Thai-version Migraine Disability Assessment (MIDAS) questionnaire: concurrent validity, test–retest reliability, internal consistency, and factors predictive for migraine-related disability

The present study was conducted with a prospective design at the Chulalongkorn Comprehensive Headache Centre at King Chulalongkorn Memorial Hospital, Faculty of Medicine, Chulalongkorn University in 2011–2012. The Chulalongkorn Institutional Review Board approved the study (approval in 2554 (2011), endorsement No. 129/2560, IRB No. 418/54).

Of the 58 participants finishing the first questionnaire, 32 mailed back the second. Of the 32 returned second questionnaires, 1 had some missing data. Therefore, 31 participants were eligible for this study (Figure 1). The baseline demographic and socioeconomic data including median years of age, male to female ratio, education, and occupation; migraine characteristics including median duration of migraine history, pain intensity, pain score, and associated symptoms; and comorbidity between those who were eligible for further analysis (n = 31) and those who were ineligible (n = 27) were not significantly different. Eligible patients were mostly women (97%), with a median age of 49 years and employed (94%) and about half of patients had less than or equal to postsecondary nontertiary education (48%) and comorbidity (48%). For migraine characteristics in eligible patients, the median duration of migraine history was 10 years with a high proportion (84%) of moderate-to-severe pain intensity and mean pain score with a median NRS score of 6, a moderate pain score. Most (87%) had migraine-associated symptoms including nausea (68%), vomiting (45%), photophobia (52%), or phonophobia (48%) and 84% had an episodic type of migraine. Migraine without aura accounted for 68%. When comparing the migraine characteristics between 3 months, 2 months, and 1 month before completing the first Thai-version MIDAS Questionnaire (Table 2), there was no significant difference in median frequency of headache per month with a median of approximately 3–4 attacks, mean pain duration per attack of a median of 2 h, and mean pain score with a median NRS score of 5. Migraine characteristics observed in 3 months, 2 months, and 1 month, and 2 weeks between before completing the first and the second Thai-version MIDAS Questionnaire were not significantly different (Table 3). This confirmed the stable condition of migraine characteristics of the patients before performing the first and second questionnaires because unstable conditions would interfere with the test–retest reliability.

Figure 1

There have been 9 non-English versions of the MIDAS questionnaire, with Italian [31], Japanese [32], Turkish [33, 34], Taiwanese [35], Thai [27, 28], Arabic [36], Greek [37], German [38], and Spanish [39] versions that have been previously reported (Table 7). The development process for a non-English version from an original English version questionnaire included forward and backward translation. A guideline for clear and user-friendly translation, adaptation, and validation of instruments for use in cross-cultural health care research was published in 2011 [40]. In brief, step 1 for translation of the original instrument into the target language (forward translation) must be done by at least 2 independent certified bilingual, bicultural translators whose mother language is the desired target language [40]. In addition, the 2 translators must have a distinct background. Step 2 is a comparison of the 2 translated versions by a third bilingual, bicultural independent translator to resolve any ambiguities/discrepancies. Step 3 is blind back-translation done by 2 other independent, bilingual, bicultural translators with a distinct background and their mother language should be the source original language [40]. Step 4 is a comparison of the 2 back-translated versions with the same purpose as the step 2 [40]. Among the 9 non-English version MIDAS questionnaires, only one, the German-version has followed the described guideline for forward and backward translation. The German-version was tested for internal consistency with a Cronbach α of 0.69 at test and 0.67 at retest and for test–retest reliability with the ICC of 0.88–0.99 in 36 patients with a chronic episodic headache at the pain center and general practice [38]. Italian, Greek, and Spanish versions had 2 bilingual and/or independent experts/speakers to perform the translation, but without mentioning the translators’ linguistic background. The Italian version was tested for internal consistency with a Cronbach α of 0.7 and test–retest reliability with a Spearman correlation coefficient of 0.54–0.77 for the 5 questions individually, and of 0.77 for MIDAS total score in 86 people with migraine without aura at a regional headache center [31]. The Greek version was tested for convergent validity using a RAND-36 questionnaire with significantly negative correlation, for internal consistency with a Cronbach α of 0.71 at test and 0.82 at retest, and for test–retest reliability with an ICC of 0.41–0.81 in 59 patients with migraine and with or without aura at 2 private clinics in Athens [37]. The Spanish version was tested for concurrent validity with headache frequency with Spearman ρ of 0.53, with headache intensity with a Spearman ρ of 0.34; for internal consistency with a Cronbach α of 0.79 and for test–retest reliability with an ICC of 0.63–0.90 in 153 university undergraduate and graduate students diagnosed as having migraine [39]. For other versions with one translator for each process, or without mentioning the process, the Japanese version was tested for criterion validity with diary-based measures with a correlation coefficient of 0.36–0.66 for the 5 questions individual and of 0.66 for MIDAS total score and test–retest reliability with Spearman correlation coefficients of 0.59–0.80 for the 5 individual questions and of 0.83 for MIDAS total score in 99 people with headache, and just more than half of them (50.5%) were diagnosed with migraine at a neurological department and an affiliated clinic [32]. A Turkish version was tested for the validity of the MIDAS total score with the number of days with headache with a Spearman ρ of 0.60, 0.47, and 0.63 at the first, the second, and the third visit, respectively; with internal consistency having a Cronbach α of 0.73, 0.87, and 0.86 at the first, the second and the third visit, respectively; and test–retest reliability with a Spearman ρ of 0.68 for MIDAS total score in people with migraine at 17 neurology clinics [33]. Another Turkish version was tested for validity with diary-based measures with correlation coefficients of 0.54–0.89 for individual questions and MIDAS total score among 3 occasions, internal consistency with a Cronbach α of 0.78–0.90 among questions and 3 occasions, and test–retest reliability with Spearman correlation coefficients of 0.66–0.81 for each of the 5 questions in 60 people with migraine at a headache center [34]. A Taiwanese version was tested for validity with the total number of headache days and headache severity with a correlation coefficient of 0.37 and 0.34, respectively; internal consistency with a Cronbach α of 0.79 and test–retest reliability with Spearman correlation coefficients of 0.48–0.73 for the 5 individual questions and of 0.67 for MIDAS total score in 31 people with migraine at a headache clinic [35]. A Thai version was tested for criterion validity with 13-week diary data with correlation coefficients of 0.32–0.74 for 5 individual questions, the number of headache days and pain severity and test–retest reliability with ICC of 0.58–0.82 for the 5 individual questions and of 0.76 for the MIDAS total score in 93 people with migraine for validity test and 30 for test–retest reliability at a headache clinic conducted between October 2010 and September 2011 [27], the study time straddling the release of the new guidelines in 2011. The other Thai version was assessed for test–retest reliability with correlation coefficients ranging from 0.36 to 0.94 and for internal consistency with an ICC of 0.95 [28]. Among all of the non-English versions, 6 versions were tested for validity, internal consistency, and test–retest reliability [33, 34, 35, 36, 37, 39]. Only 2 Turkish versions [33, 34], had moderate-to-strong validity (Spearman correlation coefficients of 0.40–0.59 considered moderate, and 0.60–0.79 considered strong) and had strong-to-very strong test–retest reliability, but both were in between acceptable and unacceptable for internal consistency (Cronbach α across 0.8).

In the present study, the process involving the forward and backward translation of the Thai-version language questionnaire was a limitation and this study conducted in 2011–2012 in the midst of the release of the new translation guidelines in 2011 [40]. However, the patients who participated were typical and would generally represent people with migraine. Most had episodic migraine with a median frequency 3–4 attacks of headache per month and were mostly stable in migraine conditions (Table 3), prerequisite conditions, for test–retest reliability test of the Thai-version MIDAS Questionnaire. The median pain duration per attack was 2 h and pain NRS score of 5 between 3 months, 2 months, and 1 month. The median pain duration of 2 h in the present study might have been modified by taking acute pain medications. One of the criteria for the diagnosis of migraine is the pain duration between 4–72 h without treatment or with unsuccessful treatment [1]. The confirmation of stable migraine conditions among these participants includes: the migraine characteristics in the past 3 months, 2 months, 1 month, and 2 weeks before completing the first and the second questionnaire (Table 3) was very similar, which would yield excellent test–retest reliability of the questionnaire, ICC of 0.94–0.95 for individual questions, ICC of 0.98 for MIDAS total score, ICC of 0.97 for question A, ICC of 0.89 for question B, and weighted κ of 0.66 (substantial agreement) for MIDAS grade. The Thai-version MIDAS Questionnaire also had acceptable internal consistency, with a Cronbach α of 0.98.

The concurrent validity of the Thai-version MIDAS Questionnaire between the MIDAS total score and mean headache severity (question B) was moderately correlated. The concurrent validity between the MIDAS total score and the mean pain duration per attack and the mean pain NRS was also moderately correlated. However, the concurrent validity between the MIDAS total score and the total number of headache days (question A) and the frequency of headache was very weak or weakly correlated. These results imply that even though the total number of headache days and the frequency of headache were higher, which includes the short pain duration per attack and the low pain score, the headache may not interfere with daily activities. By contrast, when total numbers of headache days were few, but were with long pain duration and high pain score, the migraine would interfere with daily activities causing lost days.

Interestingly, factors predictive for migraine-related disability in the present study were the median of the mean pain NRS score in the past 3 months as an independent predictive factor for migraine-related disability. The frequency of headache at any period was not predictive for migraine disability. This supports the hypothesis that the frequency of headache is not correlated with the MIDAS total score that counts missed daily activities and is not a predictor for migraine disability. To determine whether other factors that were not shown to be predictive for migraine-related disability, a study with a larger sample size should be conducted. Migraine prevention might have an effect on migraine-related disability and is a limitation of this study.

This study had some other limitations including the translation process for the Thai-version questionnaire, a lack of sample size calculation for the internal consistency test, not including migraine treatment in the analysis of factors predictive for migraine-related disability, and not adjusting P-values for multiple analyses.

The present Thai-version MIDAS Questionnaire had moderate concurrent validity, acceptable internal consistency, and excellent test–retest reliability. It would be helpful to assess clinical outcomes. Future study with a standardized translation process for the Thai language questionnaire and larger sample size are required to confirm internal consistency and determine other factors predictive for migraine-related disability.