Clinical Hub,UW Health Clinical Tool Search,UW Health Clinical Tool Search,Questionnaires,Related

Development and Validation of a Scale to Measure Symptoms of Chronic Graft vs Host Disease

Development and Validation of a Scale to Measure Symptoms of Chronic Graft vs Host Disease - Clinical Hub, UW Health Clinical Tool Search, UW Health Clinical Tool Search, Questionnaires, Related

Chronic graft-versus-host disease (cGVHD) is a seri-
ous late complication of allogeneic stem cell transplanta-
tion and the leading cause of nonrelapse death more than
2 years after the procedure [1-6]. Thirty percent to 90%
of surviving allogeneic transplant recipients develop
cGVHD, which is associated with decreased quality of
life (QOL) [7], impaired functional status [5,8], continued
need for immunosuppressive medication [6], and
increased mortality [1-6]. A higher incidence of cGVHD
is observed when patients are older, when immunosup-
pressive medications instead of T-cell depletion are used
for acute GVHD prophylaxis, or when unrelated, mis-
matched, or peripheral blood grafts serve as the stem cell
source [9-16]. Nonmyeloablative procedures are becoming
more common, but the comparable rates of acute GVHD
in patients treated with nonmyeloablative and myeloabla-
tive procedures suggest that rates of cGVHD may also be
similar [17-19].
No validated measures capturing the symptom burden
of cGVHD have been published. Observational physician-
reported data suggest that approximately 65% to 85% of
patients with cGVHD have skin involvement, 60% have
mouth involvement, 40% to 55% have liver involvement
(although this is usually asymptomatic for patients), 25% to
45% have eye involvement, 20% to 30% have nutritional
problems, and 10% to 15% have lung manifestations [20].
Thus, we developed a patient self-administered symptom
scale reflective of the multiorgan manifestations of cGVHD
based on a cohort accrued from 1998 to 2000. To be useful
in following the QOL of patients with cGVHD enrolled in
clinical studies, the symptom scale had to be sensitive to
cross-sectional and longitudinal severity of cGVHD. We
hypothesized that direct measurement of patient QOL and
Development and Validation of a Scale to Measure
Symptoms of Chronic Graft-versus-Host Disease
Stephanie J. Lee,
E. Francis Cook,
Robert Soiffer,
Joseph H. Antin
Department of Adult Oncology, Dana-Farber Cancer Institute;
Department of Medicine, Brigham and
Women’s Hospital, Harvard Medical School, Boston, Massachusetts
Correspondence and reprint requests: Stephanie Lee, MD, MPH, Center for Outcomes and Policy Research,
Dana-Farber Cancer Institute, 44 Binney St, Boston, MA 02115 (e-mail: stephanie_lee@dfci.harvard.edu).
Received April 1, 2002; accepted May 10, 2002
Chronic GVHD (cGVHD) affects 30% to 90% of surviving allogeneic transplant recipients. Thus far, no quality-of-
life instruments have been developed to measure the effect of this common complication of allogeneic transplanta-
tion on patients’ functioning and well-being. Using a prospective cohort of 107 patients with active cGVHD who
completed the symptom scale at enrollment and at intervals of 3 and 6 months, we developed a 30-item symptom
scale with 7 subscales to capture the cGVHD-specific symptom burden. The symptom scale correlated highly with
patients’ self-assessed mild, moderate, and severe cGVHD manifestations in cross-sectional analysis. Reliability
(Cronbach’s α = 0.79-0.90), test-retest (r
= 0.28-0.93), and convergent and discriminant validity compared to the
Medical Outcomes Study Short Form 36 (SF-36) and Functional Assessment of Chronic Illness Therapy with BMT
subscale (FACT-BMT) were assessed and found to be adequate. Longitudinal assessments showed that changes in
overall health status correlated best with changes in quality of life as measured by the SF-36 and FACT-BMT. In
contrast, changes in cGVHD severity were best detected by changes in the symptom scale. We recommend that
either the SF-36 or the FACT-BMT be combined with a cGVHD-specific symptom scale to measure the impact of
cGVHD on patients’ quality of life and that this endpoint be included in clinical trials testing cGVHD interventions.
The cGVHD symptom scale is a short, simple, and valid measure of cGVHD manifestations and can be used to fol-
low complication-specific symptoms using patient self-administered questionnaires.
Chronic graft-versus-host disease

Stem cell transplantation

Quality of life

Biology of Blood and Marrow Transplantation 8:444-452 (2002)
2002 American Society for Blood and Marrow Transplantation

Chronic GVHD Symptom Scale
symptom burden offers the most sensitive means of follow-
ing the clinical course of patients with cGVHD.
Patients with active cGVHD following allogeneic stem
cell transplantation at the Dana-Farber Cancer Institute and
the Brigham and Women’s Hospital Stem Cell Transplant
Program were identified through physician review of patient
lists every 3 months. Most patients were diagnosed clinically
without histopathologic confirmation. Potentially eligible
patients were mailed a cover letter, consent form, baseline
self-administered questionnaire, opt-out card, and self-
addressed stamped envelope in which to return their surveys.
If neither a survey nor an opt-out card was received within
4 weeks of the initial mailing, a second mailing was sent. If
no response was received within 3 weeks of the second mail-
ing, the patient was contacted by phone to confirm receipt of
study materials. Both incident and prevalent cases were
enrolled in the cohort between 1998 and 2000. The Institu-
tional Review Board approved the study protocol, and all
patients provided signed, informed consent for participation.
Patients who had developed cGVHD in the 3 months
preceding study enrollment were classified as newly diag-
nosed and asked to complete questionnaires every 3 months
for the first year and every 6 months thereafter. Patients
who were diagnosed with cGVHD more than 3 months
prior to enrollment were designated as established and were
surveyed every 6 months. The more intensive survey sched-
ule for newly diagnosed patients reflected our hypothesis
that the greatest changes in disease activity occur in the ini-
tial period after diagnosis.
Data Collection Instruments and Methods
The Short Form 36 (SF-36) [21,22], the Functional
Assessment of Chronic Illness Therapy with bone marrow
transplantation subscale (FACT-BMT) [23], and a checklist
of 42 potentially bothersome symptoms of cGVHD were
mailed to patients every 3 to 6 months. At each time point,
patients were also asked to provide self-assessment of their
current Karnofsky performance status (KPS), current level
of overall health (excellent, very good, good, fair, poor),
cGVHD severity (mild, moderate, or severe), and cGVHD
trajectory over the last 6 months (improved, stable, wors-
ened). If a survey or opt-out card was not returned within 3 to
4 weeks, a second cover letter and packet were mailed. If a
response was not received after an additional 4 weeks,
patients were contacted by phone. Sociodemographic infor-
mation was collected at baseline.
The SF-36 is a validated, generic, QOL instrument
measuring 8 subscales (physical, role physical, pain, general
health, vitality, social functioning, role emotional, and men-
tal health) and 2 summary scales (physical and mental). The
FACT-BMT is a validated, cancer-specific QOL instrument
with a transplantation-specific subscale designed to address
additional areas of concern to transplant recipients. This sur-
vey is composed of 4 core domains (physical, social, emotional,
and functional) and the transplantation-specific subscale
(BMT module) that result in a summary measure (FACT-
BMT) when combined. The checklist of 42 symptoms was
developed based on review of the literature and discussion
with transplantation physicians, nurses, and patients. The
symptom checklist was pilot tested on 10 patients with
cGVHD for ease of completion, clarity, and comprehensive-
ness of symptoms. Patients were asked to report whether or
not they had certain symptoms and rate how bothersome
they were on a 6-point Likert scale: “symptom not present,”
“not bothered at all,” “slightly bothered,” “moderately both-
ered,” “quite a bit bothered,” and “extremely bothered.” Test
results from 15 healthy volunteers suggested appropriate
interval scaling for response categories.
Medical records were reviewed for clinical information
by one physician (S. J. L.) who did not have knowledge of
patients’ survey responses. Data regarding type and degree of
organ involvement and clinical severity of cGVHD (limited/
extensive) [24] were abstracted from clinical staff notes that
had been recorded within 6 weeks of each survey comple-
tion date. This restriction was intended to ensure that notes
accurately reflected medical status at the time of the QOL
assessment. Grades were assigned based on data available in the
clinic notes, Eastern Cooperative Oncocology Group (ECOG)
performance status, cGVHD severity (mild/moderate/severe),
and overall health (on a 5-point Likert scale: excellent, very
good, good, fair, or poor). Mild cGVHD was defined as “signs
and symptoms of cGVHD do not interfere substantially
with function and do not progress once appropriately
treated with local therapy or standard systemic therapy
(steroids and/or cyclosporine or tacrolimus).” Moderate
cGVHD was defined as “signs and symptoms of cGVHD
interfere somewhat with function despite appropriate ther-
apy or are progressive through first-line systemic therapy
defined as steroids and/or cyclosporine or tacrolimus.”
Severe cGVHD was defined as “signs and symptoms of
cGVHD limit function substantially despite appropriate
therapy or are progressive through second-line therapy.”
These definitions have not been independently validated.
Information on relapse and death was obtained from the
clinical transplantation database maintained at the Dana-
Farber Cancer Institute.
Biostatistical Methods
Patient characteristics were reported descriptively. Sub-
scale and summary scores for the SF-36 and FACT-BMT
were calculated according to recommended methods for
handling missing data and scaling responses [21-23]. Sum-
mary scores on the SF-36 were normalized so that 50 was
the mean for the general population, with a standard devia-
tion of 10. On both the SF-36 and the FACT-BMT, a
higher score indicated better functioning.
The QOL and symptom scores of patients who reported
mild, moderate, or severe overall cGVHD symptoms at
enrollment were compared using general linear models.
Spearman correlations were calculated between the cGVHD
symptom scores and domains of the SF-36, FACT-BMT,
other patient-reported measures, and medical chart review.
Correlation coefficients were graphed (Figures 1 and 2) with
the largest dot representing strong correlation and the
smallest dot reflecting lack of correlation. Survival was
calculated from the time of enrollment in the cohort. Cox
proportional hazards modeling was used to evaluate the pre-
dictive ability of patient self-assessed severity of cGVHD at

S. J. Lee et al.
the time of enrollment into the cohort. P < .01 was consid-
ered significant because of multiple testing.
Generation and Psychometric Testing of the
Symptom Score
Factor analysis with a promax rotation was used to
reduce the 42 items on the symptom scale to 30 items. The
categories of “no symptoms” and “symptoms, but not both-
ered at all” were combined. Twelve items were eliminated
through factor analysis: dry mouth, diarrhea, ability to con-
centrate, nausea, memory loss, anorexia, hair loss, headache,
pain, numbness, mouth pain, and sexual difficulties. A score
was calculated for each factor by taking the mean of all
items completed if more than 50% were answered and nor-
malizing to a 0 to 100 scale. A summary score was similarly
calculated considering all the subscales. A higher score indi-
cated more bothersome symptoms.
Enrollment surveys were used for the majority of
the psychometric testing. Internal consistency was reported
as Cronbach’s α. Test-retest reliability was evaluated in
22 patients who were sent the symptom scale 2 to 4 weeks
after completing their enrollment surveys. Interclass corre-
lations of the symptom scale were calculated. Convergent
and discriminant validity were determined by correlation
with clinical data abstracted from the medical record, con-
current patient self-report on the validated instruments
(SF-36 and FACT-BMT), and patient self-assessed severity
of cGVHD (mild/moderate/severe).
The sensitivity of the QOL instruments to changes in
overall health and cGVHD severity was explored [25-29].
Because serial surveys were collected, we enriched for peri-
ods of change using the following algorithm: (1) for each
patient, a search was made to identify the first instance that
he/she reported a change in cGVHD or overall health so
that a 6-month change score could be calculated; or (2) for
patients who did not report any improvement or worsening
over the 24 months of possible observation, the first 2 sur-
veys 6 months apart were used to calculate change scores.
Patients were represented only once in each analysis so that
observations were independent. Correlation coefficients
were calculated between 6-month instrument change scores
and changes in overall health or cGVHD severity (catego-
rized as improved, stable, or worsened). Effect sizes were
calculated as 6-month change scores (follow-up minus base-
line) divided by the standard deviation of the baseline score
for each group. Effect sizes were reported so that a positive
value indicated an improvement in symptoms, whereas a
negative value indicated worsening symptoms.
Patient Characteristics
A total of 158 eligible patients were mailed baseline sur-
veys, and 127 (80%) of 158 returned the surveys. Of the
31nonrespondents, 15 opted out, 6 died within 2 months of
survey mailing, and 10 did not return a survey or opt-out
Figure 1. Spearman correlations between the cGVHD symptom scale and the SF-36 and FACT-BMT domains.

Chronic GVHD Symptom Scale
card despite 2 mailings and phone contact. Of the 127 respon-
dents, 20 indicated on their baseline surveys that they did
not have cGVHD. These patients were excluded from fur-
ther analysis. Thus, the cGVHD study population consisted
of 107 patients. Eighty-three patients (78%) were married
or living with a partner, and 63 (59%) had a college or post-
graduate degree. Twenty-seven percent of patients were
working full time, 15% were working part time, 7% were
homemakers, 21% were disabled, and 15% were on medical
leave. Self-assessed KPS was 80% to 100% in 62% of
patients and 70% or less in 31% of patients; 7% of patients
had missing data.
Eleven patients had developed cGVHD following donor
lymphocyte infusion for relapse (and were in remission at
the time of baseline survey completion). These patients
were considered to be without current evidence of disease
[30]. Six patients relapsed with their original malignancy
after enrollment in the cohort. Tables 1 and 2 summarize
additional clinical characteristics of the study population.
Figure 2. Spearman correlations between the cGVHD symptom scale and patient self-assessment and medical chart review.
Table 1. Patient Characteristics*
Variable n = 107
Age, median (range), y 40 (20-60)
Male, n (%) 58 (54)
Race, n (%)
White 97 (91)
Non-white 10 (9)
Married or living with partner, n (%) 83 (78)
College or postgraduate degree, n (%) 63 (59)
Disease, n (%)
AML/ALL 25 (23)
CML 48 (45)
MDS 11 (10)
MM 10 (9)
NHL/HD/CLL 10 (9)
Other 3 (3)
Disease stage, n (%)
Early 68 (64)
Intermediate 31 (29)
Advanced 8 (7)
Donor type, n (%)
Related 69 (64)
Unrelated 38 (36)
Acute GVHD prophylaxis, n (%)
T-cell depletion 37 (35)
Immune suppressive medications 70 (65)
Year of transplantation, n (%)
1998-2001 42 (39)
1996-1997 27 (25)
1994-1995 15 (14)
<1994 23 (22)
*AML indicates acute myeloid leukemia; ALL, acute lymphoblastic
leukemia; CML, chronic myeloid leukemia; MDS, myelodysplastic
syndrome; MM, multiple myeloma; NHL, non-Hodgkin’s lymphoma;
HD, Hodgkin’s disease; CLL, chronic lymphocytic leukemia.
Table 2. Chronic GVHD Characteristics
Incident/prevalent cases, n (%)
Newly diagnosed 44 (41)
Established diagnosis 63 (59)
Severity of cGVHD at enrollment, n (%)
Patient self-assessed
Mild 55 (51)
Moderate 39 (36)
Severe 13 (12)
Chart review
Limited 35 (33)
Extensive 49 (46)
Could not be determined 23 (22)
Current work status, n (%)
Working full time 29 (27)
Working part time 16 (15)
Homemaker 7 (7)
On medical leave 16 (15)
Disabled 22 (21)
Other 17 (16)
Self-reported KPS, n (%)
80%-100% 66 (62)
<70% 33 (31)
Missing 8 (7)
Follow-up of survivors, median (range), y 1.8 (0-2.8)

S. J. Lee et al.
Response rates at each time point varied from 78% to
93% of survivors. Specifically, response rates were 38 (93%) of
41 patients at 3 months, 74 (88%) of 84 patients at 6 months,
22 (81%) of 27 patients at 9 months, 52 (78%) of 67 patients
at 12 months, 46 (87%) of 53 patients at 18 months, and
34 (83%) of 41 patients at 24 months after study enrollment.
Note that only newly diagnosed patients were asked to com-
plete forms at 3 and 9 months, accounting for the smaller
Development of the Symptom Scale
Through factor analysis, 12 questions were deleted from
the original 42-item symptom survey. Seven subscales with
2 to 7 items each were derived, representing domains of
skin, eye, mouth, lung function, nutrition, psychological sta-
tus, and energy. Subscale scores could be calculated for 97%
to 98% of patients. Based on pilot testing with 10 patients
prior to the study, we estimated that it took 20 to 30 minutes
to complete the entire battery of questionnaires (141 ques-
tions). Therefore we estimated that the final 30-item symptom
scale took approximately 5 minutes to complete.
Internal consistency was high (Cronbach’s α, 0.79-0.85)
and test-retest reproducibility was good for all subscales
(0.74-0.93) except psychological status (0.55) and lung
symptoms (0.28). Intercorrelations were moderate to high
for the energy, skin, nutrition, lung function, and psycho-
logical subscales but very low for eye and mouth symptoms.
Additional psychometric properties of the subscales are
shown in Table 3.
Convergent Validity. Figures 1 and 2 show the degree of
correlation between the symptom subscales and QOL as
measured by the SF-36 and FACT-BMT, patient self-
assessment of cGVHD severity, KPS, and overall health and
physician assessment as obtained by chart review. Consistent
with prior hypotheses, the energy factor correlated most
closely with physical domains, whereas the psychological
scale correlated best with the emotional and mental
domains. Correlation with patient self-assessed overall
cGVHD severity, KPS, and overall health was moderate,
but correlation with information interpreted from the med-
ical record was poor. The eye and mouth subscales were not
correlated with summary measures of QOL.
Discriminant Validity. Table 4 and Figures 3 and 4 show
that the scales of the SF-36, FACT-BMT, and symptom
summary score and the subscales of energy, skin, nutrition,
and psychological symptoms could successfully discriminate
between patients with self-assessed mild, moderate, or
severe cGVHD at baseline. The greatest differences were
seen in physical functioning and in the symptom subscales
representing energy level, skin involvement, and nutritional
status. No differences were seen in the social, emotional, or
mental subscales of the SF-36 and the FACT-BMT or in the
lung, eye, or mouth components of the symptom scale. Dis-
criminant validity was also supported by the lack of correla-
tion between the cGVHD symptom scale and unrelated
domains of the SF-36 and FACT-BMT shown in Figure 1.
No differences in QOL or symptoms were seen when
the 35 patients with well-documented limited cGVHD were
compared to the 49 patients with extensive disease. How-
ever, 23 patients (22%) lacked adequate medical record doc-
umentation to allow classification at the time of study
Responsiveness to Change. During the study, 14 patients
reported worsening cGVHD, 33 improving cGVHD, and
20 stable cGVHD. Changes in cGVHD were not correlated
with cGVHD severity at enrollment (r
= 0.17, P = .16).
Twenty-two patients reported worsening overall health,
21improving health, and 32 stable health. Change in overall
Table 3. Descriptive Statistics of Score Distribution and Intercorrelations for the cGVHD Symptom Scale
Energy Skin Nutrition Lung Psychological Eye Mouth Summary
Items, no. 7 5 5 5 3 3 2 30
Mean 29 18 6 8 18 24 17 17
SD 26 22 14 15 23 29 26 13
Median 25 10 0 0 8 13 0 15
Range 0-93 0-85 0-100 0-95 0-92 0-100 0-100 0-56
Cronbach 0.88 0.81 0.83 0.84 0.79 0.85 0.84 0.90
Floor 17% 33% 67% 63% 43% 39% 61% 3%
Ceiling 1% 2% 1% 1% 1% 2% 2% 1%
Nonresponse 2% 3% 2% 2% 2% 2% 2% 2%
Retest 0.78 0.74 0.83 0.28 0.55 0.87 0.93 0.64
Energy 0.52* 0.47* 0.36† 0.55* 0.21 0.10 .77*
Skin 0.32† 0.28‡ 0.40* 0.06 -0.03 .53*
Nutrition 0.26‡ 0.31‡ 0.20 0.23 .55*
Lung 0.24 0.00 0.14 .42*
Psych 0.23 0.22 .72*
Eye 0.06 .49*
Mouth .42*
* P ≤ .0001.
†P ≤ .001.
‡P ≤ .01.

Chronic GVHD Symptom Scale
health was associated with overall health at enrollment (r
0.47, P = .0001) but not with cGVHD severity (r
= 0.17,
P = .14). In only 58% of cases were identical time points
used for both analyses because of our enrichment algorithm.
Table 5 shows the effect sizes and correlation coefficients of
the change scores for the SF-36, FACT-BMT, and symptom
scales with patient-reported changes in cGVHD severity
and changes in overall health. Changes in overall health
were highly correlated with changes in the subscales of the
SF-36 and FACT-BMT but not with changes in the
cGVHD symptom scale. In contrast, perceived changes in
cGVHD severity correlated with changes detected by the
summary cGVHD symptom scale (r = 0.33, P = .007) but
not by the SF-36 or FACT-BMT.
The standard deviation of the baseline summary
cGVHD symptom score was 12.9. A distribution-based
method based on 0.5 times the standard deviation of the
baseline responses was used to estimate a clinically meaning-
ful difference of 6 to 7 points on the cGVHD symptom
scale [26,27,31].
Severity of cGVHD and Survival
Survival differed between patients with mild versus
moderate or severe cGVHD. Of the 55 patients who rated
their cGVHD as “mild” on their baseline surveys, only
3 (5%) died. In contrast, 9 (23%) of 39 and 5 (38%) of
13 patients in the moderate and severe categories, respec-
tively, died. In Cox proportional hazards modeling, patients
with moderate and severe cGVHD had similar survival
rates, so these subgroups were combined. After a median of
1.8 years of follow-up for survivors, patients with moderate
or severe cGVHD had a relative risk of death of 4.2 (95%
confidence interval, 1.4-12.8; P = .01) compared to patients
with mild cGVHD.
Table 4. SF-36, FACT-BMT, and cGVHD Symptom Scores Based on Mild, Moderate, or Severe Patient Self-Rated cGVHD at Enrollment
Mild, Mean (SD) Moderate, Mean (SD) Severe, Mean (SD)
(n = 55) (n = 39) (n = 13) P
Summary physical 42 (12) 36 (10) 28 (8) .0002
Summary emotional 52 (10) 48 (11) 46 (11) .07
Physical 67 (28) 57 (26) 26 (17) <.0001
Role physical 50 (45) 30 (40) 11 (19) .004
Pain 80 (20) 70 (22) 57 (24) .002
General health 56 (24) 45 (20) 34 (26) .004
Vitality 57 (22) 40 (20) 36 (21) <.0001
Social functioning 75 (28) 66 (23) 51 (30) .01
Role emotional 84 (31) 69 (39) 38 (43) .0003
Mental health 77 (17) 73 (19) 69 (21) .3
Summary FACT-BMT 112 (20) 99 (22) 90 (23) .0006
Physical 22 (5) 19 (5) 16 (7) .0003
Social functioning 22 (5) 20 (6) 23 (4) .1
Emotional 20 (4) 19 (4) 16 (6) .01
Functional 20 (6) 17 (6) 14 (5) .001
BMT module 29 (6) 25 (7) 22 (8) .001
Summary symptoms 12 (9) 18 (13) 34 (13) <.0001
Energy 19 (20) 34 (27) 56 (24) <.0001
Skin 9 (13) 22 (20) 46 (32) <.0001
Nutrition 3 (8) 4 (7) 24 (31) <.0001
Lung 6 (16) 8 (14) 15 (15) .2
Psychological 13 (22) 16 (20) 41 (24) .0004
Eye 19 (25) 27 (33) 30 (36) .3
Mouth 16 (23) 15 (27) 29 (35) .2
Figure 3. Cross-sectional QOL at enrollment for the SF-36 physical
composite scale (PCS), SF-36 mental composite scale (MCS), FACT-
BMT summary scale, and chronic GVHD symptom scale (cGVHD Sx)
according to whether patient reported mild (n = 55), moderate (n = 39), or
severe (n = 13) cGVHD severity. The horizontal bars represent the medi-
ans; the upper and the lower bounds of the boxes represent the 75th and
25th percentiles; and the vertical lines indicate the full range of values.

S. J. Lee et al.
We report the development and validation of a 30-item,
7-subscale symptom scale for patients with cGVHD for
evaluation of adverse effects on skin, vitality, lung, nutri-
tional status, psychological functioning, eye, and mouth.
The psychometric properties of this scale show that reliabil-
ity, convergent and discriminant validity, and sensitivity to
change are adequate to recommend this scale for further
development to assess symptomatology in patients with
cGVHD. Comparisons with standardized QOL measures
show that the symptom scale is more responsive to changes
in patient-perceived cGVHD severity than are generic or
cancer-specific instruments. Thus, we recommend that this
symptom scale be added to one of the validated QOL
instruments to fully capture the symptom burden and trajec-
tory of patients with cGVHD.
With patient self-assessed cGVHD as the gold standard,
the energy, skin, nutritional status, and psychological sub-
scales showed adequate discriminant validity. Energy, skin,
nutrition, and psychological subscales seemed to be driving
cross-sectional cGVHD severity assessments, whereas
changes in cGVHD severity across time showed greater cor-
relation with changes in energy and lung symptoms. In con-
trast, eye, mouth, and lung symptoms, although undeniably
aspects of cGVHD, were not correlated with patient or
physician assessment of cGVHD severity. We hypothesize
several reasons for this lack of sensitivity. First, the symptoms
addressed on the eye subscale are dry eyes, need to use eye
drops frequently, and difficulty seeing clearly, all symptoms
that may result from total body irradiation and cataracts, not
only the sicca syndrome associated with cGVHD. Indeed,
even patients with mild cGVHD reported a high level of
bothersome eye symptoms. Second, cGVHD involvement of
the lung is relatively uncommon, occurring in approximately
10% to 15% of patients. Its rarity may have masked its
importance in determining cGVHD severity in the popula-
tion even though affected patients are highly symptomatic. A
third possibility is that eye and mouth symptoms, although
prevalent and bothersome, may not be viewed by patients as
determinants of cGVHD severity.
We found that patient self-assessment of cGVHD, in
contrast to physician information in the medical record,
was reasonably well correlated with patient self-reports of
other aspects of health. This finding suggests that direct
patient report provides a richer measure of the impact of
cGVHD on QOL and functional limitations and may be a
more sensitive measure of cGVHD activity than physician
Several limitations should be noted. First, we achieved
an 80% participation rate despite several attempts to enroll
patients and a 78% response to the 6-month follow-up
Figure 4. Cross-sectional QOL at enrollment for the domains of the
FACT instrument according to whether patient reported mild (n = 55),
moderate (n = 39), or severe (n = 13) cGVHD severity. The horizontal
bars represent the medians; the upper and the lower bounds of the
boxes represent the 75th and 25th percentiles; and the vertical lines
indicate the full range of values.
Table 5. Sensitivity of Scales to Changes in cGVHD Severity and Overall Health*
Effect Size Effect Size
Worse Stable Improved Worse Stable Improved
Scale and subscale cGVHD cGVHD cGVHD R
P Health Health Health R
n 142033 223221
Summary Physical –0.15 –0.24 0.09 .15 .25 –0.26 0.18 0.21 .26 .03
Summary Emotional –0.13 –0.24 –0.05 .06 .63 –0.59 –0.04 0.25 .34 .004
FACT-BMT –0.23 –0.07 0.05 .16 .21 –0.67 0.11 0.36 .56 <.0001
Summary symptoms –0.35 0.26 0.31 .33 .007 0.20 0.16 –0.04 .10 .41
Energy –0.10 –0.01 0.24 .25 .04 0.00 0.30 –0.03 .01 .94
Skin –0.34 0.55 0.18 .14 .27 0.19 0.23 –0.05 .12 .32
Nutrition 0.02 0.16 0.31 .05 .71 –0.26 0.19 –0.02 .03 .83
Lung –0.41 –0.30 0.28 .26 .04 0.21 0.20 0.08 .00 1.00
Psychological 0.0 0.0 0.14 .07 .59 –0.04 0.03 0.02 .03 .81
Eye –0.21 0.11 –0.17 .01 .97 –0.06 –.035 –0.08 .02 .85
Mouth 0.0 0.21 0.34 .17 .16 0.13 0.23 0.00 .05 .66
*Change scores are divided by standard deviation of baseline scores for each group. SF-36, FACT, and the symptom scale have been adjusted so
that positive effect sizes indicate improvement of symptoms and negative effect sizes indicate worsening.

Chronic GVHD Symptom Scale
survey. Results may not be representative of our entire
cGVHD population, but should have internal validity. Also,
despite efforts to have physicians identify affected patients,
20 (16%) of the 127 patients returning baseline surveys
reported that they did not have cGVHD. This result could
indicate either that patients are unaware they have cGVHD
or that their cGVHD resolved by the time we contacted
them. Without knowledge as to which possibility was the
case, we excluded this group. Second, we would have liked
to correlate more objective medical information and labora-
tory results with QOL and symptom data. However, we did
not mandate concurrent physician visits at the time of sur-
vey completion and, in retrospect, found that many patients
were not seen at our center within 6 weeks of each follow-
up period. In addition, abstraction of a core set of medical
information from clinic notes proved difficult because of
limited documentation. We recommend that future studies
investigating the relationship between self-reported QOL
and physical manifestations of cGVHD collect such data
prospectively and concurrently. Finally, we did not collect
survey data on a control group without cGVHD but other-
wise matched for time since transplantation and other char-
acteristics. In retrospect, having such a group assembled
would have provided valuable comparative information.
These studies are ongoing.
In summary, we recommend that either the SF-36 or
the FACT-BMT be combined with the cGVHD symptom
scale for use in future cGVHD studies when patient QOL
and symptoms are being assessed. Given the significant
impact of cGVHD on patients’ daily experience, we suggest
that QOL should be considered an important endpoint in
any study of cGVHD interventions, and a 6-to 7-point
change on the cGVHD symptom scale from a preinterven-
tion survey should be considered clinically meaningful.
This work was supported in part by the Amy Strelzer-
Manasevit Scholars Program and the National Institutes of
Health Grant No. CA75267-04.
The authors wish to thank Marisa Bueno, BA, and
Christina Caron, BA, for their help with study coordination.
We especially wish to thank our patients for their participation.
1. Sullivan KM, Witherspoon RP, Storb R, et al. Alternating-day
cyclosporine and prednisone for treatment of high-risk chronic
graft-v-host disease. Blood. 1988;72:555-561.
2. Sullivan KM, Witherspoon RP, Storb R, et al. Prednisone and
azathioprine compared with prednisone and placebo for treatment
of chronic graft-v-host disease: prognostic influence of prolonged
thrombocytopenia after allogeneic marrow transplantation. Blood.
3. Wingard JR, Piantadosi S, Vogelsang GB, et al. Predictors of
death from chronic graft-versus-host disease after bone marrow
transplantation. Blood. 1989;74:1428-1435.
4. Loughran TP Jr, Sullivan K, Morton T, et al. Value of day 100
screening studies for predicting the development of chronic graft-
versus-host disease after allogeneic bone marrow transplantation.
Blood. 1990;76:228-234.
5. Duell T, van Lint MT, Ljungman P, et al. Health and functional
status of long-term survivors of bone marrow transplantation:
EBMT Working Party on Late Effects and EULEP Study Group
on Late Effects: European Group for Blood and Marrow Trans-
plantation. Ann Intern Med. 1997;126:184-192.
6. Socie G, Stone JV, Wingard JR, et al. Long-term survival and late
deaths after allogeneic bone marrow transplantation: Late Effects
Working Committee of the International Bone Marrow Trans-
plant Registry. N Engl J Med. 1999;341:14-21.
7. Sutherland HJ, Fyles GM, Adams G, et al. Quality of life fol-
lowing bone marrow transplantation: a comparison of patient
reports with population norms. Bone Marrow Transplant. 1997;
8. Syrjala KL, Chapko MK, Vitaliano PP, Cummings C, Sullivan
KM. Recovery after allogeneic marrow transplantation: prospec-
tive study of predictors of long-term physical and psychosocial
functioning. Bone Marrow Transplant. 1993;11:319-327.
9. Majolino I, Saglio G, Scime R, et al. High incidence of chronic
GVHD after primary allogeneic peripheral blood stem cell trans-
plantation in patients with hematologic malignancies. Bone Mar-
row Transplant. 1996;17:555-560.
10. Storek J, Gooley T, Siadak M, et al. Allogeneic peripheral blood
stem cell transplantation may be associated with a high risk of
chronic graft-versus-host disease. Blood. 1997;90:4705-4709.
11. Urbano-Ispizua A, Garcia-Conde J, Brunet S, et al. High inci-
dence of chronic graft versus host disease after allogeneic periph-
eral blood progenitor cell transplantation: the Spanish Group of
Allo-PBPCT. Haematologica. 1997;82:683-689.
12. Solano C, Martinez C, Brunet S, et al. Chronic graft-versus-host
disease after allogeneic peripheral blood progenitor cell or bone
marrow transplantation from matched related donors: a case-con-
trol study: Spanish Group of Allo-PBT. Bone Marrow Transplant.
13. Scott MA, Gandhi MK, Jestice HK, Mahendra P, Bass G, Marcus
RE. A trend towards an increased incidence of chronic graft-
versus-host disease following allogeneic peripheral blood pro-
genitor cell transplantation: a case controlled study. Bone Marrow
Transplant. 1998;22:273-276.
14. Ustun C, Arslan O, Beksac M, et al. A retrospective comparison
of allogeneic peripheral blood stem cell and bone marrow trans-
plantation results from a single center: a focus on the incidence of
graft-vs.-host disease and relapse. Biol Blood Marrow Transplant.
15. Blaise D, Kuentz M, Fortanier C, et al. Randomized trial of bone
marrow versus lenograstim-primed blood cell allogeneic trans-
plantation in patients with early-stage leukemia: a report from
the Societe Francaise de Greffe de Moelle. J Clin Oncol. 2000;18:
16. Cutler C, Giri S, Jeyapalan S, Paniagua D, Viswanathan A, Antin
JH. Acute and chronic graft-versus-host disease after allogeneic
peripheral-blood stem-cell and bone marrow transplantation: a
meta-analysis. J Clin Oncol. 2001;19:3685-3691.
17. Khouri IF, Keating M, Korbling M, et al. Transplant-lite: induc-
tion of graft-versus-malignancy using fludarabine-based non-
ablative chemotherapy and allogeneic blood progenitor-cell
transplantation as treatment for lymphoid malignancies. J Clin
Oncol. 1998;16:2817-2824.
18. Giralt S, Thall PF, Khouri I, et al. Melphalan and purine analog-
containing preparative regimens: reduced-intensity conditioning
for patients with hematologic malignancies undergoing allogeneic
progenitor cell transplantation. Blood. 2001;97:631-637.

S. J. Lee et al.
19. McSweeney PA, Niederwieser D, Shizuru JA, et al. Hemato-
poietic cell transplantation in older patients with hematologic
malignancies: replacing high-dose cytotoxic therapy with graft-
versus-tumor effects. Blood. 2001;97:3390-3400.
20. Lee SJ, Klein JP, Barrett AJ, et al. Severity of chronic graft-vs.-
host disease: association with treatment-related mortality and
relapse. Blood. 2002;100:406-414.
21. Ware JE, Snow KK, Kosinski M, Gandek B. SF-36 Health Survey:
A Manual and Interpretation Guide. Boston, Mass: The Health
Institute, New England Medical Center; 1993.
22. Ware JE, Kosinski M, Keller SD. SF-36 Physical and Mental
Health Summary Scales: A User’s Manual. Boston, Mass: The
Health Institute, New England Medical Center; 1994.
23. McQuellon RP, Russell GB, Cella DF, et al. Quality of life meas-
urement in bone marrow transplantation: development of the Func-
tional Assessment of Cancer Therapy-Bone Marrow Transplant
(FACT-BMT) scale. Bone Marrow Transplant. 1997;19:357-368.
24. Shulman HM, Sullivan KM, Weiden PL, et al. Chronic graft-
versus-host syndrome in man: a long-term clinicopathologic study
of 20 Seattle patients. Am J Med. 1980;69:204-217.
25. Jaeschke R, Singer J, Guyatt GH. Measurement of health status:
ascertaining the minimal clinically important difference. Control
Clin Trials. 1989;10:407-415.
26. Lydick E, Epstein RS. Interpretation of quality of life changes.
Qual Life Res. 1993;2:221-226.
27. Juniper EF, Guyatt GH, Willan A, Griffith LE. Determining a
minimal important change in a disease-specific quality of life
questionnaire. J Clin Epidemiol. 1994;47:81-87.
28. Osoba D, Rodrigues G, Myles J, Zee B, Pater J. Interpreting the
significance of changes in health-related quality-of-life scores.
J Clin Oncol. 1998;16:139-144.
29. Fischer D, Stewart AL, Bloch DA, Lorig K, Laurent D, Holman H.
Capturing the patient’s view of change as a clinical outcome
measure. JAMA. 1999;282:1157-1162.
30. Craddock C, Szydlo RM, Klein JP, et al. Estimating leukemia-
free survival after allografting for chronic myeloid leukemia: a
new method that takes into account patients who relapse and are
restored to complete remission. Blood. 2000;96:86-90.
31. Sloan J. Asking the obvious questions regarding patient burden.
J Clin Oncol. 2002;20:4-6.
Please let us know whether you have been bothered by any of the following problems in the past month.
Not at all Slightly Moderately Quite a bit Extremely
a. Abnormal skin color 01234
b. Rashes 01234
c. Thickened skin 01234
d. Sores on skin 01234
e. Itchy skin 01234
f. Dry eyes 01234
g. Need to use eyedrops frequently 01234
h. Difficulty seeing clearly 01234
i. Need to avoid certain foods due to mouth pain 01234
j. Ulcers in mouth 01234
k. Receiving nutrition from an intravenous line or feeding tube 01234
l. Frequent cough 01234
m. Colored sputum 01234
n. Shortness of breath with exercise 01234
o. Shortness of breath at rest 01234
p. Need to use oxygen 01234
q. Difficulty swallowing solid foods 01234
r. Difficulty swallowing liquids 01234
s. Vomiting 01234
t. Weight loss 01234
u. Joint and muscle aches 01234
v. Limited joint movement 01234
w. Muscle cramps 01234
x. Weak muscles 01234
y. Loss of energy 01234
z. Need to sleep more/take naps 01234
aa. Fevers 01234
bb. Depression 01234
cc. Anxiety 01234
dd. Difficulty sleeping 01234