Migraine screen questionnaire: further psychometric evidence from categorical data methods

DOI: https://doi.org/10.1186/s12955- 020- 01361- 9

Abstract-Summary Psychometric investigations of tools used in the screening of migraine including the migraine screen questionnaire (MS-Q), using an adequate statistical approach is needed.

We assessed the psychometric properties of the migraine screen questionnaire

(MS-Q) using categorical data methods.

The cumulative variance rule (> 40%), the Kaiser’s criteria (Eigenvalue> 1), the Scree test and, the parallel analysis (minimum rank) identified a 1-factor model for the MS-Q with the factor loadings in the range of 0.78–0.84.

Fit indices favored a 1-factor model of the MS-Q as indicated by comparative fit index (0.993), weighted root mean square residual (0.048), root mean square error of approximation (0.067), the goodness of fit index (1.00), and non-normed fit index (0.987).

Discriminative validity was supported by significantly higher scores for the total

and all the MS-Q items except one among those with complaints of attention.

The categorical methods support the psychometric validity of the MS-Q in the

study population.

Extended: Discriminative validity was assessed by the Mann Whitney U test. The findings of this study, along with those of previously published diagnostic accuracy studies in clinical populations, provide strong evidence for its use in screening migraine in both clinical as well as research settings.

Introduction A brief, reliable, and valid questionnaire will be helpful to the primary care physi- cians in the screening of migraine and decrease its under-diagnosis.

Migraine screen questionnaire (MS-Q) is a brief measure of migraine screening with favorable diagnostic validity, test-retest reliability, and internal consistency-as determined by the Cronbach’s alpha [171].

The psychometric characterization of migraine and headache tools has been inclined towards an examination of test-re-test reliability, concurrent validity, and internal consistency [170, 171, 271–274].

Given the value of the MS-Q as a potential and clinically useful screening tool, further investigation of its measurement properties including internal consistency,

526

3 Diagnosis

factorial validity, and discriminative validity, especially taking account of the cate- gorical nature of the MS-Q item score is needed.

The present study examined the factorial validity, internal consistency, and dis- criminative validity of the MS-Q according to categorical data assumptions in uni- versity students.

Material and Methods Spearman correlation between the MS-Q item and total score indicated homogene- ity and item discrimination.

The sample size adequacy and suitability of the MS-Q score for factor analysis were determined by Bartlett’s test of Sphericity, Determinant, Kaiser-Meyer-Olkin (KMO) Test of Sampling Adequacy (95% confidence interval), communality and inter-item tetra-choric correlations.

Tetra-choric correlations (estimated using bootstrap sampling) for inter-item scores of the MS-Q were used for factor analysis because these are dichotomous variables.

H-latent is a measure of correlations between the factor and the continuous latent response score that is supposed to underlie the observed categorical scores of the MS-Q items.

H-observed is a measure of correlations between the factors and the observed

item scores and is necessarily lower than the H-latent [275].

It is a measure of the absolute loadings of the MS-Q item scores on the second

factor of minimum rank factor analysis.

Results The MS-Q scores in the studied university students fulfilled the conditions for the factor analysis.

There were adequate linear combinations between the MS-Q item scores, as

indicated by the results of Bartlett’s test of sphericity (< 0.001) [276].

No problems of multicollinearity and singularity were present in the MS-Q item

scores, as suggested by the determinant score (0.282) [276].

This further supported the factorability of the MS-Q scores by substantiating evidence of the absence of both problems of multicollinearity and singular- ity [277].

A ROC curve analysis with the dichotomous variable of attention (No problem/ Attention problem) and the MS-Q total score as test variable revealed an area under the curve of 0.66 with a 95% confidence interval of 0.58–0.74.

Discussion Evidence showed that the unidimensional model of the MS-Q had adequate facto- rial validity, excellent internal consistency, strong internal homogeneity, and suffi- cient discriminative validity and item discrimination in the university students.

All the five items were relevant for the construct validity of the MS-Q in the

study population, as implied by the communality criteria [278].

This is the first study examining the factorial validity and model fit of the MS-Q; hence a direct comparison with previously studied populations cannot be performed.

3.3 Clinical Diagnosis

527

Significantly higher scores for the total and all the MS-Q items except one sup-

port a known group: discriminative validity of the MS-Q in the study population.

The present study addresses this gap for migraine screening by reporting the results of factorial validation on the MS-Q. Another merit is the use of McDonald’s omega for assessing internal consistency following the requirements of the univari- ate distribution [279].

Conclusions The study findings provide further psychometric validation by providing evidence of adequate factorial validity, excellent internal consistency, strong internal homo- geneity, and adequate discriminative validity and item discrimination in the study population.

The findings of this study, along with those of previously published diagnostic accuracy studies in clinical populations, provide strong evidence for its use in screening migraine in both clinical as well as research settings.

Acknowledgement A machine generated summary based on the work of Manzar, Md. Dilshad; Hameed, Unaise Abdul; Salahuddin, Mohammed; Khan, Mohammad Yunus Ali; Nureye, Dejen; Wakene, Wakuma; Alamri, Majed; Albougami, Abdulrhman; PandiPerumal, Seithikuruppu R.; Bahammam, Ahmed S. 2020 in Health and Quality of Life Outcomes

Characterization of migraineurs presenting interictal widespread pressure hyperalgesia identified using a tender point count: a cross-sectional study