort required inpatient admission to a hospital or a skilled nursing or rehabilitation facility, accruing a median cost of $35,259 over the 2-year follow up period. The majority of these costs occurred during the first year of follow up.
In order to collect detailed cost data and a comprehensive list of relevant potential predictors or risk factors for higher costs, prospective methods are needed. The authors used structured interviews with participants, and cross-referenced this information with other data sources. Their impressive cohort retention rate (97%) compares favorably with previous studies in which ICU survivor cohorts have reported high loss to follow up rates resulting in potentially biased outcomes.[6] The disadvantage of using prospective interview-based methods to identify costs, acknowledged by the authors, is that we cannot know what happened to patients who were recruited but died after hospital discharge. In fact, 79 out of 217 potential survivors (36%) who agreed to follow up died before the two year time point and were therefore not included in analyses. This is particularly important as healthcare costs are usually greater in the period of time before death[7]; the study results, therefore, may underestimate post-critical care costs, and the reported relationship between potential risk factors and subsequent costs may alter.
In this survivor cohort, the majority of costs were related to hospital admission rather than admission to a skilled nursing or rehabilitation facility. While it is fairly clear from other work that hospital inpatient costs dominate other healthcare costs in ICU survivor cohorts, when societal costs are measured, these can be as great, if not greater, than healthcare costs .[8] Although these non-healthcare costs are less easy to quantify, the true cost of surviving a critical illness could encompass the costs of providing social care, loss of output due to non-employment or sickness absence, loss of earnings for caregivers, and the intangible costs attributable to newly acquired disabilities. These need to be considered to fully appreciate the financial consequences of an episode of critical illness from a societal perspective.
Is it possible to predict which ICU survivors will go on to accrue high inpatient costs in order to target post-discharge interventions? Given the relatively small sample size, there is a reasonable chance that the study was underpowered to exhaustively identify risk factors associated with costs, thus giving rise to ‘false negatives’ i.e. risk factors are not found to be significantly associated with costs when, in reality, a relationship exists. Inspecting the width of confidence intervals for coefficients supports this. Despite this limitation, the study found that a combination of pre-existing factors (comorbidity, having an informal care-giver pre-critical illness, insurance status) and acute illness factors (ICU length of stay) were associated with subsequent inpatient costs.
On a more technical statistical note, the authors neatly illustrate that the more familiar multivariable regression approaches cannot be used to identify predictors of higher costs due to the skewed distribution of cost data combined with a high count of individuals who accrue no costs. The authors used a two part modelling approach which, at first glance, appears complex as the reader is faced with two sets of regression coefficients (measures of risk) for each potential risk factor. However, the authors facilitate an intuitive interpretation of the results through clear, detailed explanations enhanced by graphical representation of ‘prototypical’ patients.
With no comparator group in this study, it remains unclear how much of the costs accrued after hospital discharge is attributable to the episode of critical illness. Are these costs related to newly acquired healthcare problems which can be causally related to the episode of critical illness? In order to estimate the costs attributable to surviving critical illness, other investigators have compared healthcare resource utilisation in ICU survivors with control populations of non-critically ill individuals[9, 10] or compared pre- and post-critical illness healthcare resource utilisation.[11] Valid comparisons with a comparator population rest on the principle of ‘exchangeability’[12]: the healthcare costs in the ICU survivor population would have been the same as the healthcare costs in the control population had the individuals in the control population been admitted to and survived ICU and vice versa. The same argument must hold for comparisons of pre- and post-ICU costs within individuals to be valid. Given the difficulty meeting these criteria, well designed studies are needed to better identify which post-ICU costs are attributable to the episode of critical illness.
This study has helped to define the costs associated with ICU survivorship. If we assume that at least some of these costs are attributable to the critical illness episode, the next step is to consider if there is something that can be done to reduce these costs. A more detailed investigation of the aetiology of post-ICU costs is likely to reveal complex, multifactorial interplay of factors. Could inpatient costs be reduced or averted by proactively following up patients to identify health and social problems that are amenable to solving? Should we be even trying to reduce these costs if they, in fact, represent appropriate interventions? Focussing future studies to answer these questions will be instrumental in informing health and social care policy relating to ICU survivors.
Despite the increase in research relating to ICU survivorship, fundamental questions still need to be answered. The results presented in this carefully written manuscript fill in some of these knowledge gaps and are a timely addition to the literature. In addition, the authors’ adherence to recommended reporting guidelines [13, 14], their use of robust statistical methods, and their thoughtful presentation of data has resulted in a well-constructed, accessible manuscript.
References
1. Coopersmith, C.M., et al., A comparison of critical care research funding and the financial burden of critical illness in the United States. Crit Care Med, 2012. 40(4): p. 1072-9.
2. Amaral, A.C. and G.D. Rubenfeld, The future of critical care. Current Opinion in Critical Care, 2009. 15(4): p. 308-13.
3. Desai, S.V., T.J. Law, and D.M. Needham, Long-term complications of critical care. Critical Care Medicine, 2011. 39(2): p. 371-9.
4. Herridge, M.S., et al., Functional disability 5 years after acute respiratory distress syndrome. New England Journal of Medicine, 2011. 364(14): p. 1293-304.
5. Ruhl, A.P. and R. Lord, Healthcare Resource Utilization and Costs of 2-year Survivors of Acute Lung Injury: An Observational Cohort Study. Annals of the American Thoracic Society, 2015. xx: p. xx.
6. Williams, T.A. and G.D. Leslie, Challenges and possible solutions for long-term follow-up of patients surviving critical illness. Australian Critical Care, 2011. 24(3): p. 175-85.
7. Wunsch, H., et al., Use of intensive care services during terminal hospitalizations in England and the United States. Am J Respir Crit Care Med, 2009. 180(9): p. 875-80.
8. Lone, N.I., et al., Surviving Intensive Care: A Systematic Review of Healthcare Resource Use After Hospital Discharge. Critical Care Medicine, 2013. 41(8): p. 1832-1843.
9. Wunsch, H., et al., Three-year outcomes for Medicare beneficiaries who survive intensive care. JAMA, 2010. 303(9): p. 849-56.
10. Keenan, S.P., et al., Intensive care unit survivors have fewer hospital readmissions and readmission days than other hospitalized patients in British Columbia. Critical Care Medicine, 2004. 32(2): p. 391-8.
11. Prescott, H.C., et al., Increased 1-year healthcare use in survivors of severe sepsis. Am J Respir Crit Care Med, 2014. 190(1): p. 62-9.
12. Hernán, M.A., A definition of causal effect for epidemiological research. Journal of Epidemiology and Community Health, 2004. 58(4): p. 265-271.
13. von Elm, E., et al., Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. BMJ, 2007. 335(7624): p. 806-8.
14. Husereau, D., et al., Consolidated Health Economic Evaluation Reporting Standards (CHEERS) statement. BMJ, 2013. 346: p. f1049.