Main result of the present study is that the CHIQ is a short, CH-specific, reliable and valid measure of CH-related disability. Therefore, the version presented in Additional file 1 is to be regarded as the final, validated version. To the best of our knowledge, this is the first evaluation of a CH-specific disability questionnaire.
Reliability and validity of the CHIQ
The CHIQ shows good item statistics, internal consistency and retest-reliability. Of the eight items, item 7 (asking for self-injurious behavior) showed the lowest average score and the weakest (although still adequate) values for item selectivity, item-scale correlation and factor loading. This is likely due to the fact that only part of the patients exhibit self-injurious behavior. However, self-injurious behavior is much more frequent in CH compared to other primary headaches [24] and if present, constitutes a considerable burden to the patient. Therefore, we decided to keep this item.
The CHIQ showed significant positive correlations with attack frequency and acute medication intake frequency during the last week, as well as cluster headache pain AUC, demonstrating that it captures current CH severity. Positive correlations were also found between the CHIQ and the HIT-6, suggesting that the CHIQ also evaluates unspecific headache-related disability. Moreover, the CHIQ was also positively correlated with the DASS depression, anxiety and stress scores and negatively correlated with the SF12v2 PCS and MCS, showing the expected relations to psychosocial factors and QoL.
Disability in cluster headache
Previous studies have used non-CH-specific questionnaires like HIT-6, MIDAS, HDI or MSQ 2.1 to measure disability in CH patients [3, 4]. These studies have shown higher impairment in chronic and active episodic CH patients compared to eCH patients in remission or healthy controls [25,26,27]. However, these questionnaires were not designed and validated for CH and may underestimate CH-related disability [5]. On the one hand, they might miss specific CH features such as nocturnal attacks and self-injurious behavior. On the other hand, items asking for the frequency of severe headache days, the desire to lie down or muscle tension due to headache may not be appropriate. Further, these questionnaires use timeframes from four weeks (HIT-6, MSQ 2.1) to three months (MIDAS) or even no timeframe (HDI). CH often shows rapid changes in attack frequency, so shorter timeframes are necessary to capture current impact.
The lack of CH-specific instruments has recently prompted the development of two comprehensive questionnaires, the 28-item Cluster Headache Quality of Life Scale (CHQ) [5] and the 36-item Cluster Headache Scales (CHS) assessing several CH-related psychological factors [6]. The CHS also comprises an 11-item ‘disability’ subscale, that however focuses on generic, not on CH specific questions (e.g. “I feel limited in everyday life” or “I can practice my hobbies”) [6]. Significantly lower QoL was detected using the CHQ in chronic compared to episodic CH patients, but no distinction was made between active and in remission eCH patients [5]. The CHS showed - among others - higher scores in the subscale ‘disability’ in cCH compared to eCH patients who scored higher in the subscale “fear of attacks” [6]. Together, this suggests a higher burden of cCH compared to eCH patients and provides important detail information.
Two further studies evaluated CH-dependent burden without the use of specific questionnaires in 1165 patients (cCH n = 306) [28], respectively 1134 patients [29]. Chronic CH patients reported significantly more interictal symptoms [28]. In the other study, 55% of patients reported suicidal ideation and 50% showed self-injurious behavior during attacks [29], emphasizing the need to evaluate self-injurious behavior or suicidal thoughts.
There are also measures that try to assess burden of disease by simply scoring attack frequency, attack duration and duration of episodes, like the Cluster Headache Severity Scale (CHSS) or the CH index [30, 31], an approach that captures only one determinant of disability. To complement the existing measures by a short, CH-specific disability questionnaire, the CHIQ (1) was limited to 8 items, (2) was designed to capture CH-specific characteristics like the impact of nocturnal attacks, the unpredictability of attacks and self-injuring behavior and (3) uses a time frame of 1 week. That makes the CHIQ useful as a short, economic, CH-specific measure of disability in both clinical and research settings. In contrast, if the goal is to plan individual behavioral interventions, the more extensive CHS seems more appropriate.
CHIQ as a measure of disability
The CHIQ shows correlations with measures of unspecific headache-related disability (HIT-6), with depression, anxiety and stress scores (DASS) and with a generic QoL instrument (SF-12v2). In addition, the CHIQ correlates with CH attack and medication intake frequency, and is rated by patients as appropriate to capture disability related to CH. Moreover, active patients (cCH and active eCH) showed significantly higher CHIQ scores compared to eCH patients in remission, as expected. The CHIQ also captured a higher disability in cCH compared to active eCH, which would be expected due to the chronic nature of cCH, with ongoing attacks without remission. In addition, cCH patients also had a higher attack frequency than active eCH patients in the present study.
In contrast, previous studies using the HIT-6 did not provide a clear picture: one study reported significantly higher HIT-6 scores in cCH compared to active eCH [32], others couldn’t find this difference [27, 33]. In a recent study with 224 CH patients (eCH 70.5%) no difference could be shown for HIT-6 scores between eCH and cCH, maybe due to a rather small cCH group [33]. Also, our study revealed no difference in HIT-6 scores between cCH and active eCH patients.
Interestingly, disability as assessed by the CHIQ was not equal to zero in eCH patients in remission, but relevant disability persisted in spite of cessation of the attacks. This is consistent with previous studies [26] and might reflect a general burden of the disease, e.g. by its long-term psychosocial consequences, the fear of upcoming active headache phases or the knowledge about the disorder itself.
We also propose a preliminary CHIQ grading (grades 1 to 4) based on quartiles of our active CH patient population. This approach has the advantage of generating grades that allow differentiation over the range of observed values. Analysis of attack and medication frequency and other questionnaire scores between grades showed that the proposed CHIQ grades represent a meaningful ordinal scale. In addition, most of the eCH patients in remission fell into grade 1. Nonetheless, for the time being we refrain from assigning clinical attributes (like mild, moderate, severe disability) to CHIQ grades because no existing CH disability scale could be used for comparison. We currently plan a follow-up study including a verbal rating scale on CH disability to help us labelling (and confirming or adjusting) the preliminary CHIQ grades.
Comparison of CHIQ and HIT-6
Compared to the HIT-6, the CHIQ received better suitability ratings from participants, showed larger correlations with clinical CH characteristics (attack and acute medication frequency and CH pain AUC) and detected a difference between cCH and active eCH patients that was not detected by the HIT-6. This suggests that the CH-specific measure CHIQ better reflects CH-related disability than unspecific measures such as the HIT-6.
CHIQ as a patient-reported outcome measure
Patient-reported outcome measures (PROMs) become increasingly important in clinical practice and research and complement more objective measures such as number of attacks. They provide a patient-centered approach and are already well implemented in other primary headaches like migraine [8]. PROMs might better reflect patients’ evaluations (like change in QoL) towards a treatment than objective measurements like attack frequency or intake of acute medication [7]. In our sample, this is supported by the fact that also eCH patients in remission showed some disability in spite of having no attacks.
Strengths and limitations
One strength of the current analysis is the large CH population examined (254 total, 196 patients with active CH). 43.7% of patients had chronic CH, which is more than expected from epidemiological studies [2], likely because these patients more frequently seek medical care and participate in support organizations. Encouragingly, a substantial amount of patients were female (34.3%). In recent decades male preponderance has decreased and female patients are increasingly recognized. An US cluster headache survey showed a female proportion of 28% among 1134 CH patients [34], similar to our study.
One limitation might be the self-reported diagnosis of patients participating via the support group. However, we assessed all criteria necessary to confirm a CH diagnosis according to the ICHD-3 in the survey and included only patients who fulfilled these criteria. Further, 80% and 9% of the patients indicated to have been diagnosed by a neurologist or pain therapist, respectively. Patients were recruited in part from our tertiary outpatient headache clinic and in part via the German CH support group organization (CSG), which might have led to an overrepresentation of severely affected patients. Therefore, results might not be directly generalized to the total population of CH patients. Discriminant validity was not investigated in the present study because we decided against including another (unrelated) questionnaire in the already long survey. Responsiveness could not be evaluated in the current setting due to small patient numbers. However, this is a goal for a further study.