Intensive Care Medicine

, Volume 31, Issue 3, pp 408–415

Impact of two different comorbidity measures on the 6-month mortality of critically ill cancer patients

Authors

    • Intensive Care Unit, Instituto Nacional de CâncerCentro de Tratamento Intensivo 10° Andar
    • Faculdade de MedicinaUniversidade Federal do Rio de Janeiro
  • Jorge I. F. Salluh
    • Intensive Care Unit, Instituto Nacional de CâncerCentro de Tratamento Intensivo 10° Andar
    • Faculdade de MedicinaUniversidade Federal do Rio de Janeiro
    • Intensive Care UnitHospital Barra D’Or
  • Carlos Gil Ferreira
    • Department of Clinical ResearchInstituto Nacional de Câncer
  • Ronir R. Luiz
    • Núcleo de Estudos de Saúde ColetivaUniversidade Federal do Rio de Janeiro
  • Nelson Spector
    • Faculdade de MedicinaUniversidade Federal do Rio de Janeiro
    • Hospital Universitário Clementino Fraga FilhoUniversidade Federal do Rio de Janeiro
  • José R. Rocco
    • Faculdade de MedicinaUniversidade Federal do Rio de Janeiro
    • Hospital Universitário Clementino Fraga FilhoUniversidade Federal do Rio de Janeiro
Original

DOI: 10.1007/s00134-005-2554-z

Cite this article as:
Soares, M., Salluh, J.I.F., Ferreira, C.G. et al. Intensive Care Med (2005) 31: 408. doi:10.1007/s00134-005-2554-z

Abstract

Objective

To evaluate the impact of two different comorbidity measures on the 6-month mortality of severely ill cancer patients.

Design and setting

Prospective cohort study in a ten-bed oncological medical-surgical intensive care unit (ICU).

Patients

A total of 772 consecutive patients were included over a 45-month period. The mean age was 57.6±16.4 years, and 642 (83%) patients had solid tumors.

Measurements and results

Data were collected on admission and during ICU stay. Comorbidities were evaluated using the Charlson Comorbidity Index (CCI) and the Adult Comorbidity Evaluation (ACE-27). The ICU, hospital, and 6-month mortality rates were 34%, 47%, and 58%, respectively. The most frequent comorbidities were hypertension (33%), diabetes mellitus (8%), and chronic pulmonary disease (7%). There were important differences between the two indices regarding the comorbidity evaluation. Using the ACE-27, 389 patients (50%) had comorbid ailments that were classified as mild (31%), moderate (14%), and severe (5%) according to the comorbidity severity. According to the CCI, 212 patients (27%) had a comorbidity, and their median score was 1 (1–2). In the multivariable Cox proportional hazard models only the presence of a severe comorbidity by the ACE-27 was associated with increased mortality. The CCI was not independently associated with the outcome. Other outcome predictors were older age, poor performance status, active cancer, need of mechanical ventilation, and severity of acute organ failures.

Conclusions

Severe comorbidities must be considered in the outcome evaluation of ICU cancer patients. The ACE-27 seems to be a useful instrument for prognostic assessment in this population.

Keywords

CancerComorbidityIntensive careMortalitySurvival analysisOutcome

Introduction

The survival of critically ill cancer patients is improving, and recent studies suggest that their outcome predictors are changing [1, 2, 3, 4, 5]. However, the mortality rates of cancer patients requiring critical care remain higher than those of patients with noncancer diseases [4]. Increased mortality has been associated with the severity of the acute illness [1, 5] and organ failures [2, 3, 4, 6, 7], cancer recurrence and progression [6, 7], poor performance status (PS) [6], and with the need for mechanical ventilation (MV) [2, 4, 5, 6].

Cancer patients often have associated chronic diseases [8, 9]. These comorbidities have received increasing attention because their severity is associated with poorer outcome and may impose severe limitations to the administration of the appropriate anticancer therapies [10]. Nevertheless, information regarding the influence of comorbidities on the prognosis of critically ill cancer patients is limited, and previous studies have evaluated chronic diseases individually [2, 6, 11]. The purpose of the present study was to evaluate the impact of chronic ailments, using two different comorbidity indices, on the 6-month mortality of a large prospective cohort of severely ill cancer patients.

Patients and methods

Design and setting

This was a prospective observational cohort study conducted between May 2000 and January 2004 at the Instituto Nacional de Câncer in Rio de Janeiro, Brazil, a 200-bed public hospital for the referral of cancer patients. The intensive care unit (ICU) is an exclusively oncological ten-bed medical-surgical unit. Information about the ICU organization has been presented previously [12]. To be considered for admission to the ICU patients must usually have a potential chance of being cured or having their cancer controlled. The Institutional Review Board waived the need of informed consent. The present study did not interfere with clinical decisions related to patient care.

Selection of participants, data collection, and definitions

During the study period we evaluated every adult patient (age ≥ 18 years) with pathologically confirmed cancer requiring admission to the ICU. Patients with an ICU stay shorter than 24 h, acute coronary syndromes, and those admitted for routine postoperative care were excluded. For patients with multiple admissions only the first one was considered. Patients in complete cancer remission for over 5 years were not evaluated. At our institution bone marrow transplant patients are cared for at a separate unit, even in case of a life-threatening complication and therefore were not studied.

The following data were collected at ICU admission: age, gender, Simplified Acute Physiology Score (SAPS) II [13], Sequential Organ Failure Assessment (SOFA) score [14], source of admission, main diagnosis for ICU admission, and previous comorbidities. Patients were classified as either surgical or medical. Surgical patients were those who had undergone surgical procedure (excluding biopsies and catheter insertions) prior to the ICU admission during the current hospital stay. The type of cancer, cancer status, anticancer treatments, and PS during the week before hospital admission were also assessed. PS was evaluated accordingto the Eastern Cooperative Oncology Group scale, as follows: 0–1, asymptomatic/ambulant; 2, requirement of minor assistance for daily activities; 3–4, requirement of major assistance or chronically bedridden [15]. Leukopenia was defined as a white blood cell count below 1000/mm3. During ICU admission the need of MV for longer 24 h and the development of acute organ failures were also assessed. Individual acute organ failures were defined as a SOFA score of 3 points or more for each system [14]. Sepsis was diagnosed using the definitions established by the American College of Chest Physicians/Society of Critical Care Medicine consensus conferences [16]. The patient was considered to have an infection when there were clinical, laboratory, radiological, and microbiological findings suggesting the presence of infection and that justified the administration of antibiotics (excluding prophylaxis). At our ICU the criteria employed for the diagnosis of infection are usually those from the Centers for Disease Control [17]. Leukopenic patients presenting with fever and receiving antibiotics were considered to have an infection.

In the present study a single author (M.S.) carefully assessed every comorbid condition requiring medical care by reviewing charts and other medical records and by interviewing patients or their proxies. Comorbidities had to be present before the present hospital admission to be considered. The measurement of comorbidities was performed using the Adult Comorbidity Evaluation-27 (ACE-27) [18] and the Charlson Comorbidity Index (CCI) [19]. Cancer diagnoses were excluded from both original comorbidity indices thus allowing the indices to reflect the real impact of comorbid ailments. The ACE-27 is a new 27-item comorbidity index for use in cancer patients [18] (J.F. Piccirillo, Adult Comorbidity Evaluation-27 comorbidity data collection form, available at: http://oto.wustl.edu/clinepi/Forms/com_form.doc, assessed 28 September 2004). The ACE-27 grades a wide range of specific comorbid diseases and conditions, according to the severity of organ decompensation and prognostic impact, into three levels: grade 1 (mild), grade 2 (moderate), and grade 3 (severe). An overall comorbidity score (none, mild, moderate, or severe) is assigned based on the highest ranked single ailment. Patients with two or more grade 2 ailments occurring in different organ systems or disease groupings were also considered as having a severe overall comorbidity score. The CCI is a weighted 19-item index, and the total score is the sum of the weights of each comorbid condition presented by the patient [19].

Characterization of the study population

During the study period 1,660 adult patients were admitted to the ICU, and 803 (48%) met the eligibility criteria; 31 patients (4%) were lost at 6-month follow-up, and 772 patients constituted the study population. Among these patients 30 (4%) were readmitted to the ICU during the same hospital stay, and 22 (3%) during a subsequent hospital stay. The patients’ main characteristics are presented in Table 1. The most frequent solid tumors were lower gastrointestinal (n=110, 14%), head and neck (n=104, 13%), brain (n=104, 13%), lung (n=76, 10%), upper gastrointestinal (n=73, 9%), urogenital (n=65, 8%), breast (n=38, 5%) and others (n=72, 9%). Anticancer treatments within the previous 6 months included surgery with curative intent (n=354, 46%), chemotherapy (n=234, 30%), and radiation therapy (n=213, 28%). Anticancer treatments were employed alone or in combination according to the oncologist responsible for each patient.
Table 1

Patients’ characteristics (n=772) (ICU intensive care unit, SAPS Simplified Acute Physiology Score, SOFA Sequential Organ Failure Assessment)

Age (years)

57.6±16.4

Male sex

415 (54%)

Source of admission

  Operating room

361 (47%)

  Wards

296 (38%)

  Emergency room

102 (13%)

  Other hospital

13 (2%)

Surgery prior to ICU on current hospital admission

417 (54%)

SAPS II (points)

43.6±18.9

SOFA (points)

6.5±3.9

Type of cancer

  Solid tumor

642 (83%)

  Locoregional

505 (65%)

  Metastatic

137 (18%)

  Hematological malignancies

130 (17%)

  Non-Hodgkin’s lymphoma

69 (9%)

  Leukemias

26 (3%)

  Multiple myeloma

18 (2%)

  Hodgkin’s disease

13 (2%)

  Other

5 (1%)

Cancer status

  Controlled

456 (59%)

  Newly diagnosed

138 (18%)

  Recurrence, progression

173 (23%)

Performance status

  0–1 (ambulant)

471 (61%)

  2 (minor assistance)

106 (14%)

  3–4 (major assistance, bedridden)

195 (25%)

Leukopenia

50 (6%)

The main reasons for ICU admission were: severe sepsis/septic shock (n=290, 38%), acute respiratory failure (excluding septic patients; n=114, 15%), severe complications in the postoperative period (n=106, 14%), neurological diseases (n=98, 13%), cardiovascular diseases (n=56, 7%), nonseptic shock (n=36, 5%), cardiopulmonary arrest (n=30, 4%), and others (n=42, 5%). During ICU stay 531 (69%) needed MV for at least 24 h. Acute organ dysfunctions were identified as follows: cardiovascular (n=399, 52%), respiratory (n=335, 43%), renal (n=224, 29%), neurological (n=139, 18%), hematological (n=138, 18%), and hepatic (n=62, 8%). Patients were admitted to the ICU at a median of 2 days (1–7) after hospital admission. The median duration of ICU stay was 6 days (3–13), and the median duration of hospital stay was 18 days (10–32).

Data presentation and statistical analysis

Standard descriptive statistics were used to describe the study population. Continuous variables are presented as mean ±standard deviation or median and interquartile range. Dummy variables were created for variables that had more than two reference categories or levels. For each variable the category with the lowest mortality risk was assigned the reference value of 1. The Cox proportional hazards model was used to assess univariate associations between 6-month mortality and outcome variables. Variables with a p value less than 0.25 in the univariable analysis were selected to enter the multivariable analysis. Multivariable Cox proportional hazards regression models were used to test the independent contribution of each variable on mortality. Variables were selected on the basis of a stepwise forward procedure and comparison of the −2 log likelihoods of the models. Graphic assessment of the Cox model’s proportional hazard assumption revealed that the assumptions of the Cox proportional hazards model were generally met [20]. The potential associations in both univariate and multivariable analysis were summarized by calculating hazard ratios (HR) and corresponding 95% confidence intervals (CI). Variables were tested for possible interactions. SAPS II was not entered in the main multivariable analyses because other independent variables are included in that scoring system (age, variables used to define organ failures, underlying malignancy). Overall survival was defined as the time from the ICU admission to death from any cause within 6 months of follow-up. The survival patterns were estimated by means of unadjusted Kaplan-Meier curves, and comparisons between the levels of comorbidity severity used the log-rank test [21]. A two-tailed p value less than 0.05 was considered statistically significant.

Results

Characterization of patients’ comorbidities

The patients’ main comorbid conditions are presented in Table 2. Twenty-four patients (3%) had previous malignancies (20 solid tumors; 4 hematological malignancies). According to the ACE-27, at least one comorbid condition was identified in 389 patients (50%). The overall score was mild in 243 patients (31%), moderate in 105 (14%), and severe in 41 (5%). Of the patients with a severe ACE-27 score 15 (37%) had at least two grade 2, and 26 (63%) had grade 3 comorbid conditions. Using the CCI, 212 patients (27%) had comorbidity (CCI≥1), and their median score was 1 (interquartile range 1–2; range 1–8). The frequencies of individual comorbidities differed significantly between the scores (p<0.001). As expected, the frequency of comorbidities increased with age. Using the ACE-27 criteria, the rates of any comorbidity according to patients’ age were: under 40 years, 15%; 40–70 years, 51%; and over 70 years, 74% (p<0.001). A similar observation was also made when the CCI criteria was used: under 40 years, 8%; 40–70 years, 28%; and over 70 years, 40% (p<0.001). The frequencies of any comorbidity were similar in men and women using the criteria of ACE-27 (52% vs. 49%, p=0.602) or those of CCI (29% vs. 25%, p=0.223). However, there were important differences in comorbidity evaluation when using the CCI or the ACE-27 indices. Among the patients with severe comorbidity using the ACE-27 score (n=41) only 17 (41%) had a CCI score higher than 2 points. Moreover, no comorbid condition (CCI=0) was recorded in four (10%) patients. The patients with CCI score higher than 2 points (n=40) were coded by the ACE-27 as having severe (n=17, 43%), moderate (n=8, 20%), or mild (n=15, 38%) overall comorbidity scores.
Table 2

Main comorbid conditions (n=772). A patient could have more than one comorbid condition. Comorbidities were considered irrespective of the severity of disease

Systemic arterial hypertension

254 (33%)

Diabetes mellitus

60 (8%)

Chronic pulmonary disease

51 (7%)

Myocardial infarction, angina

42 (5%)

Previous malignancy

24 (3%)

Chronic renal failure

21 (3%)

Alcohol abuse

17 (2%)

Endocrine disease

16 (2%)

Stroke

16 (2%)

Congestive cardiac failure

15 (2%)

Hepatic cirrhosis, chronic hepatitis

12 (2%)

Cardiac arrhythmias

12 (2%)

Acquired immunodeficiency syndrome

11 (1%)

Peptic ulcers, gastritis

10 (1%)

Asthma

10 (1%)

Venous thromboembolic disease

9 (1%)

Morbid obesity

8 (1%)

Rheumatic disease

8 (1%)

Paraplegia, hemiplegia

8 (1%)

Psychiatric disease

6 (1%)

Peripheral artery disease

5 (1%)

Outcome analysis

The ICU, 30-day, and hospital mortality rates were 34% (263/772), 42% (325/772), and 47% (366/772), respectively. Decisions to withhold or withdraw treatment were made in 133 (17%) patients at a median of 4 days (2–10) after ICU admission, and all of these patients died in the ICU. The hospital mortality rates differed between the four categories of the ACE-27 score: 48% in patients with no comorbidity, 43% with mild, 47% for moderate, and 68% with a severe score (p=0.025).

The overall 6-month mortality rate was 58% (446/772) and the median follow-up was 64 days (10–182). Nonsurvivors were older (59.5±16.1 vs. 55.1±16.5 years, p<0.001) and, as expected, had higher SAPS II (51.6±18.0 vs. 32.6.0±14.0, p<0.001) and SOFA (7.9±4.0 vs. 4.6±2.9, p<0.001) points than survivors. Survivors and nonsurvivors had a similar duration of ICU stay [6 (4–12) vs. 6 (3–13), p=0.723)]; hospital stay was longer in survivors [21 (13–36) vs. 16 (8–29), p<0.001)]. The results of the univariable analysis of characteristics related to 6-month mortality are shown in Table 3. Kaplan-Meier curves of both CCI and ACE-27 strata are shown in Fig. 1.
Table 3

Univariable analysis of factors associated with 6-month mortality (n=772) (ACE-27 Adult Comorbidity Evaluation 27, CCI Charlson Comorbidity Index)

Variables

n

Six-month mortality (%)

Hazard-ratio (95% CI)

p

Age

0.002

  <40 years

124 (16%)

45

1.00

  40–70 years

474 (61%)

58

1.45 (1.09–1.93)

  >70 years

174 (23%)

66

1.77 (1.29–2.44)

Gender

0.127

  Women

357 (46%)

55

1.00

  Men

415 (54%)

60

1.16 (0.96–1.40)

Type of cancer

<0.001

  Solid tumors

642 (83%)

54

1.00

  Hematological malignancies

130 (17%)

78

1.81 (1.45–2.26)

Surgical patient

<0.001

  No

355 (46%)

77

1.00

  Yes

417 (54%)

42

0.38 (0.31–0.46)

Performance status

<0.001

  0, 1 (ambulant)

471 (61%)

46

1.00

  2 (minor assistance)

106 (14%)

66

1.79 (1.37–2.34)

  3, 4 (major assistance, bedridden)

195 (25%)

81

2.59 (2.11–3.18)

Cancer status

<0.001

  Controlled

456 (59%)

43

1.00

  Newly diagnosed

138 (18%)

75

2.45 (1.92–3.11)

  Recurrence, progression

173 (23%)

83

3.15 (2.54–3.91)

Leukopenia

<0.001

  No

722 (94%)

56

1.00

  Yes

50 (6%)

86

2.36 (1.72–3.24)

Need of mechanical ventilation

<0.001

  No

241 (31%)

34

1.00

  Yes

531 (69%)

69

2.90 (2.28–3.69)

Number of acute organ failures

<0.001

  0

209 (27%)

30

1.00

  1–2

339 (44%)

55

2.35 (1.76–3.14)

  >2

224 (29%)

89

6.30 (4.71–8.41)

Comorbidity

  Overall ACE-27 score

<0.001

    None

383 (50%)

58

1.00

    Mild

243 (32%)

53

0.88 (0.71–1.09)

    Moderate

105 (14%)

60

1.06 (0.81–1.41)

    Severe

41 (5%)

78

1.79 (1.26–2.59)

  CCI

<0.001

    0

560 (73%)

54

1.00

    1–2

172 (22%)

65

1.37 (1.11–1.71)

    >2

40 (5%)

75

1.80 (1.24–2.62)

Fig. 1

Kaplan-Meier survival curves on the Charlson Comorbidity Index (CCI; log-rank test=15.94, p<0.001) and Adult Comorbidity Evaluation-27 (ACE-27; log-rank test=13.81, p=0.003)

As expected, patients with a severe ACE-27 score were older (65.2±10.6 vs. 57.2±16.6 years, p=0.002) than other patients. There were no differences regarding the SAPS II (46.8±17.7 vs. 43.4±18.9 points, p=0.267) or SOFA scores (7.1±4.0 vs. 6.5±3.9 points, p=0.293). The frequencies of end-of-life decisions were 27% and 17% (p=0.144) in patients with severe and nonsevere comorbidity score, respectively.

Age, gender, type of cancer, surgical status, PS, cancer status, leukopenia, need of MV, number of severe organ failures, and comorbidity scores were entered in the multivariable Cox proportional hazards regression model. Each comorbidity score was studied separately and, therefore, two models were constructed. The ACE-27 was evaluated in the first model. Since the mortality rates for no, mild, and moderate levels of ACE-27 score were similar (Table 3), the overall comorbidity score was dichotomized according to the presence or absence of a severe score. This was also made to meet the proportionality assumption of Cox models that was not satisfied when the variable had the original four levels. Seven variables were independently associated with increased 6-month mortality (Table 4). The CCI was studied in the second model and was not independently associated with increased mortality. The other prognostic factors selected in this model were similar to those identified in the model containing the ACE-27 (Table 4). The CCI was also studied as a dichotomous variable according to the presence or absence of a score higher than two points and was also not selected in the Cox models.
Table 4

Multivariable analysis of factors associated with 6-month mortality (n=772) (ACE-27 Adult Comorbidity Evaluation 27, CCI Charlson Comorbidity Index, NA not applicable, NS not significant)

Comorbidity by ACE-27

Comorbidity by CCI

Hazard ratios (95% CI)

p

Hazard ratios (95% CI)

p

Age

  ≤40 years

1.00

1.00

  40–70 years

1.66 (1.24–2.23)

0.001

1.64 (1.22–2.20)

<0.001

  >70 years

2.07 (1.49–2.88)

<0.001

1.99 (1.42–2.78)

<0.001

Surgical patient

0.69 (0.55–0.86)

0.002

0.69 (0.55–0.86)

0.001

Performance status

 0, 1 (ambulant)

1.00

1.00

 2 (minor assistance)

1.22 (0.92–1.61)

0.165

1.20 (0.91–1.59)

0.204

 3, 4 (major assistance, bedridden)

1.45 (1.14–1.83)

0.002

1.44 (1.14–1.83)

0.002

Cancer status

 Controlled

1.00

1.00

 Newly-diagnosed

1.46 (1.11–1.91)

0.006

1.43 (1.09–1.88)

0.010

 Recurrence, progression

2.20 (1.72–2.82)

<0.001

2.21 (1.72–2.83)

<0.001

Need of mechanical ventilation

1.34 (1.00–1.78)

0.048

1.36 (1.02–1.81)

0.037

Number of acute organ failures

  0

1.00

1.00

  1–2

1.77 (1.29–2.43)

<0.001

1.75 (1.27–2.40)

0.001

  >2

3.89 (2.73–5.53)

<0.001

3.82 (2.68–5.44)

<0.001

Comorbidity

  Severe ACE-27 score

    No

1.00

NA

    Yes

1.52 (1.06–2.20)

0.024

NA

  Charlson Comorbidity Index

    0

NA

NS

    1–2

NA

NS

    >2

NA

NS

Finally, we introduced the SAPS II in a multivariable analysis along with all the variables selected in the model containing the ACE-27. The presence of a severe ACE-27 score was still significantly associated with an increased mortality (HR=1.76, CI 95%=1.22–2.53, p=0.003). As expected, the effects of age, surgical status, organ failures, and need of MV were confounded by the introduction of SAPS II, and none attained statistical significance.

Discussion

This study evaluated the prognostic impact of chronic ailments that coexist with cancer in a large series of critically ill cancer patients. Our results indicate reasonable overall ICU (34%) and hospital (47%) mortality rates. These results are similar to those reported in recent studies [4, 5, 6, 22]. Most of the previous studies have focused on the ICU and/or hospital mortality, and information regarding long-term outcomes of these patients is limited. The reported 6-month mortality rates range from 75% to 79% [7, 23]. Two studies have reported that 77% of the patients died within the first year following the ICU admission [5, 24]. In our study the overall 6-month mortality rate was 58%. However, comparisons between different studies may be hampered by the wide heterogeneity of case-mix.

Comorbidities influence the prognosis [25, 26, 27] and are associated with increased costs [28] for patients requiring ICU admission. Some of the scores used to predict the outcome of ICU patients incorporate information about only a limited number of severe chronic comorbid conditions including cancer itself [13, 29, 30]. Instruments that provide more detailed information about comorbidities may improve the outcome evaluation [25, 31]. The CCI is the most widely used instrument to measure comorbidity and was developed in a study of 559 patients admitted with various medical conditions [19]. Higher CCI scores have been associated with increased mortality of general critically ill patients [26, 27]. The ACE-27 is a new specific score developed to measure comorbidities in cancer patients [18]. It has been confirmed to be a valid instrument to measure comorbidity in the outcome evaluation of cancer patients [8, 9].

We observed significant discrepancies in comorbidity classification between the two indices, even at higher levels of comorbidity. The frequencies of any comorbidity were very different whether the CCI or ACE-27 was used. Such differences can be attributed to the high prevalence of hypertension (33%) in our population since hypertension is not considered in the CCI. Disparities of comorbidity grades between the CCI and ACE-27 have been reported previously [32]. In our study only the presence of a severe ACE-27 score was independently associated with an increased mortality. There was no association between the CCI and outcome in multivariable analysis. It is possible that applying comorbidity indices that were derived from populations with different characteristics, such as the CCI, may lead to unreliable results. In addition, the weights attributed by the index may not be applicable to populations with different case-mix and outcomes [33]. Another apparent advantage of the ACE-27 over the CCI is that different levels of severity of the disease are attributed to a more comprehensive range of comorbidities. Hypertension [34] and deep venous thrombosis [35] that are associated with worse prognosis for cancer patients are taken in consideration by the ACE-27 but not by the CCI. Another point that merits further consideration is that we did not observe significant differences in the mortality rate of patients with mild or moderate scores and patients without comorbidity. The effects of chronic comorbid ailments on the outcome may take time to manifest and also vary according to mortality burden of the index cancer [8, 9]. Transposing to the scenario of critically ill cancer patients, we can hypothesize that the effects of nonsevere comorbidities would be minimized by the well-recognized characteristics associated with worse outcomes, such as poor PS, cancer recurrence/progression, severity of organ dysfunctions and need of MV.

Surgical patients had better outcomes than medical ones, and this finding is in accordance with previous studies [2, 5]. Although we have not evaluated patients admitted for routine scheduled postoperative monitoring, roughly one-half of our patients underwent surgeries for tumor resection within the 6 months before ICU admission. We may speculate patients considered for elective surgery had fewer severe comorbidities, and therefore a selection bias regarding this aspect cannot be ruled out.

This study had a few other limitations. Some of the items of the ACE-27 refer to severe acute decompensations and may impose limitations to their interpretation in a population of critically ill cancer patients. For example, grade 3 for renal system refers to acute renal failure requiring acute dialysis or in the context of multiple organ failure. We did not evaluate the impact of comorbidities in patients’ health-related quality-of-life. The severity of preadmission comorbidities is correlated with poor health-related quality-of-life [36]. Since this was a single-center study, we cannot exclude possible selection biases concerning differences in standards of care such as do-not-resuscitate orders and admission/discharge ICU policies. In addition, we did not evaluate bone marrow transplant patients because they are treated at a separate unit with different standards of care.

In conclusion, the ACE-27 seems to be a useful instrument to measure comorbidity for critically ill cancer patients. Our findings indicate that coexistence of severe comorbid ailments must be taken into account in clinical trials and counseling of patients or their families. This is particularly important for the elderly patient, given the higher prevalence of comorbidities. However, it should be reemphasized that any characteristic must not be used as unique parameter when considering the benefits of intensive care.

Copyright information

© Springer-Verlag 2005