• terminal cancer;
  • quality of life;
  • survival prediction;
  • symptom assessment;
  • multicenter study;
  • palliative care


  1. Top of page
  2. Abstract
  6. Acknowledgements


It remains unclear whether health-related quality of life (HRQoL) measurements from patients and staff can be combined with medical data to predict survival in patients with terminal cancer.


The correlations between survival and potential health-related quality-of-life (HRQoL) prognostic variables were explored in 2 independent cohorts of patients with terminal cancer (248 patients in Cohort 1 and 756 patients in Cohort 2) after adjusting for clinical and demographics variables using Cox regression models.


At the onset of the terminal phase (Cohort 1), the hazards of dying increased by 28% in the presence of dyspnea and by 68% in the presence of nausea/emesis; however, the most important predictors of worse survival were the presence of liver metastases (hazard ratio [HR], 2.5; 95% confidence interval [95% CI], 1.8–3.8), lung tumor (HR, 2.4; 95% CI, 1.7–3.4), and tumor burden (HR, 2.0; 95% CI, 1.4–2.7). In contrast, for patients who were seen in later stages of their terminal disease (Cohort 2), dyspnea (HR, 1.5; 95% CI, 1.1–1.9) and the coexistence of weakness with a diagnosis of digestive tumors (HR, 5.2; 95% CI, 1.2–21.8), breast tumors (HR, 3.1; 95% CI, 1.6–6.2), and genitourinary tumors (HR, 3.5; 95% CI, 1.6–7.8) were more predictive of survival than the type of tumor primary. Emotional functioning along with anxiety, spiritual distress, and lack of insight were not associated consistently with survival in both cohorts.


Health care professionals should focus on physical HRQoL indicators, such as nausea and emesis, dyspnea, and weakness, to gather prognostic clues in patients with terminal cancer. These symptoms may reflect consequences of cancer cachexia and the progress of patients toward this terminal syndrome. Psychosocial distress did not appear to be associated consistently with survival; however, future studies should clarify further the prognostic significance of “positive attitudes”, such as hope and optimism, in patients with advanced cancer. Cancer 2004. © 2004 American Cancer Society.

Over the last 20 years, the evidence that duration of survival is associated positively with measures of health-related quality of life (HRQoL) in heterogeneous populations of patients with advanced cancer has grown.1–4 More specifically, studies have suggested that such an association may exist among patients with advanced malignancies of the breast,4–10 esophagus,11 colorectum,2, 12 skin (melanoma),13, 14 and lung.2, 15–20

This trend also may occur among patients with terminal cancer21; however, the magnitude and clinical relevance of these effects remain unclear.22 Some authors have found that the presence or intensity of some physical symptoms (e.g., cognitive impairment, anorexia, xerostomia, dysphagia, and dyspnea) have independent prognostic importance in patients with a median survival of < 2 months.23 Others have argued, however, that psychosocial measurements have dubious predictive value24–28 and that HRQoL measures (e.g., the Spitzer Quality of Life Index) cannot be used on their own to establish a prognosis in individual patients.29

It is possible the validity and generalizability of these discrepant results may have been affected by methodological issues, such as a lack of precise criteria to define the onset of the terminal phase, difficulty in recruiting representative cohorts among patients with advanced or terminal cancer, and missing potential prognostic factors in the design of the study. There is a tendency to employ heterogeneous and convenience samples30, 31 rather than well defined and representative inception cohorts32 as a result of the first two problems of identification and sampling. The last problem may result in a failure to evaluate important prognostic factors, such as disease-related characteristics and performance status.26–28, 33, 34 In addition, the prognostic factors considered often differ across studies, preventing possible evaluation of effects found by individual studies on these domains.23 The current study was designed to overcome these limitations and to determine whether patient-rated and staff-rated QoL measurements in patients with cancer, who were seen at two different stages of their terminal phase, could be combined with medical data to predict survival.


  1. Top of page
  2. Abstract
  6. Acknowledgements


Data were collected prospectively on two separate cohorts of patients. Cohort 1 included 248 patients who were accrued sequentially between July and December, 1996, at the Cross Cancer Institute, Edmonton, Alberta. The institute is the only referral center for oncologic treatment in Northern Alberta and has a catchment population of approximately 1.5 million. Patients were eligible for the study if they had lung, breast, gastrointestinal, or prostate cancer; if they were age > 18 years; and if they had entered the terminal phase of their illness within the previous 30 days. The terminal phase was defined (for the purpose of this study) as the time when further clinical treatments to arrest disease progression were deemed unavailable or ineffective.35 Patients in Cohort 2 included 756 new referrals that were made between July and December, 1994, to 1 of 6 multiprofessional, palliative home-care teams in Southern Ireland.35–37 These teams were selected to ensure a national representation and were the only palliative home-care services covering a population of approximately 2 million. The main reasons for referrals to these home-care services were symptom control and psychosocial support for the patient and family.

Cohort Measures

Survival was recorded from the date each patient was recruited into the study. Patients who remained alive at the end of the study were censored at that time.

Cohort 1.

Baseline QoL measurements, along with clinical and demographic variables, were recorded at the time of entry. The demographic and clinical variables measured in Cohort 1 were age, gender, primary tumor type, metastatic site, tumor burden (expressed as the total number of cancerous lesions for each primary tumor types, except for patients with prostate cancer38), weight loss, and performance status (Karnofsky performance status [KPS] scale39). The HRQoL variables were provided by the core Quality of Life Questionnaire QLQ-C30 of the European Organization for Research and Treatment of Cancer (EORTC).40 This patient-rated questionnaire includes 30 items that evaluate 5 functional domains (physical, role, emotional, cognitive, and social functioning), 3 symptom domains (fatigue, nausea/emesis, and pain), global health status/QoL, and 6 single items (dyspnea, insomnia, anorexia, constipation, diarrhea, and financial impact of disease and treatment). Separate domains and symptoms are scored on 2-point, 4-point, or 7-point scales and are normalized to lie between 0 and 100. High scores for the functional and QoL domains and lower scores on the symptom scales/items represent greater well being. The reliability and validity of the EORTC QLQ-C30 was demonstrated in 305 patients with nonresectable lung cancer from centers in 13 countries. This study demonstrated a Cronbach α coefficient ≥ 0.70 for all multiitem scales, with the exception of role functioning. Validity was shown by three findings. First, whereas all interscale correlations were statistically significant, the correlation was moderate, indicating that the scales were assessing distinct components of the QoL construct. Second, most of the functional and symptom measures discriminated clearly between patients who differed with regard to clinical status, as defined by the Eastern Cooperative Oncology Group Performance Status scale, weight loss, and treatment toxicity. Third, there were statistically significant changes, in the expected direction, in physical and role functioning, global QoL, fatigue, and nausea/emesis for patients who had improved or worsened performance status during treatment. The reliability and validity of the questionnaire were highly consistent across the three language-cultural groups studied: patients from English-speaking countries, Northern Europe, and Southern Europe.

Cohort 2.

The demographic and clinical variables measured in Cohort 2 were age, gender, primary tumor type, and performance status on the KPS scale. QoL variables were measured with the Support Team Assessment Schedule (STAS).41, 42 The STAS is a staff-rated questionnaire that contains nine core items relating to physical symptoms (pain and symptom control), psychological functioning of patients and caregivers (patient and family anxiety and insight), and communication (between patient and family, between professionals, and between professionals and patients). Each item is rated on a scale from 0 (best) to 4 (worst), according to the effect, as described by the patient. The face and content validity of STAS was established by observation, professional discussion, and patient assessment of important items. Criterion validity was established by comparing patient and staff ratings, which showed moderate correlations (ρ ranged from 0.45 to 0.66; all P values were < 0.0005). In a test of constant validity, it was shown that QoL (using a modified Spitzer index) was correlated with similar STAS items in patients 6 weeks before death. Reliability, as assessed by comparing the ratings of different staff, showed agreement within 1 score in 88% or more patients: The Cohen κ value ranged from 0.48 to 0.87, and the Cronbach α value ranged from 0.68 to 0.89, depending on the stage of illness.41, 42 STAS has been used widely since in palliative care in many settings and countries. In this study, we considered core items, such as pain, other symptoms, and patient insight, as well as additional items, such as spiritual distress, dyspnea, nausea, emesis, constipation, weakness/lethargy, and patient well being.

Statistical Analysis

Symptom variables, such as dyspnea, weakness, and pain, were dichotomized according to the quartile values, ensuring that the dichotomization was 1) biologically plausible; 2) of prognostic importance; and, as far as possible, 3) consistent with previous reports. It was found that, for most HRQoL variables, averages (means and medians) represented the best cut-off values and the most appropriate way to compare results from categorical and continuous scales in the two HRQoL instruments. The α coefficients were calculated for each of the total scales and for each of the subscales for the two cohorts to determine these measures performance prior to further analysis.

Next, preliminary univariate analyses were conducted on both HRQoL and non-HRQoL variables. The Cox regression method43 was used to examine variables as single, main-effect associations with survival. Dependencies among the explanatory variables were explored through 2 × 2 tables and logistic regression analyses. Multivariate models for survival analysis were implemented in two stages.43 First, separate Cox regression models were derived for the HRQoL and non-HRQoL variables. Next, the best sets of HRQoL and non-HRQoL were included in a further Cox regression model. Unselected variables and interactions were added, one at a time, to the latter models to detect any significant contribution. In addition, we ran stepwise forward and backward regressions models with all of the HRQoL and non-HRQoL variables in the model. We repeated these analyses, stratifying both samples by primary tumor type. Variables that showed a different effect on survival according to tumor type were included in interaction terms with the type of tumor and were tested in the final models. The proportionality of hazards associated with all independent predictors of survival was checked first by visual inspection of the log-cumulative-hazard plots and then by appropriate tests of time interactions. We used SPSS software (version 8.0)44 for all statistical analyses.


  1. Top of page
  2. Abstract
  6. Acknowledgements

Baseline Characteristics

The Cronbach α coefficient ranged from 0.83 to 0.85 for EORTC scales in Cohort 1 and from 0.64 to 0.69 for STAS scales in Cohort 2, with an overall α coefficient of 0.85 for the EORTC scale and 0.69 for the STAS scale. Tables 1 and 2 present demographic and clinical characteristics of patients in Cohorts 1 and 2, respectively. The minimum follow-up was 14 months in Cohort 1 and 6 months in Cohort 2. At the end of the respective studies, 229 patients (92%) had died in Cohort 1, and 508 patients (68%) had died in Cohort 2. Patients in Cohort 2 had a much shorter survival, with a hazard of more than double that in Cohort 1. The median survival was 15.3 weeks for Cohort 1 and 6.1 weeks for Cohort 2. Patients in Cohort 2 were slightly older and had a worse performance status compared with patients in Cohort 1. The average weight loss for patients in Cohort 1 was 8.1 kilograms over 6 months prior to accrual into this study.

Table 1. Univariate and Multivariate Cox Regression Models in Cohort 1
PredictorNo. (%)Univariate analysisMultivariate analysis
HR95% CIP valueHR95% CIP value
  • HR: hazard ratio; 95% CI: 95% confidence interval; QOL: quality of life; —: not applicable and/or not statistically significant; mets: metastases; KPS: Karnofsky performance status; T-COV: time covariate.

  • a

    It was found that the hazards associated with different levels of performance status were time-dependent and decreased gradually over time.

Non-QOL variables       
 Age ≥: 65 yrs'112 (45.2)1.10.8–1.4>0.20
 Male gender103 (41.5)1.10.8–1.4>0.20
 Primary tumors       
  Prostate 21 (8.5)1.10.7–1.7>0.20
  Breast 70 (28.2)0.90.6–1.2<0.001
  Lung 77 (31.0)1.61.2––3.4<0.001
  Digestive 80 (32.3)0.70.5–0.9>0.20
 Brain mets 57 (23.0)1.41.0–1.90.04
 Lung mets 84 (33.9)1.10.8–1.5>0.20
 Liver mets 74 (29.8)1.71.3–2.2<0.0012.51.8–3.4<0.001
 Bone mets 97 (39.1)1.10.8–1.4>0.20
 Lymph node mets 80 (32.3)0.90.7–1.2>0.20
 Visceral metastases 32 (12.9)1.10.7–1.6>0.20
 Skin metastases 10 (4.0)0.80.4–1.6>0.20
 Tumor burden ≥5 lesions173 (69.8)1.81.3–2.4<0.0012.01.4–2.7<0.001
  ≥80 70 (28.2)1.01.0
  60–70113 (45.6)2.11.3–3.40.0042.11.2–3.40.006
  ≤50 65 (26.2)4.82.7–8.4<0.0013.92.2–7.0<0.001
QOL variables       
 Emotional function below average116 (46.8)1.31.0–1.60.06
 Physical function below average137 (55.2)1.51.2–2.00.002
 Role function below average112 (45.2)1.51.1–1.90.004
 Social function below average120 (48.4)1.41.1–1.80.02
 Cognitive function below average120 (48.4)1.51.1–1.90.003
 Overall QOL below average114 (46.0)1.51.1–1.90.003
 Fatigue above average128 (51.6)1.51.2–2.00.001
 Nausea/emesis above average 91 (36.7)1.71.3–2.3<0.0011. 71.2–2.2<0.001
 Pain above average112 (45.2)1.51.2–2.00.002
 Dyspnea above average132 (53.2)1.31.0––1.60.09
 Insomnia above average 89 (35.9)1.00.8–1.3>0.20
 Anorexia above average114 (46.0)1.71.3–2.2<0.001
 Constipation above average 91 (36.7)1.61.2–2.1<0.001
 Diarrhea above average 65 (26.2)0.90.7–1.3>0.20
 Financial impact above average 61 (24.6)0.80.6–1.10.15
Interaction terms       
 KPS × T-COVa       
  ≥80 × T-COV1.01.0
  60–70 × T-COV0.990.97–1.0>0.200.990.98–1.0>0.20
  ≤50 × T-COV0.960.94–0.980.0020.960.94–0.990.002
Table 2. Univariate and Multivariate Cox Regression Models in Cohort 2
PredictorNo. (%)Univariate analysisMultivariate analysis
HR95% CIP valueHR95% CIP value
  • HR: hazard ratio; 95% CI: 95% confidence interval; QOL: quality of life; —: not applicable and/or not statistically significant; KPS: Karnofsky performance status; T-COV: time covariate.

  • a

    It was found that the hazards associated with different levels of performance status were time-dependent and decreased gradually over time.

Non-QOL variables       
 Age: ≤65 yrs460 (60.8)1.21.0––2.50.07
 Female gender375 (49.6)1.21.0––0.90.01
 Primary tumors       
  Other118 (15.6)1.01.0
  Breast 64 (8.5)0.70.5––2.2>0.20
  Lung163 (21.6)1.10.9–1.4>–1.4>0.20
  Gastrointestinal237 (31.3)1.10.9––1.3>0.20
  Genitourinary108 (14.3)0.80.6–1.1>–2.5>0.20
  ≥60274 (36.2)1.01.0
  40–50178 (23.5)2.21.5–3.1<0.0012.61.7–3.9<0.001
  ≤30 88 (11.6)7.44.8–11.5<0.00112.97.8–21.4<0.001
QOL variables       
 Other symptoms above average330 (47.7)1.61.3–1.9<0.001
 Dyspnea above average141 (18.7)1.31.0––1.90.008
 Weakness above average175 (23.1)2.01.6––1.4>0.20
 Impairment in well being above average240 (31.7)1.31.1–1.60.007
 Nausea/emesis above average179 (23.7)1.10.9–1.4>0.20
 Constipation above average288 (38.1)1.00.8–1.2>0.20
 Pain above average310 (41.0)1.00.9–1.3>0.20
 Anxiety above average322 (42.6)1.00.8–1.2>0.20
 Lack of insight above average211 (27.9)0.90.7–1.1>0.20
 Spiritual distress above average 88 (11.6)1.20.9–1.6>0.20
Interaction terms       
 KPS × T-COVa       
  ≥60 × T-COV1.01.0
  40–50 × T-COV0.970.95–0.98<0.0010.910.86–0.980.006
  ≤30 × T-COV0.910.88–0.96<0.0010.680.55–0.83<0.001
 Age × tumor type       
  ≤65 yrs × other1.0
  >65 yrs × breast0.20.07–0.80.03
  >65 yrs × lung1.90.9–4.1>0.20
  >65 yrs × gastrointestinal1.00.5–2.0>0.20
  >65 yrs × genitourinary0.30.1–0.70.004
 Weakness × tumor type       
  Weakness at/below average × other tumor1.0
  Weakness above average × breast5.21.2–21.80.02
  Weakness above average × lung1.90.9–4.10.09
  Weakness above average × gastrointestinal3.11.6–6.20.001
  Weakness above average × genitourinary3.51.6–7.80.002

Statistical Analysis

Logistic regression analysis showed that patient reports of fatigue were correlated independently in Cohort 1 with anorexia (odds ratio [OR], 6.0; 95% confidence interval [95% CI], 2.8–12.5; P ≤ 0.001); nausea and emesis (OR, 6.9; 95% CI, 1.87–25.1; P = 0.004); and, to a lesser extent, dyspnea (OR, 3.2; 95% CI, 1.5–6.8; P = 0.002). In Cohort 2, staff reports of weakness were correlated with dyspnea (OR, 1.4; 95% CI, 1.2–1.6; P ≤ 0.001) and with nausea and emesis (OR, 1.5; 95% CI, 1.2–1.7; P ≤ 0.001). Lower well being was associated independently in Cohort 1 with anorexia (OR, 6.3; 95% CI, 2.7–14.6; P ≤ 0.001); emotional functioning (OR, 9.8; 95% CI, 2.7–35.2; P = 0.001); and, to a lesser extent, with fatigue (OR, 3.0; 95% CI, 1.4–6.0; P ≤ 0.001) and sleep disturbance (OR, 3.0; 95% CI, 1.3–7.0; P ≤ 0.001). In Cohort 2, worse levels of well being were correlated independently with weakness (OR, 3.2; 95% CI, 2.3–4.4; P ≤ 0.001) and nausea (OR, 5.0; 95% CI, 3.2–6.4; P ≤ 0.001). Results of the univariate and multivariate survival analyses for Cohorts 1 and 2 are presented in Tables 1 and 2, respectively.

In Cohort 1, the hazard of dying was almost tripled for patients with liver metastases, it was doubled for patients with higher tumor burden, and it was 2.4 times greater for patients with lung cancer. There was a significant association with shorter survival as the KPS score decreased, with the hazard essentially doubling at each deteriorating functional level. However, nonproportionality of hazards was detected across the 3 categories of KPS scores (≤ 50, 60–70, and ≥ 80), and a time-by-performance status interaction term was introduced. Consequently, the hazard ratio (HR) for patients with KPS scores ≤ 50 in Cohort 1 decreased by 4% compared with baseline for each week of follow-up. The hazard of dying increased by 28% in the presence of dyspnea and by 68% in the presence of nausea/emesis. All of these effects were of approximately the same magnitude across the different types of primary tumors. Only a weak association at the univariate analysis level was found between emotional functioning and survival. No association was found between nausea/emesis and constipation.

In Cohort 2, the most powerful predictors of shorter survival were KPS and, for some tumors, weakness. Like Cohort 1, worse survival was associated strongly with low KPS scores. Patients with KPS scores ≤ 30 at the beginning of their follow-up had an HR of 12.9 relative to the reference category (KPS ≥ 60). However, the HRs for patients with KPS scores of 40–50 and ≤ 30 decreased by 9% and 32%, respectively, for each week of follow-up. Among the non-HRQoL variables, gender and age also were associated with survival. The hazard of dying increased by 37% for male patients and by 40% for patients age < 65 years. Among the HRQoL variables, above average dyspnea increased the hazard of dying by 50%. Significant interactions of tumor type with weakness and age were found. Staff-reported weakness greatly increased the hazard of dying for patients with digestive (three-fold), breast (five-fold), and genitourinary (four-fold) tumors but had no effect on patients with respiratory tumors. In contrast, for patients with breast and genitourinary tumors, age > 65 years showed a decrease (four-fold) in the hazard of dying. No association with survival was found for high levels of anxiety, spiritual distress, or lack of insight.

In summary, at the onset of the terminal phase (Cohort 1), the severity of nausea/emesis was associated independently with the length of survival, but this association was not as strong as that of disease-related and clinical characteristics. In contrast, for patients who were seen in later stages of their terminal disease (Cohort 2), the presence of dyspnea and weakness in patients who had digestive, breast, and genitourinary tumors was more predictive of shortened survival than the type of primary tumor. In both cohorts, performance status at entry was associated strongly with survival; however, patients with a KPS score ≤ 50 revealed significant erosion of this association over time. Indicators of psychosocial distress, such as anxiety and spiritual distress, generally were not associated with survival.


  1. Top of page
  2. Abstract
  6. Acknowledgements

Results of the current study suggest that physical measures, rather than psychological measures, of QoL can be useful aids in predicting the survival of patients with terminal cancer. These findings are consistent with several studies, which found that the prognostic relevance of multidimensional HRQoL assessments was due largely to the physical component of these assessments and to their impact on different global HRQoL scores.1–12, 14–21, 23–28, 30–31 In particular, our data confirmed an independent prognostic value for nausea, weakness, and dyspnea.

A significant prognostic value for nausea and emesis in patients with advanced and end-stage cancer was reported by two recent studies.21, 45 The pathogenesis of nausea remains multifactorial in patients with terminal cancer, but it often is associated with constipation—a problem that can be treated effectively and/or prevented in this patient population. Our data indicate that, in Cohort 1, nausea and emesis were not associated with constipation. Chronic nausea in patients with advanced cancer frequently reflects dysfunctions in the autonomic nervous system.46 These dysfunctions have been attributed mainly to the cachexia syndrome.47 Furthermore, nausea and weakness should be considered symptoms of the complex inflammatory changes related to multiple cytokines and tumor byproducts, which largely are responsible for the cachexia syndrome.48 Supporting these hypotheses, our data showed significant weight loss in patients who were newly diagnosed with advanced cancer (Cohort 1). Our findings suggest that the presence and intensity of nausea reflect the severity of the cancer cachexia, particularly in an early phase of the terminal disease.

In our cohorts, it was found that weakness and dyspnea, along with anorexia, sleep disturbance, and low emotional functioning, also were correlated highly with the level of well being. Anorexia and weakness traditionally have been associated with the cachexia syndrome.49, 50 It is known that the severity of dyspnea does not correlate well with the presence of specific etiologic factors (e.g., low levels of hemoglobin or lung/pleural disease) but, rather, correlates with the degree of muscle weakness in patients with terminal cancer.51 The prognostic role of dyspnea, along with other symptoms of cancer cachexia, such as anorexia, dysphagia, and weight loss, has been documented well in this patient population.23 The presence of anorexia, dysphagia, and weight loss characterizes a common clinical pathway, termed the “terminal cancer syndrome”, through which patients with different types of end-stage malignancies go.35, 52 Our findings, added to these observations, suggest that both weakness and dyspnea, rather than disease characteristics, reflect the overall severity of the cancer cachexia and predict survival, particularly in the later stages of terminal cancer.

The strong but time-dependent association of low KPS with survival may explain in part why previous work either did not support the prognostic role of performance status in terminal cancer patients53 or have found performance status independently associated with survival only in patients with shorter life expectancies.54 The presence or severity of certain symptoms, such as weakness, dyspnea, and anorexia, may help to identify the patients for whom performance status assessments have greater prognostic value.


One of the cohorts in this study did not record the assessment of clinical characteristics, such as weight loss, tumor burden, and metastatic sites. However, the data on the extent of disease and metastatic site in patients who were referred to home-care service may be inaccurate, because such patients normally are not investigated extensively. Professionals may make different assessments of QoL compared with the assessments given by the dying patients themselves. In particular, professionals may under-report some symptoms compared with patients' self-assessments.55 However, in patients with far advanced illness, the reliability and completeness of assessments provided by the patient can become a problem, leading to sample bias.56 Our data suggest that staff-rated symptom assessments provide prognostic information similar to that reported for patient-rated assessments. Furthermore, recent data show good approximations for proxy ratings by, e.g., family members or nurses, for symptoms like fatigue, appetite, drowsiness, and shortness of breath in patients with terminal cancer.57 Our study did take into consideration psychosocial measures of QoL (i.e., emotional functioning, anxiety, insight, communication, and spiritual distress), which may not be the most meaningful for prognostication of survival. Some authors have suggested that psychosocial reaction to the disease or coping strategies, such as hope or optimism, may have a positive influence on survival in patients with melanoma17 and in patients with head and neck cancer.58 A more recent study, however, found no evidence that high levels of optimism prior to treatment enhanced survival in patients with nonsmall cell lung cancer.59 Finally, the two studies used different measures to assess HRQoL. The QLQ-C30 and STAS assessed symptoms and psychological aspects differently, and it is possible that some of the differences between the two studies were due not to disease stage but to the different measures used. However, the finding of some similarities—particularly in the symptoms of nausea, emesis, dyspnea, and weakness—and the concentration of findings among physical rather than psychological symptoms suggest, at the least, that these physical symptoms are important.


These results have potential implications for both clinical and research practice. Health care professionals who are interested in survival should look more carefully at certain physical symptoms as prognostic clues from HRQoL assessments in terminal cancer patients. Regardless of the setting, our findings indicate that patients who present with chronic nausea and emesis, dyspnea, and weakness are more advanced in the progress toward terminal cachexia compared with patients who do not present with these symptoms. Routine quantification of the severity of these symptoms using scales, such as the QLQ-C30 or STAS, may help physicians to recognize that life is short and, if it has not already been done, to facilitate consultation/referral to programs that assist patients and their families with end-of-life concerns. Nausea and emesis, dyspnea, and weakness also deserve particular attention, because they probably reflect the biologic consequences of the cancer-cachexia syndrome. Therefore, careful assessment of these symptoms is warranted both for determining the clinical relevance of the molecular changes induced by the cancer cachexia and for measuring outcomes of therapeutic interventions aimed at reversing or alleviating those changes.60 According to our data, patients and families should be reassured that they do not feel they or their relative are going to die sooner if they are in psychosocial distress. However future studies should clarify further the prognostic significance of positive attitudes, such as hope and optimism.

Disease characteristics, such as the type of tumor and the number of metastases, retain important prognostic value at the onset of terminal disease and may clarify the survival effect of certain symptoms in later stages of the terminal phase. The evaluation of performance status should be considered as one of many criteria rather than the main criterion to establish prognosis in patients with terminal disease.

Future studies should validate the inception criteria we used to accrue cancer patients at the onset of their terminal disease38 and should define common inception points for the accrual of patients who are in later stages of their terminal phase. Quality of life assessments may assist in the selection of homogeneous cohorts of patients with terminal cancer in which the prognostic role of demographic and disease-related variables may be defined better.


  1. Top of page
  2. Abstract
  6. Acknowledgements

The authors thank Drs. Mark Atkinson, Mervyn Dean, and Neil MacDonald for their invaluable peer review of this article.


  1. Top of page
  2. Abstract
  6. Acknowledgements
  • 1
    Coates A, Porzsolt F, Osoba D. Quality of life in oncology practice: prognostic value of EORTC QLQ-C30 scores in patients with advanced malignancy. Eur J Cancer. 1997; 33: 10251030.
  • 2
    Loprinzi CL, Laurie JA, Wieand HS, et al. Prospective evaluation of prognostic variables from patient-completed questionnaires. North Central Cancer Treatment Group. J Clin Oncol. 1994; 12: 601607.
  • 3
    Sloan JA, Loprinzi CL, Laurine JA, et al. A simple stratification factor prognostic for survival in advanced cancer: the good/bad/uncertain index. J Clin Oncol. 2001; 19: 35393546.
  • 4
    Shadbolt B, Barresi J, Craft P. Self-rated health as a predictor of survival among patients with advanced cancer. J Clin Oncol. 2002; 20: 25142519.
  • 5
    Coates AS, Gebski V, Bishop JF, et al. Improving the quality of life in advanced breast cancer. A comparison of continuous and intermittent strategies. N Engl J Med. 1987; 317: 14901495.
  • 6
    Coates AS, Byrne M, Bishop JF, et al. Intermittent versus continuous chemotherapy for breast cancer [letter]. N Engl J Med. 1988; 318: 1468.
  • 7
    Coates AS, Gebski V, Signorini D, et al. Prognostic value of quality of life scores during chemotherapy for advanced breast cancer. Australian-New Zealand Breast Cancer Trials Group. J Clin Oncol. 1992; 10: 18331838.
  • 8
    Butow PN, Coates AS, Dunn SM. Psychosocial predictor of survival: metastatic breast cancer. Ann Oncol. 2000; 11: 469474.
  • 9
    Shimozuma K, Sonoo H, Ichihara K, et al. The prognostic value of quality of life scores: preliminary results of an analysis of patients with breast cancer. Surg Today. 2000; 30: 255261.
  • 10
    Coates AS, Hurny C, Peterson HF, et al. Quality-of-life scores predict outcome in metastatic but not early breast cancer. J Clin Oncol. 2000; 18: 37683774.
  • 11
    Blazeby JM, Brookes ST, Alderson D. The prognostic value of quality of life scores during treatment for oesophageal cancer. Gut. 2001; 49: 227230.
  • 12
    Earlam S, Glover C, Fordy C, et al. Relation between tumor size, quality of life and survival in patients with colorectal liver metastases. J Clin Oncol. 1996: 14: 171175.
  • 13
    Coates AS, Thomson D, McLeod GR, et al. Prognostic value of quality of life score in trial of chemotherapy with or without interferon in patients with metastatic melanoma. Eur J Cancer. 1993; 29A: 17311734.
  • 14
    Butow PN, Coates AS, Dunn SM. Psychosocial predictors of survival in metastatic melanoma. J Clin Oncol. 1999; 17: 22562263.
  • 15
    Kaasa S, Mastekaasa A, Lund E. Prognostic factors for patients with inoperable non-small cell lung cancer, limited disease: the importance of patients' subjective experience of disease and psychosocial well-being. Radiol Oncol. 1989; 15: 235242.
  • 16
    Ruckdeschel JC, Piantadosi S, The Lung Cancer Study Group. Quality of life assessments in lung surgery for brochogenic carcinoma. Theor Surg. 1991; 6: 201205.
  • 17
    Ganz PA, Lee JJ, Siau J. Quality of life assessment: an independent prognostic variable for survival in lung cancer. Cancer. 1991; 67: 31313135.
  • 18
    Buccheri GF, Ferrigno D, Tamburini M, Brunelli C. The patient's perception of his own quality of life might have an adjunctive prognostic significance in lung cancer. Lung Cancer. 1995; 12: 4558.
  • 19
    Herndon JE II, Fleishman S, Kornblith AB, et al. Is quality of life predictive of the survival of patients with advanced nonsmall cell lung carcinoma? Cancer. 1999; 85: 333340.
  • 20
    Montazeri A, Milroy R, Hole D, et al. Quality of life in lung cancer patients: an important prognostic factor. Lung Cancer. 2001; 31: 233240.
  • 21
    Tamburini M, Brunelli C, Rosso S, Ventafridda V. Prognostic value of quality of life scores in terminal cancer patients. J Pain Symptom Manage. 1996; 1: 3241.
  • 22
    Maltoni M, Pirovano M, Nanni O, et al. Prognostic factors in terminal cancer patients. Eur J Palliat Care. 1996; 3: 122125.
  • 23
    Viganò A, Dorgan M, Bruera E, Buckingam J, Suarez-Almazor ME. Survival prediction in terminal cancer patients: a systematic review of the medical literature. Palliat Med. 2000; 14: 363374.
  • 24
    Cassileth BR, Lusk EJ, Miller DS, et al. Psychosocial correlates of survival in advanced malignant disease? N Engl J Med. 1985; 312: 15511555.
  • 25
    Cassileth BR, Walsh WP, Lusk EJ. Psychosocial correlates of cancer survival: a subsequent report 3 to 8 years after cancer diagnosis. J Clin Oncol. 1988; 6: 17531759.
  • 26
    Ringdal GI, Ringdal K, Kvinnsland S, Gotestam KG. Quality of life of cancer patients with different prognoses. Qual Life Res. 1994; 3: 143154.
  • 27
    Ringdal GI, Gotestam KG, Kaasa S, et al. Prognostic factors and survival in a heterogeneous sample of cancer patients. Br J Cancer. 1996; 73: 15941599.
  • 28
    Ringdal GI, Ringdal K. A follow-up study of the quality of life in cancer patients with different prognoses. Qual Life Res. 2000; 9: 6573.
  • 29
    Addington-Hall JM, MacDonald LD, Anderson HR. Can the Spitzer's Quality of Life Index help to reduce prognostic uncertainty in terminal care? Br J Cancer. 1990; 62: 695699.
  • 30
    Dancey J, Zee B, Osoba D, et al. Quality of life scores: an independent prognostic variable in a general population of cancer patient receiving chemotherapy. Qual Life Res. 1997; 6: 151158.
  • 31
    Chang VT, Thaler H, Polyak TA, et al. Quality of life and survival. Cancer. 1998; 83: 173179.
  • 32
    Laupacis A, Wells G, Richardson WS, et al. Users' guides to medical literature, V. How to use an article about prognosis. JAMA. 1994; 272: 234237.
  • 33
    Roud PC. Psychosocial variables associated with the exceptional survival of patients with advanced malignant disease. Int J Psychiatr Med. 1986/1987; 16: 113122.
  • 34
    Watson M, Haviland JS, Greer S, et al. Influence of psychological response on survival in breast cancer; a population cohort study. Lancet. 1999; 354: 13311336.
  • 35
    Hodgson C, Higginson IJ, McDonnell M, Butters E. Family anxiety in advanced cancer: a multicentre prospective study in Ireland. Br J Cancer. 1997; 76: 12111214.
  • 36
    Edmonds P, Higginson I, Altmann D, Sen-Gupta G, McDonnell M. Is the presence of dyspnea a risk factor for morbidity in cancer patients? J Pain Symptom Manage. 2000; 19: 1522.
  • 37
    Higginson IJ, Costantini M. Communication in end-of-life cancer care: a comparison of team assessments in three European countries. J Clin Oncol. 2002; 20: 36743682.
  • 38
    Viganò A, Bruera E, Suarez-Almazor ME. Terminal cancer syndrome: myth or reality. J Palliat Care. 1999; 15: 3239.
  • 39
    Karfnosky DA, Burchenal JH. The clinical evaluation of chemotherapeutic agents in cancer. In: MacLeodCM, editor. Evaluation of chemotherapeutic agents. New York: Columbia University Press, 1949: 199205.
  • 40
    Aaronson NK, Ahmedzai S, Bergman B et al. The European Organization for Research and Treatment of Cancer QLQ-C30: a quality of life instrument for use in international clinical trials in oncology. J Natl Cancer Inst. 1993; 85: 365376.
  • 41
    Higginson I, McCarthy MA. comparison of two measures of quality of life: their sensitivity and validity for patients with advanced cancer. Palliat Med. 1994; 8: 282290.
  • 42
    Higginson I, McCarthy M. Validity of the Support Team Assessment Schedule: views of patients and families. Palliat Med. 1993; 7: 219228.
  • 43
    Parmar MKB, Machin D. Survival analysis: a practical approach. Chichester: John Wiley & Sons, 1994.
  • 44
    SPSS Inc. SPSS for Windows. Base system user's guide, release 8.0. Chicago: SPSS, Inc, 1998.
  • 45
    Viganò A, Bruera E, Jhangri GS, et al. Clinical survival predictors in patients with advanced cancer. Arch Intern Med. 2000; 160: 861868.
  • 46
    Perreira J, Bruera E. Chronic nausea. In: BrueraE, HigginsonIJ, editors. Cachexia-anorexia in cancer patients. Oxford: Oxford University Press, 1996.
  • 47
    Bruera E, Chadwick S, MacDonald N, et al. Study of cardiovascular autonomic insufficiency in advanced cancer patients. Cancer Treat Rep. 1986; 70: 13831387.
  • 48
    Bruera E, Strasser F. Anorexia-cachexia. In: BrueraE, RipamontiC, editors. Gastrointestinal symptoms in advanced cancer patients. Oxford: Oxford University Press, 2002: 3980.
  • 49
    Viganò A, Watanabe S, Bruera E. Anorexia and cachexia in advanced cancer patients. Cancer Surveys. 1994; 21: 99115.
  • 50
    Krech RL, Walsh D. Symptoms of pancreatic cancer. J Pain Symptom Manage. 1991; 6: 360367.
  • 51
    Ripamonti C, Bruera E. Dyspnea: pathophysiology and assessment. J Pain Symptom Manage. 1997; 13: 220232.
  • 52
    Reuben DB, Mor V, Hiris J, et al. Clinical symptoms of survival in patients with terminal cancer. Arch Intern Med. 1988; 148: 15861591.
  • 53
    Bruera E, Miller MJ, Kuehn N, et al. Estimated survival of patients admitted to a palliative care unit: a prospective study. J Pain Symptom Manage. 1992; 7: 8286.
  • 54
    Allard P, Dionne A, Potvin D. Factors associated with length of survival among 1081 terminally ill cancer patients. J Palliat Care. 1995; 11: 2024.
  • 55
    Slevin MI, Plant H, Lynch D, et al. Who should measure quality of life, the doctor or the patient? Br J Cancer. 1988; 57: 109112.
  • 56
    Maguire P. Monitoring the quality of life in cancer patients and their relatives. In: SymingtonT, WilliamsAE, McVieJG, editors. Cancer: assessment and monitoring. London: Churchill Livingstone, 1980: 4052.
  • 57
    Viganò A, Atkinson M, Radice D, et al. Symptom assessment in terminally ill cancer patients: clinical and research insights from the Edmonton Symptom Assessment System (ESAS). Second Congress of the European Association for Palliative Care Research Network, Lyon, France, May 23–25, 2002.
  • 58
    Allison PJ, Guichard C, Fung K, Gilain L. Dispositional optimism predicts survival status 1 year after diagnosis in head and neck cancer. J Clin Oncol. 2003; 21: 543548.
  • 59
    Schofield P, Ball D, Smith J, Borland R, et al. Optimism and survival in lung carcinoma patients. Cancer. 2004; 100: 12761282.
  • 60
    Viganò A, Bruera E. Enteral and parenteral nutrition in cancer patients. In: HigginsonI, BrueraE, editors. Cachexia-anorexia syndrome in advanced cancer. London: Oxford Medical Publications, 1996: 110127.