Simple Prognostic Model for Patients With Advanced Cancer Based ...

Clinical Research Practices

Original Contribution

Simple Prognostic Model for Patients With Advanced Cancer Based on Performance Status By Raymond W. Jang, MD, Valerie B. Caraiscos, MD, PhD, Nadia Swami, Subrata Banerjee, MD, Ernie Mak, MD, Ebru Kaya, MBBS, Gary Rodin, MD, John Bryson, MD, Julia Z. Ridley, MD, MEd, Lisa W. Le, and Camilla Zimmermann, MD, PhD

Abstract Purpose: Providing survival estimates is important for decision making in oncology care. The purpose of this study was to provide survival estimates for outpatients with advanced cancer, using the Eastern Cooperative Oncology Group (ECOG), Palliative Performance Scale (PPS), and Karnofsky Performance Status (KPS) scales, and to compare their ability to predict survival.

Methods: ECOG, PPS, and KPS were completed by physicians for each new patient attending the Princess Margaret Cancer Centre outpatient Oncology Palliative Care Clinic (OPCC) from April 2007 to February 2010. Survival analysis was performed using the Kaplan-Meier method. The log-rank test for trend was employed to test for differences in survival curves for each level of performance status (PS), and the concordance

Introduction Despite continued advances in cancer treatment, many cases of cancer remain incurable. Accurate prognostic information can help physicians decide whether to initiate or continue anticancer therapy,1 facilitate transitions to hospice care,2 and enable appropriate advance care planning and end-of-life decision making.3 Patients with advanced cancer and their family members often ask questions related to prognosis, and open, empathic discussions about this topic may improve satisfaction with care.4 Multiple clinical and physiological factors have been found to be related to prognosis for patients with advanced cancer.5 These include symptoms such as anorexia, dyspnea, and fatigue6,7; disease characteristics such as cancer diagnosis, site of metastasis, and comorbidity8; laboratory values such as hypoalbuminemia, leukocytosis, and anemia6,9; the clinician’s overall prediction of survival5,10; and performance status.11 Various models have been designed, based on combinations of the factors mentioned above and others, to assist clinicians with prognostication.6,8-10,12 However, these models tend to be complex, requiring input of laboratory values or calculations that can be time consuming and impractical for rapid outpatient assessment of prognosis in a cancer clinic.6,9 As well, some of these models were designed to determine the probability of surviving less than 2 months,6,12 and may not be appropriate for delineating longer survival times in patients with advanced cancer in an outpatient setting. Copyright © 2014 by American Society of Clinical Oncology

index (C-statistic) was used to test the predictive discriminatory ability of each PS measure.

Results: Measures were completed for 1,655 patients. PS delineated survival well for all three scales according to the logrank test for trend (P ⬍ .001). Survival was approximately halved for each worsening performance level. Median survival times, in days, for each ECOG level were: EGOG 0, 293; ECOG 1, 197; ECOG 2, 104; ECOG 3, 55; and ECOG 4, 25.5. Median survival times, in days, for PPS (and KPS) were: PPS/KPS 80-100, 221 (215); PPS/KPS 60 to 70, 115 (119); PPS/KPS 40 to 50, 51 (49); PPS/KPS 10 to 30, 22 (29). The C-statistic was similar for all three scales and ranged from 0.63 to 0.64.

Conclusion: We present a simple tool that uses PS alone to prognosticate in advanced cancer, and has similar discriminatory ability to more complex models.

Performance status represents a global assessment of the patient’s level of function. It is assessed routinely in oncology outpatient settings, and is used to assess eligibility of patients for clinical trials, determine level of fitness for cancer treatment, follow the response to treatment, and assess eligibility for services such as home care. The Eastern Cooperative Oncology Group (ECOG) scale13 is most commonly employed in oncology settings, although the Karnofsky Performance Status (KPS)14 is also used; the Palliative Performance Scale (PPS) tends to be utilized in palliative care settings.15 Although these three measures are known to be correlated with survival,16-19 there has been no study (to our knowledge) assessing and comparing their predictive validity in relation to actual survival time. The purpose of this study was to assess the survival time associated with levels of performance status in outpatients with advanced cancer for the ECOG, PPS, and KPS measures, and to determine the validity of each scale in estimating survival. Ultimately, we wished to develop a simple tool, based on performance status alone, that could be used to estimate prognosis for patients with advanced cancer in an outpatient setting.

Methods Study Setting and Participants The study was conducted at the Princess Margaret Cancer Centre, a comprehensive cancer center within the University Health

S E P T E M B E R 2014

•

jop.ascopubs.org

e335

Downloaded from jop.ascopubs.org on September 22, 2015. For personal use only. No other uses without permission. Copyright © 2014 American Society of Clinical Oncology. All rights reserved.

University of Toronto; and Princess Margaret Cancer Centre, University Health Network, Toronto, Ontario, Canada

Jang et al

Performance Status Measures The KPS scale was developed in 1948 to evaluate the ability of patients with cancer to tolerate chemotherapy.14 It is composed of 11 categories, which range from normal activity (100) to death (0). The ECOG scale, which was developed in 1960, is a simpler measure composed of six categories that range from normal activity, at a score of 0, to death, at a score of 5.13 The PPS is an 11-category scale based on the KPS, which was first published in 1996 to aid decision making and communication in palliative care settings.15 It takes into account ambulation, activity, evidence of disease, self-care, intake, and level of consciousness to assess patient function, and ranges from normal at 100, to death at 0.

Study Procedures This study was part of a larger study that also assessed empirically the interconversion of the three performance status scales.21 Ten physicians, who were aware of the study aims, completed the PPS, ECOG and KPS for each of their respective patients at the end of the initial consultation. The three measures were distributed as a questionnaire package to the physicians, who were instructed to circle the appropriate assessment score for each scale. The physicians worked exclusively in an oncology setting and were previously familiar with the measures. They did not receive any formal training on completion of the measures; however, written information about the scales was provided at the beginning of the study. The order of the three measures in the package was randomized weekly. All physicians involved had completed a fellowship in palliative medicine and/or had greater than 5 years of experience in a palliative oncology setting. Death data were updated to April 2013, and were obtained from multiple sources, to ensure as complete a data set as possible. The main source of information was the Princess Margaret Cancer Registry, which receives updates from the Ontario Cancer Registry; supplementary sources included individual hospital patient records and obituary information. e336

JOURNAL

OF

ONCOLOGY PRACTICE

•

Statistical Analyses Demographic and clinical variables were summarized using descriptive statistics. Survival time was either measured from the first visit date to the date of death, or censored at the last outpatient clinic follow-up date. KPS and PPS were grouped into four categories, based on the similarity of the score descriptions and on previous empirical research21: 80 to 100 (full ambulation), 60 to 70 (reduced ambulation), 40 to 50 (mainly sitting/lying), and 10 to 30 (totally bedbound). Separate survival analyses were performed for the ECOG, KPS, and PPS scales, using the Kaplan-Meier method. The log-rank test for trend was used to test a difference in survival curves for each performance measure. The concordance index (C-statistic)22 was used to measure the predictive ability of the initial scores for the three performance status scales. The C-statistic measures the probability that for a randomly chosen pair of patients, the predicted and observed outcomes are concordant. The C-statistic provides a global measure of the sensitivity and specificity of a model and is equal to the area under the curve of a receiver operating characteristic curve. A value of 0.5 indicates no predictive discrimination and a value of 1.0 indicates perfect separation of patients with different outcomes. In computing the C-statistic, a Cox proportional hazards model was built using the performance status score as the predictor. As we did not have separate training and validation sets, the C-statistic might be overoptimistic or overestimated. Therefore, a bootstrap-validated estimation procedure was used to obtain the optimism-corrected C-statistic.23 When calculating the predictive ability, bootstrapping with 2,000 iterations was performed to validate the discriminatory ability of the performance status scores. Analyses were performed with the R statistical software package (version 2.14),24 and verified using SAS statistical software (version 9.3).25

Results From April 2007 to February 2010, 1,711 patients were seen in the OPCC. Performance status was not recorded in 56 patients; hence, 1,655 patients were included in the analysis. Patient characteristics and information on performance status are shown in Table 1. The median age was 65 years old, and 51% were female. Most of the patients (1,506 of 1,655; 91%) had died at the time of analysis. Most of the patients had solid tumor malignancies, with a wide variety of tumor sites being represented. The range of performance status was broad, except for a small number of patients with either perfect performance status (ECOG 0 or KPS/PPS 90 to 100) or very poor performance status (ECOG 4 or KPS/PPS 20 to 30). The median survival for all patients was 133 days (95% CI, 123 to 144 days). The median survival and the 95% CI according to performance status are shown in Appendix Table A1 (online only) and Figure 1. For all three performance measures, the median survival was lower for each worsening performance level. For the ECOG measure, the median survival was 293 days (95% CI, 242 to 403 days) for patients with ECOG 0; 197

V O L . 10, I S S U E 5

Copyright © 2014 by American Society of Clinical Oncology


Network (UHN) in Toronto, Canada. The Oncology Palliative Care Clinic (OPCC) is part of a larger palliative care program at UHN.20 Patients are referred to the OPCC by their medical, radiation, or surgical oncologist for assessment and management of pain and other symptoms and for palliative care planning. In the OPCC, patients are first assessed by a registered nurse, who takes an initial history, and then by a palliative care physician, who completes a full assessment lasting 1 to 2 hours. Eligible patients were all those attending the OPCC for an initial consultation between April 1, 2007, and February 28, 2010. The OPCC accepts all patients with advanced cancer, regardless of treatment status or performance status; the majority have metastatic disease, and approximately one third are still receiving active treatment for their cancer. Patients attending the clinic are 18 years or older and may have any type of malignancy.

Simple Prognostic Model Based on Performance Status

Table 1. Patient Demographics and Characteristics Characteristics (N ⴝ 1,655)

No.

%

ECOG PS 0

Age, years Median

65

Range

21-97

Sex (female)

1 2

846

51

149

9

1,506

91

GI

439

27

Lung

301

18

Genitourinary

201

12

Breast

160

10

Gynecological

156

9

Head and neck

89

5

CNS

88

5

3

Vital status Alive Deceased

Hematological

PPS

84

5

137

8

0

68

4

1

640

39

2

568

34

3

339

21

4

38

2

80-100

508

31

60-70

784

48

40-50

315

19

10-30

24

1

Other

80-100 60-70 40-50 10-30

KPS 80-100

ECOG (n ⫽ 1,653) 60-70 40-50 10-30

PPS (n ⫽ 1,631)

KPS (n ⫽ 1,633) 80-100

514

31

60-70

783

48

40-50

317

19

10-30

19

1

Abbreviations: ECOG, Eastern Cooperative Oncology Group; KPS, Karnofsky Performance Status; PPS, Palliative Performance Scale.

days (95% CI, 183 to 219 days) for ECOG 1; 104 days (95% CI, 90 to 118 days) for ECOG 2; 55 days (95% CI, 46 to 66 days) for ECOG 3; and 25.5 days (95% CI, 17 to 51 days) for ECOG 4. The survival curves for ECOG and PPS are shown in Figure 2. The survival curve for KPS is not shown but is similar to the PPS curve presented in Figure 2. Performance scores delineated survival well using any of the performance measures. This was highly significant by the log-rank test for trend (P ⬍ .001 for all three performance measures). Discriminatory ability using the C-statistic was similar for all three performance status scores (0.64 for ECOG, 0.63 for PPS, and 0.63 for KPS).

Discussion Prognostication is important for patients, their families, and health care professionals in preparing for the future and in Copyright © 2014 by American Society of Clinical Oncology

0

1

2

3

4

5

6

7

8

9

10 11 12

Median Survival Time (months) Figure 1. Graphical representation of median survival and 95% CI by performance status. ECOG PS, Eastern Cooperative Oncology Group performance status; KPS, Karnofsky Performance Status; PPS, Palliative Performance Scale.

determining eligibility for health care resources. However, clinical prognostication is difficult and fraught with error, often resulting in overestimation of survival.10 Previous research on prognostic models in advanced cancer has emphasized accuracy over parsimony, and most models contain a large number of variables.6,8,9,12 In our study, we found that performance status alone, as measured by ECOG, KPS, or PPS, was effective in delineating the survival of outpatients with advanced cancer, as seen by the Kaplan-Meier survival curves and the highly significant log-rank test for trend. Various models have been developed to assess survival in patients with advanced cancer. The Palliative Prognostic Index (PPI) utilizes the PPS, oral intake, edema, dyspnea at rest, and delirium to predict 3-week and 6-week survival.12 This scale requires conversion of these variables to a partial score. The Prognosis in Palliative Care Study (PiPS-A and PiPS-B) models use 10 to 12 variables to predict survival, with PiPS-B models requiring blood work results to aid in prognostication.9 The Palliative Prognostic (PaP) Score uses clinical prediction of survival, KPS, anorexia, dyspnea, total WBC count, and lympho-


•

jop.ascopubs.org

e337


Tumor site

4

Jang et al

B

1.0 o oo

Overall Survival (probability)

o

0.8

ECOG = 0 ECOG = 1 ECOG = 2 ECOG = 3 ECOG = 4

o o

0.6 o

Overall Survival (probability)

A

o

o

0.4

o o

o

o

o

oo o o

0.2

o

5

10

o

15

20

o

25

o o oo o

30

PPS = 80-100 PPS = 60-70 PPS = 40-50 PPS = 10-30

o o o

0.6 o

0.4

o o o

o

o oo oo oo oo oo o o o o

o

0.2

o

oo oo o oooo o o

0.8

o

oo

35

40

0

Follow-Up Time (months)

5

10

15

o o ooo

20

o o oo oo

25

30

oo

35

40

Follow-Up Time (months)

Figure 2. Survival curves for all patients by (A) Eastern Cooperative Oncology Group performance status (ECOG) and (B) Palliative Performance Scale score (PPS).

cyte percentage to predict probability of 30-day survival in patients with advanced solid tumors.6 Categories within each parameter are assigned a partial score, from which a total score is derived. The Delirium-PaP (D-PaP) incorporates delirium into the PaP model, to predict 30-day survival.26 The number of variables involved, the necessity of blood work, and the complexity of partial score assignment make these models difficult to use in a fast-paced outpatient setting. Chow et al8 developed a three-part model that uses performance status (KPS ⱕ 60), non– breast cancer diagnosis, and nonbony metastases to stratify survival in patients with advanced cancer in an outpatient radiation oncology clinic. This model is less complex, but it has not been tested in other patient settings. Although the models in our study contain only performance status, their predictive abilities were comparable to those of more complex models. The C-statistic, which measures predictive ability, was 0.64 for the models using the ECOG, and 0.63 for those using PPS or KPS, indicating modest predictive performance. The C-statistic for the model developed by Chow et al8 was similar, at 0.66 for the training data, and 0.65 and 0.63 for the temporal and external validation data sets, respectively. The more complicated PiPS models had a C-statistic between 0.67 and 0.69, using bootstrapping resampling instead of an external validation set.9 A recent prospective study of patients with advanced cancer being admitted to hospice11 examined the relative predictive performance of the PaP score,6 D-PaP,26 PPI,12 and PPS.18 The C-statistic values were highest for PaP and D-PaP at 0.72 and 0.73, respectively, and were slightly lower for the PPS and PPI at 0.63 and 0.62, respectively; of note, the C-statistic for PPS was the same as in our study. Although performance status alone may not lead to the best discriminative ability, the results are comparable to more complex models, while avoiding the necessity for complex calculations and blood work. Our study also provides survival estimates for each performance level. For patients with an ECOG of 4, the estimated median survival was approximately 25 days. For each improved performance level, the estimated survival was approximately twice of the performance level below it (eg, ECOG 3 had a e338

JOURNAL

OF

ONCOLOGY PRACTICE

•

median survival of approximately 50 days). A similar pattern was also found for KPS and PPS, with cut-points of 80 to 100 (approximately 200 days), 60 to 70 (approximately 100 days), 40 to 50 (approximately 50 days), and 10 to 30 (approximately 25 days). The survival estimates for our patients are longer compared with those of other studies, likely because our study participants were outpatients rather than inpatients in palliative care units or hospices.16,28,29 For example, in our study the median survival for a PPS level of 60 to 70 was 115 days, whereas in a meta-analysis, the median survival times for patients with PPS levels of 70 and 60 were 78 days and 48 days, respectively.29 In this meta-analysis, five of the six contributing studies included patients in a hospice or a palliative care unit, and patients without cancer were also included. Our study adds to the current literature on prognosis for patients with advanced cancer in two ways. We show that a simple prognostic model, based on performance status alone, delineates survival well and has predictive ability comparable with more complex models. Further, the model can be a useful prognostic tool in clinics for outpatients with advanced cancer. A clinician need only remember that an ECOG of 4 corresponds to a median survival of approximately 25 days in order to easily calculate the median survivals for the other ECOG levels, which increase by 50% for each level. As for any model, this estimate requires adjustment by the clinician, based on clinical impression and experience.30 However, unlike for other models, there is no need to collect and enter other clinical or laboratory information into an algorithm. Rather, the survival estimate can be increased or decreased according to tumor type8 and clinical judgement. There was a potential risk for measurement bias in this study, as the assessors were not formally trained in using the measures. However, the scales are easy to use, and most practicing oncologists and palliative care physicians do not receive training in using these scales. Furthermore, physicians did not have to remember the categories, as they were given printed versions of the scales on which to circle their ratings. Only physicians completed the scales for this study; however, previous research has shown that inter-rater agreement between phy-

V O L . 10, I S S U E 5



0

o

oo oo o

1.0 ooo o


Acknowledgment Supported by the Canadian Cancer Society (Award No. 020509; C.Z.) and by the Ontario Ministry of Health and Long Term Care. R.W.J. was supported by a 2011 Young Investigator Award from the Conquer Cancer Foundation, and a 2013 Young Investigator Award by the Multinational Association of Supportive Care in Cancer. C.Z. was supported by the Rose Chair in Supportive Care, Faculty of Medicine, University of Toronto. Presented in part at the 2013 Multinational Association of Supportive Care in Cancer/International Society of Oral Oncology International Cancer Care Symposium, Berlin, Germany, June 27-29, 2013. Authors’ Disclosures of Potential Conflicts of Interest The authors indicated no potential conflicts of interest. Author Contributions Conception and design: Raymond W. Jang, Camilla Zimmermann Financial support: Camilla Zimmermann Administrative support: Nadia Swami Collection and assembly of data: Raymond W. Jang, Nadia Swami, Subrata Banerjee, Ernie Mak, Ebru Kaya, John Bryson, Julia Z. Ridley, Camilla Zimmermann Data analysis and interpretation: Raymond W. Jang, Valerie B. Caraiscos, Gary Rodin, Lisa W. Le, Camilla Zimmermann Manuscript writing: All authors Final approval of manuscript: All authors Corresponding author: Camilla Zimmermann, MD, PhD, Department of Psychosocial Oncology and Palliative Care, Princess Margaret Cancer Centre, 610 University Avenue, 16 to 712, Toronto, ON, Canada, M5G 2M9; e-mail: [email protected].

DOI: 10.1200/JOP.2014.001457; published online ahead of print at jop.ascopubs.org on August 12, 2014.

References 1. Earle CC, Landrum MB, Souza JM, et al: Aggressiveness of cancer care near the end of life: Is it a quality-of-care issue? J Clin Oncol 26:3860-3866, 2008 2. Finlay E, Casarett D: Making difficult discussions easier: Using prognosis to facilitate transitions to hospice. CA Cancer J Clin 59:250-263 3. Glare P, Sinclair C, Downing M, et al: Predicting survival in patients with advanced disease. Eur J Cancer 44:1146-1156, 2008 4. Heyland DK, Allan DE, Rocker G, et al: Discussing prognosis with patients and their families near the end of life: Impact on satisfaction with end-of-life care. Open Med 3:e101-e110, 2009 5. Maltoni M, Caraceni A, Brunelli C, et al: Prognostic factors in advanced cancer patients: Evidence-based clinical recommendations–A study by the Steering Committee of the European Association for Palliative Care. J Clin Oncol 23:62406248, 2005

12. Morita T, Tsunoda J, Inoue S, et al: The Palliative Prognostic Index: A scoring system for survival prediction of terminally ill cancer patients. Support Care Cancer 7:128-133, 1999 13. Oken MM, Creech RH, Tormey DC, et al: Toxicity and response criteria of the Eastern Cooperative Oncology Group. Am J Clin Oncol 5:649-655, 1982 14. Karnofsky DA, Burchenal JH: The clinical evaluation of chemotherapeutic agents in cancer, in MacLeod C (ed): Evaluation of Chemotherapeutic Agents. New York, NY, Columbia University Press, 1949, pp 191-205 15. Anderson F, Downing GM, Hill J, et al: Palliative performance scale (PPS): A new tool. J Palliat Care 12:5-11, 1996 16. Chuang R-B, Hu W-Y, Chiu T-Y, et al: Prediction of survival in terminal cancer patients in Taiwan: Constructing a prognostic scale. J Pain Symptom Manage 28:115-122, 2004

6. Pirovano M, Maltoni M, Nanni O, et al: A new palliative prognostic score: A first step for the staging of terminally ill cancer patients. Italian Multicenter and Study Group on Palliative Care. J Pain Symptom Manage 17:231-239, 1999

17. Maltoni M, Pirovano M, Scarpi E, et al: Prediction of survival of patients terminally ill with cancer: Results of an Italian prospective multicentric study. Cancer 75:2613-2622, 1995

7. Chow E, Fung K, Panzarella T, et al: A predictive model for survival in metastatic cancer patients attending an outpatient palliative radiotherapy clinic. Int J Radiat Oncol Biol Phys 53:1291-1302, 2002

18. Lau F, Downing GM, Lesperance M, et al: Use of Palliative Performance Scale in end-of-life prognostication. J Palliat Med 9:1066-1075, 2006

8. Chow E, Abdolell M, Panzarella T, et al: Predictive model for survival in patients with advanced cancer. J Clin Oncol 26:5863-5869, 2008

19. Buccheri G, Ferrigno D, Tamburini M: Karnofsky and ECOG performance status scoring in lung cancer: A prospective, longitudinal study of 536 patients from a single institution. Eur J Cancer 32A:1135-1141, 1996

9. Gwilliam B, Keeley V, Todd C, et al: Development of prognosis in palliative care study (PiPS) predictor models to improve prognostication in advanced cancer: Prospective cohort study. BMJ 343:d4920, 2011

20. Zimmermann C, Seccareccia D, Clarke A, et al: Bringing palliative care to a Canadian cancer center: The palliative care program at Princess Margaret Hospital. Support Care Cancer 14:982-987, 2006

10. Christakis NA, Lamont EB: Extent and determinants of error in doctors’ prognoses in terminally ill patients: Prospective cohort study. BMJ 320:469-472, 2000

21. Ma C, Bandukwala S, Burman D, et al: Interconversion of three measures of performance status: An empirical analysis. Eur J Cancer 46:3175-3183, 2010

11. Maltoni M, Scarpi E, Pittureri C, et al: Prospective comparison of prognostic scores in palliative care cancer populations. Oncologist 17:446-454, 2012


22. Harrell FE: Regression Modeling Strategies: With Applications to Linear Models, Logistic Regression, and Survival Analysis. New York, NY, Springer-Verlag, 2001


•

jop.ascopubs.org

e339


sicians and nurses is good.31 Another limitation of our study is that we did not have separate training and validation sets to test the predictive ability of the performance scales. However, we used the bootstrapping technique to increase the robustness of our predictions. It is noteworthy that all patients were enrolled at their first appointment in a palliative care clinic, and that even those with the best performance status survived only approximately 1 year. The survival for the ECOG 0 category is variable, as patients with advanced cancer in this category have optimal performance status and some may live many years. This variability in survival, together with the small sample size for this performance status group, accounts for the wide CI for this subgroup. As well, prognosis may differ with type of tumor; due to the large numbers required, it was not feasible to obtain separate survival estimates for the different tumor types in our study. We also did not have many patients with very poor performance status, as our study only included outpatients. As such, our findings best apply to ambulatory patients with advanced cancer whose clinical prognosis is a year or less; however, this is also the population for which prognostication is the most uncertain and of greatest importance, particularly in terms of end-of-life planning. In summary, our study demonstrates that performance status alone is a good prognostic factor in outpatients with advanced cancer and that ECOG, PPS, and KPS all have similar predictive ability. Importantly, it also provides a simple tool that can be used in clinics to assist clinicians in prognostication for outpatients with advanced cancer, enabling more proactive advance care planning. Future research is indicated to validate this model in other oncology settings.

Jang et al

23. Gonen M: Analyzing Receiver Operating Characteristic Curves with SAS. Cary NC, SAS Institute, 2007 24. The R Project for Statistical Computing 2013. www.r-project.org/ 25. Reference deleted 26. Scarpi E, Maltoni M, Miceli R, et al: Survival prediction for terminally ill cancer patients: Revision of the palliative prognostic score with incorporation of delirium. Oncologist 16:1793-1799, 2011 27. Reference deleted

28. Virik K, Glare P: Validation of the palliative performance scale for inpatients admitted to a palliative care unit in Sydney, Australia. J Pain Symptom Manage 23:455-457, 2002 29. Downing M, Lau F, Lesperance M, et al: Meta-analysis of survival prediction with Palliative Performance Scale. J Palliat Care 23:245-252; discussion 252-254, 2007 30. Stone PC, Lund S: Predicting prognosis in patients with advanced cancer. Ann Oncol 18:971-976, 2007 31. Zimmermann C, Burman D, Bandukwala S, et al: Nurse and physician inter-rater agreement of three performance status measures in palliative care outpatients. Support Care Cancer 18:609-616, 2010


e340

JOURNAL

OF

ONCOLOGY PRACTICE

•

V O L . 10, I S S U E 5



Appendix Table A1. Survival Time According to Performance Status (all patients) Performance Status

Median Survival (days)

95% CI (days)

ECOG 293

242 to 403

197

183 to 219

2

104

90 to 118

3

55

46 to 66

4

25.5

17 to 51


0 1

PPS 80-100

221

197 to 244

60-70

115

105 to 131

40-50

51

44 to 60

10-30

22

12 to 102

KPS 80-100

215

190 to 241

60-70

119

106 to 132

40-50

49

43 to 59

10-30

29

22 to 144

Abbreviations: ECOG, Eastern Cooperative Oncology Group; KPS, Karnofsky Performance Status; PPS, Palliative Performance Scale.



•

jop.ascopubs.org

e341