FROM:
Quality of Life Research 2003 (Dec); 12 (8): 963–794
~ FULL TEXT
M Kosinski, M S Bayliss, J B Bjorner, J E Ware Jr, W H Garber, A Batenhorst, R Cady, C G H Dahlöf, A Dowson, S Tepper
Mark Kosinski,
QualityMetric Incorporated,
640 George Washington Highway,
Lincoln, RI 02865, USA
Background: Migraine and other severe headaches can cause suffering and reduce functioning and productivity. Patients are the best source of information about such impact.
Objective: To develop a new short form (HIT-6) for assessing the impact of headaches that has broad content coverage but is brief as well as reliable and valid enough to use in screening and monitoring patients in clinical research and practice.
Methods: HIT-6 items were selected from an existing item pool of 54 items and from 35 items suggested by clinicians. Items were selected and modified based on content validity, item response theory (IRT) information functions, item internal consistency, distributions of scores, clinical validity, and linguistic analyses. The HIT-6 was evaluated in an Internet-based survey of headache sufferers (n = 1,103) who were members of America Online (AOL). After 14 days, 540 participated in a follow-up survey.
Results: HIT-6 covers six content categories represented in widely used surveys of headache impact. Internal consistency, alternate forms, and test-retest reliability estimates of HIT-6 were 0.89, 0.90, and 0.80, respectively. Individual patient score confidence intervals (95%) of app. +/–5 were observed for 88% of all respondents. In tests of validity in discriminating across diagnostic and headache severity groups, relative validity (RV) coefficients of 0.82 and 1.00 were observed for HIT–6, in comparison with the Total Score. Patient–level classifications based in HIT–6 were accurate 88.7% of the time at the recommended cut–off score for a probability of migraine diagnosis. HIT–6 was responsive to self–reported changes in headache impact.
Conclusions: The IRT model estimated for a 'pool' of items from widely used measures of headache impact was useful in constructing an efficient, reliable, and valid 'static' short form (HIT–6) for use in screening and monitoring patient outcomes.
Keywords: Headache impact, Headache impact test (HIT™), Health–related quality of life (HRQOL), HIT–6, HIT, Item response theory (IRT), Migraine, Reliability, Validation
From the FULL TEXT Article:
Introduction
Headache disorders are highly prevalent and can
be severely disabling. [1, 2] Essential to research
and the clinical management of headache and migraine
is a method for standardizing the information
gathered from patients about the impact on
the patient’s life. Over the past decade there has
been a proliferation of standardized questionnaires
that measure the impact of headache or migraine
on functional status and well–being. To be most
useful in clinical practice and research, questionnaires
that measure the impact of headache or
migraine should meet psychometric and clinical
criteria for reliable and valid measurement. To be
practical these questionnaires must also be short.
However, short form surveys currently used to
evaluate the impact of headache or migraine on
patient functioning are restricted in the range of
measurement, focusing on extreme disability,
which are not relevant for everybody. [3]
Furthermore,
short form surveys that measure a limited
range are not useful in measuring improvement
because of problems with ceiling effects and they
typically lack the measurement precision for making
decisions at the level of the individual patient. [3]
One solution to these problems is to develop
short form surveys that are tailored to each patient’s
level of functioning. The Internet–based
headache impact test (DynHA™ HIT™), a computerized
adaptive questionnaire, has demonstrated
evidence of precision, reliability, validity, and
clinical relevance in the measurement of headache
impact. [3] However, since practical barriers still
exist in using the Internet, particularly at the point
of care, it became obvious that a brief paper and
pencil version of HIT was needed that would meet
accepted standards of reliability and validity and
maintain comparability to the Internet–based HIT.
The challenge was to develop a paper and pencil
version of HIT brief enough to be practical, yet
long enough to comprehensively measure a wide
range of headache impact and cover all the content
areas included in the total item pool that captures
the clinical, personal and socially relevant effects
of headache. To meet this objective we took a
three–pronged approach to developing the HIT
short form, taking into account psychometric evidence,
clinical relevance, and linguistic translation
of the items selected from the item pool for the
short form. This approach was carried out in independent
development and validation phases.
The purpose of this paper is to provide a brief
account of the development phase of HIT–6 and
focus primarily on the validation phase.
The objectives of the validation phase were to
(1) calibrate the final version of the items in HIT–6 to the
total HIT item pool,
(2) develop scoring algorithms
to place the HIT–6 scores on the same scale
as scored from the total HIT item pool,
(3) evaluate
the test–retest, alternate forms and internal
consistency reliability of HIT–6 scale scores,
(4) evaluate the construct validity of the HIT–6 scale,
(5) test the clinical validity of HIT–6 scale scores in
relation to headache severity and migraine diagnosis,
and
(6) test the responsiveness of HIT–6
scale scores in relation to self–reported changes in
headache impact.
Methods
Development of HIT–6
Psychometric relevance
The starting point for selecting the HIT–6 items
was the 54 items in the HIT item pool previously
analyzed by IRT methods. [3, 4] IRT information
functions and content validity were considered in
selecting a subset of items for a fixed–length survey
of headache impact. Using data from the National
Survey of Headache Impact (NSHI) the best candidate
items were evaluated on the basis of IRT
information functions and content validity (in
relation to widely used surveys and clinician
judgment). Information functions express the
contribution of each item to the overall test precision
for various levels of headache impact. We
selected a subset of 10 candidate items with IRT
information functions that spanned a wide range
of headache impact defined by the entire item
pool, where more than 90% of recent headache
sufferers scored. These 10 items also represented
the six main content areas covered in widely used
surveys (pain, social functioning, role functioning,
vitality, cognitive functioning, and psychological
distress).
Clinical relevance
The next phase of development consisted of an
independent review of the 10 candidate items by a
panel of clinicians involved in the treatment of
migraine headaches. The panel of clinicians recommended
35 newly developed items to be considered
for the short form. Many of these
suggested items covered similar content areas as
captured by the original 10 candidate items, except
that they were worded differently. Some of
the suggested items covered content areas not
captured by the original 10 candidate items and
were regarded as clinically useful in gauging the
severity of migraine during a typical physicianpatient
interview. For example, one of the new
items asked the patient ‘When you have a headache,
how much of the time do you wish you could
lie down?’
A survey was fielded by telephone interview
(n = 459) and over the Internet (n = 601) that included
the original 10 candidate items and the
additional 35 items suggested by the panel of clinicians.
Participants for the telephone survey were
sampled from a prescription database and had a
prescription for migraine medication during the
previous year. Otherwise the design of this study
was the same as the previous study (NSHI) used to
develop the HIT item pool. We evaluated whether
the additional items contributed to filling in measurement
gaps and/or extended the range of impact
defined by the original 10 candidate items. In
addition, based on the self–assessment of headache
severity, the frequency of migraine symptoms, and
work loss productivity, we were able to also evaluate
the clinical validity of each item. From this
study, six items were selected to be included in the
new HIT short form (HIT–6). Four of the HIT–6
items came from the original 10 and two came
from the items suggested by the clinicians. These
six items covered the six content areas represented
in the total HIT item pool, covered more than
50% of the range of headache impact measured
by the total item pool, and were among the most
valid items in discriminant validity tests involving
criterion measures of headache severity, frequency
of migraine symptoms and work loss
productivity.
Linguistic translation
The final phase in the development of the HIT–6
short form consisted of a linguistic translation of
the six items into 27 languages. The translation of
HIT–6 items followed the same methodology used
to translate the SF–36 Health Survey, including
forward and backward translations of items and
an independent review. [5] This translation process
resulted in modifications to all HIT–6 items and
response options. [6]
Validation of HIT-6
Sample
HIT-6 was validated using data compiled from a
general population survey of recent headache
sufferers collected via the Internet using AOL’s
Opinion Place. Research on other health status
questionnaires have found answers collected via
the internet to be very similar to answers collected
via paper and pencil. [7] AOL subscribers completed
surveys by logging on to the AOL Opinion
Place. Potential respondents entering Opinion
Place were randomly assigned to one of many
ongoing survey projects. Participants eligible for
his study: (1) were aged 18–65 years of age, (2) had
a headache in the past 4 weeks that was not due to
a cold, flu, a head injury, or a hangover, (3) agreed
to be contacted again in 2 weeks to complete another
survey. Participants received AOL reward
points as an incentive for participation.
Procedure
At Time 1, the survey consisted of the HIT-6,
computer adaptive administration of five items
from the total HIT item pool (DynHA HIT), a
battery of items on the frequency of six migraine
symptoms, headache severity, the SF-8 Health
Survey [7], and a number of new headache impact
items (κ = 23) written by clinicians to enhance the
total HIT item pool. The symptom and severity
questions were based on a clinically validated,
computer-assisted telephone interview used in
previous studies [8, 9] and used in this study to validate the HIT-6. The SF-8 Health Survey was included in this study to evaluate the relationship
between HIT-6 and generic measures of health
status. DynHA HIT was included to evaluate alternative
forms reliability of the HIT-6.
At Time 2, the questionnaire consisted of the
HIT-6, DynHA HIT, three items to assess changes
in headache impact since Time 1, and the SF-8
Health Survey. The sequence of administration of
the items remained the same from Time 1 to Time
2, with the HIT-6 items always administered first
so as to minimize any response effects on the test–
retest reliability study.
Calibration and scoring HIT-6
First, we estimated an IRT scale (IRT-Total HIT)
based on all available item responses, including the
DynHA HIT (total of 34 headache impact items).
The IRT-Total HIT scale was scored to have a
mean of 50 and standard deviation of 10 in the
general population of recent headache sufferers. [3]
Next we examined which of two choices of item
category weights to assign to HIT-6 item responses
that could be summed to closely approximate the
IRT-Total HIT score. In method 1 the following
item category weights were assigned to each HIT-6
item response: never = 6, rarely = 8, sometimes = 10, very often = 12, and always = 14. In
method 2 the following item category weights were
assigned to each HIT-6 item response: never = 6,
rarely = 8, sometimes = 10, very often = 11, and
always = 13. We examined the concordance between
the IRT-Total HIT scale scores and HIT-6
scale scores by correlational methods, by plots,
and by calculating the score difference between the
two scales to determine the choice of item category
weights to use in scoring HIT-6.
Reliability analyses
Reliability was assessed using internal consistency,
alternate-forms, and test–retest methods. For internal
consistency reliability, Cronbach’s a was
estimated for HIT-6 items at Time 1 and Time 2.
Alternate forms reliability was estimated by computing
the intra-class correlation coefficient between
HIT-6, DynHA HIT, and IRT-Total HIT
scale scores at Time 1 and between HIT-6 and
DynHA HIT scale scores at Time 2. Test–retest
reliability was assessed by computing the intraclass
correlation between HIT-6 scale scores at
Time 1 and Time 2. The test–retest reliability analyses
were carried out first using all individuals
who completed Time 1 and Time 2 questionnaires
and second only using a subset of individuals who
had reported no change in headache impact during
that interval.
To evaluate the precision of HIT-6 scores at the
level of the individual patient, measurement error
was estimated for various HIT-6 score levels using
IRT methods (EAP score estimation) [10] and the
item parameters established for HIT-6 items. [4]
Construct validity
Construct validity was evaluated in terms of HIT-6
correlations with generic health-related quality of
life (HRQOL) measures that included the SF-8
scales and physical and mental health summary
measures. We hypothesized that HIT-6 would
correlate negatively with general health status (the
greater the headache impact the lower the general
health status). We also expected that HIT-6 would
correlate stronger with the physical measures than
with the mental measures of general health status.
Construct validity was also evaluated in clinical
tests of discriminant validity using the method of
known groups validity. [11] The method of known
groups validity compares mean scale scores across
groups know to differ on a clinical criterion measure.
In this study, two clinical criterion measures
were used. The first was self-ratings of headache
pain severity. Participants were asked to rate the
severity of their headache pain on a scale from 0
(no pain) to 10 (pain as bad as it can be). Participants
were categorized as mild if they answered 1,
2, or 3, moderate if they answered 4, 5, 6, or 7, and
severe if they answered 8, 9, or 10. The second
clinical criterion was the probability of migraine
diagnosis, which was based on questions concerning
the frequency of six migraine symptoms
used in previously conducted epidemiology studies
of migraine. [12] A series of one-way analysis of
variance (ANOVA) were conducted to test performance
of HIT-6 relative to the IRT-Total HIT
scale in distinguishing between participants that
differed in headache severity and probability of
migraine diagnosis. RV coefficients were computed
by dividing the F-statistics of each scale by the
largest F-statistics observed among all scales in
each test. [13, 14] We expected to observe higher
scores (greater headache impact) on both HIT
scales among patients with more severe headache
and among patients with a higher probability of a
migraine diagnosis. However, RV coefficients were
expected to be higher for the IRT-Total HIT scale
in comparison to HIT-6.
Screening
The accuracy of HIT-6 in patient level screening
was investigated at the recommended DynHAHIT
cut-point score (>56) for a migraine diagnosis. [15] The agreement in classification at the recommended cut-point score was compared across
HIT-6 and the IRT-Total HIT scale.
Responsiveness of HIT-6
The responsiveness of HIT-6 scores was evaluated
by comparing average score changes across groups
of study participants differing in self-reported
change in headache impact. For these analyses, we
used data from the 1998 NSHI [3] that included
the developmental version of HIT-6. Three self-reported
change measures were in the follow-up
survey (Appendix A). Respondents were classified
into three groups, ‘better’, ‘same’, or ‘worse’ depending
upon their responses to the three questions.
Changes in HIT-6 and IRT-Total HIT scale
scores were calculated by subtracting baseline
scores from follow-up scores. The method of
known groups validity and ANOVA methods were
used to compare changes in HIT-6 and IRT-Total
HIT scale scores across groups differing in selfreported
change. RV statistics were calculated to
compare the performance of HIT-6 relative to the
IRT-Total HIT scale in distinguishing between
groups.
Reading level
The reading level of the HIT-6 was evaluated using
the Flesch–Kincaid Reading Ease and Grade
Level index. This Flesch–Kincaid evaluates readability
based on the average number of syllables
per word and the average number of words per
sentence. The Reading Ease procedure assigns a
score between 0 and 100, with higher scores
indicating more difficult material. The Flesch–
Kincaid Grade Level index indicates a school
grade reading level. For example a score of 8.0
means that an eighth grader would understand the
questions. [16]
Results
Sample
At Time 1 1,103 eligible respondents completed the
survey. The majority of participants were female
(73%) and the average age of the sample was
37 years. Approximately 14 days (on average) after
completion of the Time 1 survey, 540 participants
completed the survey at Time 2. The
majority of the participants at Time 2 were female
(72%) and the average age of the sample was
37.5 years. Despite 50% attrition, the samples
were relatively the same in age and gender across
the two time points, suggesting that the subset of
the baseline sample completing the follow-up
questionnaire was representative.
Calibration and scoring HIT-6
Table 1
Figure 1
Figure 2
Table 2
Figure 3
Table 3
Table 4
Table 5
Table 6
|
As shown in Table 1, the correlation between the
two methods of computing a sum score for HIT-6
was high (0.993) as were the correlations between
each of the HIT-6 sum scores and the IRT-Total
HIT scale score (method 1: r = 0.906; method 2:
0.903). Figures 1 and 2 show that HIT-6 scoring
method 2 (Figure 2) had a closer match with the
IRT-Total HIT scale score. Method 1 overestimated
headache impact at the high impact range
(Figure 1) compared to method 2. Also, in evaluating
the correspondence between HIT-6 scale
scores and the IRT-Total HIT scale there were
more data points outside the 95% confidence interval
drawn around the identity line for the HIT-6
scale scored under method 1 (Figure 1) compared
to method 2 (Figure 2). Lastly, the mean difference
between HIT-6 scored with method 1 and the IRTbased
score was 2.76 (range )11.26 to 17.93) while
the mean difference between HIT-6 scored with
method 2 and the IRT-based score was 0.59 (range)11.26 to 11.92) (Table 2). In light of these results,
the focus of the remaining results for HIT-6 was
based on scoring method 2.
Figure 3 illustrate the range of headache impact
covered by HIT-6 item parameters in relation to
the range of headache impact measured by the
total HIT item pool. As shown, the item parameters
estimated for HIT-6 items covered 53% of the
range of headache impact measured by the total
HIT item pool.
Reliability analyses
Results of reliability analyses are presented in
Table 3. Internal consistency (Cronbach’s a) reliability of HIT-6 scale scores was 0.89 at Time 1
and 0.90 at Time 2. The intra-class correlation
between HIT-6 and the total HIT score (alternate
forms reliability) was 0.90 and between HIT-6 and
HIT-DynHA (alternate forms reliability) was 0.84.
The intra-class correlation between HIT-6 scores
at Time 1 and Time 2 (test–retest reliability) for the
total sample (n = 540) was 0.78. The intra-class
correlation between HIT-6 scores at Time 1 and
Time 2 (test–retest reliability) for the ‘stable’
sample (n = 245) was 0.80.
Table 4 presents the 95% confidence intervals
around HIT-6 and IRT-based scale scores across
eight levels of the IRT-based scale. As shown, for
scores between 45 and 69 (91% of those studied),
where HIT-6 was most precise, 95% confidence
intervals of approximately five points were observed
for individual patient scores. In comparison,
95% confidence intervals of approximately
three points were observed for individual patient
scores between 45 and 70 on the IRT-Total HIT
score.
Construct validity
Correlations between HIT scales (HIT-6 and IRTTotal
HIT scale) and SF-8 scales and summary
measures are presented in Table 5. As expected, all
correlations with SF-8 scales and summary measures
were negative. Correlations between the IRTTotal
HIT scale and SF-8 scales and summaries
were stronger than those observed between HIT-6
and SF-8 scales and summaries. The pattern of
correlations between both HIT scales (HIT-6 and
total HIT) and SF-8 scales and summaries was
nearly identical, with the highest correlations observed
between HIT scales and the SF-8 role
physical (RP) and social functioning (SF) scales
and the lowest correlations observed between HIT
scales and the SF-8 bodily pain (BP) and mental
health (MH) scales. In relation to the SF-8 physical
(PCS) and mental (MCS) summary measures, both
HIT scales correlated higher with PCS than MCS.
Table 6 presents results from tests of the validity
of HIT-6 and the IRT-Total HIT scale in discriminating
among groups known to differ in
headache severity and the probability of a migraine
diagnosis. In the test involving headache
pain severity as the criterion measure, both scales
show large and statistically significant differences
in mean scores across the headache pain severity
groups, with the more severe pain groups scoring
greater headache impact. For each headache pain
severity group, the mean scores on both scales
were within one-half a point of each other. Significance
tests showed both scales to perform
equally well at discriminating across the headache
pain severity groups. In tests involving the probability
of a migraine diagnosis as the criterion
measure, both scales showed large and statistically
significant differences in mean scores across the
diagnostic groups, with the migraine group scoring
greater headache impact than the non-migraine
group. The mean HIT-6 and total HIT scores were
within one-half a point of each other for the nonmigraine
group and the migraine group. Significance
testing shows that HIT-6 was 18% less efficient
than the IRT-Total HIT scale in
discriminating between migraine and non-migraine
groups, as determined by the F and RV
statistics.
Screening
Table 7
Table 8
|
Results of tests of the accuracy of HIT-6 relative
to the cut-point score of >56 on the IRT-Total
HIT scale used in patient-level screening for the
probability of a migraine diagnosis are presented
in Table 7. The HIT-6 correctly classified 88.8% of
the sample at the recommended cut-point score for
the IRT-Total HIT scale, with sensitivity and
specificity statistics of 93.1 and 79.4%, respectively.
Responsiveness
Table 8 presents the results of analyses that compare
the relative performance of HIT-6 and IRT Total
HIT scale scores in responding to self-reported
change. As shown, both scales showed
large and statistically significant differences in
mean change scores across the self-reported
change groups. Scale scores improved on average
among those respondents who self-reported improved
headache impact, while scores declined on
average among those respondents who self-reported
worsening headache impact. Average score
changes were generally less than one-point among
those respondents who self-reported no change in
headache impact. Significance tests showed that
the HIT-6 scale performed equally well or better at
discriminating across the self-reported change
groups.
Reading level
Results of the Flesch–Kincaid readability test
found the HIT-6 to have a Flesch–Kincaid Reading
Ease score of 66.2 and a Flesch–Kincaid Grade
Level Index of 7.4.
Discussion
The IRT model estimated for a ‘pool’ of items
from widely used measures of headache impact
was very useful in constructing an efficient ‘static’
short form survey of headache impact for use in
screening and monitoring patient outcomes. The
items in HIT-6 cover the content areas found in
widely used measures of headache impact, including
pain, social-role limitations, cognitive
functioning, psychological distress, and vitality.
Although the HIT-6 is shorter than most widely
used measures of headache impact, it demonstrated
satisfactory reliability and validity for use
in group-level studies throughout the entire range
of headache impact studied. HIT-6 was almost as
accurate as the IRT-Total HIT scale in identifying
persons above the established cut-point score for
probable diagnosis. Using IRT-based scoring
methods, the HIT-6 was shown to be precise enough
to detect differences on average of approximately
five points in headache impact at the level
of the individual for all but the most extreme score
levels.
Based on the IRT item parameters estimated for
HIT-6 items, HIT-6 was shown to cover 53% of
the range of headache impact measured by entire
HIT item pool (54 items), which spans from 1.5
standard deviations below (mild impact) to 2.5
standard deviations above (severe impact) the
norm score of 50 observed among recent headache
sufferers. In this range, 91% of the total general
population of recent headache sufferers scored on
the scale defined by the total HIT item pool. In
comparison to the total HIT item pool, HIT-6
showed the biggest gap in measurement at the mild
headache impact range. The total HIT item pool
measures down to nearly 4 standard deviations
below the normative score of 50, while HIT-6
measures only 1.5 standard deviations below the
normative score, a gap of 2.5 standard deviation
units. The implication is that HIT-6 is may be
prone to ceiling problems in measuring patients
with very mild headache impact. However, this
may not be too problematic since less than 10% of
recent headache sufferers in the general population
scored more than 1.5 standard deviations below
(mild impact) the normative score on the scale
defined by the total HIT item pool. Furthermore,
despite the relatively limited range of headache
impact covered by HIT-6 compared to the range
defined by the entire HIT item pool, this short
form is more inclusive than other instruments
currently being used to asses headache impact. [3]
Using classical psychometric methods to assess
reliability HIT-6 was shown to exceed the minimum
standard (>0.70) for group level comparisons.
Internal consistency reliability estimates
approached levels recommended for comparisons
of individual patients (>0.90) despite the heterogeneity
of item content. test–retest reliability estimates
suggest that HIT-6 scores are stable over
time among respondents showing no change in
headache impact. Under the IRT framework,
where reliability and measurement error varies
across the score distribution, HIT-6 was extremely
reliable across a fairly wide range of headache
impact scores. HIT-6 was most precise between
scores of 45 and 70, where the 95% confidence
interval around an individual patient score was
approximately five points on average (corresponding
to reliability greater than 0.90).
Correlations between HIT-6 scale scores and
measures of HRQOL were negative. As expected,
respondents with higher scores on HIT-6 indicating
greater headache impact had lower HRQOL
scores. This is consistent with many studies
showing diminished HRQOL attributed to headache
and migraine. [17–20] The relationship between
HIT-6 and HRQOL scales was at best
moderate, with correlations all below 0.40, which
may be a function of the heterogeneity of content
of HIT-6 items. The pattern of correlations suggested
that HIT-6 was more related to measures of
physical health, particularly in the role functioning
domains, than MH. This pattern was true for both
HIT-6 and IRT-Total HIT scores. However,
across al HRQOL scales, the IRT-Total HIT score
showed substantially larger correlations with
HRQOL than those observed between HIT-6 and
HRQOL. This difference may in part be explained
by the differences in reliability and the range of
impact measured by the two HIT scales.
Results of discriminant validity testing showed
that HIT-6 reached the same statistical conclusions
about group differences, as did the HIT scale
based on the total item pool. In the test involving
migraine diagnosis, the RV coefficient for HIT-6
was nearly 20% below that observed for the HIT
scale based on the total item pool. In the test involving
headache severity, HIT-6 performed as
well as the HIT scale based on the total item pool.
In both tests, HIT-6 scores showed more variability
among the groups expected to have mild
headache impact (no migraine and mild severity
groups). This is consistent with the fact that HIT-6
is less precise in measuring mild headache impact
and therefore can be expected to have greater
measurement error than the HIT scale based on
the total item pool. Consequently, HIT-6 showed
more variability in scores for groups with mild
headache impact, which likely contributed to the
superior performance of the HIT scale based on
the total item pool in the test involving migraine.
Another important finding from the discriminant
validity tests was the observations that mean
scores on HIT-6 were within one-half a point of
mean scores based on the total item pool for each
of the criterion groups (diagnosis and severity).
The implication of these findings is that they lend
support to the interchangeability of scores based
on HIT-6 and HIT scores based on the total item
pool, at least in group level comparisons. Because
of the high degree of correspondence in scores,
norms and interpretation guidelines established for
HIT will be useful for interpreting scores on HIT-
6. However, more studies are needed to better
understand the interchangeability of scores at the
level of individual patients, particularly at the mild
headache impact range where HIT-6 is less precise.
Many barriers exist that contribute to the lack
of detection and appropriate care of migraine
headaches [21] , one of which concerns the length of time patients have to consult with their doctor. Studies of consultation in general practice [21, 22],
where migraine patients most often consult initially [23], suggest that the length of time of
consultation, which typically ranges from 5 to
8 min, is insufficient for patients to adequately
describe their symptoms and the degree of resulting
disability associated with their headaches.
Furthermore, there are no objective markers or
diagnostic tests that define migraine. One of the
potential uses of HIT-6 in clinical practice may be
as a first stage screen for the diagnosis of migraine
headache. In this study, HIT-6 was found to correctly
classify nearly 90% of those respondents
who scored higher than the cut-point score (>56)
for a probable migraine diagnosis on the IRTTotal
HIT scale. More importantly, HIT-6 can be
administered and scored in 2–3 min time, which is
well within the typical range of consultation time
in general practice. However, clinic based studies
to examine the positive predictive value of HIT-6
are warranted.
The evidence from this study showed that HIT-6
was responsive to self-reported change in headache
impact. Across all three criterion measures of
change, HIT-6 was equal to or better than the HIT
scale based on the total item pool in discriminating
between groups of patients differing in self-reported
changes in headache impact. Of importance
was the finding that average changes in scores for
each criterion group were very similar for both
HIT scales, which lends further support to the
point raised above about the interchangeability of
HIT-6 and HIT scores. In this study, HIT-6 scale
scores declined (decreased headache impact) by
approximately three points on average among
persons self-reporting improvement on all three
criterion measures of change. A decline of this
magnitude is roughly 3/10ths of a standard deviation,
which in terms of effect size is considered
small. While further study is necessary to better
understand the responsiveness of HIT-6 in clinical
studies, the results of this study offer preliminary
estimates of change in HIT-6 scale scores that
could be considered meaningful from the patient’s
perspective.
The evidence presented from this study suggested
that we successfully achieved our goals of
developing a brief measure of headache impact
that is
(1) psychometrically sound; and
(2) clinically relevant.
Our efforts resulted in a six-item
questionnaire that proved to be reliable and valid
for group-level comparisons, patient-level screening,
and responsive to changes in headache impact.
The HIT-6 items were shown to cover a substantial
range of headache impact as defined by a much
larger pool of items and include content areas
found in most widely used tools for measuring
headache impact. Modifications made to HIT-6
items resulted in an instrument that was more
easily translated into other languages. Translations
of HIT-6 are now available in 27 languages
in total through QualityMetric and studies are
currently being conducted to evaluate the performance
of the translated forms in clinical studies.
Appendix A
Self-reported changes in headache impact items.
References:
Headache Classification Committee of the International Headache Society.
Classification and diagnostic criteria for headache
disorders, cranial neuralgias, and facial pain.
Cephalalgia 1992; 12: 229–237.
Solomon GD, Price KL.
Burden of migraine: A review of its socioeconomic impact.
Pharmacoeconomics 1997; 11(Suppl 2): 1–10.
Ware JE, Bjorner JB, Kosinski M.
Practical implications of item response theory and computerized adaptive testing:
A brief summary of ongoing studies of widely used headache impact scales.
Med Care 2000; 38(Suppl II, 9): 1173–1182.
Bjorner JB, Kosinski M, Ware JE Jr.
Calibration of an item pool for assessing the burden of headaches:
An application of item response theory to the Headache Impact Test (HIT™).
Qual Life Res 2003; 12: 913–933.
Bullinger M, Alonso J, Apolone G, et al.
Translating health status questionnaires and evaluating their quality:
The International Quality of Life Assessment (IQOLA) Project.
J Clin Epidemiol 1998; 51(11): 913–923.
Gandek B, Alacoque J, Uzun V, Andrew-Hobbs M.
Translating the Short-Form Headache Impact Test (HIT-6)
in 27 countries: Methodological and conceptual issues.
Qual Life Res 2003; 12: 975–979.
Ware JE, Kosinski M, Dewey JE, Gandek B.
SF-8 Health Survey Manual – How to Score and Interpret
Single-Item Health Measures: A Manual
for Users of the SF-8 Health Survey.
Lincoln, RI: QualityMetric Incorporated, 2001.
Stewart WF, Lipton R, Liberman J.
Variation in migraine prevalence by race.
Neurology 1996; 47: 52–59.
Stewart WF, Lipton R, Kolonder K, Liberman J, Sawer J.
Reliability of the migraine disability assessment score
in a population-based sample of headache sufferers.
Cephalgia 1999; 19: 107–114.
Bock RD, Mislevy RJ.
Adaptive EAP estimation of ability in a microcomputer environment.
Appl Psychol Meas 1982; 6: 431–444.
Kerlinger FN.
Foundations of Behavioral Research.
New York: Holt, Rinehart and Winston, 1973.
Stewart WF, Lipton RB, Celentano DD, Reed ML.
Prevalence of migraine headache in the United States:
Relation to age, income, race and other
sociodemographic factors.
J Am Med Assoc 1992; 267(1): 64–69.
McHorney CA, Ware JE, Raczek AE.
The MOS 36-Item Short-Form Health Survey (SF-36). II.
Psychometric and clinical tests of validity in
measuring physical and mental health constructs.
Med Care 1993; 31: 247–263.
Ware JE, Kosinski M, Bayliss MS, McHorney CA, Rogers WH, Raczek AE.
Comparison of methods for scoring and statistical analysis of SF-36
health profiles and summary measures: Summary of
results from the Medical Outcomes Study (MOS).
Med Care 1995; 33(Suppl 4): AS264–AS279.
Bjorner JB, Ware JE, Kosinski M, Diamond M, Tepper S.
Validation of the Headache Impact Test using
patient-reported symptoms and headache severity.
In: Olesen J, Steiner TJ, Lipton RB (eds),
Reducing the Burden of Headache.
Oxford University Press, 2003.
Kincaid JP, Fishburne RP, Rogers RL, Chissom BS.
Derivation of New Readability Formulas (Automated Readability Index,
Fog Count and Flesch Reading Ease Formula) for Navy Enlisted Personnel.
Memphis, TN: Naval Air Station, 1975.
Lipton RB, Hamelsky SW, Kolodner KB, Steiner TJ, Stewart WF.
Migraine, quality of life, and depression:
A population-based case–control study.
Neurology 2000; 55(5): 629–635.
Meletiche DM, Lofland JH, Young WB.
Quality-of-life differences between patients with
episodic and transformed migraine.
Headache 2001; 41(6): 573–578.
Monson MJ, Lainez MJ.
Quality of life in migraine and chronic daily headache patients.
Cephalalgia 1998; 18(9): 638–643.
Osterhaus JT, Townsend RJ, Gandek B, et al.
Measuring the functional status and well-being of
patients with migraine headaches.
Headache 1994; 34: 337–343.
Edmeads J, Lainez JM, Brandes JL, Schoenen J, Freitag F.
Potential of the Migraine Disability Assessment (MIDAS) Questionnaire
as a public health initiative and in clinical practice.
Neurology 2001; 56: S29–S34.
Carr-Hill R, Jenkins-Clarke S, Dixon P, et al.
Do minutes count? Consultation lengths in general practice.
J Health Ser Res Policy 1998; 3: 207–213.
Hu XH, Markson LE, Lipton RB, Stewart WF, Berger ML.
Burden of migraine in the United States:
Disability and economic costs.
Arch Intern Med 1999; 159(8): 813–818.
/
Return to HEADACHE
Since 2-12-2023
|