Outcomes of Critically Ill Cancer Patients in a ... - ATS Journals

Outcomes of Critically Ill Cancer Patients in a University Hospital Setting JOHN P. KRESS, JEFFREY CHRISTENSON, ANNE S. POHLMAN, DARREN R. LINKIN, and JESSE B. HALL Department of Medicine, University of Chicago, Chicago, Illinois

Critically ill cancer patients constitute a large percentage of admissions to tertiary care medical intensive care units (ICUs). We sought to describe outcomes of such patients, and to evaluate how conditions commonly seen in these patients impact mortality. A total of 348 consecutive medical ICU cancer patients were evaluated. Subgroup comparisons included the three most common cancer types (leukemia, lymphoma, lung cancer), as well as three different treatments/conditions (bone marrow transplant [BMT] versus non-BMT, mechanical ventilation [MV] versus non-MV, neutropenic versus non-neutropenic). There were no mortality differences between patients with leukemia, lymphoma, or lung cancer. By logistic regression, mortality predictors were: MV, hepatic failure, and cardiovascular failure for the group as a whole (41% overall mortality); MV and allogeneic (as compared with autologous) BMT for the BMT group (39% overall mortality); hepatic failure, cardiovascular failure, and persistent acute respiratory distress syndrome (ARDS) for the MV group (67% overall mortality); and MV for the neutropenic group (53% overall mortality). Neutropenia showed no independent association with mortality in the group as a whole or any subgroup analyzed. We conclude that respiratory, hepatic, and cardiovascular failure predict mortality, whereas neutropenia does not. Additionally, we have noted an encouraging improvement in survival in many groups of critically ill cancer patients. Kress JP, Christenson J, Pohlman AS, Linkin DR, Hall JB. Outcomes of critically ill cancer patients in a university hospital setting. AM J RESPIR CRIT CARE MED 1999;160:1957–1961.

The management of critically ill cancer patients can be difficult for many reasons (1, 2). These patients are often extremely ill, requiring extensive resuscitative measures. Many are young and have undergone toxic therapies with a hope for cure and return to an active life. Cancer patients may be accustomed to dealing with poor odds and are often viewed as “fighters” not easily willing to give up (3, 4). Because of this, clinicians may be reluctant to cease aggressive measures, even when intuition suggests death is certain. Previous analyses have described populations of cancer patients, and have sought reliable outcome predictors. Patients have been subgrouped by categories such as age (5, 6), oncologic diagnosis (7), bone marrow transplant (BMT) status (8– 10), and severity of illness (11). Because new therapies and management strategies may impact outcomes, previously described factors associated with a poor prognosis (e.g., neutropenia, mechanical ventilation [MV]) may no longer hold true. Decisions to withhold and/or withdraw therapies versus attempts to cure underlying conditions are aided by accurate, up-to-date analyses of outcomes (12). Patients who will almost

(Received in original form December 7, 1998 and in revised form May 27, 1999) Correspondence and requests for reprints should be addressed to Jesse B. Hall, M.D., Section of Pulmonary and Critical Care Medicine, MC 6026, University of Chicago, 5841 S. Maryland Avenue, Chicago, IL 60637. E-mail: jhall@medicine. bsd.uchicago.edu Am J Respir Crit Care Med Vol 160. pp 1957–1961, 1999 Internet address: www.atsjournals.org

certainly die may then be identified early, so that futile care (13) can be minimized or avoided. Accordingly, we performed a detailed, retrospective analysis of outcomes for medical oncology patients admitted to our intensive care unit (ICU) over a 3 1⁄2-year period. We sought to evaluate how conditions common to this group of patients impacted hospital mortality. We examined subgroups of patients (e.g., common cancer types and BMT recipients) with a goal of identifying particular conditions that affected their outcome. With this analysis, we sought to better predict outcomes of cancer patients in the ICU, identifying cases where aggressive treatment for cure is merited, as well as those where such care is likely futile.

METHODS Patients and Data Abstraction After receiving institutional review board approval, we reviewed the medical records of all hospitalized adults with an oncologic diagnosis admitted to the medical ICU at the University of Chicago Hospital. Patients were identified by crossing the ICU database (from July 1, 1993—when a computerized hospital database began archiving data— to December 31, 1996) with oncologic International Classification of Diseases–Ninth Revision (ICD-9) codes. All charts were reviewed by three persons experienced in critical care chart abstraction (two critical care physicians and one critical care nurse–clinician). Information from charts was logged onto a standardized computer data collection form. All charts were reviewed independently by two separate data extractors. For those patients with more than one hospital admission over the study period, only the first admission was included in the analysis to assure independence of observations. All patients had an active oncologic diagnosis, defined as the presence of one or more of

1958

AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE

the following: (1) positive histologic diagnosis at the time of ICU admission, (2) metastatic disease, (3) ongoing chemotherapy or complications from recent (less than 6 mo) chemotherapy (e.g., neutropenia, graft-versus-host disease, mucositis), and/or (4) residual tumor or incomplete surgical resection. In our hospital, all cancer patients are eligible for ICU admission unless explicit advanced directives precluding such an admission are present. The hospital is a 480-bed tertiary care center with separate medical and surgical ICUs. The medical ICU is a closed unit managed by house staff, a critical care fellow, and a critical care attending physician.

Variable Definition Demographic data (age, sex, oncologic diagnosis, ICU admitting diagnosis) were recorded for all patients. Other data collected included Acute Physiology and Chronic Health Evaluation (APACHE) II (14) and Simplified Acute Physiology Score (SAPS) II (15) scores, type of respiratory failure, duration of MV, daily multisystem organ failure (MSOF) scores, BMT status and type of BMT (allogeneic versus autologous), presence of metastatic disease, and status with regard to “do not resuscitate” orders. Organ failures were defined in the following manner: renal failure—creatinine . 3.4 mg/dl; hepatic failure— bilirubin . 3.5 mg/dl and/or alkaline phosphatase . 350 U/L; cardiovascular failure—need for dobutamine, norepinephrine, or epinephrine at any dose, or dopamine at . 5 mg/kg/min; hematologic failure— white blood cell count , 2.0/ml, platelet count , 40,000/ml, and/or International Normalized Ratio (INR) . 2; respiratory failure—need for intubation and MV (16). The type of respiratory failure was divided into four categories: (1) Acute hypoxemic respiratory failure (AHRF)—any alveolar filling process identified by chest X-ray (e.g., pneumonia, pulmonary edema from any cause, alveolar hemorrhage). Included in this group were patients with acute respiratory distress syndrome (ARDS) as defined by the American–European Consensus Conference (17); (2) Ventilatory failure—any process associated with either a decreased drive to breathe, impaired neuromuscular competence of the respiratory system, or an excessive respiratory load (18). This group included reactive airways disease exacerbations, respiratory muscle weakness (any cause), as well as other causes of ventila· . tion–perfusion (V /Q) mismatch resulting in ventilatory failure; (3) Postprocedure-related respiratory failure—atelectasis owing to sedative medications leading to respiratory failure; (4) Shock-related respiratory failure—any state of hypoperfusion leading to a need for MV. When more than one cause of respiratory failure was present, the patient was classified by the primary reason for MV as discerned from the ICU progress notes by the data abstractors. Additional dichotomous variables evaluated were: requirement for MV through an endotracheal tube, presence of neutropenia (absolute neutrophil count < 500/ml) while in the ICU, and hospital mortality. Patients with the three most common malignancies (leukemia, lymphoma, lung cancer) as well as those who had received a BMT were also identified and evaluated using the same variables outlined previously.

VOL 160

1999

were initially analyzed using a Mann-Whitney U test or Kruskal-Wallis test as appropriate. These data are listed as median (25th–75th percentile). Nominal data were analyzed by chi-square analysis with Yates continuity correction or Fisher exact test where appropriate. The initial analysis included the group of cancer patients as a whole. Variables found significant at p , 0.15 by univariate analysis were then subjected to stepwise forward logistic regression analysis. Analysis to rule out interdependence between variables was included in the logistic regression analysis. Because APACHE II and SAPS II scores have been previously noted to be strongly associated with mortality, these variables were not included in the model. Leukemia, lymphoma, and lung cancer patients were compared by one-way analysis of variance (ANOVA) with post hoc analysis using the Scheffe test for interval data and chi square for nominal data. Patients were also analyzed according to the details of their treatment/condition (BMT, MV, neutropenia). Direct comparisons were made between BMT patients versus non-BMT patients, MV versus non-MV, and neutropenic versus non-neutropenic. These three analyses were performed independently, but the groups were not mutually exclusive (e.g., a BMT patient could also be neutropenic). Each of these three groups initially underwent univariate analysis similar to that described previously. Once again, variables found significant at p , 0.15 (except APACHE II and SAPS II scores) were subjected to stepwise forward logistic regression. A p value , 0.05 was used to indicate statistical significance. All statistical analyses were performed on a personal computer using SPSS version 8.0 software.

RESULTS Demographics

A total of 386 medical oncology patients were admitted to the ICU over the period of our data collection. Thirty-eight patients had more than one hospital admission, leaving 348 patients for the analysis. Patients had a wide range of oncologic diagnoses, with leukemia (30%), lymphoma (18%), and lung cancer (16%) being the most common. Other cancer diagnoses included head and neck (13%), breast (8%), gynecologic (6%), renal (2%), colorectal (2%), and prostate (2%). There were 211 patients with solid tumors and 137 with hematologic malignancies. One hundred thirty-nine of 211 patients (66%) with solid tumors had known metastases. The reasons for ICU admission were broad with the most common being ventilatory failure (12%), acute hypoxemic respiratory failure (12%), sepsis (12%), gastrointestinal bleeding (10%), shock (10%), monitoring (9%), airway obstruction (5%), cardiac arrest (5%), and pneumonia (4%). Many patients had more than one ICU admitting diagnosis.

Statistical Analysis Parametric interval data were initially analyzed using a two-tailed Student’s t test. These data are listed as mean 6 SD. Nonparametric data

TABLE 2 LOGISTIC REGRESSION ANALYSIS FOR ALL ONCOLOGY PATIENTS

TABLE 1 ONCOLOGY PATIENT SUMMARY* n Age Male/Female APACHE II SAPS II MV, % Neutropenia, % Overall mortality, % Mortality if zero organ failures, % Mortality if single organ failures, % Mortality if multiple organ failures, %

348 58 (46–68) 167/181 19 (13–24) 42 (34–53) 44 22 41 15† 39† 67†

* Age, APACHE II, and SAPS II are reported as median (25th–75th quartiles). † p , 0.0001 by one-way ANOVA comparing zero, single, and multiple organ failures.

All Oncology Patients (n 5 348)

Mortality Odds Ratio† (95% CI)

Respiratory failure* Type 1 Type 2 Type 3 Type 4 Hepatic failure Cardiovascular failure

7.7 (3.3–17.9)‡ 5.3 (2.5–11.3)‡ 2.2 (0.7–6.6) 8.1 (3.2–20.5)‡ 4.9 (2.4–9.9)‡ 4.3 (1.9–9.8)§

* Respiratory failure: Type 1 5 acute hypoxemic respiratory failure; Type 2 5 ventilatory failure; Type 3 5 postprocedure–related respiratory failure; Type 4 5 shock-related respiratory failure. † Odds ratios (OR) for respiratory failure types are referenced to patients without respiratory failure. ‡ p , 0.0001. § p , 0.001.

1959

Kress, Christenson, Pohlman, et al.: Outcomes of Critically Ill Cancer Patients TABLE 3 MOST COMMON CANCER TYPES–UNIVARIATE COMPARISON

n Age, yr, mean 6 SD Male/Female APACHE II* SAPS II* MV, % Neutropenia, % Mortality, %

Leukemia

Lymphoma

Lung Cancer

p Value

80 50 6 16†§ 40/40 22 (16–27) 44 (35–53) 41 58‡i 48

47 57 6 15 26/21 21 (15–26) 46 (37–57) 53 30¶ 49

44 61 6 10 21/23 19 (11–22) 41 (27–49) 46 5 48

, 0.001 0.75 0.002 0.05 0.43 , 0.001 0.99

Definition of abbreviation: MV 5 mechanically ventilated. * APACHE II and SAPS II reported as median (25th–75th quartiles). † p , 0.05 compared with lymphoma group (Scheffe test). ‡ p , 0.01 compared with lymphoma group (Scheffe test). § p , 0.001 compared with lung cancer group (Scheffe test). i p , 0.0001 compared with lung cancer group (Scheffe test). ¶ p , 0.01 compared with lung cancer group (Scheffe test).

ity was 39%, not different from the non-BMT patients. Variables associated with mortality by univariate analysis in BMT patients (p , 0.15) included: MV, MSOF score on ICU admission, and type of BMT (autologous BMT mortality was lower). When these variables were placed in the logistic regression model, MV and type of BMT remained significant. Univariate and multivariate analyses are summarized in the first column of Tables 4 and 5, respectively. Twenty BMT patients required MV, with nine survivors. All nine ventilated BMT survivors received autologous BMT (nine of 13 survived), whereas none of the seven ventilated allogeneic patients survived. Because of the small numbers and zero survival in the allogeneic group, logistic regression was not possible for the mechanically ventilated BMT patients; however, the mortality difference between allogeneic and autologous groups was significant by Fisher exact test (p 5 0.005). MV versus Non-MV

The breakdown by type of respiratory failure was as follows: (1) acute hypoxemic respiratory failure—n 5 51 (44 of the patients in this group had ARDS); (2) ventilatory failure— n 5 46; (3) postprocedure-related respiratory failure—n 5 18; (4) shock-related respiratory failure—n 5 38. Analysis of All 348 Cancer Patients

Overall mortality for the group of 348 cancer patients was 41%. As the number of organ failures increased from zero to > 2, mortality significantly increased from 15% for zero organ failures to 67% for > 2 organ failures. Table 1 summarizes descriptive statistics for the 348 cancer patients. Variables associated with mortality by univariate analysis (p , 0.15) included: type of respiratory failure, duration of MV, and the following organ failures: renal, hepatic, cardiovascular, and hematologic. When these variables were placed in the logistic regression model, all types of respiratory failure (except postprocedure-related respiratory failure [type 3]), hepatic failure, and cardiovascular failure remained significant. Table 2 summarizes these results. Common Cancer Types (Leukemia, Lymphoma, Lung Cancer)

The leukemia patients were the youngest of the three groups and had the highest incidence of neutropenia. The lung cancer patients were older and had lower severity of illness scores. Neither incidence of MV nor mortality was different for any of the three groups. Table 3 shows a summary of the three most common cancer types. BMT versus Non-BMT

The BMT patients were younger than non-BMT patients. Despite a higher incidence of neutropenia, BMT patient mortal-

MV patients had higher severity of illness scores and higher mortality. The incidence of neutropenia was also higher than in non-MV patients. Variables associated with mortality by univariate analysis in MV patients (p , 0.15) included: neutropenia, leukemia, and the following organ failures: renal, hepatic, cardiovascular, hematologic, and persistent ARDS (greater than 1 d). When these variables were placed in the logistic regression model, hepatic failure, cardiovascular failure, and persistent ARDS remained significant. These analyses are summarized in the second column of Tables 4 and 5, respectively. Neutropenic versus Non-neutropenic

By univariate analysis, neutropenic patients were younger, with higher severity of illness scores and higher mortality than non-neutropenic patients. Variables entered into the logistic regression model included: age, MV, duration of neutropenia, and the following organ failures: renal, hepatic, cardiovascular, and persistent ARDS. By logistic regression, only age and MV remained significant. These analyses are summarized in the third column of Tables 4 and 5, respectively. All neutropenic patients received granulocyte colony stimulating factor. Among surviving neutropenic patients, 21 of 36 (58%) were neutropenic at the time of ICU discharge; in nonsurviving neutropenic patients, 33 of 41 (80%) were still neutropenic at ICU discharge (p 5 0.06). ICU mortality was 40% in the neutropenic patients. “Do Not Resuscitate” Orders

A total of 130 patients received “do not resuscitate” orders during their ICU stay, with 106 eventually dying (82%). Two hundred eighteen patients never received “do not resuscitate” orders, and 183 of these survived to hospital discharge (84%).

TABLE 4 BMT VERSUS NON-BMT, MV VERSUS NOT MV, NEUT VERSUS NOT NEUT–UNIVARIATE COMPARISONS BMT n 44 Age, yr* 47 (36–53) Male/Female 21/23 APACHE II* 17 (15–24) SAPS II* 44 (37–51) MV, % 46 Neut, % 41 Mort, % 39

Non-BMT

p Value

304 60 (49–70) , 0.0001 146/158 0.97 19 (13–24) 0.62 42 (33–53) 0.34 44 0.83 19 , 0.001 41 0.79

MV 153 58 (47–68) 69/84 22 (16–27) 51 (41–62) NA 28 67

Non-MV

p Value

Neut

195 77 58 (46–69) 0.82 51 (41–63) 98/97 0.34 36/41 17 (12–22) , 0.0001 24 (19–28) 37 (28–45) , 0.0001 52 (43–62) NA NA 55 18 0.03 NA 20 , 0.0001 53

Non-Neut

p Value

271 59 (48–69) , 0.0001 131/140 0.81 17 (12–22) , 0.0001 39 (31–50) , 0.0001 41 0.03 NA NA 37 0.01

Definition of abbreviations: MV 5 mechanically ventilated; Neut 5 neutropenia; Mort 5 mortality; NA 5 not applicable. * Age, APACHE II, and SAPS II are reported as median (25th–75th quartiles).

1960

AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE

VOL 160

1999

TABLE 5 LOGISTIC REGRESSION ANALYSIS FOR DIFFERENT TREATMENT GROUPS

Age MV Auto-BMT (versus Allo-BMT) Hepatic failure CV failure ARDS . 1 d

BMT (n 5 44) Mortality Odds Ratio (95% CI)

MV (n 5 153) Mortality Odds Ratio (95% CI)

Neutropenic (n 5 77) Mortality Odds Ratio (95% CI)

NS 13.7 (1.6–120.6)* 0.05 (0.005–0.4)† NS NS NS

NS NA NA 4.0 (1.1–14.8)* 4.1 (1.6–10.8)† 3.5 (1.3–9.1)*

1.04 (1.0–1.07)* 10.0 (3.4–29.9)‡ NA NS NS NS

Definition of abbreviations: Allo-BMT versus Auto-BMT 5 allogeneic versus autologous BMT; MV 5 mechanially ventilated; NA 5 not applicable; NS 5 not significant. * p , 0.05. † p , 0.01. ‡ p , 0.0001.

DISCUSSION The results of this large series show an acceptable outcome for patients with active oncologic problems admitted to the medical ICU, with an overall hospital mortality of 41%. Previous studies have reported a wide mortality range for medical oncology patients. This is likely due to the heterogeneous nature of malignancies in these reports. Our mortality was similar to previous reports (5, 6, 19), and higher than the overall mortality of 20 to 30% in our medical ICU. Our database included patients with a wide variety of cancer types as well as reasons for ICU admission. We believe this distribution is similar to that seen in most tertiary care centers. We sought to determine conditions common to critically ill cancer patients associated with increased mortality. Respiratory failure was consistently associated with mortality in the group as a whole, as well as in the subgroups evaluated. The association between MV and mortality has been previously described by other investigators (20–24). A review of the literature reveals mortality rates between 72 to 98% for MV oncology patients. Studies that include hematologic malignancies (25–27) report mortalities in the 80% to upper 90% range, with 70% mortalities typically reported for solid tumor malignancies. Our 67% mortality rate for MV patients included a population in whom nearly half had hematologic malignancies and/or BMT. To our knowledge, this is the lowest reported mortality for MV oncology patients to date, apart from case reports and small series. These more recent data raise the possibility that newer treatment and ventilatory strategies for MV oncology patients may be leading to improvements in outcome. Certainly, our results do not prove this notion, and further prospective studies are needed to better define which, if any, changes in management strategies are associated with improved outcome. From our data we can conclude that respiratory failure in critically ill cancer patients, except when caused by sedative medications given around the time of a procedure, was independently associated with an increased mortality. Hepatic and cardiovascular failure were also independent predictors of mortality in this group of 348 patients. For the subgroup of 153 patients requiring MV, hepatic and cardiovascular failure, as well as persistent ARDS predicted increased mortality. The BMT patients had surprising results—a 39% hospital mortality—not different from the non-BMT group, despite a higher incidence of neutropenia. Twenty of the 44 BMT patients required MV, with nine surviving to hospital discharge. This observation is quite striking, considering previous reports of more dismal outcomes in this group (2, 8–10, 19, 28). Dura-

tion of MV did not prognosticate survival in mechanically ventilated BMT patients, similar to findings reported by Rubenfeld and Crawford (12). Others have noted improved survival of mechanically ventilated BMT patients over time and suggest that improvements in transplantation techniques, better infection prophylactic measures, and the use of hematopoietic growth factors (first available in 1991) may be responsible (12). Clearly, there are subgroups of mechanically ventilated BMT patients that have a reasonable chance for survival, and our 45% survival is encouraging. However, caution is needed in evaluating our results because of the small number of patients. The improved outcome of autologous versus allogeneic BMT patients is interesting. One may speculate that avoiding complications from total body irradiation and graft-versushost disease may impact on survival in these patients during critical illness. Earlier comparisons of outcomes of critically ill allogeneic versus auto-BMT patients have found no difference in survival (8, 29). More recently, Price and coworkers (28) found allogeneic BMT associated with increased mortality in a logistic regression analysis of blood and marrow transplant patients; however, they found no mortality difference between ventilated allogeneic and autologous BMT recipients (mortality 20/24 versus 12/16 for allogeneic and autologous, respectively). Though our numbers are very small, survival in nine of 13 ventilated autologous BMT patients is a striking contrast. We are not aware of a previous report of different survival rates between ventilated allogeneic and autologous BMT patients. Larger studies are needed to clarify this question. Neutropenia, largely limited to patients with hematologic malignancies and/or BMT, is often viewed as a poor prognostic indicator. Surprisingly, we found neutropenia was not independently associated with mortality in any group analyzed. In particular, it did not show any additional impact upon mortality in the MV group. Few studies have looked directly at the effects of neutropenia on survival. Blot and coworkers noted a 55% ICU mortality in 107 neutropenic patients (hospital mortality not recorded) (30). This study covered a 4-yr period (1987 through 1991) before hematopoietic growth factors were available—a possible, but unproven, reason for our lower ICU mortality. Lloyd-Thomas and coworkers reported 60 ICU patients with hematologic malignancy (31). For those neutropenic in the ICU, most nonsurvivors remained neutropenic up to death. Conversely, all hospital survivors had normal white blood cell counts at time of ICU discharge. Whereas neutropenia at ICU discharge was more common among our nonsurvivors, more

Kress, Christenson, Pohlman, et al.: Outcomes of Critically Ill Cancer Patients

than half our neutropenic survivors remained neutropenic at time of ICU discharge. In sharp contrast to the results reported by Lloyd-Thomas and coworkers in 1988, our more recent results suggest that recovery of neutrophil counts while in the ICU is not a prerequisite to hospital survival. Decisions regarding “do not resuscitate” orders were more than 80% accurate in predicting both survivors and nonsurvivors. Movements to avoid futile resuscitation efforts have been noted previously by others (32). The movement toward withholding care in futile cases may translate into a greater likelihood of avoiding ICU care altogether for those with extremely poor prognoses. Such a movement may partly explain the observed improvement in survival that we and others have noted. This study has several limitations. Some subgroup analyses have small numbers of patients. These results cannot supplant those of larger studies; rather, we view our data as complementary to these studies. Our work has confirmed that subgroups with good outcomes do exist and that outcomes in such groups may be even better than previously reported. One other weakness is the lack of long-term follow-up in our patients. Nevertheless, hospital mortality is the endpoint followed in most studies. In summary, MV and hepatic and cardiovascular organ failure are powerful, independent predictors of mortality in critically ill medical oncology patients, irrespective of the type of cancer. On the contrary, neutropenia does not appear to have any independent effect on mortality. Patients in categories previously associated with very poor outcomes, such as BMT, neutropenia, and even MV, may in fact have improved prognoses when compared with earlier studies. The absence of organ failures may help to predict a subgroup of these patients with a reasonable chance for recovery, for whom aggressive resuscitative measures may be reasonable. References 1. Crawford, S. W., and F. B. Petersen. 1992. Long-term survival from respiratory failure after marrow transplantation for malignancy. Am. Rev. Respir. Dis. 145:510–514. 2. Schuster, D. P. 1992. Everything that should be done—not everything that can be done. Am. Rev. Respir. Dis. 145:508–509. 3. Brown, D., J. A. Roberts, T. E. Elkins, D. Larson, and M. Hopkins. 1994. Hard choices: the gynecologic cancer patient’s end-of-life preferences. Gynecol. Oncol. 55:355–362. 4. Weeks, J. C., E. F. Cook, S. J. O’Day, L. M. Peterson, N. Wenger, D. Reding, F. E. Harrell, P. Kussin, N. V. Dawson, A. F. Connors, Jr., J. Lynn, and R. S. Phillips. 1998. Relationship between cancer patients’ predictions of prognosis and their treatment preferences. J.A.M.A. 279:1709–1714. 5. Chalfin, D. B., and G. C. Carlon. 1990. Age and utilization of intensive care unit resources of critically ill cancer patients. Crit. Care Med. 18: 694–698. 6. Schapira, D. V., J. Studnicki, D. D. Bradham, P. Wolff, and A. Jarrett. 1993. Intensive care, survival, and expense of treating critically ill cancer patients. J.A.M.A. 269:783–786. 7. Hauser, M. J., J. Tabak, and H. Baier. 1982. Survival of patients with cancer in a medical critical care unit. Arch. Intern. Med. 142:527–529. 8. Paz, H. L., P. Crilley, M. Weinar, and I. Brodsky. 1993. Outcome of patients requiring medical ICU admission following bone marrow transplantation. Chest 104:527–531. 9. Denardo, S. J., R. K. Oye, and P. E. Bellamy. 1989. Efficacy of intensive care for bone marrow transplant patients with respiratory failure. Crit. Care Med. 17:4–6. 10. Crawford, S. W., D. A. Schwartz, F. B. Petersen, and J. G. Clark. 1988.

11.

12.

13. 14.

15.

16.

17.

18.

19.

20.

21.

22.

23. 24. 25.

26.

27.

28.

29.

30.

31.

32.

1961

Mechanical ventilation after marrow transplantation: risk factors and clinical outcome. Am. Rev. Respir. Dis. 137:682–687. Abbott, R. R., M. Setter, S. Chan, and K. Choi. 1991. APACHE II: prediction of outcome of 451 oncology admissions to a community hospital. Ann. Oncol. 2:571–574. Rubenfeld, G. D., and S. W. Crawford. 1996. Withdrawing life support from mechanically ventilated recipients of bone marrow transplants: a case for evidence-based guidelines. Ann. Intern. Med. 125:625–633. Schneiderman, L. J., N. S. Jecker, and A. R. Jonsen. 1990. Medical futility: its meaning and ethical implications. Ann. Intern. Med. 112:949–954. Knaus, W. A., E. A. Draper, D. P. Wagner, and J. E. Zimmerman. 1985. APACHE II: a severity of disease classification system. Crit. Care Med. 13:818–829. LeGall, J. R., S. Lemeshow, and F. Saulnier. 1993. A new Simplified Acute Physiology Score (SAPS II) based on a European/North American study. J.A.M.A. 270:2957–2963. Hebert, P. C., A. J. Drummond, J. Singer, G. R. Bernard, and J. A. Russell. 1993. A simple multiple system organ failure scoring system predicts mortality of patients who have sepsis syndrome. Chest 104:230–235. Bernard, G. R., A. Artigas, K. L. Brigham, J. Carlet, K. Falke, L. Hudson, M. Lamy, J. R. Legall, A. Morris, and R. Spragg. 1994. The American–European consensus conference on ARDS: definitions, mechanisms, relevant outcomes, and clinical trial coordination. Am. J. Respir. Crit. Care Med. 149:818–824. Schmidt, G. A., and J. B. Hall. 1998. Acute-on-chronic respiratory failure. In J. B. Hall, G. A. Schmidt, and L. D. H. Wood, editors. Principles of Critical Care, 2nd ed. McGraw-Hill, New York. 565–578. Studnicki, J., D. V. Schapira, J. V. Straumfjord, R. A. Clark, J. Marshburn, and D. C. Werner. 1994. A national profile of the use of intensive care by Medicare patients with cancer. Cancer 74:2366–2373. Ewer, M. S., M. K. Ali, M. S. Atta, R. C. Morice, and P. V. Balakrishnan. 1986. Outcome of lung cancer patients requiring mechanical ventilation for pulmonary failure. J.A.M.A. 256:3364–3366. Ludwigs, U. G., S. Baehrendtz, M. Wanecek, and G. Matell. 1991. Mechanical ventilation in medical and neurological diseases: 11 years of experience. J. Intern. Med. 229:117–124. Dees, A., J. L. Ligthart, W. L. J. van Putten, A. S. T. Planting, and G. Stoter. 1990. Mechanical ventilation in cancer patients: analysis of clinical data and outcome. Netherlands J. Med. 37:183–188. Sculier, J. P., and E. Markiewicz. 1991. Medical cancer patients and intensive care. Anticancer Res. 11:2171–2174. Snow, R. M., W. C. Miller, D. L. Rice, and M. K. Ali. 1979. Respiratory failure in cancer patients. J.A.M.A. 241:2039–2042. Schuster, D. P., and J. M. Marion. 1983. Precedents for meaningful recovery during treatment in a medical intensive care unit: outcome in patients with hematologic malignancy. Am. J. Med. 75:402–408. Ashkenazi, Y. J., B. S. Kramer, and E. Harman. 1986. Short-term outcome among patients with leukemia and lymphoma admitted to a medical intensive care unit. Southern Med. J. 79:1086–1088. Estopa, R., A. Torres Marti, N. Kastanos, A. Rives, A. Agusti-Vidal, and C. Rozman. 1984. Acute respiratory failure in severe hematologic disorders. Crit. Care Med. 12:26–28. Price, K. J., P. F. Thall, S. K. Kish, V. R. Shannon, and B. S. Andersson. 1998. Prognostic indicators for blood and marrow transplant patients admitted to an intensive care unit. Am. J. Respir. Crit. Care Med. 158: 876–884. Afessa, B., A. Tefferi, H. C. Hoagland, L. Letendre, and S. G. Peters. 1992. Outcome of recipients of bone marrow transplants who require intensive care support. Mayo Clin. Proc. 67:117–122. Blot, F., M. Guiguet, G. Nitenberg, B. Leclercq, B. Gachot, and B. Escudier. 1997. Prognostic factors for neutropenic patients in an intensive care unit: respective roles of underlying malignancies and acute organ failures. Eur. J. Cancer 33:1031–1037. Lloyd-Thomas, A. R., I. Wright, T. A. Lister, and C. J. Hinds. 1988. Prognosis of patients receiving intensive care for life threatening medical complications of haematological malignancy. B.M.J. 290:1025– 1029. Prendergast, T. J., and J. M. Luce. 1997. Increasing incidence of withholding and withdrawal of life support from the critically ill. Am. J. Respir. Crit. Care Med. 155:15–20.