2294
The Performance Status Scale for Head and Neck
Cancer Patients and the Functional Assessment of
Cancer Therapy-Head and Neck Scale
A Study of Uti/ity and Validity
Marcy A. List, Ph.D.'
Linda L. D'Antonio, Ph.D.2
David F. Cella, Ph.D.3
Amy Siston, BSC.'
Patricia Mumby, Ph.0.'
Daniel Haraf, M.D.'
Everett Vokes, M.D.'
' University of Chicago Cancer Research Center,
Chicago, Illinois.
Loma Linda University School of Medicine &
Loma Linda University Cancer Institute, Loma
Linda, California.
Rush Presbyterian St. Luke's Medical Center,
Chicago, Illinois.
Presented in part at the Fourth Research Work-
shop on the Biology, Prevention and Treatment
of Head and Neck Cancer, Arlington, VA, Sep-
tember 8-11,1994 and at the American Society
for Clinical Oncology, Los Angeles, CA, May 20-
23, 1995.
Supported in part by the National Cancer Insti-
tute's Cancer Center Support Grant CA 14599.
Address for reprints: Marcy A. List, Ph.D., Asso-
ciate Director, Cancer Control, University of Chi-
cago Cancer Research Center, 5841 S.Maryland
MC 1140, Chicago IL 60637.
Received November 22, 1995; revision received
February 15, 1996; accepted February 15, 1996.
0 1996 American Cancer Society
BACKGROUND. The goal of this investigation was to examine the relationship be-
tween, and application of, two disease specific quality of life (QL) measures cur-
rently being employed for head and neck cancer patients: the Functional Assess-
ment of Cancer Therapy-Head and Neck Scale (FACT-H&N)and the Performance
Status Scale for Head and Neck Cancer Patients (PSS-HN).
METHODS. The FACT-H&N and PSS-HN were administered to 151 head and neck
cancer patients with a range of disease sites, treatment status (on vs. off treatment),
and treatment modalities (surgeiy, radiation, and chemotherapy).
RESULTS. FACT-H&Nsubscale and total scores and PSS-HN subscale scores proved
sensitive to patients groups (showed significant and clinically meaningful differ-
ences) on the basis of treatment status (on vs. off treatment) and global perfor-
mance status (Karnofsky scores). The pattern of correlations between FACT-H&N
and PSS-HN subscales supported the scales' construct (convergent vs. divergent)
validity. The strongest and most significant associations were observed between
PSS-HN Normalcy of Diet and Eating in Public, and the head and neck subscale
(HNS)of FACT-H&N,both of which were designed to measure the unique problems
of head and neck cancer patients. More modest associations were observed be-
tween subscales measuring physical and functional areas of performance, social
functioning, and emotional well-being.
CONCLUSIONS. The FACT-HNS was found to be reliable and valid when applied to
head and neck cancer patients. It clearly adds information to that collected by the
parent (core) instrument. The PSS-HN also provides unique information, indepen-
dent of that provided by the Karnofsky or the FACT-H&N. This study supported
the multidimensional nature of QL for head and neck cancer patients, and thus
the importance of assessing disease specific concerns in addition to general health
status when assessing functional and QL outcome. Cancer 1996; 772294-301.
0 1996 Atnerican Cancer Society.
KEYWORDS head and neck cancer, quality of life, performance status, measurement
validity, Functional Assessment of Cancer Therapy-Head and Neck, Performance
Status Scale for Head and Neck Cancer.
Q uality of life (QL) and functional status have been recognized as im-
portant outcome variables in the evaluation of head and neck cancer
treatment. In particular, disfigurement and residual impairments of
eating and speaking are of paramount concern and may secondarily affect
a broad spectrum of social, family, and work related roles and responsibil-
ities.'.' The nature and extent of disability may depend, in part, on the
type of treatment. Surgery, radiotherapy, and combined chemoradiother-
apy for advanced disease carry unique acute as well as chronic toxicities.

.For example, while surgery for larynx cancer may result
in significant communication impairment: radiotherapy or
concomitant chemoradiotherapy may result in dry mouth,4
stiffening or constriction of local tissues, and subsequent
problems with chewing and swallowing solid foods.5-'
In response to this awareness, recent years have been
witness to the development of a number of disease spe-
cific QL and performance outcome instrument~."'~Two
of the more widely employed measures include a perfor-
mance specific instrument, the Performance Status Scale
for Head and Neck Cancer (PSS-HN), and a multidimen-
sional QL index, the Functional Assessment of Cancer
Therapy-Head and Neck Scale (FACT-H&N).
The PSS-HN, developed and tested by List et al., was
designed to evaluate performance in areas of functioning
most likely affected by head and neck cancer and its treat-
ment, specifically eating, speaking, and eating in public."
The PSS-HN has been shown to have adequate inter-rater
reliability and to be sensitive to differences in perfor-
mance and change over time.15.16The FACT-G (General),
developed by Cella et al., is a multidimensional QL instru-
ment specifically designed for use with cancer patients.
Scale construction and initial validation have been de-
scribed previously." The FACT-G consists of a 28-item
core which can be supplemented by site andlor treat-
rnent specific subscales, including a head and neck can-
cer specific subscale (HNS).
Both PSS-HN and FACT-H&Nare quick, easy to com-
plete, and currently being used in clinical investigations"
a.nd research settings. While the content and efficacy of
these instruments continue to be examined, to date there
are very few published data on the validity of the HNS of
FACT-H&N" and only limited data on the PSS-HN.'0~'"'8
Such disease specific data are critical to appropriate inter-
pretation of scale scores and comparisons across groups
of patients; additionally, at present, there are no disease
specific norms for interpreting an individual patient's
scores in a clinical setting. Preliminary reports on the
PSS-HN and the FACT have been promising and it is now
timely to comprehensively assess the overall performance
of these tools. The goal of this investigation was to exam-
ine application of the PSS-HN and the FACT-H&N in a
large heterogeneous population of head and neck cancer
patients, to present descriptive statistics for samples of
head and neck patients, to further investigate the validity
of the measures, and to describe the relationships among
subscales.
METHODS
Patients and Procedures
One hundred fifty one patients who had received a variety
of types of treatment at the University of Chicago Hospi-
tals (n = 101) and the Loma Linda University Medical
Center (n = 50) were included in this study. Participants
had to be able to speak English, had to have no known
PSS-HN and FACT-H&N: Utility and ValidityList et al. 2295
evidence of brain metastasis, delirium, or psychosis, and
had to give informed consent. The final sample included
56 patients on treatment (representing all patients cur-
rently enrolled on 2 concomitant chemoradiotherapy
protocols-no refusals), 45 patients with no evidence of
disease (NED) assessed at least 1 year post completion
of concomitant chemoradiotherapy treatment ( 3 refusals
from a total of 48 eligible), and 50 patients (47 NED)
assessed at least 3 months following surgical manage-
ment (13 refusals from a total of 63 eligible and available
during the time frame of the study). The PSS-HN and the
FACT-H&N were administered to all patients at the time
of a clinic or hospital visit.
Measures
The PSS-HN is a clinician-rated instrument consisting of
three subscales: Normalcy of diet, Understandability of
Speech, and Eating in Public (some patient input incor-
porated in ratings of eating and diet). Each is rated from
0 to 100, with higher scores indicating better perfor-
mance.'"," The FACT-H&N is a self-report instrument
consisting of 28 general + 11 head and neck specific
items, each rated on a 0 to 4 Likert type scale. Items are
then combined" to describe patient functioning in six
areas: physical well-being, social and family well-being,
relationship with doctor, emotional well-being, func-
tional well-being, and head and neck related symptoms
(HNS).Higher subscale scores represent better QL. While
Version 2 of the FACT-H&N was employed in the current
study, Version 3 is now available for use.
Statistical Methods
First, patient scores were summarized to provide descrip-
tive data on the scales, including means, medians, and
standard deviations. Second, the sensitivity of both the
PSS-HN and the FACT-H&N subscales was examined by
comparing groups of patients who were expected to be
functioning at different levels (discriminant validity). Be-
tween group differences were assessed using analysis of
variance (ANOVA) procedures" for continuous variables
(FACT-H&N), and nonparametric analyses2' (Mann-
Whitney) for ordinal data (PSS-HN). Spearman correla-
tion coefficients were used to examine the relationship
between measures.
RESULTS
Patient Characteristics
Participating patients represent a wide range of sociode-
mographic characteristics, disease sites, stages, and treat-
ment modalities. Seventy percent (70%) of the group were
male, mean age was 58 years, and two-thirds had at least
a high school education. Sixty-four percent were white
and 3 1%African American. The majority of patients (85%)
had advanced disease at diagnosis and the 3 most fre-
quent disease sites were larynx (32%), oropharynx (28%),
2296 CANCER June 1,1996 / Volume 77 / Number 11

TABLE 1 TABLE 2
Selected Patient Demographics and Disease Variables (n = 151) FACT-H&N Subscale and Total Scores’

Race Disease site FACT-H&N subscale Mean SD Range Coefficient (Y

_ _ _ ~ ~
African American 46 (31%) Larynx 49 134%) ~ ~ ~~~

White 96 (64%) Oropharynx 42 (28%) Physicdl 21.6 5.7 6-28 0.79

Other 9 (5%) Oral cavity 24 (16%) Functional 17.9 6.1 4-28 0.75
Age Hrpopharynx 11 (7%) Social 21.6 5.6 4-28 0.59
Mean 58 yrs Nasopharynx 5 (3%) Emotional 16.9 2.9 8-20 0.59
Range 17-82 Otherb 18 (12) Relatianship with doctor 7.2 1.3 3-8 0.83
Gender Stage Total (FACT-G) 85.1 15.9 45-111 0.89
Male 106 I 6 (4%) Head and Neck Subscale (11-item) 25.9 6.1 6-43 0.63
Female 45 (30%) I1 11 (7%) Head and Neck Subscale (9-itemib 19.2 7.2 3-35 0.70
Karnofsky ratinga 111 36 (24%)
40 I(l%) Iv 92 (61%) FACT.H&N: Functional Assessment of Cancer Therapy-Head and Neck SD: standard deviation.
50 3 (3%) Missing 6 (4%) n = 151 head and neck cancer patients.
60 10 (10%) Treatment status Nine-item Head and Neck Subscale excludes items 41 and 42 related to smolant! and alcohol.
70 29 (29%) Treatment completed 95 (63%)
80 14 (14%) Primary surgery ti- rt 50 (33%)
90 21 (21%) Primary chemoradiotherapy 45 (30%)
100 22 ( ~ 2 9 ~ ) On chemoradiotherapy 56 (37%)
Median score: 80
study, level 90 on the PSS-HN has been changed from
the original description’”,15of “peanuts” to “full diet with
Ratings available lor 101 U of C patients. liquid assist” (manuscript in preparation).
Includes unlaiown primary (31, missing (41. other (11-paranasal sinus, parotid, and salivar,, Eland). Table 2 presents means, standard deviations, ranges,
and coefficient as (internal consistency) for FACT-H&N
subscale and total scores. Scores for the HNS were com-
puted twice, first according to FACT-H&N Manual-Ver-
and oral cavity (16%).Of the 95 patients who had com- sion .P3based on an 1 1 -item subscale, and second, based
pleted treatment, 50 had primary surgery ? radiotherapy on a 9-item version of the subscale, excluding items 41
and 45 had primary concomitant chemoradiotherapy. All and 42 related to smoking and alcohol consumption. The
on treatment patients were receiving combination che- 9-item scoring is based on conceptual grounds and sup-
moradiotherapy (including: 1500 centigray (cGy)/week, ported by psychometric assessment of the internal con-
hydroxyurea, 5-fluorouracil 2 cisplatin for 5 cycles). sistency (9-item HNS a = 0.70; ll-item HNS a = 0.63).
Global performance ratings, based on Karnofsky perfor- Investigators are encouraged to employ the revised HNS
mance status,”’ were available for the group of 101 Uni- of FACT and the revised PSS-HN normalcy of diet scale.
versity of Chicago patients. Median Karnofsky score was
80 and the majority of patients demonstrated reasonably Sensitivity and Differentiating Known Groups
good overall functioning. Selected and more detailed pa- The FACT-H&N subscale and total scores, and the PSS-
tient demographics and disease characteristics are pre- HN subscale scores were able to differentiate patients
sented in Table 1. according to overall performance status (Karnofsky) and
treatment status (on treatment vs. off treatment at least
FACT-H&N and PSS-HN Scores 3 months).
Overall there was considerable variability on Normalcy Overall performance status was limited to the group
of Diet, some variability on Eating in Public, and little on of 101 University of Chicago patients for whom Karnof-
Understandability of Speech. Median scores and percent s k p ratings had been made at the same time as the
scoring at or below 50 were: Normalcy of Diet: 50, 56%; QL assessment. Patients were dichotomized into good
Eating in Public: 100, 28%; and Understandability of performance versus poor performance based on a Kar-
Speech: 100, 13%. High scores on the speech subscale nofsky median split (median: 80, defined as able to work,
are likely attributable to sample selection. Approximately normal activity with effort, and some signs or symptoms
two-thirds of the group were on, or had recently com- of disease).
pleted, chemoradiotherapy protocols, regimens which Figure 1 presents the distribution of PSS-HN subscale
generally have little impact on speech; most surgically scores (diet, speech and eating in public) by Karnofsky
treated patients were more than 1 year post treatment global performance status. All subscales significantly differ-
and, as needed, had acquired alternative speech mecha- entiated patients with good global functioning from those
nisms, e g , electrolarynx or transesophageal puncture. with poor global functioning and differences were in ex-
Fifty-one patients (all on treatment) were in the hospital pected directions; patients with lower Karnofsky ratings
at the time of assessment and thus were not scored on showing greater impairment in head and neck specific areas
the eating in public scale. Based in part on data from this of functioning. The greatest difference was in the types of

TABLE 3
FACT-H&N Mean Scores (Standard Deviation) by Overall Performance Status
Karnofskyperformance
status n Physical
Good performancea 58 22.8 (5.6)
Poor performanceb 42' 17.9 (5.6)
P value < 0.0001
Sample excludes 50 patients from Lama Linda as there was no performance status measure available. FACT-H&N Functional Assessment of Cancer Therapy-Head and Neck; n. number
' Karnofsky 80 or greater.
" liarnofsky less than 80.
' One patient missing FACT-H&N data.
" Nine-item head and neck subscale
foods patients were eating (diet scale).Although the median
speech score was 100 in both groups, 83% of the patients
in the good performance group were understandable all of
Ihe time (scored 100) compared with only 53% of the pa-
tients in the poor performance group.
Table 3 presents mean FACT-H&N subscale and total
scores by Karnofsky performance status. With the excep-
tion of the emotional well-being subscale, all FACT-
H&N subscales, and total FACT-G significantly distin-
bgished good performance from poor performance pa-
tients, again in expected directions. While there was
considerable variability in the magnitude of mean differ-
ences, except for relationship with doctor, all appeared
clinically meaningful (2-3 point difference on 5-7-item
subscales)?4
All patients (n = 151) were grouped according to
treatment status: on-treatment versus off-treatment at
least 3 months. Treatment status comparisons are pre-
sented in Figure 2 (PSS-HN differences) and Table 4
(FACT-H&Ndifferences). Treatment groups did not differ
in ethnicity or sex, they did, however, differ on age (60
vs. 55, P = 0.01); off-treatment patients were older. Age
was treated as a covariate in subsequent analyses.
Normalcy of Diet and Understandability of Speech
SIgnificantly differentiated patients by treatment status.
(Fig. 2). For diet, differences were in the expected direc-
tion with off-treatment patients more likely to be eating
solid foods (higher scores). In contrast, speech scores
were higher for on-treatment patients. This finding may
be attributable to a potential sample selection bias.
Whereas all on-treatment patients were receiving chemo-
radiotherapy, the off treatment group included surgically
treated patients who likely had greater speech distur-
bance. Finally, although the distribution of eating in pub-
lic scores is suggestive of treatment status differences, the
on treatment group includes only 5 of 56 on treatment
patients; all other on treatment patients were in hospital
arid eating in public was not relevant.
Table 4 presents FACT-H&Nsubscales and total score
by treatment status. FACT-G (total), physical well-being,
PSS-HN and FACT-H&N: Utility and Validity/List et al. 2297
Relationship
Functional Social with physician Emotional HNS~ FACT-G
19.3 (5.9) 21.7 (6.1) 7.3 (1.2) 17.4 (2.8) 21.7 (6.1) 90.1 (14.5)
13.8 (5.6) 13.5 (4.9) 6.8 (1.5) 16.4 (2.8) 13.5 (4.9) 74.9 (15.3)
< 0.0001 < 0.0001 0.05 0.08 < 0.0001 < 0.0001
functional well-being and head and neck subscales were
significantly higher (better) in the group of off-treatment
patients. In order of magnitude, the largest mean differ-
ences were on the head and neck subscale (mean: 6.7),
physical well-being (4.31, and functional well-being (3.9).
Although between group differences on relationship with
doctor reached statistical significance, the magnitude of the
difference (0.4) may not be clinically meaningful.
Correlations Among Measures
Convergent and discriminant validity were evaluated by
correlating FACT-H&N subscales with PSS-HN subscales
(Spearman correlations are presented in Table 5). Ex-
pected coefficient magnitudes were driven by the princi-
ple of construct (convergent vs. divergent) ~ a l i d i t y ? ~ . ' ~
High correlation coefficients were expected between sub-
scales measuring similar constructs or areas of function-
ing, and lower correlations were expected between sub-
scales measuring dissimilar constructs. Consistent with
expectations, the highest correlations were between the
PSS-HN Normalcy of Diet and Eating in Public subscales
and the HNS of FACT-H&N (r = 0.66, P < 0.0001; r =
0.42, P < 0.0001, respectively). Both the PSS-HN scale
and the HNS were developed to assess the specific and
unique concerns and symptoms experienced by head and
neck cancer patients. The Normalcy of Diet scale was
also modestly but significantly correlated with the 2 other
FACT subscales measuring functioning, that is, physical
(r = 0.29, P < 0.0001) and functional (r = 0.31, P = 0.0001)
well-being. In contrast, diet was not correlated with social
or emotional well-being or relationship with doctor.
Eating in Public was modestly but significantly correlated
with physical (r = 0.22, P < 0.05) and social (r = 0.22, P
< 0.05) well-being, and Understandability of Speech with
social well-being (r = 0.16, P < 0.05). In all instances
correlations were in expect directions, better perfor-
mance associated with better QL. The generally low corre-
lations between Understandability of Speech and FACT
subscales may be attributable, in part, to the limited
range of scores on the speech scale and/or the fact that
2298 CANCER June 1,1996 / Volume 77 / Number 11
FIGURE 1. Performance status shown (0Good Karnofsky; Poor
Karnofsky).
speech is the only subscale rated solely from the perspec-
tive of the rater (no patient input). Ratings based on dif-
ferent methods of assessment will, under equivalent other
circumstances, produce lower coefficients of association
than ratings based on similar methods of assessment."
Karnofsky and PSS-HN
Since both Karnofsky and the PSS-HN provide data about
patients' performance status, these ratings were com-
pared to determine the degree of overlap. A modest de-
gree of association between the scales was evidenced by
moderate correlation coefficients (Karnofsky and: diet, r
= 0.56, P < 0.0001; speech, r = 0.34, P < 0.0001; eating
 PSS-HN and FACT-H&N: Utility and Validity/List et al. 2299

TABLE 4
FACT-H&N Mean Scores (StandardDeviation)bv Treatment Status
Relationship
Treatment status n Physical Functional Social with physician Emotional HNS FACT-6

Off- treatment 95 23.2 (4.9) 19.3 (5.7) 21.3 (5.91 7.3 (1.2) 16.9 (2.9) 21.3 (6.9) 88.1 (15.7)
On-treatment 56 18.9 (6.1) 15.3 (6.1) 21.9 (4.9) 6.9 (1.3) 16.8 (2.9) 15.6 (6.4) 79.7 (15.1)
P value < 0.0001 < 0.0001 0.61 0.05 0.84 < 0.0001 0.003

FI\CT-H&NFunctional Assessment of Cancer Therapy-Head and Neck n: number; MD: ; HNS: Head and Neck Scale.
The off treatment group was significantly older than the on treatment group (59.9 yrs vs. 54.9 ys,P = 0.011; age was used as a coviiate in the above analyses.

TABLE 5
Spearman Correlations between PSS-HN Subscales and FACT-H&N Subscale and Total Scores

Relationship
Diet Speech Physical Functional Social with physician Emotional HNSa FACl'-6

Diet 0.29' 0.3Id 0.07 0.09 0.01 0.66d 0.24'

Speech 0.23' -0.02 -0.04 0.16b -0.08 -0.02 0.17b 0.01
E!ating in Public 0.47d 0.4Id 0.22b 0.10 0.22b -0.12 -0.04 0.42' 0.2Ob

Nine-item scale; excludes items 41 and 42 related to smoking and alcohol.

P f:0.05.
' P < 0.005.
P < 0.0001.

in public, r = 0.31, P < 0.05) and significant PSS-HN DISCUSSION

differences between groups based on Karnofsky cut-off This study supports the validity and clinical application
scores (described above). However, the PSS-HN also pro- of the PSS-HN and the FACT-H&N. Both scales are reli-
vides independent, unique information as demonstrated able when applied to patients with head and neck cancer,
by the range of scores (and thus levels of functioning) on and both are sensitive to differences in functioning, e.g.,
the PSS-HN scores within Karnofsky groups. For example, based on overall performance and treatment status. Al-
in the group of patients with overall good performance though significantly related to Karnofsky performance
status (80 or more on the Karnofsky), 43% were unable status, treatment status, and each other, both the FACT-
to eat solid foods ( 5 50 on the diet scale), and 13% ate H&N and PSS-HN offer unique QL data which can con-
only in selected places and only with selected people ( 5 tribute meaningful information to the care of patients
50 on eating in public scale). with head and neck cancer. Data have relevance to pa-
Patients were then dichotomized into high versus low tient preparation and education, rehabilitation, and eval-
PSS-HN scores (> 50 vs. 5 50) and groups were examined uation of treatment impact and efficacy.
for differences in Karnofsky scores. There were no sig- The data presented here represent only the second"'
nificant differences in mean or median Karnofsky ratings published report on the FACT HNS. The HNS was de-
between groups categorized by the diet scale or the eating signed to assess disease and treatment related experi-
in public scale. For example, patients eating only liquid ences of head and neck cancer patients. The subscale
or soft foods did not differ on overall functioning (ability proved sensitive to differences among these patients on
to work, live independently) from patients eating solid the basis of treatment status and overall performance
foods. There were significant differences, however, when status, and was strongly correlated with PSS-HN. It clearly
patients were grouped by speech (> 50 vs. 5 50). Speech adds information to that collected by the parent (core)
is likely to have more of an impact on day to day activities, instrument, the FACT-G.
including employment status, and thus, shows greater The pattern of correlations between FACT-H&N and
association to Karnofsky ratings. Considered together, the PSS-HN subscales is consistent with expectations based
data confirm that compared with Karnofsky, the PSS-HN on presumed underlying scale constructs and thus fur-
is more sensitive to the unique problems of head and thers supports the validity of the measures. In addition
neck cancer patients and provides independent and addi- to strong correlations among subscales measuring similar
tional information. functional arenas, the PSS-HN measures of speech and
2300 CANCER June 1,1996 / Volume 77 / Number 11
eating in public were associated with social well-being
but not with emotional well-being or relationship with
doctor. The Eating in Public subscale was intended as a
measure of socialization and, thus, its association with
the FACT social well-being subscale is encouraging.
In contrast to the high sensitivity of the FACT physi-
cal, functional, and social well-being subscales to differ-
ences in performance status, there was no significant re-
lationship between the emotional well-being subscale
and performance status. Such a finding is reasonable
from the standpoint of construct validity; performance
status is a physical activity measure and thus, would be
expected to relate more strongly to other physical, rather
than mental measures. Conversely, many studies have
found significant, if somewhat modest relationships be-
tween physical and mental well-being. This relationship
has been demonstrated in general medical samples,2’
breast cancer patientsz8and lung cancer patients.” Using
the same measure as employed in the current study
(FACT-GI, Cella et al.“ found a significant association
between physical and emotional well-being in a group of
465 cancer patients with mixed diagnosis.
In light of contradictory data, one is compelled to
question the lack of relationship between patient activity
level (performance status) and self-reported emotional
well-being in the sample of head and neck cancer patients
described here. Based in part on clinical impression, the
authors suggest that this group of patients may have a
relatively blunted reporting of distress. This hypothesis is
supported by the fact that, when compared with other
groups of cancer patients, the current sample score
higher on the emotional well-being subscale, that is, they
report less d i ~ t r e s s . ” , Alternatively,
’~,~~ it may be that dis-
tress in this cohort of patients is determined by factors
other than physical ability and activity, factors such as
past alcohol use, social support, and/or other personal
demographic characteristics. Further examination of
these issues is warranted.
One concern with the FACT-H&N in this sample are
the relatively low internal consistency (a)coefficients for
the social and emotional well-being subscales. These “re-
liability” coefficients (a = 0.59) (Table 2) are actually best
understood as indicators of the unidimensionality of the
presumed underljing (latent) trait being measured by the
set of questions. Although 0.70 has been suggested as the
minimum value of a~ceptability,~’ many experts avoid
making definitive recommendations and it is not unusual
to find published scales with lower a?’ It is well recog-
nized that the a coefficient is highly dependent upon test
length, with brief, patient friendly scales at a disadvan-
tage. In addition, group comparison validity can be
achieved with relatively low a coefficients, i.e., 0.50 to
0.70 range.”’ Indeed, this circumstance was observed in
the current data; the social well-being scale was sensitive
to performance status, treatment status, and the PSS-HN
eating in public scale, despite its “low” coefficient of uni-
dimensionality. In contrast, the emotional well-being
subscale which had a similar a (0.59) coefficient, did not
show sensitivity to known group differences (discrimi-
nant validity).
As discussed above, this difference may be related to
the apparent distinctiveness of this patient population
with respect to self-reported emotional status. In other
groups of cancer patients, a coefficients for the emotional
subscale are consistently higher: a = 0.74 for mixed can-
cer diagnoses;“ a = 0.69 for lung cancer patient^;'^ and
a = 0.72 to 0.85 for various samples of HIV-infected pa-
. ~ ~ current data represent the only known a
t i e n t ~ The
coefficients derived from a sample consisting solely of
head and neck cancer patients. Thus, while further repli-
cation is needed, it may be that patients with head and
neck cancer are an atypical population with respect to
self-reported emotional health. Caution is therefore rec-
ommended in interpreting data from the emotional well-
being subscale in head and neck cancer patients, particu-
larly in individual patient assessment applications.
In conclusion, both the FACT-H&N and the PSS-HN
are reliable, valid, and easily administered tools which
can be used to describe performance status and QL of
head and neck cancer patients. Designed as a general QL
measure for cancer patients, the FACT provides informa-
tion about multiple dimensions and scores can be com-
pared with those of other patient groups, at the same
time as it provides for the addition of disease specific
items. The one caution about interpretation of the FACT
or other self-report emotional health measures in head
and neck cancer patients, appears to be the somewhat
unusual and complicated relationship between mental
and physical health in this group. Further examination
of this issue is underway. As a disease specific functional
measure, the PSS-HN provides greater detail about the
unique functional impairments of head and neck pa-
tients. Both measures can be administered repeatedly
over time with little patient or health care provider bur-
den. Together they provide independent and comple-
mentary information critical to the longitudinal evalua-
tion of treatment and rehabilitative regimens as well as
the early identification of patients at risk for persistent
dysfunction.
REFERENCES
Shapiro PA, Kornfeld DS. Psychiatric aspects of head and
neck surgery. Psych Clin N Am 1987; 10:87- 100.
Rapoport Y, Kreitler S, Chaitchik S, Algor R, Weissler K. Psy-
chosocial problems in head-and-neck patients and their
change with time since diagnosis. Ann OncoZ1993;4:69-73.
Natvig K. Laryngectomies in Norway. Study No. 5: problems
of everyday life. J Otolaryngol 1983;13:15-22.
Silverman S. Oral cavity toxicity. In: John MI, Flam MS,
Legha SS, Phillips TL, Editors. Chemoradiation: an Inte-
grated Approach to Cancer Treatment. Philadelphia: Lea &
Febiger, 1993:511-23.

5. Humphreys B, Mathog R, Miller P, Rosen R, Muz J, Nelson
R. Videofluoroscopic and scintigraphic analysis of dysphagia
in the head and neck cancer patient. Laryngoscope 1987;
9725-32.
ti. Teichgraeber J , Browman J, Goepfert H. Functional analysis
of treatment of oral cavity cancer. Arch Otolaryngol Head
Neck Surg 1986;112959-65.
7. Bjordal K, Kaasa S, Mastekaasa A. Quality of life in patients
treated for head and neck cancer: a follow-up study 7 to 11
years after radiotherapy. Int J Oncol Biol Phys 1994;28:847-
56.
8. Aaronson NK, Ahmedzai S , Bullinger M, Crabeels D, Estape
J, Filiberti A, et al. The EORTC core quality of life question-
naire: interim results of an international field study. In:
Osoba D, Editor. Effect of Cancer on Quality of Life. Boston:
CRC Press, 1991:185-203.
9. Aaronson NK, Ahmedzai S, Bergman B, Bullinger M, Cull A,
Duez NJ, et al. The European Organization for Research and
Treatment of Cancer QLQ-C30: a quality of life instrument
for use in international clinical trials in oncology. J Natl
Cancer Inst 1993;85:365-76.
10 List MA, Ritter-Sterr C, Lansky SB. A performance status
scale for head and neck cancer patients. Cancer 1990;
66:564-9.
11. 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. I Clin
Oncol 1993;11:570-9.
12. Browman GP, Levine MN, Hodson DI, Sathya J, Russell R,
Skingley P, et al. The head and neck radiotherapy question-
naire: a morbiditylquality of life instrument for clinical trials
of radiation therapy in locally advanced head and neck can-
cer. J Clin Oncol 1993;11:863-72.
13. Hassan SJ, Weymuller EA. Assessment of quality of life in
head and neck cancer patients. Head Neck 1993;15:485-96.
14. Niezgoda HE, Parer JL. A validation study of the domains of
the core EORTC quality of life questionnaire. Qual Life Res
1993;2:319-25.
15. List MA, Ritter-Sterr C, Baker T, Colangelo L, Matz G, Pau-
loski BR, et al. A longitudinal assessment of quality of life
in laryngeal cancer patients. Head Neck 1996;18:l-10.
16. List MA, Mumby P, Haraf D, Siston A, Malone D, Mick R, et
al. Quality of life and performance outcome in survivors
of concomitant chemoradiotherapy protocols for head and
neck cancer. Proc A m Soc Clin Oncol 1995;14297.
17. Scott CB, Stetz J, Bruner DW, Wasserman TH. Radiation
PSS-HN and FACT-H&N: Utility and Validity/List et al. 2301
Therapy Oncology Group quality of life assessment: design,
analysis, and data management issues. Qual Life 1994;
3: 199- 206.
18. D'Antonio LL, Zimmerman GJ, Cella DF, Long S. Quality of
life and functional status measures in patients with head
and neck cancer. Arch Otol Head Neck Surg. in press.
19. Cella DF. Manual for the Functional Assessment of Cancer
Therapy (FACT) Measurement System (version 3). Chicago:
Rush Medical Center, 1994.
20. Winer BJ, Brown DR, Michels KM. Statistical Principles of
Experimental Design. 3rd ed. New York McGraw-Hill, 1991.
21. Hollander M, Wolfe DA. Nonparametric Statistical Methods.
1st ed. New York: John Wiley & Sons, 1973.
22. 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, 1949:199-205.
23. Cella DF. Manual for the Functional Assessment of Cancer
Therapy (FACT) Measurement System (version 2). Chicago:
Rush Medical Center, 1993.
24. Cella DF, Bonomi AE, Lloyd SR, Tulsky DS, Kaplan E, Bo-
nomi P. Reliability and validity of the Functional Assessment
of Cancer Therapy-Lung (FACT-L)quality of life instrument.
Lung Cancer 1995;12:199-220.
25. Cronbach LJ, Meehl PE. Construct validity in psychological
tests. Psycho! Bul 1955:52:281-302.
26. Campbell D, Fiske D. Convergent and discriminant validity
by the, multitrait-multimethod matrix. Psycho1 Bul 1959;
56:81- 105.
27. Hays RD, Stewart AL. The structure of self-reported health
in chronic disease patients. Psych Assess 1990;2:22-30.
28. Ganz PA, Hirji K, Sim MS, Schag CA, Fred C, Polinsky MI.
Predicting psychosocial risk in patients with breast cancer.
Med Care 1993;31:419-31.
29. Cella DF, Orofiamma B, Holland JC, Silberfarb PM, Tross S,
Feldstein M, et al. Relationship of psychological distress,
extent of disease, and performance status in patients with
lung cancer. Cancer 1987;60:1661-7.
30. Nunnally JC. Psychometric Theory. 2nd ed. New York
McGraw Hill, 1978.
31. DeVellis RF. Scale Development: Theory and Applications.
Newbury Park: Sage Publications, 1991.
32. Helmstadter GC. Principles of Psychological Measurement.
New York Appleton-Century-Crofts, 1964.
33. McCain N, Cella DF. Correlates of stress in HIVdisease. West
J Nurs Res 1995;17:141-55.