The DCCT Research Group

Reliability and Validity

of a Diabetes Quality-of-Life
Measure for the Diabetes Control
and Complications Trial (DCCT)

We have developed a diabetes quality-of-life (DQOL) complex shifts in patterns of morbidity, asymptomatic
measure oriented toward the patient with insulin- changes in physiologic measures, treatment side effects,
dependent diabetes mellitus (IDDM). The DQOL was and differential burdens on patient and family life-styles.
assessed for its reliability and validity in a group of In these clinical trials, patient perception of impact or
patients with IDDM (n = 192). We found that the DQOL satisfaction with level of well-being and disability is a
and its four scales had high degrees of internal
consistency (Cronbach's r = .66-.92) and excellent test-
particularly important consideration in the final evalu-
retest reliability (r = .78-.92). Using conceptually ation of treatment. Indeed, these perceptions may be
relevant measures of psychiatric symptoms, perceived critical in determining whether patients adhere to such
well-being and adjustment to illness, we also treatments.
demonstrated convergent validity of the DQOL. This The Diabetes Control and Complications Trial (DCCT)
instrument was initially designed for use in the Diabetes is a controlled, randomized clinical trial comparing the
Control and Complications Trial, a multicenter efficacy of two different treatment regimens on the ap-
controlled clinical trial evaluating the effects of two pearance and progression of chronic complications of
different diabetes treatment regimens on the
insulin-dependent diabetes mellitus (IDDM). The ration-
appearance and progression of early vascular
complications. However, the DQOL may also be useful ale for and design of the DCCT have been presented in
in evaluating the quality of life in other groups of detail (8). In brief, patients entering this trial are ran-
patients with IDDM. Diabetes Care 11:725-32, 1988 domly assigned to one of two treatment groups, standard
diabetes treatment or an experimental treatment regi-
men requiring extensive patient reeducation and effort
to maintain blood glucose levels as near as possible to
the levels of people without diabetes while minimizing

Q
uality of life, along with more traditional hypoglycemia. The experimental regimen requires self-
physiologic measures of health status, has monitoring of blood glucose four times per day and either
been recognized increasingly as a useful cri- three or more daily injections of insulin or use of an
terion for evaluating medical outcome (1-3). insulin pump; the doses are adjusted on the basis of
As a consequence, quality-of-life assessment blood glucose monitoring results and anticipated diet
has emerged as an important approach to evaluate the and exercise. Experimental group subjects are also re-
effects of alternative medical treatment strategies (1-7). quired to make frequent contact with health providers
The utility of quality-of-life assessment has become es- to carry out these tasks. Although standard treatment in
pecially apparent in the evaluation of treatments for the DCCT is demanding, the experimental treatment
chronic illnesses, where outcomes typically involve regimen is even more so. Quality-of-life-related out-
comes are being assessed to compare the relative per-
sonal burden of participation in the two treatment groups
Prepared by A. Jacobson, I. Barofsky, P. Cleary, and L. Rand for the DCCT over the anticipated 10-yr duration of the study.
Research Croup. A complete list of the DCCT Research Croup appeared in Many measures have been developed to assess quality
Diabetes Care 10:1-19, 1987.
Address correspondence and reprint requests to Box NDIC/DCCT, Bethesda, of life (4-6). The strategies underlying these measures
MD 20892. vary considerably. Because of growing interest in com-

DIABETES CARE, VOL. 11, NO. 9, OCTOBER 1988 725

DIABETES QUALITY-OF-LIFE MEASURE

paring and contrasting the functional health status of cruited. The sample consisted of 136 adults (age 18-41
individuals across illness groups, measures such as the yr) and 56 adolescents (age 13-17.9 yr). (Demographic
Sickness Impact Profile (5) and the Quality of Weil-Being information was not obtained on 2 adults due to clerical
Scale (6) have been developed to study the general pop- error.) Of the 190 subjects with complete data, 114 were
ulation. Investigators planning to evaluate patients in male and 76 were female. The mean duration of dia-
clinical trials have suggested that such strategies may be betes was 8 yr. There was 1 subject with >15 yr of
insensitive to small but clinically meaningful changes IDDM (23 yr). The sample consisted primarily of Cau-
induced by treatment (1,2,9). In response to this con- casian subjects from Hollingshead social classes I—IV
cern, many trials have proposed the use of measures (A. Hollingshead, unpublished observations). Forty per-
specific to the goals and end points of the trial. To ac- cent of the adult patients were married.
complish this, multiple measures of functional health Measures. The DQOL is a multiple-choice assessment
status may be required. Each trial may have important designed for ease of administration and for use with both
differences, thereby requiring the development of trial- adolescents and adults. The items do not identify spe-
oriented measures to be used in conjunction with pre- cific types of treatment (e.g., insulin pump) or self-mon-
viously tested assessments (1,2,7). The development of itoring (urine or blood testing) so that it is applicable to
new measures is required when quality-of-life assess- patients using different methods of diabetes manage-
ment has not been performed previously with a partic- ment. It has four primary scales (satisfaction, impact,
ular group of patients. We faced this issue in planning diabetes worry, social/vocational worry) with 46 core
an assessment for IDDM patients in the DCCT. Follow- items for use in all patients. The APPENDIX presents the
ing strategies used by other investigators, we identifed items for each scale. Patients are asked to rate the com-
previously used measures of some end points of concern mon issues of having diabetes in terms of their current
to the trial—psychological distress (10) and cognitive functioning. Because the questions are posed from three
function (8). However, no measures were available that perspectives (the impact generated by diabetes, patient
addressed patient perception of impact and satisfaction satisfaction with him/herself, and worry about antici-
with specific features of diabetes treatment. Thus, we pated effects of diabetes), the DQOL can be considered
elected to develop an additional psychosocial index ori- as a battery of related subtests. Each subtest or scale
ented toward diabetes. The Diabetes Quality-of-Life assesses quality of life from a different vantage point.
(DQOL) measure was designed to address patient-per- Responses to questions are made with a 5-point Likert
ceived personal burden of the trial. It was designed for scale. Satisfaction is rated from 1 (very satisfied) to 5
use in the DCCT but with applicability to a wider range (very dissatisfied). Impact and worry scales are rated
of patients with IDDM. In this article, we summarize from 1 (no impact and never worried) to 5 (always im-
the development of the DQOL and present an evalua- pacted and always worried). In addition to the core items,
tion of specific aspects of is reliability and validity. there are 16 questions included in the DQOL that assess
schooling experience and family relationships for pa-
tients living with their parents. These auxiliary items can
be used to provide additional information about the life
MATERIALS AND METHODS experiences of adolescent patients. In this article, we
present the reliability and validity data for the total DQOL
Subjects. To examine the reliability and validity of the and the core items of the four scales that comprise the
DQOL, a study was undertaken in a sample of conven- DQOL.
tionally treated patients with IDDM similar to the DCCT The strategy for scale development is similar to that
patient group in age, duration of diabetes, and level of
early vascular complications. Each of the 21 participat-
ing centers submitted a list of 40 non-DCCT patients TABLE 1
who could be subjects for the DQOL evaluations. Ten Characteristics of sample
subjects were randomly selected from each of the lists,
and 192 of those 210 subjects consented to participate Adults Adolescents
(n = 134) (n = 56)
in the study. Patients from all 21 centers participated.
The entry criteria for the DCCT and the DQOL study Mean age (yr) 28 ± 7 16 ± 1
population were the following: 7) age >13 and <40 yr; Sex (% male) 60 59
2) pubertal development at or beyond Tanner stage II Social class (%)
(11,12); 3) duration of IDDM >1 and <15 yr; 4) treat- I 41 20
ment with one or two injections of insulin per day; 5) II 21 14
<130% ideal body weight; 6) for adolescent subjects III 14 24
(age 13-1 7 yr), no history of failure to maintain normal IV 20 37
growth and development during the previous 2 yr; 7) V 4 4
Mean duration of IDDM (yr) 8.5 ± 4.3 6.4 ± 3.0
generally good health, without advanced complications
Race (% Caucasian) 93 91
of diabetes. Marital status (% married) 40 0
Table 1 shows the characteristics of the sample re-

726 DIABETES CARE, VOL. 11, NO. 9, OCTOBER 1988

 THE DCCT RESEARCH GROUP

TABLE 2
Between-group comparisons of scores on DQOL scales* for adults and adolescents

Adult males Adult females Adolescent males Adolescent females

(n == 82) (n == 54) (n - 33) (n == 23)

Mean SD Mean SD Mean SD Mean SD

Total DQOL score 2.0 0.4 2.1t 0.4 1.8 0.4 2.0 0.4
Satisfaction 2.0 0.6 2.2 0.6 2.0 0.5 2.0 0.6
Impact 2.0 0.4 2.It 0.4 1.9 0.4 2.0 0.5
Worry: diabetes related 1.9 0.6 2.2 0.6 1.6 0.6 1.9 0.7
Worry: social/vocational 1.9 0.7 2.0 0.7 1.5 0.5 1.7 0.7

*Five-point Likert Scale; see MATERIALS AND METHODS.

tP < .05; adult males vs. females and adolescent males vs. females made with two-tailed t tests.

recommended by Guyatt et al. (9). The items that make measure overlap, each test was selected to assess spe-
up the DQOL were derived from the following sources: cific DQOL scales.
7) review of the literature on diabetes to identify typical The SCL is the most recent revision of the Hopkins
concerns of individuals with diabetes and problems that Symptom Checklist. It is a widely used measure of psy-
impact on their lives (14,15); 2) the clinical experience chiatric symptoms, has nine scales, and has been sub-
of health professionals knowledgeable about the treat- jected to multiple assessments of reliability and validity
ment of diabetes; and 3) patients who have IDDM. Un- (10,17). Psychiatric symptoms were assessed because
like other quality-of-life measures, the DQOL includes of their expected relationship to the diabetes worry and
worry scales because concerns or worries have been social/vocational worry scales. We anticipated the SCL
described as an important way that diabetes can influ- global severity index would be most strongly correlated
ence the patient and family (15,16). During its devel- to the worry scales compared with the ABS and PAIS
opment, drafts of the DQOL were given to IDDM scales. The PAIS distress scale provides a second inde-
patients for review as to meaning, relevance, and read- pendent indicator of psychological upset. Thus, we ex-
ability. Drafts were also reviewed by experienced dia- pected the SCL and PAIS distress scales to have similar
betologists and diabetes nurses who were members of correlations with the worry scale.
the DCCT research group. With this process, an initial The ABS, developed by Bradburn (18), consists of 10
item pool was culled to select items of most relevance items that measure positive and negative aspects of psy-
to patients with IDDM undergoing treatments of differ- chological well-being. It has been used in various large-
ing intensity. scale population surveys and, although not devoid of
To determine the validity of the DQOL, three assess- limitations, has been recommended as the best available
ment instruments were selected because they measure indicator of well-being in general population surveys
related aspects of quality of life, thereby providing an (19). The ABS was used in this study because it is a
estimate of convergent validity. These were the Symp- conveniently administered assessment that provides both
tom Checklist-90-R(SCL), the Bradburn Affect Balance a global and a derived (affect-balance) estimate of well-
Scale (ABS), and the Psychosocial Adjustment of Illness being or satisfaction. Well-being or satisfaction is a well-
Scale (PAIS). Although the constructs assessed by each recognized dimension along which quality of life is reg-

TABLE 3
Between-group comparisons of scores on three scales for adults and adolescents

Adult males Adult females Adolescent males Adolescent females

(n == 82) (n == 54) (n == 33) (n == 23)

Mean SD Mean SD Mean SD Mean SD

SCL T score
Global severity index 56.5 14.9 54.7 12.7 41.1 11.0 50.5* 15.1
PAIS raw score
Vocational environment 1.4 1.6 1.9 2.2 1.4 1.7 1.8 1.6
Domestic environment 1.8 2.2 2.4 2.6 1.9 2.3 2.7 3.0
Affect Balance Scale 3.6 2.8 3.0 2.7 3.3 2.8 2.7 3.6

SCL T score, Symptom Checklist (standardized transformation); PAIS, Psychological Adjustment to Illness Scale.
*P < .05, Student's two-tailed t test for adult males vs. females, adolescent males vs. females.

DIABETES CARE, VOL. 11, NO. 9, OCTOBER 1988 727

DIABETES QUALITY-OF-LIFE MEASURE

TABLE 4 Statistical analysis. Two forms of reliability were as-

Intercorrelation among DQOL scales sessed for the total score and the four scales of the DQOL:
7) internal consistency with Cronbach's a (22) and 2)
Worry: test-retest by use of Pearson product-moment correla-
Satisfaction Impact diabetes related
tions. Convergent validity was assessed by correlational
Adults analyses of DQOL total and scale scores with the SCL,
Satisfaction ABS, and PAIS. Because many statistical tests were per-
Impact .60* formed, only those correlations with P < .0001 were
Worry: diabetes related .26 .58* considered significant. The t tests (2-tailed) used for
Worry: social/vocational .37 .52* .68* comparisons between sexes were considered significant
Adolescents with the conventional P < .05.
Satisfaction
Impact .64*
Worry: diabetes related .27 .54*
Worry: social/vocational .34 .41 .54* RESULTS

* P < .0001. Psychosocial measures: descriptive findings. The

scores for the DQOL were examined by sex and for
adolescents and adults separately. Table 2 displays find-
ularly assessed and is one of the principal components ings of these analyses. Among the adults, the DQOL
of the DQOL. The ABS is scored on a scale of - 1 0 to mean total and individual scale scores showed that sub-
+ 10, with +10 representing the highest satisfaction. jects were generally satisified and not worried and that
We expected that the ABS would be most strongly as- diabetes made only a modest impact on their lives. There
sociated with the DQOL satisfaction scale. were two small but significant differences between sexes
The PAIS assesses the personal impact of an illness on such that adult males reported less of an impact of di-
different aspects of an individual's life. The PAIS has abetes and fewer diabetes-related worries than adult fe-
seven scales that evaluate health-care orientation; so- males. Among adolescents, the DQOL scores were
cial, vocational, and domestic environments; extended slightly more positive, with adolescent males and fe-
family and sexual relationships; and psychological dis- males showing no difference in self-reported quality of
tress. Recent studies suggest favorable properties of re- life.
liability and validity (20,21). The PAIS is available in Table 3 depicts results of selected scales from the SCL,
interview and self-report formats. The self-report version PAIS, and ABS for adults and adolescents. Among the
was used in this study (20,21). We anticipated that the adults, both males and females typically had mean scores
PAIS scales, with the exception of the distress scales, that were not suggestive of impairment. For example,
would correlate most closely with the DQOL impact the ABS scores showed a positive sense of well-being in
scale. both males and females. In addition, SCL T scores ranged
In summary, we expected that each of the three mea- from 51.1 to 58.4, just above the 50th percentile for
sures would show significant correlations with the over- general populations. The only difference between adult
all DQOL score and, because the DQOL scales are not males and females was found on the PAIS psychological
independent, would be positively correlated with each distress scale, with males being significantly less dis-
of these scales. We anticipated that these correlations tressed than females (P < .01). No other gender differ-
would typically fall in the range of .3-.7. This range ences were found on the SCL, PAIS, and ABS.
was anticipated because there is no measure that exactly Adolescents also reported a generally positive sense
evaluates quality of life. However, we expected varying of well-being and adjustment, with adolescent males
patterns of correlation of each DQOL scale with the showing somewhat better adjustment than females. SCL
measures of validity. The satisfaction scale was expected T scores for males were in the 38.6-44.7 range, whereas
to have particularly strong correlations with the ABS score
compared with the other measures. The impact scale
was expected to have especially strong correlations with TABLE 5
the ABS scale and the PAIS scales (except for the PAIS Internal consistency of satisfaction, impact, and worry
distress scale). The worry scales were anticipated to be subscales for adults and adolescents
most strongly linked to the SCL and PAIS distress scales. Cronbach's a
Procedures. The subjects were seen in two sessions. In Number
the first session, they were given the DQOL, SCL, ABS, of items Adults Adolescents
and PAIS in random order. Approximately 1 wk later,
the DQOL was administered again to the same subjects. DQOL total 46 0.92 0.92
The 1 -wk interval was selected with the expectation that Satisfaction 15 0.88 086
quality of life would be unlikely to change in that period Impact 20 0.77 0.85
4 0.67 0.66
but that any shorter period of time would elicit responses Worry: diabetes related
Worry: social/vocational 7 0.83 0.87
that were recalled from the prior testing.

728 DIABETES CARE, VOL. 11, NO. 9, OCTOBER 1988

 THE DCCT RESEARCH GROUP

TABLE 6 Table 6 shows the test-retest results. The Pearson prod-

Test-retest correlations of DQOL total score and satisfac- uct-moment correlations for both adults and adolescents
tion, impact, and worry scales were consistently in the .80-.90 range (P < .0001).
Validity. The product-moment correlations of the DQOL
Adults Adolescents
total and scale scores with the SCL, ABS, and the PAIS
DQOL total 0.92 0.92 scores for the adult subjects are depicted in Table 7. The
Satisfaction 0.89 0.86 satisfaction and impact scales and the DQOL total score
Impact 0.89 0.89 consistently showed significant correlations with the SCL,
Worry: diabetes related 0.80 0.88 ABS, and PAIS scales in the expected direction. The
Worry: social/vocational 0.78 0.88 diabetes worry and social/vocational scales were sig-
nificantly related only with the SCL and with the psy-
P < .0001, according to Pearson product-moment correlations. chological distress subscale of the PAIS.
Table 8 shows the DQOL correlations in the adoles-
cent sample. The satisfaction and impact scales and the
those for females were in the 47.0-52.4 range. The DQOL total scores had several significant correlations
t tests comparing adolescent males and females on with the SCL, ABS, and PAIS scales. As in the adult
the SCL scales revealed that males reported significantly sample, the worry scales were most strongly correlated
fewer symptoms on six of the nine scales and on the with the measures of psychological distress.
global severity index. There were no sex differences on
the PAIS or ABS.
Relationships between DQOL scales. Table 4 shows
the intercorrelations between the core DQOL scales DISCUSSION
among adults and adolescents. As expected, the strong-
est associations were found between the impact and

T
his study supports the reliability and validity of
satisfaction scales and between the two worry scales. the DQOL measure developed for use in the
Impact of diabetes appeared to be more consistently DCCT. Specifically, we found that the DQOL
associated with both diabetes and social/vocational scales had high degrees of internal consistency,
worries than was the satisfaction scale. as measured by Cronbach's a. There was also strong
Internal consistency. Table 5 shows the results of the evidence for the test-retest reliability of the measure.
analysis of the DQOL for internal consistency. The find- Using conceptually relevant measures of well-being (ABS),
ings were similar for both adults and adolescents with psychiatric symptoms (SCL), and adjustment to illness
the satisfaction, impact, social/vocational worry scales, (PAIS), we have shown evidence for convergent validity
and the total DQOL scores showing a high level of in- of the DQOL. Because no measure specifically assesses
ternal consistency. The diabetes worry scale revealed diabetes-related quality of life, we do not have a "gold
somewhat lower levels of internal consistency. standard" to use to assess validity. These theoretically
Test-retest reliability. The mean time interval between relevant measures have provided us with a network of
the two DQOL measurements was 9 days, and the me- evaluations approximating quality of life in a population
dian was 7 days. Eighty percent of the sample took the of patients having IDDM without major complications.
second DQOL within 6-14 days of the first DQOL. Specifically, we found the expected pattern of modestly

TABLE 7
Correlations of DQOL scores of adults with scores of other scales

Worry: diabetes Worry: social/

DQOL total Satisfaction Impact related vocational

SCL
Global severity index .60* .49* .50* .40* .50*
ABSt -.57* -.55* -.47* -.27 -.28
PAIS
Health-care orientation .53* .57* .40* .25 .25
Vocational environment .53* .47* .51* .25 .28
Domestic environment .58* .51* .58* .23 .28
Sexual relationships .35* .33 .40* .08 .06
Extended family relationships .34* .28 .35* .06 .21
Social environment .46* .42* .44* .12 .29
Psychological distress .63* .51* .55* .46* .46*

DQOL, Diabetes Quality-of-Life measure; SCL, Symptom Checklist; ABS, Affect Balance Scale; PAIS, Psychosocial Adjustment to Illness Scale.
* P < .0001.
tBecause of direction of scoring, the negative correlations signify positive relationships between the ABS and DQOL.

DIABETES CARE, VOL. 11, NO. 9, OCTOBER 1988 729

DIABETES QUALITYOFLIFE MEASURE

TABLE 8
Correlation of DQOL scores of adolescents with scores of other scales

Worry, diabetes Worry: social/

DQOL total Satisfaction Impact related vocational

SCL
Global severity index .77* .59* .66* .49 .66*
ABSt -.67* -.60* -.54* -.25 -.55*
PAIS
Health-care orientation .74* .80* .65* .24 .26
Vocational environment .44 .30 .34 .39 .26
Domestic environment .72* .62* .68* .29 .44
Sexual relationships* .36 .23 .25 .35 .53
Extended family relationships .38 .30 .34 .08 .38
Social environment .59* .52* .54* .35 .29
Psychological distress .81* .61* .71* .59* .67*

DQOL, Diabetes Quality-of-Life measure; SCL, Symptom Checklist; ABS, Affect Balance Scale; PAIS, Psychosocial Adjustment to Illness Scale.
* P < .0001.
tBecause of direction of scoring, the negative correlations signify positive relationships between the ABS and DQOL.
*Small sample (n = 20) responded.

strong and consistently positive correlations of these of medical treatments have employed a battery of tests
measures with the overall DQOL score. to capture a wide range of issues relevant to the person's
We anticipated that there would be different patterns functional health status. For example, Bombardier et al.
of relationship of the DQOL scales with each of the (7), in evaluating the effects of auranofin, an anti-inflam-
measures. These assumptions were partially supported. matory agent, on rheumatoid arthritis, included assess-
For example, we found that the worry scales were most ments of perceived well-being, general health, and in-
strongly linked to the distress-oriented measures, i.e., dices specific to arthritis (e.g., pain). Similarly, the DQOL
the SCL global severity index and the psychological dis- may be useful as part of a battery of measures selected
tress scale of the PAIS. However, we found that these to capture general health concerns and specific areas
distress indices were also consistently associated with relevant to diabetes. We have used this approach in the
patient perceptions of satisfaction with and impact of DCCT. Because it has been shown that the experimental
diabetes. This suggests that the impact and satisfaction treatment regimen increases the frequency and severity
scales may serve as broader gauges of diabetes-related of hypoglycemia and, in turn, may possibly be associ-
quality of life, whereas the worry scales are more spe- ated with impairments in cognitive function (23), a spe-
cific to level of patient-perceived psychological distress cific set of neurobehavioral tests is incorporated as one
or symptomatology. This view of the impact and satis- of the psychosocial outcome measures (8). The battery
faction scales is further supported by the similar patterns also incorporates the SCL-90-R, which measures psy-
of association these two scales showed with the PAIS chiatric symptoms (17), to assess possible effects of pro-
and ABS scales. Use of the PAIS helped identify specific longed experimental treatment on the emotional func-
areas of personal experience that are most closely tied tioning of subjects. Evaluation of DCCT patients with
to adolescent perceptions of their quality of life. Health- the DQOL and SCL after 1-yr follow-up has indicated
care environment, domestic relationships, and psycho- no negative effects of experimental treatment (23).
logical distress were more closely linked to quality of This study of the reliability and validity of the DQOL
life than sexual and extended family relationships. has certain limitations. We have relied primarily on the
The DQOL diabetes-related worry scale showed a use of self-report measures to provide initial evidence
lower internal consistency score than did the impact and regarding the validity of the DQOL. We might have used
satisfaction scales. Although this internal consistency samples of patients with different levels of medical status
score was relatively lower, it was actually just below (e.g., patients with visual impairments vs. those with-
the generally acceptable range in scale development. out). We elected not to use such heterogeneous samples
The lower level of internal consistency among worry in order to approximate characteristics of subjects under
items may reflect a tendency of patients to focus on a study in the DCCT. We also could have utilized inter-
particular concern, thereby exhibiting relatively less worry view-based measures of patient status but believed that
about other issues. Another possible explanation is that this was not feasible for the first evaluation of the DQOL.
the diabetes-related worry scale consisted of only 4 items, The approach to development and assessment of the
whereas the satisfaction scale had 15 items, the impact psychometric properties of the DQOL is very similar to
scale 20 items, and the social/vocational worry scale 7 the procedures recommended by Guyattet al. (9). Clearly,
items. an important next step in the process of validation should
Recent studies evaluating quality of life as outcomes include patients with divergent functional health char-

730 DIABETES CARE, VOL. 11, NO. 9, OCTOBER 1988

 THE DCCT RESEARCH GROUP

acteristics. Subsequent studies should also address the 3. How satisfied are you with the time it takes to de-
predictive validity of the DQOL. Because the validation termine your sugar level?
of any measure is a repetitive process of successive ap- 4. How satisfied are you with your current treatment?
proximations, we expect that other studies that employ 5. How satisfied are you with the flexibility you have
the DQOL will add to our understanding of the psy- in your diet?
chometric properties of the measure. 6. How satisfied are you with the burden your dia-
The choice of study subjects who approximated the betes is placing on your family?
characteristics of the DCCT study patients allowed us to 7. How satisfied are you with your knowledge about
draw conclusions about the particular merits of using your diabetes?
the DQOL in the DCCT. Because of the eligibility cri- 8. How satisfied are you with your sleep?
teria of this study, the sample is relatively homogeneous 9. How satisfied are you with your social relationships
with a lower variance on psychosocial variables than is and friendships?
likely to be found in a less selected population of pa- 10. How satisfied are you with your sex life?
tients with IDDM. Even so, our findings provide support 11. How satisfied are you with your work, school, and
for the reliability and validity of the DQOL for the DCCT. household activities?
Our ability to identify relationships between the DQOL 12. How satisfied are you with the appearance of your
and the other psychosocial measures in this homoge- body?
neous sample also suggests the potential value of the 13. How satisfied are you with the time you spend ex-
DQOL in assessing patients with a wider range of clin- ercising?
ical characteristics. The ease of administration and use 14. How satisfied are you with your leisure time?
of items that address relevant concerns without refer- 15. How satisfied are you with life in general?
ence to specific treatment strategies also make the DQOL
broadly applicable to diverse groups of diabetic pa- Impact
tients. Thus, the DQOL may be useful in various clinical
settings. It could be used as a screening measure to 1. How often do you feel pain associated with the
detect unstated or unaddressed concerns of patients about treatment for your diabetes?
their diabetes. Responses to specific items could provide 2. How often are you embarrassed by having to deal
cues for further discussions between patients and their with your diabetes in public?
medical provider about personal experiences of diabetes 3. How often do you have low blood sugar?
and treatment. Increasing provider understanding of the 4. How often do you feel physically ill?
patient's perspective can improve the therapeutic alli- 5. How often does your diabetes interfere with your
ance and the patient's participation in treatment (15). family life?
Therefore, DQOL may be applicable in clinical settings 6. How often do you have a bad night's sleep?
as well as in other trials of new diabetes treatments. 7. How often do you find your diabetes limiting your
social relationships and friendships?
8. How often do you feel good about yourself?
ACKNOWLEDGMENTS 9. How often do you feel restricted by your diet?
10. How often does your diabetes interfere with your
The DCCT is supported by the Division of Diabetes, sex life?
Endocrinology, and Metabolic Diseases of the National 11. How often does your diabetes keep you from driv-
Institute of Diabetes and Digestive and Kidney Diseases, ing a car or using a machine (e.g., a typewriter)?
NIH, through cooperative agreements and a research 12. How often does your diabetes interfere with your
contract. Additional support or technical assistance has exercising?
been provided by the National Institute of Neurologic 13. How often do you miss work, school, or household
and Communicative Disorders and Stroke; the National duties because of your diabetes?
Heart, Lung, and Blood Institute; the National Eye In- 14. How often do you find yourself explaining what it
stitute; and the Division of Research Resources, NIH. means to have diabetes?
David M. Nathan is chairman of the DCCT Editorial 15. How often do you find that your diabetes interrupts
Board. your leisure-time activities?
16. How often do you tell others about your diabetes?
17. How often are you teased because you have dia-
betes?
APPENDIX 18. How often do you feel that because of your diabetes
you go to the bathroom more than others?
Satisfaction
19. How often do you find that you eat something you
1. How satisfied are you with the amount of time it shouldn't rather than tell someone that you have
takes to manage your diabetes? diabetes?
2. How satisfied are you with the amount of time you 20. How often do you hide from others the fact that
spend getting checkups? you are having an insulin reaction?

DIABETES CARE, VOL. 11, NO. 9, OCTOBER 1988 731

DIABETES QUAUTY-0F-L1FE MEASURE

Worry: Social/Vocational ness Impact Profile: development and final revision of a

health status measure. Med Care 19:787-805, 1981
1. How often do you worry about whether you will 6. Bush JM, Kaplan RM: Health-related quality of life mea-
get married? surement. Health Psychol 1:61-80, 1982
2. How often do you worry about whether you will 7. Bombardier C, Ware J, Russell I, Larson M, Chalmers A,
have children? Read J: The Auranofin Cooperating Group: auranofin
therapy and quality of life in patients with rheumatoid
3. How often do you worry about whether you will
arthritis: results of a multicenter trial. Am j Med 81:565-
not get a job you want? 78, 1986
4. How often do you worry about whether you will 8. DCCT Research Group: Diabetes Control and Compli-
be denied insurance? cation Trial (DCCT): design and methodological consid-
5. How often do you worry about whether you will erations for the feasibility phase. Diabetes 35:530-45,
be able to complete your education? 1986
6. How often do you worry about whether you will 9. Guyatt G, Bombardier C, Tugwell P: Measuring disease-
miss work? specific quality of life in clinical trials. Can Med Assoc I
7. How often do you worry about whether you will 134:889-95, 1986
be able to take a vacation or a trip? 10. Derogatis LR, Rickels K, Rock A: The SCL-90-R. Admin-
istration, Scoring and Procedures Manual I. Baltimore,
Worry: Diabetes Related MD, Clin. Psychom. Res., 1977
11. Marshall WA, Tanner JM: Variations of pubertal patterns
How often do you worry about whether you will in girls. Arch Dis Child 44:291-303, 1969
pass out? 12. Marshall WA, Tanner JM: Variations of pubertal patterns
in boys. Arch Dis Child 45:13-23, 1970
How often do you worry that your body looks dif-
14. Sullivan BJ: Adjustment of diabetic adolescent girls: de-
ferent because you have diabetes? velopment of the Diabetic Adjustment Scale. Psychosom
How often do you worry that you will get compli- Med 41:119-26, 1979
cations from your diabetes? 15. Jacobson AM, Hauser ST: Behavioral and psychological
How often do you worry about whether someone aspects of diabetes. In Diabetes Mellitus. Theory and Prac-
will not go out with you because you have diabe- tice. 3rd ed. Ellenberg M, Rifkin H, Eds. New York, Med.
tes? Exam., 1983
16. Banion CR, Miles MS, Carter MC: Problems of mothers
in management of children with diabetes. Diabetes Care
6:548-51, 1983
REFERENCES 17. Derogatis LR, Rickels K, Rock A: The SCL-90 and the
MMPI: a step in validation of a new self-report scale. Br
1. Croog S, Levine S, Tecta M, Brown B, Bulpitt C, Jenkins ) Psychiatry 128:280-89, 1976
C, Klerman G, Williams G: The effects of antihypertensive 18. Bradburn NM: The Structure of Psychological Well-Being.
therapy on the quality of life. N Engl / Med 314:1657- Chicago, IL, Aldine, 1969
64, 1986 19. McDowell I, Praught E: On the measurement of happi-
2. Mackenzie C, Carlson M, Digioia D, Kelley K: A patient ness: an examination of the Bradburn Scale in Canada
specific measure of change in maximal function. Arch Health Survey. Am J Epidemiol 116:949-58, 1982
Intern Med 46:1325-29, 1986 20. Derogatis LR: Psychosocial Adjustment to Illness Scale
3. Evans R, Manninen D, Garrison L, Hart G, Blagg C, Gut- {PAIS and PAIS-SR). Scoring, Procedures and Administra-
man R, Hull A, Lowrie E: The quality of life of patients tion Manual I. Baltimore, MD, Clin. Psychom. Res., 1983
with end-stage renal disease. N Engl I Med 312:553-40, 21. Derogatis LR: The Psychosocial Adjustment of Illness Scale
1985 (PAIS). / Psychosom Res 30:77-91, 1983
4. Brook RH, Ware RE Jr, Davies-Avery A, Stewart AL, Don- 22. Nunnally J: Introduction to Psychological Measurement.
ald CA, Rogers WH, Williams KN: Overview of adult New York, McGraw-Hill, 1970
health status measures fielded in Rand's Health Insurance 23. The DCCT Research Group: Diabetes Control and Com-
Study. Med Care 17 (Suppl. 7):1-131, 1979 plications Trial (DCCT): results of feasibility study. Dia-
5. Bergner M, Bobbitt RA, Carter WB, Gibson BS: The Sick- betes Care 10:1-19, 1987

732 DIABETES CARE, VOL. 11, NO. 9, OCTOBER 1988