1162

The Memorial Symptom Assessment Scale Short Form

(MSAS-SF)
Validity and Reliability

Victor T. Chang, M.D.1
Shirley S. Hwang, R.N., A.O.C.N., M.S.2
Martin Feuerman, M.S.3
Basil S. Kasimis, M.D., D.Sc.1
Howard T. Thaler, Ph.D.4
1
University of Medicine and Dentistry of New Jer-
sey, New Jersey Medical School, Section Hema-
tology/Oncology, VA New Jersey Health Care Sys-
tem at East Orange, East Orange, New Jersey.
2
University of Medicine and Dentistry of New Jer-
sey, Section Hematology/Oncology, Patient Care
Services, VA New Jersey Health Care System at
East Orange, East Orange, New Jersey.
3
Information Services and Technology, Academic
Computing Services, UMDNJ/New Jersey Medical
School, Newark, New Jersey.
4
Department of Epidemiology and Biostatistics,
Memorial Sloan-Kettering Cancer Center, New
York, New York.
Presented in part at the Annual Scientific Meeting
of the American Society of Clinical Oncology, Den-
ver, Colorado, May 17–20, 1997. Chang VT,
Hwang SS, Corpion C, Feuerman M. Validation of
the Memorial Symptom Assessment Scale Short
Form (MSAS-SF) [abstract 16]. Proc Am Soc Clin
Oncol 1997;16:47a. Also presented in part at the
Third International Conference on Comprehensive
Cancer Care, New York, New York, November
12–14, 1999, Chang VT, Hwang SS, Corpion C,
Feuerman M. Validation of the Memorial Symptom
Assessment Scale Short Form [abstract 71]. J Pain
Symptom Manage 1999;18:S20.
BACKGROUND. The Memorial Symptom Assessment Scale Short Form (MSAS-SF),
an abbreviated version of the Memorial Symptom Assessment Scale, measures
each of 32 symptoms with respect to distress or frequency alone. A physical
symptom subscale (PHYS), psychologic symptom subscale (PSYCH), and global
distress index (GDI) can be derived from the Short Form. We validated the MSAS-
SF in a population of cancer patients.
METHODS. Two hundred ninety-nine cancer patients examined at the Section of
Hematology/Oncology completed the MSAS-SF and the Functional Assessment
Cancer Therapy (FACT-G). The Karnofsky performance status (KPS), extent of
disease (EOD), and demographic data were assessed. The Cronbach alpha coeffi-
cient was used to assess internal reliability. MSAS-SF subscales were assessed
against subscales of the FACT-G, the KPS, and EOD to determine criterion validity.
Test–retest analysis was performed at 1 day and at 1 week.
RESULTS. The Cronbach alpha coefficients for the MSAS-SF subscales ranged from
0.76 to 0.87. The MSAS-SF subscales showed convergent validity with FACT sub-
scales. Correlation coefficients were ⫺0.74 (P ⬍ 0.001) for the PHYS and FACT-G
physical well-being subscales, ⫺0.68 (P ⬍ 0.001) for the PSYCH and FACT emo-
tional well-being subscales, and ⫺0.70 (P ⬍ 0.001) for GDI and FACT summary of
quality-of-life subscales. The MSAS-SF subscales demonstrated convergent validity
with performance status, inpatient status, and extent of disease. The test–retest
correlation coefficients for the MSAS-SF subscales ranged from 0.86 to 0.94 at 1 day
and from 0.40 to 0.84 for the 1 week group.
CONCLUSIONS. The MSAS-SF is a valid and easy to use instrument for symptom
assessment. Cancer 2000;89:1162–71.
Published 2000 by the American Cancer Society.*
KEYWORDS: neoplasm, symptom, Memorial Symptom Assessment Scale, veteran,
quality of life.

The authors thank Dr. Russell Portenoy for helpful

discussions. Ms Chris Corpion and Ms. Michelle
Rindos assisted with interviews.
Martin Feuerman’s current address: Office of Ac-
ademic Affairs, Winthrop University Hospital, Mine-
ola, New York.
Address for reprints: Victor T. Chang, M.D., UMDNJ
New Jersey Medical School, Section Hematology/
Oncology (111), VA New Jersey Health Care Sys-
tem at East Orange, 385 Tremont Avenue, East
S ymptom assessment is a fundamental component of quality-of-
life (QOL) assessment in palliative care. Previous approaches to
symptom assessments have been based upon one dimensional symp-
tom checklists,1,2 which often measure symptom severity or symptom
distress. The Memorial Symptom Assessment Scale (MSAS)3 is a val-
idated multidimensional symptom assessment instrument that cap-
Orange, NJ 07018;Fax: (973) 395-7089; Email: Received January 10, 2000; revision received May
CHANG.VICTOR_T@EAST-ORANGE.VA.GOV 15, 2000; accepted May 15, 2000.
The views expressed herein do not necessarily *This article is a US Government work, and as
reflect the views of the Department of Veterans such, is in the public domain in the United States
Affairs or of the U.S. Government. of America.

Published 2000 by the American Cancer Society*

 Memorial Symptom Assessment Scale Short Form Validation/Chang et al. 1163

TABLE 1 cluded one that would be simple to read and follow,

Demographics of Study Population quick and easy to complete and analyze, and based on
a categoric or visual analog scale.11 A shorter version
Overall Inpatient Outpatient
Characteristic (n ⴝ 299) (n ⴝ 149) (n ⴝ 150) of the MSAS has been developed with one-third the
number of questions and has been studied in patients
Gender (male) 295 146 149 with acquired immune deficiency syndrome.12 We re-
Age (yrs) port the validation of the MSAS Short Form (MSAS-SF)
Median (range) 66 (27–89) 64 (27–88) 69 (30–89)
in a population of cancer patients at a Veterans Ad-
Mean ⫾ SD 65.2 ⫾ 11.8 63.5 ⫾ 11.9 66.9 ⫾ 11.4
Tumor type (%) ministration (VA) hospital.
Head and neck 21 (7) 14 (10) 7 (5)
Lung 82 (27) 48 (32) 34 (23) MATERIALS AND METHODS
GI 55 (19) 21 (14) 34 (23) Patient Selection and Instruments
GU 93 (31) 41 (28) 52 (34)
The Section of Hematology/Oncology of the VA New
Hematologic 40 (13) 20 (13) 20 (13)
Miscellaneousa 8 (3) 5 (3) 3 (2) Jersey Health Care System at East Orange is the refer-
KPS ral center in New Jersey for veterans with cancer.
Median (range) 70 (20–100) 60 (20–100) 80 (20–100) Patients who participated in Institutional Review
Mean ⫾ SD 68.9 ⫾ 19.3 63.5 ⫾ 18.7 74.4 ⫾ 18.4 Board approved studies (longitudinal cancer care sur-
vey,13 fatigue survey,14 development of a cancer pain
SD: standard deviation; GI: gastrointestinal; GU: genitourinary; KPS: Karnofsky performance status.
a
Miscellaneous inpatients included: sarcoma, 2; melanoma, 1; adenocarcinoma unknown primary, 1; prognostic scale,15 and the test–retest study) com-
breast, 1. Miscellaneous outpatients: sarcoma, 2; adenocarcinoma unknown primary, 1. pleted the MSAS-SF and Functional Assessment Can-
cer Therapy (FACT-G)16 as part of an interview. All
patients gave informed consent before participating.
tures patient rated severity, frequency, and distress as- Each of these studies enrolled a consecutive group of
sociated with 32 highly prevalent symptoms. Since its patients. Patients who enrolled in more than one
introduction, the MSAS has found applications in stud- study were only counted once, and data from other
ies of patients with ovarian carcinoma,4 breast carcino- studies were eliminated. Concurrent demographic
ma,5,6 head and neck carcinoma,7 and cancer pain.8 –10 items, including age, gender, primary cancer diagno-
Comprehensive symptom assessments may be sis, extent of disease, and observer rated Karnofsky
time consuming and difficult for patients to complete, performance status (KPS)17 also were gathered.
and there is a need for shorter symptom assessment For the test–retest study, a convenience group of
instruments. In a review of QOL tools for patients with patients with stable KPS and no acute medical prob-
cancer, proposed criteria for an ideal instrument in- lems was selected. Two convenience groups of rela-

TABLE 2
Summary Scores for MSAS-SF and FACT-G

All patients (n ⴝ 299) Inpatient (n ⴝ 149) Outpatient (n ⴝ 150)

t testa (P
Scale Mean SD Range Mean SD Range Mean SD Range value)

FACT-G
PWB 20.03 6.14 0–28 18.9 6.00 4–28 21.1 6.09 0–28 ⬍ 0.002a
SFWB 18.98 5.68 2–28 19.4 5.65 5–28 18.6 5.71 2–28 NS
EWB 16.42 4.06 0–20 15.9 4.44 0–20 16.9 3.59 4–20 ⬍ 0.03a
FWB 14.70 7.18 0–28 13.9 6.96 0–26 15.5 7.33 0–28 ⬍ 0.03a
SUMQOL 77.06 16.73 26–112 74.9 16.92 32–112 79.1 16.33 26–110 ⬍ 0.04a
MSAS-SF
GDI 1.15 0.81 0–3.2 1.25 0.82 0–3.2 1.06 0.79 0–3.06 ⬍ 0.03a
PHYS 1.02 0.76 0–3.07 1.12 0.74 0–3.07 0.92 0.76 0–2.93 ⬍ 0.02a
PSYCH 0.84 0.84 0–3.5 0.91 0.88 0–3.5 0.76 0.79 0–3.1 NS
TMSAS 0.77 0.53 0–2.51 0.84 0.53 0–2.5 0.70 0.51 0–2.45 ⬍ 0.02a
No. of symptoms 10.5 5.7 0–26 11.1 5.72 0–26 9.8 5.75 0–26 ⬍ 0.046a

MSAS-SF: Memorial Symptom Assessment Scale Short Form; FACT-G: Functional Assessment Cancer Therapy; SD: standard deviation; PWB: physical well-being; SFWB: social/family well-being; NS: not significant;
EWB: emotional well-being; FWB: functional well-being; SUMQOL: total FACT-G QOL scores; GDI: global distress index; PHYS: physical symptom distress; PSYCH: psychologic symptom distress; TMSAS: total
symptom distress.
a
Student t test results between inpatients and outpatients, two-tailed P value.
1164 CANCER September 1, 2000 / Volume 89 / Number 5

tively stable patients were studied. One group of 23 TABLE 3A

patients had test–retest assessments performed the Frequency of Symptoms by MSAS-SF
next day of the MSAS-SF, KPS, and FACT. A second
Symptom Frequency (%)
group of 22 patients had test–retest assessments per-
formed at a 6 – 8 day interval. This study was approved Pain 214 (72)
by the VA New Jersey Health Care System Institutional Lack of energy 210 (70)
Review Board, and patients gave informed consent Dry mouth 165 (55)
Shortness of breath 149 (50)
before participating.
Feeling drowsy 148 (49)
The KPS is an 11-point rating scale ranging from 0 Lack of appetite 134 (45)
to 100 (0, dead; 100, normal function) used to assess Weight loss 133 (44)
patients’ physical functional level related to cancer Difficulty sleeping 127 (42)
and its treatment. Cough 121 (40)
Worrying 118 (39)
The FACT-G (version 3) is a validated 28-item
Constipation 115 (38)
general patient-rated measure of quality of life for Numbness and tingling 113 (38)
cancer patients with any tumor type. Each item is Feeling irritable 106 (35)
scored from 0 to 4 anchored from “not at all” to “very Feeling sad 104 (35)
much.” There are 5 subscales: functional well-being Food taste 94 (31)
Feeling nervous 91 (30)
(FWB) (7 items), physical well-being (PWB) (7 items),
Difficulty concentrating 87 (29)
social/family well-being (SFWB) (7 items), relation- Dizziness 84 (28)
ship with M.D. (RMD) (2 items), and emotional well- Urination 80 (27)
being (EWB) (5 items) with total QOL score ranging Sweats 79 (26)
from 0 to 112. The FACT-G has been used widely in Nausea 75 (25)
Feeling bloated 73 (24)
clinical trials, is easy to complete, and has demon-
Itching 73 (24)
strated sensitivity according to performance status Don’t look like self 68 (23)
and extent of disease. Changes in skin 68 (23)
The MSAS-SF is a patient-rated instrument in Sexual interest 64 (21)
which patients rate symptom distress associated with Swelling of arms/legs 59 (20)
Difficulty swallowing 57 (19)
26 physical symptoms and the frequency of 4 psycho-
Diarrhea 41 (14)
logic symptoms during the past 7 days. Each symptom Vomiting 33 (11)
is scored from 0 to 4 ranging from “no symptom” to Hair loss 28 (9)
“very much.” If the symptom is not present, a value of Mouth sores 26 (9)
zero is assigned. Distress is rated on a 5-point (0 – 4)
MSAS-SF: Memorial Symptom Assessment Short Form.
Likert scale (not at all, 0.8; a little bit, 1.6; somewhat,
2.4; quite a bit, 3.2; very much, 4.0). Frequency of
psychologic symptoms is scored as rarely (1), occa-
sionally (2), frequently (3), and almost constantly (4). Statistical Analysis
The sequence of symptoms on the short form differs Cronbach alpha test was performed to assess internal
from that of the standard form. consistency of the MSAS-SF.18 Pearson pairwise cor-
MSAS-SF subscales include: the global distress in-
relation coefficients were calculated between the sub-
dex (GDI) (4 psychologic symptoms: feeling sad, wor-
scales of the MSAS-SF and of the FACT-G. Correlations
rying, feeling irritable, and feeling nervous, and 6
between physical well-being subscales of the FACT
physical symptoms: lack of energy, pain, lack of appe-
and the physical symptom distress subscale of the
tite, feeling drowsy, constipation, dry mouth); the
MSAS-SF and between the emotional well-being sub-
physical symptom distress score (PHYS) comprises 12
prevalent physical symptoms (lack of energy, pain, scale of the FACT and the psychologic symptom sub-
lack of appetite, feeling drowsy, constipation, dry scale of the MSAS-SF were examined to further assess
mouth, nausea, vomiting, change in taste, weight loss, criterion validity. Spearman correlation coefficients
feeling bloated, and dizziness): the psychologic symp- were used to assess the test–retest coefficient for the
tom distress score (PSYCH) includes 6 prevalent psy- MSAS-SF. Kruskal–Wallis analysis of variance was
chologic symptoms (worrying, feeling sad, feeling ner- used to compare differences in the MSAS-SF subscales
vous, difficulty sleeping, feeling irritable, and difficulty and FACT subscales by KPS. One-way analysis of vari-
concentrating). The number of symptoms is derived ance was performed to compare MSAS-SF and FACT
from screening for the presence of 32 symptoms at scores by extent of disease. Statistical analysis was
each interview. performed with the STATA program.19
 Memorial Symptom Assessment Scale Short Form Validation/Chang et al. 1165

TABLE 3B
Distress for the Most Frequently Reported Symptoms

How much did it distress you?

Not at all/a little Somewhat Quite a bit/very much

Physical symptom N n % n % n %

Pain 214 50 23 46 21 118 55

Lack of energy 210 62 29 42 20 106 50
Dry mouth 165 82 50 28 17 55 33
Short of breath 149 60 40 25 17 64 43
Drowsy 148 70 47 41 27 37 25
Lack appetite 134 43 32 34 25 57 42
Weight loss 133 57 43 25 19 51 38
Difficulty sleeping 127 41 32 23 18 63 50
Cough 121 68 56 28 23 25 21
Constipation 115 33 29 29 25 53 46

How frequently did you experience it?

Rarely Occasionally/frequently Almost constantly

Psychologic symptoms Incidence n %a n % n %

Worrying 118 13 11 96 81 9 8
Irritable 106 19 18 82 77 5 5
Sad 104 6 6 92 88 6 6
Nervous 91 16 17 63 69 12 13

a
Percentages are of those patients who answered yes to a symptom.

TABLE 4 longitudinal study at the third visit, 65 patients from

Pearson Correlations between MSAS-SF, FACT-G Subscales, and KPS
the pain study at the first visit, 159 patients from a
FACT-G general survey of inpatients and outpatients, and 34
patients from the test–retest study. There were 150
Scale PWB SFWB EWB FWB SUMQOL KPS
outpatients and 149 inpatients. These studies were a
MSAS-SF series of studies designed to assess aspects of symp-
PHYS ⫺0.74 ⫺0.02a ⫺0.40 ⫺0.51 ⫺0.61 ⫺0.66 tom distress and enrolled consecutive patients. The
PSYCH ⫺0.48 ⫺0.13 ⫺0.68 ⫺0.56 ⫺0.65 ⫺0.35 test–retest group was a convenience sample. The me-
GDI ⫺0.74 ⫺0.05a ⫺0.61 ⫺0.60 ⫺0.70 ⫺0.62
dian age was 66 years (range, 27– 89 years). Primary
TMSAS ⫺0.72 ⫺0.11a ⫺0.56 ⫺0.56 ⫺0.60 ⫺0.62
NS ⫺0.64 ⫺0.08a ⫺0.49 ⫺0.50 ⫺0.61 ⫺0.59 cancer sites were head and neck (21%, 7%), lung (82%,
KPS ⫺0.55 ⫺0.07a 0.50 0.50 0.47 27%), gastrointestinal (55%, 19%), genitourinary (93%,
31%), hematologic (40%, 13%), and other (8%, 3%%).
P ⬍ 0.001 unless specified otherwise.
Median KPS was 70% (range, 20 –100%). Further de-
MSAS-SF: Memorial Symptom Assessment Short Form; FACT-G: Functional Assessment Cancer Ther-
apy; PWB: physical well-being; SFWB: social/family well-being; EWB: emotional well-being; FWB: tails are presented in Table 1. There was no evidence
functional well-being; SUMQOL: total FACT-G quality-of-life scores; KPS: Karnofsky performance of disease (NED) in 19 (6%) patients, localized disease
status; GDI: global distress index; PHYS: physical symptom distress; PSYCH: psychologic symptom in 7 (2%) patients, regionally advanced disease in 43
distress; TMSAS: total symptom distress; NS: no. of symptoms.
a
(15%) patients, and metastatic disease in 230 (77%)
Not significant.
patients.
Separate convenience samples of inpatients and
RESULTS outpatients were recruited for test–retest reliability
Patient Characteristics analysis. There are 2 groups: 1 day apart and 1 week
Two hundred ninety-nine patients who met the crite- apart. There were 12 outpatients and 11 inpatients in
ria as specified in the previous section were included the 2-day group, and 10 inpatients and 12 outpatients
for the analysis. These included 41 patients from the in the 1-week group.
1166 CANCER September 1, 2000 / Volume 89 / Number 5

TABLE 5
Variation of MSAS-SF and FACT-G Scores with the Extent of Diseasea

NED and local (n ⴝ 26) Regional (n ⴝ 43) Metastatic (n ⴝ 230)

Scale Mean SD Mean SD Mean SD F P value

MSAS-SF
PHYS 0.53 0.63 0.60 0.59 1.16 0.75 17.83 ⬍ 0.0001
PSYCH 0.63 0.73 0.56 0.64 0.91 0.87 4.02 ⬍ 0.02
GDI 0.72 0.75 0.71 0.64 1.29 0.80 14.64 ⬍ 0.0001
TMSAS 0.49 0.55 0.51 0.39 0.85 0.52 12.72 ⬍ 0.0001
NS 7.3 6.38 7.9 4.5 11.3 5.6 11.62 ⬍ 0.0001
FACT-G
PWB 23.92 4.53 22.77 5.16 19.07 6.16 13.27 ⬍ 0.0001
SFWB 16.82 6.27 17.59 5.93 19.49 5.49 4.15 ⬍ 0.02
EWB 17.19 2.71 17.76 2.83 16.08 4.32 3.71 ⬍ 0.03
FWB 17.77 7.43 18.12 7.13 13.71 6.89 9.96 ⬍ 0.001
SUMQOL 83.82 16.77 83.07 14.99 75.29 16.69 5.79 ⬍ 0.003

MSAS-SF: Memorial Symptom Assessment Short Form; FACT-G: Functional Assessment Cancer Therapy; NED: no evidence of disease; SD: standard deviation; PHYS: physical symptom subscale; PSYCH: psychologic
symptom subscale; GDI: global distress index; TMSAS: total symptom distress; NS: no. of symptoms; PWB: physical well-being; SFWB: social/family well-being; EWB: emotional well-being; FWB: functional well-being;
SUMQOL: total FACT-G quality-of-life scores.
a
Differences between patients with metastatic disease and patients with regional disease/NED were significant but not between the latter two groups.

TABLE 6
Variation of MSAS-SF and FACT-G Scores with the KPSa

KPS 20–50 (n ⴝ 68) KPS 60–70 (n ⴝ 78) KPS 80 (n ⴝ 104) KPS 90–100 (n ⴝ 49)

Scale Mean SD Mean SD Mean SD Mean SD F P value

MSAS-SF
PHYS 1.71 0.66 1.39 0.61 0.74 0.54 0.23 0.35 81.43 ⬍ 0.0001
PSYCH 1.23 0.99 0.98 0.76 0.73 0.76 0.29 0.50 14.89 ⬍ 0.0001
GDI 1.83 0.73 1.42 0.62 0.92 0.64 0.31 0.47 63.99 ⬍ 0.0001
TMSAS 1.19 0.49 0.98 0.41 0.59 0.41 0.23 0.27 63.48 ⬍ 0.0001
NS 14.3 5.05 13.0 4.47 8.8 4.6 4.5 4.2 56.96 ⬍ 0.0001
FACT-G
PWB 15.81 5.84 17.77 5.56 21.73 4.93 25.88 3.20 46.08 ⬍ 0.0001
SFWB 19.56 5.49 19.16 4.94 18.95 6.33 17.95 5.61 0.79 0.50
EWB 14.63 5.32 15.78 3.93 17.30 3.20 18.04 2.57 10.09 ⬍ 0.0001
FWB 10.31 5.29 12.44 5.67 15.96 7.51 21.73 4.35 38.22 ⬍ 0.0001
SUMQOL 67.19 15.89 72.13 13.43 80.81 16.44 90.65 10.91 29.06 ⬍ 0.0001

MSAS-SF: Memorial Symptom Assessment Short Form; FACT-G: Functional Assessment Cancer Therapy; KPS: Karnofsky performance status; SD: standard deviation; PHYS: physical symptom subscale; PSYCH:
psychologic symptom subscale; GDI: global distress index; TMSAS: total symptom distress; NS: no. of symptoms; PWB: physical well-being; SFWB: social/family well-being; EWB: emotional well-being; FWB:
functional well-being; SUMQOL: total FACT-G quality-of-life scores.
a
Comparisons were made with Kruskal–Wallis test, two-tailed P value.

Summary Scores for MSAS-SF and FACT-G symptoms (P ⬍ 0.046) of the MSAS-SF. Inpatients had
The summary scores for the MSAS-SF and FACT-G significantly lower FACT scores in the domains of PWB
subscales are listed in Table 2. The mean number of (P ⬍ 0.002), EWB (P ⬍ 0.03), FWB (P ⬍ 0.03), and total
symptoms was 10.5 (range, 0 –26) for the entire study FACT-G QOL (SUMQOL) scores (P ⬍ 0.04).
population, 11.1 for inpatients and 9.8 for outpatients.
Comparisons of scores between inpatient and outpa- Symptom Prevalence by MSAS-SF
tient groups are listed in Table 2. The inpatients had The most frequently reported physical symptoms
significantly higher symptom distress scores in the included pain (72%), lack of energy (70%), dry
PHYS subscale (P ⬍ 0.02), GDI (P ⬍ 0.03), total symp- mouth (55%), shortness of breath (50%), feeling
tom distress (TMSAS) (P ⬍ 0.02), and number of drowsy (49%), lack of appetite (45%), weight loss
 Memorial Symptom Assessment Scale Short Form Validation/Chang et al. 1167

TABLE 7 number of symptoms is closely correlated with the

Test–Retest Resultsa SUMQOL score on the FACT instrument. Criterion
validity was further established by demonstrating ap-
1 day (n ⴝ 23) 1 week (n ⴝ 22)
propriate and significant correlations between the
Scale r P value r P value MSAS-SF parameters and the KPS.

FACT-G
PWB 0.81 ⬍ 0.0001 0.52 ⬍ 0.01
Convergent Validity Based on the KPS, Extent of Disease,
SFWB 0.70 ⬍ 0.003 0.25 NS
EWB 0.90 ⬍ 0.0001 0.77 ⬍ 0.0001 and Inpatient Status
FWB 0.80 ⬍ 0.0001 0.75 ⬍ 0.0001 Further analysis was performed to assess the sensitiv-
SUMQOL 0.88 ⬍ 0.0001 0.58 ⬍ 0.005 ity of MSAS-SF subscales to the KPS and extent of
MSAS-SF disease. The results are listed on Tables 5 and 6. In
PHYS 0.87 ⬍ 0.0001 0.46 ⬍ 0.03
summary, MSAS-SF subscale scores showed a sharp
PSYCH 0.89 ⬍ 0.0001 0.84 ⬍ 0.0001
TMSAS 0.92 ⬍ 0.0001 0.60 ⬍ 0.003 boundary between patients with and without meta-
GDI 0.94 ⬍ 0.0001 0.40 ⬍ 0.07 static disease. Patients with metastatic disease (n
No. of symptoms 0.86 ⬍ 0.0001 0.75 ⬍ 0.0001 ⫽ 230) exhibited significantly higher number of symp-
KPS 0.95 ⬍ 0.0001 0.94 ⬍ 0.0001 toms, physical and psychologic distress subscales,
global distress subscale, and total MSAS scores com-
FACT-G: Functional Assessment Cancer Therapy; SFWB: social/family well-being; EWB: emotional
well-being; FWB: functional well-being; SUMQOL: total FACT-G quality-of-life scores; MSAS-SF: Me- pared with patients with NED/local (n ⫽ 26) and re-
morial Symptom Assessment Scale Short Form; PHYS: physical symptom subscale; PSYCH: psycholog- gional disease (n ⫽ 43) groups. Similar observations
ical symptom subscale; TMSAS: total symptom distress; GDI: global distress index; KPS: Karnofsky also were recorded for the FACT subscales and
performance status. SUMQOL scores measured by FACT-G.
a
Spearman correlation coefficients and associated probabilities are reported.
One-way analysis of variance analysis was per-
formed to compare the means in MSAS-SF and
FACT-G parameters among 4 different KPS catego-
(44%), difficulty sleeping (42%), cough (40%), and ries (KPS 20 –50%, n ⫽ 68; KPS 60 –70%, n ⫽ 78; KPS
constipation (38%). Among these symptoms, except 80%, n ⫽ 104; and KPS 90 –100%, n ⫽ 49) as conve-
for cough, greater than 50% of patients reported that nience groups. The results, shown in Table 6, indi-
the distress caused by each symptom was at least cate that all the MSAS-SF subscales scores differed
“somewhat.” However, greater than 50% of the pa- significantly between different KPS categories. Anal-
tients with pain or lack of energy reported “quite a yses by KPS quartiles yielded similar results. FACT-
bit”or “very much” symptom distress. The most fre- G subscale scores also exhibited significant sensitiv-
quent psychologic symptoms included worrying ity to KPS.
(39%), feeling irritable (35%), and feeling sad (35%). MSAS-SF and FACT subscales differed signifi-
The majority of patients experienced psychologic cantly for inpatients and outpatients except for the
symptoms “occasionally.” Complete data are listed FACT SFWB and for the MSAS psychologic subscale.
in Tables 3A and 3B.

MSAS-SF Reliability Test–Retest Reliability

All the MSAS subscales were reliable in the MSAS-SF. The test–retest study was performed to assess the co-
The Cronbach alpha coefficient was 0.80 for the GDI, efficient of stability. Spearman pairwise correlation
0.82 for the physical symptom distress (PHYS), 0.76 for coefficients are listed on Table 7. Statistically signifi-
the psychologic symptom distress (PSYCH), and 0.87 cant correlations were observed for the two test–retest
for the TMSAS. groups; however, the magnitude was different. For the
1-day test–retest group, the coefficients were strongly
Criterion Validity significant ranging from 0.70 to 0.90 for FACT-G sub-
The MSAS-SF subscale scores were assessed against scales, 0.95 for KPS, and ranging from 0.86 to 0.94 for
subscales of the FACT-G with pairwise Pearson corre- MSAS-SF scores.
lation coefficients (Table 4). Correlation coefficients However, for the 1-week test–retest group, the
were ⫺0.74 (P ⬍ 0.001) for the PHYS and FACT-G coefficients were 0.25– 0.77 for FACT-G subscales, 0.94
PWB, ⫺0.68 (P ⬍ 0.001) for the PSYCH and FACT-G for KPS, and ranged from 0.40 to 0.84 for MSAS-SF
EWB, and ⫺0.70 (P ⬍ 0.001) for GDI and FACT-G scores. For both FACT-G and MSAS-SF, significant
SUMQOL score. Of the MSAS-SF subscales, only correlations are present at 2 days and 1 week, and
PSYCH correlated significantly with FACT SFWB. The correlations are much higher at 1 day than at 1 week.
1168 CANCER September 1, 2000 / Volume 89 / Number 5

FIGURE 1. Facsimile of the revised

Memorial Symptom Assessment Scale
Short Form is shown.
 Memorial Symptom Assessment Scale Short Form Validation/Chang et al.
FIGURE 1. (continued)
Comparison of Results with Previous Studies
Values of the MSAS-SF subscales and their correlation
with FACT subscales were compared with a previous
study of 240 VA patients in which the MSAS and the
FACT were used. Although the two populations stud-
ied are somewhat different, correlation coefficients
between subscales of the FACT-G and MSAS and
FACT-G and MSAS-SF were nearly identical, except for
SFWB, in which coefficients were weaker or insignifi-
cant between the MSAS-SF and FACT-G.
DISCUSSION
Because patients with cancer have multiple physical
and psychologic symptoms, comprehensive symptom
assessment is a requisite toward good symptom con-
trol and an important component of QOL assessment.
This need has led to the development of scales that
elicit patient rated symptom reports, including the
Edmonton Symptom Assessment Scale,20,2122 the Lung
Cancer Symptom Scale,23 and the Rotterdam Symp-
tom Checklist.24 These instruments have measured
one dimension, typically severity or distress, for mul-
tiple symptoms selected by consensus panels or ex-
pert opinion.
Previous studies have shown that different dimen-
sions of symptoms may be important in symptom
assessment.25 This led to development of the Memo-
rial Symptom Assessment Scale, which showed that
summary scores from multidimensional measure-
ments of physical and psychologic symptom can be
used to fashion scales that are highly correlated with
quality of life.26 Analysis of the MSAS suggested that
physical symptom distress and frequency of psycho-
logic symptom frequency most closely correlated with
QOL measures. The MSAS-SF selects these specific
dimensions for symptom assessment. The develop-
ment of the MSAS-SF suggests how a QOL analysis
may guide the development of shorter instruments to
assess the impact of symptoms on QOL. Such instru-
1169
ments may be more useful in a clinical setting; the full
MSAS is still recommended for research purposes.
Further work is in progress.27
In this study, we found that responses to the
MSAS-SF showed internal reliability by Cronbach al-
pha coefficients. We also found that criterion validity
was present. The summary parameters for physical
and psychologic symptom distress of the short form
correlated well with corresponding parameters in the
FACT-G instrument, with inpatient/outpatient status,
and with the KSP. The close correlation of the number
of symptoms with the FACT SUMQOL score that has
been noted in previous work is confirmed in this
study.
The test–retest results show a high correlation
coefficient for reassessments the day after and a lower
but still significant correlation coefficient at 1 week.
The optimal interval for test–retest evaluation is not
yet known. Intervals have included 4 days for the
FACT and the European Organization for Research
Treatment of Cancer QLQ-C30,28 and 1 week for the
FACT brain subscale.29 Test–retest data have not been
reported yet for the MSAS-SF. We anticipated seeing
more variation at 1 week retest than 1 day retest be-
cause patients may change clinically over time. These
values suggest that the MSAS-SF is a sensitive instru-
ment and that the scale should be reset for a time
interval of 2 days instead of 1 week. There are many
possible reasons for the decreased correlation at 1
week. One reason may be that patients with advanced
disease may be more unstable, and that these changes
are captured by the MSAS-SF. However, other factors,
such as changes in the patient’s perceptions of symp-
toms, are also possible. This was a small sample size,
and further work is needed.
The population that completed this study differed
from the population we last reported30 in that patients
had a lower KPS, a larger percentage of patients had
metastatic disease, and a slightly different profile of
1170 CANCER September 1, 2000 / Volume 89 / Number 5
primary disease sites was observed. The FACT-G sub-
scale scores were lower. Higher MSAS symptom prev-
alence and higher symptom distress scores also were
observed in the overall population in comparison with
our previous survey.7 When correlations of the MSAS-
SF subscales are compared with correlations of the full
MSAS instrument with the FACT, correlation coeffi-
cients between the MSAS-SF subscales and FACT sub-
scales were nearly identical to those of the MSAS for
PWB, EWB, and FWB domains. They decreased signif-
icantly for SFWB, with only the correlation coefficient
between MSAS PSYCH and SFWB retaining statistical
significance. This suggests that for most dimensions of
quality of life, findings on the MSAS-SF are similar to
those for the parent MSAS instrument. Further study is
needed to confirm and explore the reasons for a de-
creased correlation between MSAS-SF subscale scores
and the FACT SFWB subscale.
MSAS-SF subscale scores varied significantly and
appropriately with KPS and extent of disease. Scores
greater than 1.0 for any of the MSAS-SF subscales
should suggest the presence of significant symptom
distress. Interpretation of these scores, including the
clinical meaning of magnitude of differences in MSAS
scores, will be clarified further as more populations
are studied.
The KPS, FACT, and MSAS-SF measure different
aspects of QOL. The KPS emphasizes performance and
function, the FACT provides information on the di-
mensions relevant to quality of life, and the MSAS-SF
provides a detailed symptom assessment. The KPS is
valuable for prognosis but does not indicate what
steps the practitioner can take to help patients with
poor performance status. The information from QOL
instruments and from symptom instruments comple-
ments that from the KSP.31 In a pilot study of QOL
instruments in an office setting, both patients and
physicians found discussions of QOL to be helpful.32
Similarly, the use of symptom assessment instruments
may help guide health professionals in directing pal-
liative care efforts for patients with poor performance
status.
The MSAS-SF was easy to administer and took less
than 5 minutes to complete. These findings suggest
that patients can be rapidly and comprehensively as-
sessed for symptoms with instruments such as the
MSAS-SF. For patients with limited stamina, the short
form may be an effective instrument. For a more com-
plete assessment of symptoms, however, the full MSAS
may be more informative. An occasional problem with
the MSAS-SF was that some patients equated the box
for not having a symptom with the adjacent box for
not being bothered by the symptom. To correct this
problem, the form has been revised so that the first
box is checked if the patient has the symptom (Fig, 1).
The population studied was primarily elderly male
patients with advanced disease and heterogeneous diag-
noses. As such, it differs from the initial population of
patients with carcinoma of the breast, ovary, prostate, or
colon. The findings of good correlations between the
MSAS-SF and FACT-G in a different population suggests
that the MSAS-SF may be a robust instrument across
different populations. A small fraction of patients were in
complete remission. Data from these patients were in-
cluded in the analyses because oncologists do see pa-
tients in complete remission. This mean values for this
group did not differ to much from those with local dis-
ease. Other populations, such as patients with early dis-
ease and women, should be studied with this instru-
ment. As data from different patient groups are collected
and compared, the interpretation of scores can be de-
termined more precisely. Another possible limitation
may be that scoring the form may be tedious. This can
be resolved with an automated data entry program. We
conclude that the MSAS-SF is a valid instrument for
symptom assessment in the VA population.
REFERENCES
1. Brodman K, Erdmann AJ, Lorge I, Woff HG. The Cornell
Medical Index, an adjunct to the medical interview. JAMA
1949;140:530 – 4.
2. Osoba D. Self-rating symptom checklists: a simple method
for recording and evaluating symptom control in oncology.
Cancer Treat Rev 1993;19(Suppl A):43–1.
3. Portenoy RK, Thaler HT, Kornblith AB, Lepore J, Friedland-
er-Klar H, Kiyasu E, et al. The Memorial Symptom Assess-
ment Scale: an instrument for the evaluation of symptom
prevalence, characteristics, and distress. Eur J Cancer 1994;
30A:1226 –36.
4. Kornblith AB, Thaler HT, Wong G, Vlamis V, Lepore JM,
Loseth DB, et al. Quality of life of women with ovarian
cancer. Gynecol Oncol 1995;59:231– 42.
5. Seidman AD, Portenoy RK, Yao TJ, Lepore J, Mont EK, Kor-
mansky J, et al. Quality of life in phase II trials: a study of
methodology and predictive value in patients with advanced
breast cancer treated with paclitaxel plus granulocyte colo-
ny-stimulating factor. J Natl Cancer Inst 1995;87:1316 –22.
6. Hann DM, Jacobsen PB, Martin SC, Kronish LE, Azzarello
LM, Fields KK. Quality of life following bone marrow trans-
plantation for breast cancer: a comparative study. Bone
Marrow Transplant 1997;19:257– 64.
7. Harrison LV, Zelefsky MJ, Pfister DG, Carper E, Raben A,
Kraus DH, et al. Detailed quality of life assessment in pa-
tients treated with primary radiotherapy for squamous cell
cancer of the base of the tongue. Head Neck 1997;19:169 –75.
8. Du Pen S, Du Pen AR, Hansberry J, Kraybill BM, Stillman M,
Panke J, et al. Implementing guidelines for cancer pain
management: results of a randomized controlled clinical
trial. J Clin Oncol 1999;17:361–70.
9. Ingham J, Seidman A, Yao TJ, Lepore J, Portenoy R. An
exploratory study of frequent pain measurement in a cancer
clinical trial. Qual Life Res 1996;5:503–7.
 Memorial Symptom Assessment Scale Short Form Validation/Chang et al.
10. Payne R, Mathias SD, Pasta DJ, Wanke LA, Williams R,
Mahmoud R. Quality of life and cancer pain: satisfaction
and side effects with transdermal fentanyl versus oral mor-
phine. J Clin Oncol 1998;16:1588 –93.
11. Donnelly S, Walsh TD. Quality of life assessment in ad-
vanced cancer. Palliat Med 1996;10:275– 83.
12. Vogl D, Rosenfeld B, Breitbart W, Thaler H, Passik S, McDonald
M, et al. Symptom prevalence, characteristics, and distress in
AIDS outpatients. J Pain Symptom Manage 1999;18:253– 64.
13. Hwang SS, Chang VT, Fairclough DL. Longitudinal measure-
ments for quality of life and symptoms in terminal cancer
patients [abstract]. Proc ASCO 1996;15:536.
14. Hwang SS, Chang VT, Cogswell J, Ohanian M, Kasimis B.
Fatigue (F), depression, symptom distress (SD), quality of
life (QOL) and survival (S) in male cancer patients (pts) at a
VA Medical Center [abstract]. Proc Am Soc Clin Oncol 1999;
18:594a.
15. Hwang SS, Chang VT, Corpion C, Kasimis B. Outcomes of
cancer pain management based upon AHCPR guidelines:
pain relief and quality of life (QOL) [abstract]. Proc Am Soc
Clin Oncol 1998;17:61a.
16. Cella DF, Tulsky DS, Gray G, Sarafian B, Linn E, Bonomi A,
et al. The Functional Assessment of Cancer Therapy Scale:
development and validation of the general measure. J Clin
Oncol 1993;11:570 –9.
17. Karnofsky DA, Burchenal JH. The clinical evaluation of che-
motherapeutic agents in cancer. In: MacLeod CM, editor.
Evaluation of chemotherapeutic agents. New York: Colum-
bia University Press, 1949:191–205.
18. Cronbach LJ. Coefficient alpha and the internal structure of
tests. Psychometrica 1951;16:297–334.
19. StatCorp. Stata Statistical Software: Release 5.0. College Sta-
tion, TX: Stata Corporation, 1997.
20. Bruera E, Kuehn N, Miller MJ, Selmser P, Macmillan K. The
Edmonton Symptom Assessment System (ESAS): a simple
method for the assessment of palliative care patients. J
Palliat Care 1991;7:6 –9.
21. Chang VT, Hwang SS, Feuerman M. Validation of the
Edmonton Symptom Assessment Scale. Cancer 2000;88:
2164 –71.
1171
22. McCorkle R, Young K. Development of a symptom distress
scale. Cancer Nurs 1978;1:373– 8.
23. Hollen PJ, Gralla RJ, Kris MG, Cox C, Belani CP, Grunberg
SM, et al Measurement of quality of life in patients with lung
cancer in multicenter trials of new therapies. Cancer 1994;
73:2087–98.
24. De Haes JCJM, van Knippenberg FCE, Nejit JP. Measuring
psychological and physical distress in cancer patients: struc-
ture and application of the Rotterdam Symptom Checklist.
Br J Cancer 1990;62:1034 – 8.
25. Portenoy RK, Miransky J, Thaler HT, Hornung J, Bianchi C,
Cibas-Kong I, et al. Pain in ambulatory patients with lung or
colon cancer. Prevalence, characteristics and effect. Cancer
1992;70:1616 –24.
26. Portenoy RK, Thaler HT, Kornblith AB, Lepore J, Friedland-
er-Klar H, Coyle N, et al. Symptom prevalence, characteris-
tics, and distress in a cancer population. Qual Life Res 1994;
3:183–9.
27. Chang VT, Hwang SS, Kasimis B, Thaler HT. Shorter symp-
tom assessment instruments [abstract]. Proc Am Soc Clin
Oncol 1999;18:583a.
28. Hjermstad MJ, Fossa SD, Bjordal K, Kaasa S. Test/retest
study of the European Organization for Research and Treat-
ment of Cancer Core Quality of Life Questionnaire. J Clin
Oncol 1995;13:1249 –54.
29. Weitzner MA, Meyers CA, Gelke CK, Byrne KS, Cella DF,
Levin VA. The Functional Assessment of Cancer Therapy
(FACT) Scale. Development of a brain subscale and revali-
dation of the general version (FACT-G) in patients with
primary brain tumors. Cancer 1995;75:1151– 61.
30. Chang VT, Hwang SS, Kasimis BS. Symptom and quality of
life survey of oncology patients at a VA Medical Center.
Cancer 2000;88:1175– 81.
31. Schaafsma J, Osoba D. The Karnofsky performance status
scale re-examined: a cross-validation with the EORTC C30.
Qual Life Res 1994;3:413–24.
32. Detmar SB, Aaronson NK. Quality of life assessment in daily
oncology practice: a feasibility study. Eur J Cancer 1998;34:
1181– 6.