Professional Documents
Culture Documents
Results
Seven self-report measures of constipation were identified. The content
areas evaluated by these measures varied. Only two measures had ade-
quate validity and reliability, sensitivity to change, or were tested in
more than one sample.
Conclusions
Findings from this review suggest that the Chinese Constipation Ques-
tionnaire and the Patient Assessment of Constipation-Symptom Ques-
tionnaire demonstrate adequate psychometric properties for a
constipation measure. Additional research is warranted to refine or
develop a more comprehensive self-report measure to evaluate constipa-
tion in adults.
Constipation measures
Stool frequency <3 per week* Less frequent 1–2 times per Number per week 1–2 times per <3 per week*
than normal 1–2 days 1–2 days
to <1 per month to <1 per 2 weeks
Length of time Length of time Length of time
per attempt per attempt per attempt
Stool consistency Hard ⁄ lumpy* Small ⁄ oozing Hard* or small Hard* to pellet like Hard* Hard* or pellety
Loose or watery Soft ⁄ loose ⁄ normal Small Watery
Rectal symptoms Incomplete Incomplete Incomplete Incomplete Incomplete Incomplete Incomplete
evacuation* evacuation* evacuation* evacuation* evacuation* evacuation* evacuation*
+urge to defecate +urge to defecate +urge to defecate Sensation of anorectal +urge to defecate +urge to defecate Strain*
Difficult Pain with BM blockage ⁄ obstruction* Strain*
to pass stools Pain with BM
Pain with BM Strain* Rectal bleeding
Rectal fullness
or pressure
Abdominal Bloating Distention ⁄ bloating Abdominal pain Pain, visits to Bloating Bloating
symptoms provider >6· Abdominal pain Abdominal pain
last year Discomfort in
abdomen stomach
cramps
Assistive measures Digital evacuation Digital evacuation Digital evacuation
of stool* of stool* of stool*
Support vagina
or perineum*
Medications Number of Laxatives ⁄ enemas Laxatives ⁄ enemas ⁄ Laxatives ⁄ enemas
laxatives suppositories
Number of items* Four items One item Two items Five items Three items Three items Four items
congruent
with Rome III
(X items of 6)
Construct
CCQ (7) N = 221; 111 constipated Six items, unidimensional, Literature review, NR Kendall’s DV: reported; Completed Present, sensitivity Cronbach’s ICC = 0.70
patients; derived from 30 items patient interview tau = 0.69, CV: NR and specificity = 91% alpha = 0.79 10–14 days
110 healthy controls P = 0.001
CAS (11) N = 64; 32 at-risk Nine items, Literature review NR NR NR NR Present between Cronbach’s r = 0.98
patients being treated unidimensional patients at risk and alpha = 0.70 1 h delay
for cancer; healthy controls
tau = 0.47–0.97,
NR, not reported; DV, divergent validity; CV, convergent validity; SR, self-report; PPV, positive predictive value; N.S., nonsignificant; ICC, Intra Class Coefficient; PGWB, Patient Global Well Being; SF-36, Medical Outcomes Study-
7-day delay
of the literature. Construct validity was evaluated
Test–retest
Kendall’s
by comparing 32 patients (69% female; mean
age = 53.9 years) who were at risk for developing con-
stipation from cancer treatments to 32 healthy adults
consistency
(94% female; mean age = 37.9). Significantly higher
Reliability
Internal
constipation scores were reported by the cancer
NR
patients than by the control group (t = 6.3,
Discriminates only P < 0.0001). In addition, symptom severity was evalu-
subjects without ated in two subgroups of the 32 cancer patients. Sig-
constipation
Discriminant
r = 0.56 incomplete
r = 0.52 straining,
constipation.
convergent
Divergent ⁄
Construct
Content
unidimensional
Participants
CCQ(7) 6 0–4 Likert scale, Total score 0–24; Preceding 2 weeks Not reported Hong Kong 0
0 = asymptomatic, cut-off score ‡5 and preceding Patient sample reported
4 = very severe symptoms 3 months easy to understand
Questionnaire
number of unsuccessful attempts at evacuation in a least for items. The factor matrices were used to identify
24-h period and duration of constipation). The CSS a five-factor solution or five subscales: stool symptoms,
consists of seven items that are scored using a five- rectal symptoms, abdominal symptoms, systemic symp-
point Likert scale that ranges from 0 (none of the time) toms and urgency symptoms. Systemic symptoms and
to 4 (all of the time) and one item that is rated on a urgency symptoms were found to be separate from the
0–2 scale. A total score can range from 0 (normal) to constipation symptom domain and were deleted. The
30 (severe constipation). A cut-off score of 15 suggests final 12 items were assigned to three subscales: stool,
constipation. rectal and abdominal symptoms.8
The measure was tested in a sample of 232 patients Positive correlations between three items recorded in
with constipation referred for an anorectal physiology a bowel diary (straining, incomplete evacuation and
evaluation (79% female; mean age = 64.9). Significant stool consistency) and corresponding symptoms on the
correlations, between CSS scores and objective findings PAC-SYM provide evidence of convergent validity.
from anorectal physiology tests, provide evidence of Additional evidence of the PAC-SYM’s validity comes
criterion validity (r not reported, P < 0.05). In addition, from moderate positive correlations between items on
CSS scores differentiated between patients who were the Medical Outcomes Study-Short Form 36 (SF-36)
constipated and those who were not; both groups were scale18 and the Patient Global Well Being (PGWB)
confirmed by physiological testing, with 96% accuracy. scale.19 Participants who received interventions for con-
Additional evidence to support the discriminant validity stipation were assessed at three different times using the
of the CSS comes from its ability to distinguish between PAC-SYM, to assess for both sensitivity to change and
subgroups of constipated patients and not constipated discriminant validity. The PAC-SYM distinguished
patients whose condition was confirmed by physiologi- between responders (those who reported improvement
cal tests. Physiological tests in this validation study in symptoms) and nonresponders (those who reported
were used to establish convergent validity. However, no change or worse symptoms) (t = 6.1, P < 0.001).8
no data were reported on the reliability of the CSS, its Estimates of test–retest reliability of the initial mea-
responsiveness, or its sensitivity to change. sure were high [r = 0.74, intraclass correlation coeffi-
cient (ICC) = 0.72, P < 0.01]. The Cronbach’s alpha for
the total PAC-SYM score was 0.98. Test–retest and
Patient Assessment of Constipation – Symptom. The
internal consistency estimates for the final measure
Patient Assessment of Constipation – Symptom (PAC-
were high (a = 0.89, ICC 0.75). Subscale Cronbach’s
SYM)8 was developed as a brief, easily administered tool
alphas were 0.80 for stool symptoms, 0.84 for abdomi-
to assess symptom frequency and severity of chronic
nal symptoms and 0.87 for rectal symptoms.8
constipation. The authors used a definition for constipa-
tion based on the Rome II criteria.15 This 12-item self-
report measure is divided into three symptom subscales Knowles Eccersley Scott Symptom Score. The Know-
(i.e. abdominal, rectal and stool). Items are scored on a les Eccersley Scott Symptom (KESS)10 is an 11-item
four-point Likert scale, with 4 indicating the worst measure that was developed to help diagnose constipa-
symptom severity. A total score for the PAC-SYM can tion and to discriminate among pathophysiological
range from 0 to 48. No cut-off score was reported. subgroups. The authors acknowledged that constipa-
The initial psychometric evaluation of the PAC-SYM tion is a poorly defined clinical symptom, but did not
involved patients (n = 216; 94% female; mean provide a definition of constipation. This omission
age = 42.2) from nine sites.8 Subsequently, the validity makes evaluation of content validity nearly impossi-
of the PAC-SYM was evaluated for use in older adults ble. Items on the KESS are based on a review of the
in long-term care facilities16 and in patients with low literature, items from the CSS questionnaire,6 as well
back pain and opioid-induced constipation.17 as the Rome II criteria.15 The KESS uses four- to five-
Criterion or concurrent validity was established based point Likert scales that are scored on an unweighted
on the high correlations between patient and investiga- linear integer scale. Total scores can range from 0 (no
tor global severity ratings. Construct validity of the symptoms) to 39 (high symptom severity). A cut-off
PAC-SYM was established using exploratory factor and score of ‡11 indicates constipation.
multi-trait analyses. Six factors were retained based on The KESS was tested in a sample of patients with
eigen values of >1 and factor loadings of >0.3 on at chronic constipation (n = 71; 93% female) and healthy
controls (n = 20; 90% female) with a similar, unspeci- define constipation. A pilot study of 10 people deter-
fied age distribution in each group.10 Content validity mined that the GQ was easy to understand and to
was established through a review of the literature and complete. The items are scored using two different
incorporation of elements from the Rome II criteria four-point Likert scales. The first consisted of ‘never’,
and the CSS into the KESS. In a sample of patients ‘sometimes’ (<25% of the time), ‘often’ (‡25% of the
with constipation, convergent validity was established time) and ‘always’. The second consisted of ‘never’,
based on significant correlations between KESS and ‘fewer than once a week’, ‘one or more times a week’
CSS scores (r = 0.90). In addition, the KESS was able and ‘every day’. Total and cut-off scores were not
to discriminate between constipated patients, whose reported.
diagnosis was confirmed with physiological tests, and The measure was tested in a sample of 349 partici-
healthy controls. KESS scores predicted which patients pants (52% female; age range 18–65 years). Content
had pure slow-transit constipation or rectal evacuatory validity was established through having included the
disorder better than patients with a mixed aetiology Rome criteria in the development of the measure. Con-
(ROC 55%, CI: 43–67%). However, no significant dif- vergent validity was reported through partial testing of
ferences were found in median total KESS scores by the Rome I and II criteria. The GQ was able to discrimi-
subgroup (P = 0.13). No data are available on the nate between participants with constipation and those
reliability of the KESS or on its responsiveness or sen- without it. However, the low, positive likelihood ratios
sitivity to change. limit the clinical usefulness of the GQ to identify indi-
viduals with constipation. No data were reported on the
reliability of the GQ or on its sensitivity to change.
Visual Scale Analog Questionnaire. The Visual Scale
Analog Questionnaire (VSAQ)12 is a five-item tool
designed to assess the frequency and severity of con-
Chinese Constipation Questionnaire. The Chinese
stipation symptoms in healthy individuals. The Rome
Constipation Questionnaire (CCQ)7 is a six-item self-
II definition of constipation was used to create the
report measure designed to diagnose functional consti-
items. Two of the five VSAQ items, stool consistency
pation in Chinese participants. The Rome II criteria15
and straining, are rated on a Likert scale that ranges
were used as the definition of constipation.
from 1 to 5. Three items, presence of constipation,
A sample of 111 patients (95% female; mean
duration of constipation, incomplete evacuation, are
age = 38.5) with constipation and 110 healthy con-
rated on a 0–10 scale. Methods for scoring the VSAQ
trols (64% female; mean age = 34.2) participated in a
are not reported, although a score of ‡3 on the 0–10
4-week psychometric evaluation of the CCQ. The
scale is the cut-off score for constipation.
initial version of the CCQ consisted of 24 items that
The VSAQ was tested in a sample of healthy hospital
were scored using a five-point Likert scale. Principal
employees (n = 369; 54% female; mean age = 31.6).
components analysis was used to create the final
Participants were asked to rate each bowel movement
version of the CCQ that consisted of six items with
for 7 days. Significant correlations were found
four subscales: bowel function, stool and rectal symp-
between items on the VSAQ and bowel diaries (i.e.
toms, laxative use and abdominal symptoms. A com-
straining and incomplete evacuation). The VSAQ was
posite score was generated by principal components
found to be discriminative only for those participants
analysis with a cut-off score of ‡5 being able to dis-
without constipation, not for those with constipation.
criminate between controls and constipated patients
Estimates of test–retest reliability were high in healthy
with both a sensitivity and specificity of 91%.
participants (kappa = 0.97). However, no data were
Content validity of the CCQ was established through
reported on criterion validity, internal consistency or
interviews with patients and a review of the literature.
sensitivity to change.
Evidence of construct validity was provided by nega-
tive correlations between CCQ scores and seven
Garrigues Questionnaire. The Garrigues Questionnaire domains on the SF-36 scale18 (physical functioning,
(GQ)9 is a 21-item self-report measure that was devel- role physical, bodily pain, general health, social func-
oped to define the presence of chronic constipation. tioning, role emotional and mental health), but not
Only 13 of the 21 items are related to bowel habits. vitality. In addition, the CCQ was able to discriminate
The Rome I20 and Rome II15 criteria were used to between constipated patients and healthy controls.
Estimates of test–retest reliability were high III criteria because these criteria are a diagnostic
(ICC = 0.70, P < 0.01), with patients rating themselves standard.22
unchanged from baseline and at a second retesting All but one8 of the seven measures evaluate some
10–14 days later. Cronbach’s alpha for the total CCQ aspect of stool frequency. However, only the CCQ and
score was 0.79. VSAQ assess frequency of bowel movements using the
Rome III format of <3 bowel movements per week.
The remaining measures assess stool frequency in a
Critique of the seven constipation measures
variety of ways and three measures6, 9, 10 include an
Definition of constipation. Defining a construct is an evaluation of the length of time per attempt.
important first step in the development of a measure. Stool consistency is evaluated in all but one6 of the
Historically, measurement of stool frequency was used measures. In five of the measures,7–10, 12 hard stool is
to define constipation. ‘Normal’ frequency of stool one of the descriptors and is consistent with the Rome
evacuation comprises a narrow to a broad range of III criteria. At the other end of the continuum, the
time that has large intraindividual and interindividual descriptors of stool consistency are not consistent (e.g.
variability. This self-report criterion does not describe oozing and watery).
the entire symptom experience of constipation. The In terms of rectal symptoms, all seven measures
definition of constipation has evolved to now include include a rating of incomplete evacuation, which is
abdominal and rectal symptoms. one of the Rome III criteria. Five measures6–11 have
Wide variations exist in the definition of constipa- ‘urge to defecate present, unable to defecate’ as one of
tion used in the development of measures included in the items in this content area. Three measures6, 8, 11
this review. The CSS and KESS utilized a purely path- include a rating of ‘pain with defecation’. Only three
ophysiological definition based on the underlying aeti- measures8, 9, 12 contain the other Rome criterion,
ology of constipation (e.g. slow transit and pelvic namely straining.
outlet obstruction). Interestingly, developers of the Six of the seven measures include abdominal symp-
CSS and KESS attempted to identify the subtypes of toms. The two most common symptoms assessed are
constipation more from a clinical than a mechanistic bloating and abdominal pain. These symptoms do not
perspective. In contrast, the developers of the CAS appear in the Rome III criteria for functional constipa-
provided a clear definition of constipation and used it tion, but are descriptors for irritable bowel syndrome.
to develop the instrument. The CCQ, PAC-SYM and Only three6, 9, 10 of the seven measures assess for
VSAQ were developed using the Rome II criteria,15 the use of assistive measures, which is part of the
whereas the developers of the GQ used both the Rome Rome III criteria. Only three of the measures6, 10, 12
I20 and Rome II15 criteria. evaluate the duration of constipation.
None of the seven measures includes all of the Rome
III criteria for functional constipation. The total num-
Content of the measures. How a construct is opera-
ber of items matching the Rome III criteria ranges
tionalized is represented by the content included in
from only 111 to 59 of the 6, in addition to a range of
the measure. Table 1 provides a synthesis of the
other symptoms described above and in Table 1. Our
content evaluated by each measure. We compared
review of the content in all of the measures suggests
each measure with the others in terms of its specific
that the assessment of constipation using self-report
content. We also compared the content of each mea-
measures is complex.
sure with the symptoms of functional constipation
found in the most recent version of the Rome crite-
ria21 (i.e. at least two or more of the following Psychometric properties: validity. The psychometric
symptoms at least 25% of the time: straining, lumpy properties of the seven measures are summarized in
or hard stools, sensation of incomplete evacuation, Table 2. All of the measures provide some evidence of
sensation of anorectal obstruction, manual manoeu- content validity. Most of the measures were developed
vres to facilitate evacuation, fewer than three defe- based on a review of the literature, but the CCQ and
cations per week). As no gold standard exists for the PAC-SYM were based on patient interviews and
the evaluation of the symptoms of constipation, we clinical opinions. Four measures provided evidence of
compared the content of these measures to the Rome good-to-moderate criterion validity by demonstrating
significant correlations between the measure’s scores over time. Of these two measures, only the PAC-SYM
and Rome I and II criteria,9 anorectal physiological has undergone psychometric evaluation in more than
tests6–10 or patient and investigator global severity one patient sample.16-17
ratings.8
Evidence of convergent validity was established in
General characteristics of the seven constipation
five measures6, 8–10, 12 through significant correlations
measures. Table 3 summarizes the general character-
between constipation symptoms and a variety of con-
istics and clinical utility of the seven measures. Three
stipation measures such as bowel diaries, Rome criteria
measures (43%) were developed in the United
and self-report measures. However, the authors of the
States,6, 8, 11 one in Hong Kong,7 one in Spain,9 one in
CSS did not identify the measures used to support
Turkey12 and one in the United Kingdom.10 Only the
their assertion of convergent validity.
PAC-SYM has been translated into several languages,
Only the CCQ and the PAC-SYM provided evidence
though validation studies of the translations have not
of divergent validity. Both these measures were associ-
been published.
ated with changes in QOL. Patients with higher consti-
The target population was identified in all seven of
pation scores on the CCQ reported lower scores on
the measures. For two of the measures,9–12 the target
seven domains of the SF-36.18 The PAC-SYM had
population was the general population, which consisted
moderate positive correlations with items on both the
of healthy individuals. Four measures6–8, 10 identified
SF-36 and PGWB,19 showing a relationship between
patients with constipation as the target population. Only
symptoms and QOL. The authors of the PAC-SYM con-
one measure11 was constructed as a risk assessment
cluded that their tool measured a construct distinct
measure for patients undergoing cancer treatment.
from QOL.
Participant and administrator burden are minimal
Although discriminant validity was tested in all
for all seven of the measures. Time to completion
seven of the measures, only four of them6, 7, 10, 11 pro-
ranges from 5 to 15 min because the number of items
vided evidence to support their measure’s ability to
ranges from 5 to 21. Six measures have <12 items.
discriminate between patients with and without consti-
While short measures are clinically useful, they may
pation, and only one of them8 provided evidence to
lack sensitivity and specificity.
support their measure’s ability to discriminate between
All of the measures used some form of a Likert
responders and nonresponders. The two other mea-
scale, with 3–5 points to rate items. While the Likert
sures9, 12 were only able to identify patients who did
scale is a widely used scaling technique, the require-
not have constipation, which limits the utility of these
ments for a ‘good’ scale include a minimum number
measures.
of items (usually >10 items) and an adequate number
A validity test essential for measures with evaluative
of response alternatives (usually ‡5 points) to ensure
purpose is sensitivity to change. This test, frequently
the scale’s reliability.23 Only the PAC-SYM meets both
mistaken for test–retest, analyses the sensitivity of the
these criteria.
tool to measure changes in a symptom over time. Only
A cut-off score is an important aspect of the clinical
the PAC-SYM and the CCQ provided evidence of the
utility of a measure. Without a cut-off score, it is diffi-
measure’s ability to detect changes in constipation
cult to use a measure for treatment decisions and out-
over time.
come evaluation. Five of the measures6, 7, 10–12
reported a cut-off score for the presence of constipa-
Psychometric properties: reliability. Reliability esti- tion. The scoring system for the KESS is rather compli-
mates were reported for only four7, 8, 11, 12 of the cated, which may limit its clinical utility.
seven scales. All four have Cronbach’s alphas in the
acceptable range of ‡0.70. In addition, adequate esti-
CONCLUSIONS
mates of test–retest reliability were reported for the
same four measures.7, 8, 11, 12 Only the CCQ7 meets all of our prespecified criteria for
Our analysis of their psychometric properties found a constipation self-report measure. It underwent rigor-
that only the CCQ and the PAC-SYM provide robust ous development and testing; provided robust esti-
evidence of validity and reliability. Both measures also mates of validity and reliability; described a cut-off
provide evidence of their sensitivity to detect change score that discriminated for constipation and provided
evidence of the measure’s sensitivity to detect change constipation in adults. As this review highlights, no one
over time. The major limitation of the CCQ is that it perfect measure is currently available. A measure is
was designed for a Chinese-speaking population and needed that evaluates the complex constellation of
no published reports in English-speaking populations symptoms that constipation encompasses. Using the
were identified, which suggests that it is not widely findings from this review, more comprehensive mea-
used. The PAC-SYM8 is an acceptable alternative. sures can be developed and tested in a variety of clini-
Although a cut-off score and its ability to discriminate cal populations, as well as in the general population.
between healthy and symptomatic patients have not The development of more valid, reliable and compre-
been determined, other positive aspects of this measure hensive self-report measures of constipation will facili-
include the degree of rigor used in its development, tate research on its epidemiology and management.
robust validity and reliability estimates, its sensitivity
to detect change, and use in English-speaking popula-
ACKNOWLEDGEMENTS
tions.
The results of this summary and critique suggest that Declaration of personal interests: None. Declaration of
additional research is warranted to refine or develop a funding interests: The writing of this paper was funded
more comprehensive self-report measure to evaluate in part by the Betty Irene Moore Fellowship.