| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Ann Thorac Surg 2002;73:1472-1478
© 2002 The Society of Thoracic Surgeons
a Department of Surgery, College of Physicians and Surgeons of Columbia University, New York, New York, USA
b Department of Anesthesiology, College of Physicians and Surgeons of Columbia University, New York, New York, USA
Accepted for publication January 21, 2002.
* Address reprint requests to Dr Williams, Surgical Arrhythmia Program, MHB 7-126, 177 Ft. Washington Blvd, New York, NY 10032 USA
e-mail: mw365{at}columbia.edu
 |
Abstract |
|---|
 Top Abstract IntroductionMaterial and methodsResultsCommentReferences |
|---|
Methods. We retrospectively analyzed all adult cardiac surgical patients in 1998 for ICU stays more than 14 days. Data were analyzed to create multiple organ dysfunction scores (MODS, range 0 to 24) and hospital charges. Follow-up was conducted 1 and 2 years apart for survival and quality-of-life evaluation.
Results. Forty-nine patients remained in the ICU more than 14 days, comprising 3.8% of our patients but 28% of total ICU bed time. This population had a 28.5% hospital mortality, which was greater than those in the ICU less than 14 days (5.3%, p < 0.05). By 2 years, 22 of the 35 discharged patients were alive, 16 of whom had a normal quality of life. Patients alive at 2 years had lower MODS at day 14 than those who died (2.6 ± 1.4 versus 5.5 ± 3.8; p < 0.005) as well as lower hospital costs ($223,000 ± $128,000 versus $306,000 ± $128,000; p < 0.05). No patient with an MODS of at least 6 at day 14 survived.
Conclusions. Patients remaining in the ICU for more than 14 days suffer a higher mortality at greater expense. A MODS at day 14 may help predict those who will not enjoy long-term survival and thus aid in the decision to terminate care.
|
Introduction |
|---|
|
TopAbstractIntroductionMaterial and methodsResultsCommentReferences |
|---|
the baby boomer generation enters its sixth decade. Critically ill intensive care unit (ICU) patients with prolonged hospital length of stay (LOS) appear to be another subset of patient populations responsible for a disproportionate expenditure of time and resources. Medical outcomes in these patients range widely in morbidity and mortality as well as financial expense. What might be considered by some as heroic measures to preserve life at all costs does not uniformly result in either a survival or cost benefit. It would be ideal to be able to determine which members of the prolonged ICU stay population will suffer poor health outcomes and thus modify care appropriately.
Anyone practicing in the field of cardiac surgery has experienced patients with poor prognoses who remain inthe ICU for extended periods of time only to ultimately suffer poor outcomes. Oftentimes, it seems predictable from the early postoperative period that these patients are unlikely to enjoy a favorable outcome, yet we persist in continued care at all costs. Reasons for this continued support include uncertainty of outcome, the complex dynamic created between a physician and the patient’s family over a general unwillingness to surrender to unfavorable outcomes, as well as the ethical, legal, and cultural implications of addressing end-of-life issues. These questions continue to make up one of the most difficult components of caring for critically ill cardiac surgical patients and will most likely never be met with simple answers. Developing a more comprehensive understanding of both the prevalence and outcomes of these complicated patients would facilitate this difficult decision-making process.
In the absence of a well-established predictive clinical instrument available to allow physicians and family members to make decisions regarding therapy, we decided to test a series of viable outcomes measures in our study. Our premise in combining an objective measure of physiologic measurements with an evaluation of long-term postdischarge quality of life was to provide a standard instrument to aid in difficult clinical decision-making. Our principle objective was to determine status of survival, quality of life, and cost for a population of cardiac ICU patients with prolonged LOS as well as to attempt to predict those who will suffer an unfavorable outcome. We hoped thereby to establish a predictive clinical instrument that would assist physicians and family members in making difficult decisions regarding continued aggressive medical support in the face of prolonged ICU illness. By no means are we suggesting that critical end-of-life decisions be taken lightly or depend solely on any single scoring system. Rather, we are proposing that a reproducible scoring system may serve as a useful clinical instrument for explaining prognosis and outcome to family members.
|
Material and methods |
|---|
|
TopAbstractIntroductionMaterial and methodsResultsCommentReferences |
|---|
Data collection
All data were obtained from computer and paper patient charts as well as from the prospectively obtained ICU and hospital charge databases. The long-term survival follow-up information and surveys were obtained by telephone contact of the patient or their proxy at two defined times 1 year apart. The measurements obtained on the patients identified as having a prolonged ICU LOS included a modified multiple organ dysfunction score (MODS), survival, total hospital costs, and quality-of-life indices as determined through an activity of daily living (ADL) score.
Description of the multiple organ dysfunction score
Seeking to use a viable system of multiple physiologic markers to assess the severity of multiple organ failure in patients within an ICU setting, we found that the modified MODS described by Marshall [1] provided a comprehensive assessment of organ dysfunction in the population of interest. The MODS consisted of six measurements assessing organ function (Table 1)
that we adopted almost identically for heart operation patients. The first measurement contained in the MODS expressed pulmonary function as a ratio of partial pressure of oxygen to fraction of inspired oxygen. Respiratory failure constituted an indicator of the multiple organ dysfunction syndrome, and was best evaluated by a measure of this ratio of oxygenation, reflecting the need for mechanical ventilation [2]. Renal function was assessed by serum creatinine levels. We considered renal dysfunction an important manifestation of multiple organ failure and used an unbiased biochemical index to evaluate renal function rather than using clinical techniques such as urine output [3]. Hepatic dysfunction as a component of the multiple organ dysfunction syndrome was defined as clinical hyperbilirubinemia [2]. Cardiac performance was rated using a pressure-adjusted heart rate (heart rate times central venous pressure divided by mean arterial pressure). In the critically ill, systemic hypotension is a poor prognostic indicator. However, the use of blood pressure alone is a highly variable marker, subject to transient elevation and depression as a result of pharmaceutical therapies (eg, inotropic agents and vasopressors) and resuscitative measures. In addition, blood pressure alone fails to fully represent cardiovascular function. For these reasons, we accepted the modified measure used in the literature, the pressure-adjusted heart rate [4]. We used a platelet count as a hematologic marker given that thrombocytopenia represents the most common hematologic disorder in the ICU population. Other hematologic manifestations including disseminated intravascular coagulation, leukopenia, leukocytosis, anemia, and prolonged prothrombin time to partial thromboplastin time constituted historically rare entities in this group. For this reason, platelet count satisfied all criteria as a marker of organ dysfunction [5]. Finally, a modified Glasgow Coma Score (GCS) was used to gauge neurologic function. Because of our high volume of intubated patients, we modified the GCS by eliminating the verbal component of the assessment. Resultant GCS ranged from 2 to 10. For each organ system evaluated, values ranging from 0 to 4 points were assigned. Total MODS included a possible range of 0 to 24, the higher scores representing a greater severity of organ dysfunction. An MODS of 0 in any single category represented normal organ function. Conversely, an MODS of 4 represented the poorest of organ function for a particular organ system. According to Marshall’s system [1], a score of 0 in any one category of organ function correlated with a less than 5% mortality rate, whereas a greater than 50% mortality rate correlated with a score of 4.
|
Total hospital costs
Hospital costs were obtained for the entire LOS for each patient from Columbia’s Clinical Data Warehouse. The warehouse maintains all billing for diagnoses and procedures. The value represents the sum of the actual indirect and actual direct costs.
Description of quality-of-life assessment
Quality-of-life outcomes measures were adapted from Katz’s ADLs scale [6]. Katz developed an index of independence in ADL to monitor the independence of chronically ill patients. The six activities used in this evaluation were related hierarchically. Any person acting without supervision, direction, or actual personal assistant was considered to be acting independently within any of the six categories. The original development of these six items by Katz was based on prolonged observation of function and dysfunction among disabled patients. Our adaptation of Katz’s ADL index used a modified version of the scale, which originally included three categories representing degrees of function for each of the six ADLs. In place of defining the three categories verbally, our modified version substituted point values. For the six ADLs applied, each were assigned values of 2 points, 1 point, and 0 points for activities performed independently, with assist, and unable to perform, respectively. The ADL scale was administered to surviving patients over the telephone at the first and second follow-up times. The activities on which we based our assessment consisted of bathing, dressing, toileting, standing, eating, and continence (Table 2).
A cumulative ADL score of 12 represented a normal score. Those receiving scores between 8 and 12 were considered impaired, and those with a score less than 8 were considered severely impaired.
|
2 analysis. A probability value of less than 0.05 was considered statistically significant. Kaplan-Meier curves were constructed to analyze patient survival. |
Results |
|---|
|
TopAbstractIntroductionMaterial and methodsResultsCommentReferences |
|---|
The patients had the following procedures: coronary operation (n = 17), valve operation (n = 4), coronary and valve operation (n = 6), transplant operation (n = 12), ventricular assist device implant (n = 7), aneurysm operation (n = 2), and pericardiectomy (n = 1).
Survival
In-hospital mortality for the population spending 14 days or more in the ICU was 28.5% as compared with a 5.3% hospital mortality for those patients staying less (p < 0.05). Of the 35 discharged patients, 27 (55%) were alive at the first follow-up and 22 (45%) were alive at the second follow-up (Fig 1).
This represented 45% of the original 49 patients but 81% of the patients who survived to hospital discharge. The mean duration of follow-up after discharge was 445 ± 316 days. Patients who survived 2 years after discharge had a significantly shorter total ICU LOS than their cohort who did not survive 2 years (26 ± 12 days versus 39 ± 25 days; p < 0.05; Fig 2).
|
|
There were no individual variables that were univariately predictive except for the hematology score for 2-year survival. The presence of thrombocytopenia discriminated between those who did not survive and those who survived 2 years (0.9 ± 1.3 x 106/mL versus 0.3 ± 0.6 x 106/mL; p < 0.05).
Interestingly, although the MODS reveals no difference at day 0 or day 7, patients showing an improvement by day 14 tended to survive, whereas those not surviving tended to show no improvement (Figs 3,4).
|
|
|
|
Activities of daily living result
At the first follow-up, 19 patients regained a normal quality of life, 5 had an impaired quality of life, and 3 had a severely altered quality of life. The 3 patients with the severely altered quality of life at 1 year had all died by 2 years. At the second follow-up 16 patients had a normal quality of life, 6 had an impaired quality of life, and none had a severely impaired quality of life. The 16 normal patients represented 33% of the original 49 patients, but 60% of the patients who survived to hospital discharge.
Of the 22 surviving patients, 4 were actively working whereas the others were retired. All patients except one lived at home. One patient resided in a nursing home despite an ADL score of 12.
|
Comment |
|---|
|
TopAbstractIntroductionMaterial and methodsResultsCommentReferences |
|---|
Patients remaining in the ICU for prolonged periods of time after cardiac operations generate disproportionate costs that dramatically increase the health-care budget. Each day of ICU stay costs from 3 to 6 times a non-ICU day [7]. A recent Canadian study determined that ICU-day resource utilization per ICU patient remained relatively constant regardless of the particular diagnosis [7]. The high percentages of poor health outcomes after such prolonged stays may indicate that such expenditures are excessive. Third-party health-care payers such as Medicare and other insurance agencies are particularly interested in health outcomes after hospital or ICU stay [8]. Despite extremely high likelihood of death, these patients generate very high charges. In our assessment, we determined that the average patient generated more than $250,000 in hospital costs alone, excluding out-of-hospital expenses. This expense was significantly greater in those who did not survive to hospital discharge (presumably related to their longer LOS). The importance of this difference is better illustrated by examining the cost per year of life saved, which was $155,000/y in those surviving, particularly relevant considering most of these patients regained a normal or near normal quality of life. Those who did not survive cost more than $1,000,000 per year of life saved. This analysis is obviously more exaggerated by the larger denominator in those surviving, but it does put into perspective the costs of these patients and further validates the necessity of proactive determination of which patients will not have favorable outcomes. Although a difference in cost does exist, this analysis, like all other health-care expenditure analyses, falls short in its inability to determine what is a worthwhile expense in attempting to preserve a patient’s life.
Ideally we wish to create a screening scale that could be used to help determine whether care should be continued or whether the level of care should be altered. We applied the MODS in the hopes of providing this clinical tool. To determine the utility of such a model we determined the sensitivity, specificity, negative predictive value, and positive predictive value of the MODS for either a mortality at 2 years, or a mortality or severely altered quality of life at 1 year. As the score increased we had improving specificity and sensitivity, but the sensitivity still stayed somewhat low, a phenomenon resulting from the death of several patients with low MODS. In trying to create a tool that may potentially determine continuation of care, one would preferably err on the side of specificity rather than sensitivity. In our study, by selecting a MODS score of 7 or greater at 14 days, all patients died by 2 years and 95% died or had a severely altered quality of life at 1 year. We are not advocating that a singular scoring system is of sufficient weight to make the determination of continued support; however, it can be used as an additional tool in aiding this decision and will ultimately be strengthened by prospective analysis.
As physicians, we generally find the decision to terminate health care a difficult one to make, and thus we persist despite objective evidence of clinical failure. Although we could not demonstrate a difference in MODS at day 0 or 7, there was a difference at day 14. Of interest is that patients who ultimately die, in contrast to those surviving, fail to show any improvement in MODS at day 14. Anecdotally this would also appear to be the case, as it seems most patients who do not improve or continue to deteriorate over prolonged periods rarely have a favorable outcome. This is well known to the ICU staff who persist in taking care of these patients, often for several months waiting for a terminal event.
We are not the first group to use screening scales in an attempt to predict mortality. Barie and colleagues [9] used both the MODS as described by Marshall and acute physiology and chronic health evaluation (APACHE) II scores for predicting mortality in critically ill surgical patients and did not find a difference between the two systems, most probably because both predicted the same events. This study did not include cardiac patients nor did it focus on patients with prolonged ICU LOS. Others have completed similar studies using mortality prediction models [10, 11]. In the cardiac realm there have also been studies to predict mortality, most notably the Parsonnet score [8]. Unfortunately, this valuable index cannot be applied retrospectively. Higgins and associates [12] examined 135 factors to predict a prolonged ICU LOS (defined as > 120 hours) in coronary artery bypass grafting patients. Ultimately by multivariable analysis they identified 11 predictors (96 by univariate). This scoring scale is too cumbersome for rapid application and has not been correlated with mortality. Although there have been several other cardiac studies analyzing predictors for mortality, some of which included ICU LOS [11, 13], few have looked at long-term survival [14]. The most extensive study with more than 13,000 patients defined six variables as predictors of mortality, ICU LOS, and overall stay [15]. Their six-variable index appeared to be quite effective, but it was dependent on preoperative variables and was not useful in our situation when we are trying to predict mortality once the patient is in the ICU for more than 14 days. To our knowledge there is no other study that has followed the patients long-term as well as determined quality-of-life indices.
The scoring system we used does have some limitations. For example, patient creatinine levels were recorded without considering whether or not patients were receiving dialysis. In our patient population, some patients received ventricular assist devices, reducing the value of a pressure-adjusted heart rate in this population, by providing erroneously low scores. Although the scoring system is promising, it ultimately must be verified in a prospective fashion and perhaps modified as needed before its clinical application can be advised. Another limitation of the study was our inability to account for costs generated outside of the hospital, such as those incurred by outpatient facilities, physicians’ services, home health care, nursing home services, and other outside professional care-giving organizations. Finally, the score does not take into account preoperative variables that may be interesting in determining changes from baseline but are not necessary in determining the state of a patient’s health at a single point in time as a predictor for ultimate outcome.
On a final note, this study does not account for the difficulty of health-care providers and families at addressing these issues. In the context of this study, it is simple to place everything in financial and clinical terms, but ultimately many decisions must be made on a patient by patient basis. An added year of life to one patient may be a triumph of our specialty, whereas to another it may be unnecessary prolonged physical and psychological pain to all those involved. The purpose of this study is to place the problem in perspective and to propose a potential tool that may aid in the decision process.
In conclusion, our retrospective review revealed that cardiac surgical patients with ICU LOS more than 14 days represented less than 4% of our total patient population, accounted for a third of the total ICU days, and had a mortality sixfold greater than their shorter stay cohort. Their individual hospital charges averaged more than $400,000, and patients who died within 2 years of operation consumed $1.1 million per year of life saved. However, of patients who survived to hospital discharge, 63% were still alive at 2 years, and of these, 86% enjoyed a normal quality of life. Thus, the high expenditures and variable outcome would advocate proactive prediction of which patients should have continued aggressive care. The MODS at 14 days may be an acceptable predictor for helping determine which patients will survive. Any patient with a score of 7 or higher is unlikely to survive long term, and those who do not show improvement from day 7 to 14 are also less likely to enjoy long-term survival. Health-care providers have a duty to actively guide families as to whether continued care in these patients is worthwhile. Looking beyond hospital mortality to long-term survival and quality of life after discharge should play an integral role in this decision.
|
References |
|---|
|
TopAbstractIntroductionMaterial and methodsResultsCommentReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  G. Soppa, C. Woodford, M. Yates, R. Shetty, M. Moore, O. Valencia, N. Fletcher, and M. Jahangiri Functional status and survival after prolonged intensive care unit stay following cardiac surgery Interact CardioVasc Thorac Surg, June 1, 2013; 16(6): 750 - 754. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. Hassan, C. Anderson, A. Kypson, L. Kindell, T. B. Ferguson, W. R. Chitwood Jr, and E. Rodriguez Clinical Outcomes in Patients With Prolonged Intensive Care Unit Length of Stay After Cardiac Surgical Procedures Ann. Thorac. Surg., February 1, 2012; 93(2): 565 - 569. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. Wagener, M. Minhaz, S. Wang, O. Panzer, H. Wunsch, H. R. Playford, and R. N. Sladen The Surgical Procedure Assessment (SPA) score predicts intensive care unit length of stay after cardiac surgery J. Thorac. Cardiovasc. Surg., August 1, 2011; 142(2): 443 - 450. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
E. Lagercrantz, D. Lindblom, and U. Sartipy Survival and Quality of Life in Cardiac Surgery Patients With Prolonged Intensive Care Ann. Thorac. Surg., February 1, 2010; 89(2): 490 - 495. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Gaudino, F. Girola, M. Piscitelli, L. Martinelli, A. Anselmi, C. Della Vella, R. Schiavello, and G. Possati Long-term survival and quality of life of patients with prolonged postoperative intensive care unit stay: Unmasking an apparent success J. Thorac. Cardiovasc. Surg., August 1, 2007; 134(2): 465 - 469. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 O. Hein, J. Birnbaum, K. Wernecke, W. Konertz, and C. Spies Intensive Care Unit Stay of More than 14 Days after Cardiac Surgery is Associated with Non-cardiac Organ Failure Journal of International Medical Research, November 1, 2006; 34(6): 695 - 703. [Abstract] [PDF]
|
|
|
|
 |
|
 P. Gersbach, H. Tevaearai, J.-P. Revelly, P. Bize, R. Chiolero, and L. K. von Segesser Are there accurate predictors of long-term vital and functional outcomes in cardiac surgical patients requiring prolonged intensive care? Eur J Cardiothorac Surg, April 1, 2006; 29(4): 466 - 472. [Full Text] [PDF]
|
|
|
|
 |
|
 S. V. Ghotkar, A. D. Grayson, and W. C. Dihmis Effect of Prolonged Intensive Care Stay on Survival following Coronary Surgery Asian Cardiovascular and Thoracic Annals, December 1, 2005; 13(4): 345 - 350. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. Hellgren and E. Stahle Quality of Life After Heart Valve Surgery With Prolonged Intensive Care Ann. Thorac. Surg., November 1, 2005; 80(5): 1693 - 1698. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J.-Y. Dupuis Clinical Predictions and Decisions to Perform Cardiac Surgery on High-Risk Patients Seminars in Cardiothoracic and Vascular Anesthesia, June 1, 2005; 9(2): 179 - 186. [Abstract] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| ANN THORAC SURG | ASIAN CARDIOVASC THORAC ANN | EUR J CARDIOTHORAC SURG |
| J THORAC CARDIOVASC SURG | ICVTS | ALL CTSNet JOURNALS |
8) Patients Who Have Either Died or Maintain a Poor Quality of Life (ADL < 8)