Content uploaded by G Michael Downing
Author content
All content in this area was uploaded by G Michael Downing
Content may be subject to copyright.
1066
JOURNAL OF PALLIATIVE MEDICINE
Volume 9, Number 5, 2006
© Mary Ann Liebert, Inc.
Use of Palliative Performance Scale
in End-of-Life Prognostication
FRANCIS LAU, Ph.D.,
1
G. MICHAEL DOWNING, M.D.,
2
MARY LESPERANCE, Ph.D.,
3
JACK SHAW, B.Sc.,
1
and CRAIG KUZIEMSKY, B.Com.
1
ABSTRACT
Background: Current literature suggests clinicians are not accurate in prognostication when
estimating survival times of palliative care patients. There are reported studies in which the
Palliative Performance Scale (PPS) is used as a prognostic tool to predict survival of these pa-
tients. Yet, their findings are different in terms of the presence of distinct PPS survival pro-
files and significant covariates.
Objective: This study investigates the use of PPS as a prognostication tool for estimating
survival times of patients with life-limiting illness in a palliative care unit. These findings
are compared to those from earlier studies in terms of PPS survival profiles and covariates.
Methods: This is a retrospective cohort study in which the admission PPS scores of 733 pal-
liative care patients admitted between March 3, 2000 and August 9, 2002 were examined for
survival patterns. Other predictors for survival included were age, gender, and diagnosis.
Results: Study findings revealed that admission PPS score was a strong predictor of sur-
vival in patients already identified as palliative, along with gender and age, but diagnosis
was not significantly related to survival. We also found that scores of PPS 10% through PPS
50% led to distinct survival curves, and male patients had consistently lower survival rates
than females regardless of PPS score.
Conclusion: Our findings differ somewhat from earlier studies that suggested the presence
of three distinct PPS survival profiles or bands, with diagnosis and noncancer as significant
covariates. Such differences are likely attributed to the size and characteristics of the patient
populations involved and further analysis with larger patient samples may help clarify PPS
use in prognosis.
INTRODUCTION
F
OR PATIENTS WITH LIFE
-
LIMITING ILLNESS
, know-
ing how much time remains is often impor-
tant for decision making regarding goals of care,
treatment options and dealing with closure on
personal and family matters. A study by Kirk et
al.
1
on “what cancer patients and their families
receiving palliative care want to be told” has
found prognosis and hope as most important in
patient and family communication. Having accu-
rate and timely prognostic information becomes
the basis for one’s transition from curative treat-
ments to considering appropriate supportive care
and likely decline to death. This remains a clini-
cal challenge with wide variability seen in patient
1
Health Information Science,
3
Department of Mathematics and Statistics, University of Victoria, Victoria, British
Columbia, Canada.
2
Victoria Hospice Society, Victoria, British Columbia, Canada.
diagnoses, stages, degrees of symptom distress
and pursuit of treatments. Furthermore, current
literature suggests clinicians are not accurate in
prognostication when estimating survival times
of terminally ill patients.
2
To illustrate, Christakis
and Lamont
3
conducted a prospective cohort
study in five Chicago outpatient hospice pro-
grams to compare survival estimates of 343 clin-
icians against actual survival of 468 terminally ill
patients. They found that overall, clinicians over-
estimated survival by a factor of 5.3; the inaccu-
racy was not restricted to particular types of clin-
icians or patients.
The 2001 systematic review by Chow et al.
4
showed that the use of prognostic tools can im-
prove the accuracy of clinician survival predic-
tion estimates. The review identified several
prognostic indicators, with performance status
most strongly correlated to survival duration.
This was followed by the “terminal syndrome”
with such factors as anorexia, weight loss, and
dysphagia. One reported prognostic tool that
measures the functional status of palliative care
patients is the Palliative Performance Scale
i
(PPS).
The PPS, adapted from the Karnofsky Perfor-
mance Scale’s functional dimensions of ambula-
tion, activity level and evidence of disease,
5
and
adds self-care, oral intake, and level of con-
sciousness. PPS is divided into 11 categories from
0% to 100% in 10% increments. A patient with
PPS 0% is dead; with PPS 100% is mobile and
healthy. Since it was first published by Anderson
et al.
6
in 1996 as a tool for measuring changing
functional status in palliative care patients, PPS
has been adopted and used in a variety of health-
care settings in different countries.
Despite its extensive clinical use, a 2005 MED-
LINE search found only four journal articles on
the use of PPS for survival prediction. One study
7
involved patients with cancer in a palliative care
unit (PCU), while the other three included het-
erogeneous populations from inpatient hospices
8
and community-based hospice programs.
9,10
Even though all four studies confirmed that PPS
is a strong predictor of survival duration in those
already identified as hospice or palliative care pa-
tients, there are important differences in the find-
ings that require further investigation. For in-
stance, in their 1999 study of terminally ill
patients with cancer, Morita et al.
7
reported the
presence of three distinct PPS profiles or bands
at PPS 10%–20%, PPS 30%–50%, and PPS 60% or
more, respectively. In contrast, the 2001 study by
Virik and Glare
8
revealed no such distinct group-
ings. Whereas in the two 2005 studies, Harrold et
al.
9
computed mortality rates using three equal
divisions of PPS scores at PPS 10%–20%, PPS
30%–40%, and PPS 50% or more, and Head et al.
10
grouped PPS 10%–20% and PPS 60%–70% be-
cause of insufficient sample size.
In this paper, we describe the findings of a ret-
rospective cohort study examining PPS as a prog-
nostication tool for palliative care patients ad-
mitted to the Victoria Hospice Society (VHS)
Palliative Care Unit in British Columbia, Canada.
Our results are compared to those published in
the literature in terms of PPS survival profiles and
other variables associated with PPS in predicting
survival times. We conclude with a discussion of
the implications of these findings and suggest
next steps to further validate PPS as a prognosti-
cation tool for terminally ill palliative care pa-
tients.
METHODS
Design and sample
This retrospective cohort study included all pa-
tients admitted to the Victoria Hospice Society
(VHS) Palliative Care Unit (PCU) between March
3, 2000 and August 9, 2002. Admission to the 17-
bed PCU included both tertiary acute assessment
and residential extended care levels. PPS scores
were recorded by unit nurses within the first 24
hours of admission. If any data were incomplete,
including gender, age, date of birth, date of death,
admission date, discharge date, primary diagno-
sis category, or specific disease, the patient was
excluded from analysis. For patients with multi-
ple admissions, only the earliest unit admission
data were included in the analysis. All data were
routinely collected by hospice staff not directly
involved with this study. Data collection in-
volved a computer-based (MS Access
®
, Mi-
crosoft, Inc., Redmond, WA) hospice patient reg-
istration service intake and research outcome
database. The cohort was obtained from a 28-
month study period and is considered represen-
tative of the types of patients admitted to VHS.
The outcome for this study was survival time.
Survival time was defined as the length of time
in days from the patient’s earliest PCU admission
date until death. Patients for whom the date of
death was unknown, survival time was measured
PPS IN EOL PROGNOSTICATION 1067
as the earliest PCU admission date until the last
PCU discharge. These patients had survival times
that are called censored.
Ethical approval for use of the data was re-
ceived from the Victoria Hospice Society, Ethics
Review Committee of Vancouver Island Health
Authority and the University of Victoria Ethics
Committee (UVic/VIHA Joint Ethics Applica-
tion, Protocol Number 2005-96).
Data analysis
Descriptive statistics including mean, median,
and frequency distributions were used to de-
scribe the characteristics of the cohort in relation
to their admission PPS scores. Kaplan-Meier sur-
vival curves were computed and graphed. Log-
rank tests were used to test for significant differ-
ences between survival curves, and additionally,
to test for differences between survival curves for
groups of PPS categories: (PPS 10% and PPS 20%),
(PPS 30%, PPS 40%, and PPS 50%), and (PPS 50%
and PPS 60%). Within the cohort, subgroup sur-
vival analysis and log-rank tests were computed
for patients with cancer as the primary diagno-
sis. The Cox proportional hazards model was
used to identify the relationship between the
hazard ratio for death with admission PPS, age
group, gender, primary diagnosis category and
specific cancer disease group as covariates. Mor-
tality rates were computed using the Kaplan-
Meier curves for specific time periods and strat-
ified by PPS categories. All calculations were
performed using SPSS version 12 (SPSS Inc.,
Chicago, IL).
RESULTS
Patient characteristics
The characteristics of the PCU patients admit-
ted during the study period are shown in Table
1. There were 972 admissions for 831 patients. Of
these 831 patients, 758 (91%) had PPS scores
recorded within 24 hours of admission. Sixteen
patients who had survival times greater than 365
days were considered atypical and excluded from
LAU ET AL.
1068
T
ABLE
1. P
ATIENT
C
HARACTERISTICS
Variable Result
# of patients considered for final analysis 733
Male 335 (45.7%)
Female 398 (54.3%)
# of patients per age group (based on earliest PCU admission date)
19–44 years 37 (5%)
45–64 194 (26.5%)
65–74 175 (23.9%)
75–84 233 (31.8%)
85 94 (12.8%)
Mean age 70.25 (s.e. 0.502)
Median age 73
# of patients with cancer/noncancer primary diagnosis category
Cancer 647 (88.3%)
Noncancer 86 (11.7%)
# of patients with most common cancer types
a
(at time of admission)
Lung 152 (20.7%)
Colorectal 75 (10.2%)
Breast–Female 74 (10.1%)
Prostate 38 (5.2%)
Other Cancer 308 (42.0%)
# of patients with noncancer diagnosis
Heart disease 8
Lung disease 19
Renal failure 6
Neuralogic disease 6
Liver disease 4
Other 43
a
Based on data from the Canadian Cancer Society Canadian Cancer Statistics 2005.
PCU, palliative care unit.
the study. A further 9 patients were excluded be-
cause of missing data. This resulted in a final co-
hort of 733 patients with a mean age of 70.25 years
of which 335 were male (45.7%) and 398 (54.3%)
female. Primary diagnoses included 647 (88.3%)
patients with cancer and 86 (11.7%) patients with-
out cancer. In this study, patients with cancer
were grouped according to the four most com-
mon types of cancer reported in Canada.
Overall survival patterns
The mean, median, and range of the survival
times of the patients by predictor variables PPS,
age, gender, and diagnosis are shown in Table 2.
Seven patients were censored with unknown
death dates. The overall mean survival time ex-
cluding censored patients was 27 days (95% con-
fidence interval [CI] 24, 30), median of 10 days
(95% CI 9, 11) and a range of 1–348 days. None
of the censored patients had admission PPS
higher than 60%.
Survival curves by PPS
Survival curves by admission PPS. The Kaplan-
Meier survival curves stratified by admission PPS
score are shown in Figure 1. The log-rank test for
the equality of survival curves is highly signifi-
cant at p0.0001, suggesting there are significant
differences among the curves over the PPS cate-
gories. The subset of 561 patients with admission
scores of PPS 30%, PPS 40%, and PPS 50% showed
significantly different survival curves (log-rank
test p0.0001), as did the subset of 155 patients
with scores of PPS 10% and PPS 20% (log-rank
test p0.0004). The tail for PPS 20% was mainly
responsible for this difference and is discussed
later. There was no significant difference between
PPS 50% and PPS 60% survival curves, however,
there were only 17 patients with PPS 60% in this
study.
Survival curves by admission PPS in patients with
cancer. Survival curves and log-rank tests were
performed for patients having cancer as the pri-
mary diagnosis. The results for this subgroup
were consistent with those of the entire cohort;
the log-rank test for the equality of survival
curves was highly significant at p0.0001; the
subset of 505 patients with admission scores of
PPS 30%, PPS 40% and PPS 50% showed signifi-
cantly different survival curves (log-rank test p
0.0001) as did the subset of 127 patients with
scores of PPS 10% and PPS 20% (log-rank test p
PPS IN EOL PROGNOSTICATION 1069
T
ABLE
2. S
URVIVAL
T
IMES BY
A
GE
, G
ENDER
, D
IAGNOSIS
,
AND
PPS
No. of patients Log rank
Variable Mean (95% CI) Median (95% CI) Range (%) pvalue
Overall 27 (24, 30) 10 (9, 11) 1,–348 733 (100%)
Age 0.0166
19–44 30 (9, 51) 8 (3, 13) 1–278 37 (5%)
45–64 34 (27, 42) 16 (11, 21) 1–348 194 (26.5%)
65–74 24 (19, 30) 9 (6, 12) 1–216 175 (23.9%)
75–84 25 (19, 31) 10 (8, 12) 1–344 233 (31.8%)
8520 (12, 28) 7 (5, 9) 1–248 94 (12.8%)
Gender 0.0009
Female 31 (26, 35) 12 (9, 15) 1–348 335 (45.7%)
Male 23 (18, 28) 8 (6, 10) 1–347 398 (54.3%)
Diagnosis 0.7036
Cancer 26 (23, 30) 11 (9, 13) 1–348 647 (88.3%)
Noncancer 32 (18, 46) 5 (2, 8) 1–347 86 (11.7%)
Admission PPS 0.001
PPS 10% 2 (1, 2) 1 (1, 1) 1–12 66 (9.0%)
PPS 20% 6 (3, 8) 2 (2, 2) 1–81 89 (12.2%)
PPS 30%
a
18 (14, 22) 9 (7, 11) 1–295 218 (29.7%)
PPS 40%
a
36 (29, 43) 17 (13, 21) 1–347 225 (30.7%)
PPS 50%
a
51 (40, 62) 27 (18, 36) 1–287 118 (16.1%)
PPS 60% 64 (26, 101) 40 (20, 60) 6–348 17 (2.3%)
a
PPS of 30%, 40%, and 50% had censored patients.
PPS, Palliative Performance Scale; CI, confidence interval.
Survival times (in days)
0.0003). There was no significant difference be-
tween PPS 50% and PPS 60% survival curves,
however, only 15 patients had a PPS score of 60%.
Survival by PPS and covariates
Significant hazard ratios. The Cox proportional
hazards model was used to examine the rela-
tionship between the hazard ratio of death with
the predictor variables included in the study: PPS
score, age group, gender, and primary diagnosis.
The results shown in Table 3 indicate that age on
PCU admission, initial PPS score and gender are
significantly related to the hazard for death. Pa-
tients in age categories 45–64 (95% CI 0.518–0.865;
p0.002) had significantly lower hazards than
patients aged 85 and above. Female patients had
lower hazards than male patients (p0.026) al-
though the 95% confidence interval for the rela-
tive hazard very nearly includes the value one
(0.709–0.978). Patients with initial scores of PPS
10%, PPS 20%, or PPS 30% had significantly larger
hazards than those with initial scores of PPS 60%
(p0.001) with 95% confidence intervals for the
relative hazards much larger than one.
Mortality rates over time. Mortality rates over
time, stratified by PPS category, are shown in
Table 4. The mortality rates show the cumulative
percentage of patients who died at specific time
periods up to 1 year. Table 4 shows that the mor-
tality rates of patients with admission scores of
PPS 10% and PPS 20% differ substantially from
those with PPS 30% or more. For example, all pa-
tients with PPS 10% and 94% of patients with PPS
20% died within 2 weeks of admission.
DISCUSSION
Our findings indicate that initial PPS score
upon PCU admission is a strong predictor of sur-
vival for palliative care patients. Gender and age
(but not primary diagnosis) were found to be sig-
nificant predictor variables as well; no clinically
significant interaction effects among these vari-
ables were noted from Cox regression analysis.
These findings are different from those of Har-
rold et al.
8
that report an interaction between PPS
and cancer/noncancer diagnosis in a Cox model
for survival time. Their cohort consisted of 214
patients with and 252 patients without cancer.
However, because our data included only 11.7%
patients without versus 54% patients with cancer
by Harold et al., this likely limits the ability to de-
tect any difference as a result of our smaller sam-
ple proportion.
Comparing differences between PPS 10% and
PPS 20%, Figure 1 shows a tailing effect in the
PPS 20%, which accounts for the main statistical
LAU ET AL.
1070
FIG. 1. Kaplan-Meier survival codes by admission Palliative Performance Scale (PPS).
difference from PPS 10% with clinically signifi-
cant differences in mortality rates at 7 days and
30 days. That is, some patients at a very low func-
tional level and apparently imminently dying do
survive longer than others. Which specific dis-
eases, cancer types or complications contributed
to these differences warrant further investigation
and clinical understanding.
Our cohort had median survival times of 1 day
at PPS 10% (CI 1,1), 2 days at PPS 20% (CI 2,2),
increasing to 9 days at PPS 30% (CI 7,11), then to
17, 27, and 40 days at scores of PPS 40%, PPS 50%,
and PPS 60%, respectively. Based on the mortal-
ity rates computed from this cohort, it can be ex-
pected that approximately 50% of patients with
PPS 10% would not survive past 1 day upon ad-
mission, approximately 90% would not survive
beyond the fourth day, and 99% would not sur-
vive by the seventh day; patients with PPS 30%
would have greater than 50% chance of survival
for 7 days or longer. Although the data demon-
strate that male patients have shorter survival
times than females given the same initial PPS
score (mean 23 versus 31 days), the clinical rele-
vance of this difference is unclear. A larger sam-
ple size is needed to clinically differentiate gen-
der-specific survival times.
While these findings support earlier studies
that PPS is indeed a strong predictor of survival
in terminally ill palliative care patients, there are
PPS IN EOL PROGNOSTICATION 1071
T
ABLE
3. H
AZARD
R
ATIOS FOR
A
GE
, D
IAGNOSIS
, PPS,
AND
G
ENDER
Hazard
Variable ratio Lower Upper pvalue
Age at admit category 0.021
(vs 85)
19–44 0.986 0.668 1.456 0.944
45–64 0.669 0.518 0.865 0.002
65–74 0.792 0.612 1.024 0.075
75–84 0.772 0.604 0.988 0.040
Primary diagnosis category 0.077
(vs noncancer)
Breast–Female 0.831 0.595 1.160 0.277
Colorectal 1.088 0.790 1.499 0.604
Lung 1.263 0.959 1.664 0.097
Other Cancer 1.139 0.888 1.460 0.305
Prostate 0.944 0.630 1.414 0.778
Admission PPS 0.001
(vs PPS 60%)
PPS 10% 18.022 10.138 32.037 0.001
PPS 20% 8.252 4.771 14.271 0.001
PPS 30% 2.717 1.618 4.561 0.001
PPS 40% 1.661 0.994 2.776 0.053
PPS 50% 1.204 0.709 2.045 0.493
Gender (vs male) 0.833 0.709 0.978 0.026
PPS, Palliative Performance Scale; CI, confidence interval.
95% CI for hazard ratio
T
ABLE
4. M
ORTALITY
R
ATES
O
VER
T
IME
Death on or before
(days) 13571430456090180365
% Mortality (all patients) 11 26 35 42 59 77 84 89 94 98 100
PPS
10% (n66) 52859499100——— — — —
20% (n89) 30 69 83 88 94 96 96 99 100 — —
30% (n218) 6 18 29 44 68 86 91 96 97 99 100
40% (n225) 1 13 19 24 45 70 77 83 91 96 100
50% (n118) 1 6 10 14 30 52 64 75 85 95 100
60% (n17) 0006124165657694100
PPS, Palliative Performance Scale.
also differences across these studies that warrant
further investigation. In particular, our study
showed that scores of PPS 10%, PPS 20%, PPS
30%, and PPS 40% display distinct hazard ratios
and survival curves. This is in contrast to pub-
lished reports by Morita,
7
Harrold
9
and Head
10
each reporting various groupings of PPS scores
into bands or profiles. For example, in their anal-
ysis, Harrold et al.
9
used three bands at PPS
10%–20%, PPS 30%–40%, and PPS 50%–70%. Sim-
ilar to ours, however, the 2002 study by Virik and
Glare
8
consisting of 135 patients found distinct
PPS survival curves with no evidence of group-
ing into bands.
There are several possible reasons and limita-
tions that may explain the differences observed
across these studies. First, the patient cohort
characteristics may differ because of disease
stage, symptom crises, or complications and age.
For instance, Virik and Glare
8
compared the
overall median survival times of their patients
with those from Morita et al.
7
(13 versus 27 days)
and concluded that their patients were likely
closer to the terminal stage of their illness. A sec-
ond reason for discrepancies may relate to the
location of care or the type of the PCUs. For ex-
ample, the VHS PCU combines tertiary and res-
idential patients. Tertiary admissions are usu-
ally for symptom crises or complex issues and
may not be easily compared to other palliative
settings. Although VHS also has a large pallia-
tive home clientele with many home deaths,
only PCU deaths were included in this study. It
is unclear whether survival patterns would be
different across various locations. A third re-
lated limitation for PPS in survival estimation is
that this study neither proves nor disproves the
utility of PPS as a general predictor of patient
prognosis since patients identified as palliative
or hospice are a subset of the general popula-
tion. Fourth, although not implied in these stud-
ies, misinterpretation of how to use the PPS tool
can occur because it is a clinician’s judgment for
each level using relatively broad criteria. Be-
cause PPS was originally published, this issue
has occasionally arisen, prompting clarification
and minor refinement by VHS. Although no dif-
ferent in content and structure from the original
PPS, PPSv2 is now the official version used at
the Victoria Hospice. PPSv2 includes slight
wording clarifications and definitions of terms
not in the original version. See Appendix 1 for
an example of the tool.
CONCLUSION
Since its introduction in the 1990s, PPS has be-
come a popular tool for assessing changing func-
tional status of palliative care patients. It has clin-
ical utility as an easy-to-use communication tool.
Despite its widespread use, the validity and pre-
dictive accuracy of PPS as a prognostication tool
is not well understood. The four studies pub-
lished to date have shown PPS to be a good pre-
dictor of survival in palliative patients with can-
cer but varying results in terms of the presence
of distinct survival profiles, the influence of other
predictor variables, and its applicability to het-
erogeneous hospice populations beyond patients
with cancer. A larger analysis using 10 years of
patient data, including information on disease,
age, setting of care, and rate of change in PPS, is
underway and may provide further insight and
clarity.
ACKNOWLEDGMENTS
Funding support for this study has been pro-
vided by the Canadian Institutes for Health Re-
search New Emerging Team grant and the Telus
Strategic Alliance Grant.
REFERENCES
1. Kirk P, Kirk I, Kristjanson LJ: What do patients re-
ceiving palliative care for cancer and their families
want to be told? A Canadian and Australian qualita-
tive study. BMJ 2004;328:1343.
2. Vigano A, Dorgan M, Bruera E, Suarez-Almazor ME:
The relative accuracy of the clinical estimation of the
duration of life for patients with end of life cancer.
Cancer 1999;86:170–176.
3. Christakis NA, Lamont EB: Extent and determinants
of error in doctors’ prognoses in terminally ill pa-
tients: Prospective cohort study. BMJ 2000;320:469–
473.
4. Chow E, Harth T, Hruby G, Finkelstein J, Wu J, Dan-
joux C: How accurate are physicians’ clinical predic-
tions of survival and the available prognostic tools in
estimating survival times in terminally ill cancer pa-
tients? A systematic review. Clin Oncol 2001;15:203–
218.
5. Karnofsky D, Albelman W, Craver L, Burchena lJ: The
use of nitrogen mustards in the palliative treatment
of carcinoma. Cancer 1948;1:634–656.
6. Anderson F, Downing MG, Hill J, Casorso L, Lerch
N: Palliative performance scale (PPS): A new tool. J
Palliat Care 1996;12:5–11.
LAU ET AL.
1072
7. Morita T, Tsunoda J, Inoue S, Chihara S: Validity of
the palliative performance scale from a survival per-
spective. J Pain Symptom Manage 1999;18:2–3.
8. Virik K, Glare P: Validation of the palliative perfor-
mance scale for inpatients admitted to a palliative care
unit in Sydney Australia. J Pain Symptom Manage
2002;23:455–456.
9. Harrold J, Rickerson E, Carroll JT, McGrath J, Morales
K, Japo J, Casarett D: Is the palliative performance
scale a useful predictor of mortality in a heteroge-
neous hospice population? J Palliat Med 2005;8:503–
509.
10. Head B, Ritchie CS, Smoot TM: Prognostication in
hospice care: Can the palliative performance scale
help? J Palliat Med 2005;8:492–502.
Address reprint requests to:
Francis Lau, Ph.D.
Health Information Science
University of Victoria
P.O. Box 3050 STN CSC
Victoria, British Columbia V8W 3P5
Canada
E-mail: fylau@uvic.ca
PPS IN EOL PROGNOSTICATION 1073
A
PPENDIX
1.
Palliative Performance Scale (PPSv2)
version 2
PPS Activity & evidence of
Level Ambulation disease Self-care Intake Conscious level
100% Full Normal activity & work Full Normal Full
No evidence of disease
90% Full Normal activity & work Full Normal Full
Some evidence of disease
80% Full Normal activity with Effort Full Normal or Full
Some evidence of disease reduced
70% Reduced Unable Normal Job/Work Full Normal or Full
Significant disease reduced
60% Reduced Unable hobby/house work Occasional assistance Normal or Full
Significant disease necessary reduced or Confusion
50% Mainly Sit/Lie Unable to do any work Considerable assistance Normal or Full
Extensive disease required reduced or Confusion
40% Mainly in Bed Unable to do most activity Mainly assistance Normal or Full or Drowsy
Extensive disease reduced Confusion
30% Totally Bed Unable to do any activity Total Care Normal or Full or Drowsy
Bound Extensive disease reduced Confusion
20% Totally Bed Unable to do any activity Total Care Minimal to Full or Drowsy
Bound Extensive disease sips Confusion
10% Totally Bed Unable to do any activity Total Care Mouth care Full or Drowsy
Bound Extensive disease only Confusion
0% Death — — — —
Instruction for Use of PPS (see also definition of terms)
1. PPS scores are determined by reading horizontally at each level to find a ‘best fit’ for the patient which is then
assigned as the PPS% score.
2. Begin at the left column and read downwards until the appropriate ambulation level is reached, then read across
to the next column and downwards again until the activity/evidence of disease is located. These steps are re-
peated until all five columns are covered before assigning the actual PPS for that patient. In this way, ‘leftward’
columns (columns to the left of any specific column) are ‘stronger’ determinants and generally take precedence
over others.
Example 1: A patient who spends the majority of the day sitting or lying down due to fatigue from advanced
disease and requires considerable assistance to walk even for short distances but who is otherwise fully con-
scious level with good intake would be scored at PPS 50%.
Example 2: A patient who has become paralyzed and quadriplegic requiring total care would be PPS 30%.
Although this patient may be placed in a wheelchair (and perhaps seem initially to be at 50%), the score is
30% because he or she would be otherwise totally bed bound due to the disease or complication if it were
not for caregivers providing total care including lift/transfer. The patient may have normal intake and full
conscious level.
Example 3: However, if the patient in example 2 was paraplegic and bed bound but still able to do some self-
care such as feed themselves, then the PPS would be higher at 40 or 50% since he or she is not ‘total care.’
3. PPS scores are in 10% increments only. Sometimes, there are several columns easily placed at one level but one
or two which seem better at a higher or lower level. One then needs to make a ‘best fit’ decision. Choosing a
‘half-fit’ value of PPS 45%, for example, is not correct. The combination of clinical judgment and ‘leftward prece-
dence’ is used to determine whether 40% or 50% is the more accurate score for that patient.
4. PPS may be used for several purposes. First, it is an excellent communication tool for quickly describing a pa-
tient’s current functional level. Second, it may have value in critera for workload assessment or other measure-
ments and comparisions. Finally, it appears to have prognostic value.
Definitions of Terms for PPS
As noted below, some of the terms have similar meanings with the differences being more readily apparent as one
reads horizontally across each row to find an overall ‘best fit’ using all five columns.
1. Ambulation
The items ‘mainly sit/lie,’ ‘mainly in bed,’ and ‘totally bed bound’ are clearly similiar. The subtle differences are re-
lated to items in the self-care column. For example, ‘totally bed bound’ at PPS 30% is due to either profound weak-
ness or paralysis such that the patient not only can’t get out of bed but is also unable to do any self-care. The differ-
ence between ‘sit/lie’ and ‘bed’ is proportionate to the amount of time the patient is able to sit up vs need to lie down.
‘Reduced ambulation’ is located at the PPS 70% and PPS 60% level. By using the adjacent column, the reduction of
ambulation is tied to inability to carry out their normal job, work occupation or some hobbies or housework activi-
ties. The person is still able to walk and transfer on their own but at PPS 60% needs occasional assistance.
2. Activity & Extent of disease
‘Some,’ ‘significant,’ and ‘extensive’ disease refer to physical and investigative evidence which shows degrees of
progression. For example in breast cancer, a local recurrence would imply ‘some’ disease, one or two metastases in
the lung or bone would imply ‘significant’ disease, whereas multiple metastases in lung, bone, liver, brain, hyper-
calcemia or other major complications would be ‘extensive’ disease. The extent may also refer to progression of dis-
ease despite active treatments. Using PPS in AIDS, ‘some’ may mean the shift from HIV to AIDS, ‘significant’ implies
progression in physical decline, new or difficult symptoms and laboratory findings with low counts. ‘Extensive’ refers
to one or more serious complications with or without continuation of active antiretrovirals, antibiotics, etc.
The above extent of disease is also judged in context with the ability to maintain one’s work and hobbies or activi-
ties. Decline in activity may mean the person still plays golf but reduces from playing 18 holes to 9 holes, or just a
par 3, or to backyard putting. People who enjoy walking will gradually reduce the distance covered, although they
may continue trying, sometimes even close to death (eg. trying to walk the halls).
3. Self-Care
‘Occasional assistance’ means that most of the time patients are able to transfer out of bed, walk, wash, toilet and
eat by their own means, but that on occasion (perhaps once daily or a few times weekly) they require minor assis-
tance.
‘Considerable assistance’ means that regularly every day the patient needs help, usually by one person, to do some
of the activities noted above. For example, the person needs help to get to the bathroom but is then able to brush his
or her teeth or wash at least hands and face. Food will often need to be cut into edible sizes but the patient is then
able to eat of his or her own accord.
‘Mainly assistance’ is a further extension of ‘considerable.’ Using the above example, the patient now needs help get-
ting up but also needs assistance washing his face and shaving, but can usually eat with minimal or no help. This
may fluctuate according to fatigue during the day.
‘Total care’ means that the patient is completely unable to eat without help, toilet or do any self-care. Depending on
the clinical situation, the patient may or may not be able to chew and swallow food once prepared and fed to him or
her.
LAU ET AL.
1074
4. Intake
Changes in intake are quite obvious with ‘normal intake’ referring to the person’s usual eating habits while healthy.
‘Reduced’ means any reduction from that and is highly variable according to the unique individual circumstances.
‘Minimal’ refers to very small amounts, usually pureed or liquid, which are well below nutritional sustenance.
5. Conscious Level
‘Full consciousness’ implies full alertness and orientation with good cognitive abilities in various domains of think-
ing, memory, etc. ‘Confusion’ is used to denote presence of either delirium or dementia and is a reduced level of
consciousness. It may be mild, moderate or severe with multiple possible etiologies. ‘Drowsiness’ implies either
fatigue, drug side effects, delirium or closeness to death and is sometimes included in the term stupor. ‘Coma’ in this
context is the absence of response to verbal or physical stimuli; some reflexes may or may not remain. The depth of
coma may fluctuate throughout a 24 hour period.
©Copyright Notice
The Palliative Performance Scale version 2 (PPSv2) tool is copyright to Victoria Hospice Society and replaces the first PPS published in 1996
[J Pall Care 9(4): 26–32]. It cannot be altered or used in any way other than as intended and described here. Programs may use PPSv2 with
appropriate recognition. Available in electronic Word format by email request to judy.martell@caphealth.org
Correspondence should be sent to Medical Director, Victoria Hospice Society, 1900 Fort St, Victoria, BC, V8R 1J8, Canada
PPS IN EOL PROGNOSTICATION 1075