If the Morning-Evening Questionnaire (MEQ) is able to predict the actigraphy-based acrophase, how does its reduced, five-item version (rMEQ) perform?
A. Montaruli,L. Galasso,F. Carandente,J. Vitale,E. Roveda,A. Caumo
DOI: https://doi.org/10.1080/07420528.2017.1306708
2017-04-20
Chronobiology International
Abstract:Dear Editor, We have recently shown that the score yielded by the Morning-Evening Questionnaire (MEQ) (Horne & Ostberg, 1976) can be used to predict the acrophase measured by actigraphy in a sample of 54 college students of North Italy (Roveda et al., 2017). The MEQ score was very well correlated with the acrophase (r = 0.84, p < 0.001), and the linear regression equation linking the MEQ score to the acrophase is: acrophase = 1238.7–5.49·MEQ, thus indicating that a one-point increase in the MEQ score is accompanied, on average, by a decrease of approximately 5 min in the acrophase. Since the online publication of the article, we were contacted by potential users of this equation who do not have at their disposal theMEQ score, but only its reduced version (rMEQ) based on 5 over 19 items (Adan & Almiral, 1991). These inquiries motivated the present follow-up which is aimed to address the following questions. If the MEQ score is capable of satisfactorily predicting the acrophase, what is the performance of its shorter version rMEQ? Does the predictive ability of rMEQ remain good enough or the loss of information inherent in the item reduction take its toll? Such questions are clearly relevant for the investigators planning to use a questionnairebased approach to gain insight into the acrophase. Therefore, we returned to the original database (Roveda et al., 2017; Vitale et al., 2015) and calculated the rMEQ with 5 items, whose numbers in the MEQ were 1, 7, 10, 18, and 19 (Adan & Almiral, 1991). The rMEQ score was 13.8 ± 4.5 (mean ± SD, n = 54). Then, the exact same regression analysis procedures were applied as previously described (Roveda et al., 2017) so that a new linear equation was generated enabling the rMEQ to predict the acrophase. The scattergram of the data together with the best-fit regression line is showed in Figure 1. The equation of the regression line is: acrophase = 1182.4–16.24·rMEQ. The slope of the regression line (−16.24 min per unit of rMEQ) was significantly different from 0 (p < 0.001). Its value indicates that a one-point increase in the rMEQ score was accompanied, on average, by a decrease of approximately 16 min in the acrophase. Pearson’s r was 0.79 (p < 0.001), and thus only slightly lower than the value r = 0.84 that measured the strength of the relationship between theMEQ and the acrophase (Roveda et al., 2017). If we make reference to the coefficient of determination (i.e., R), the MEQ explained 70% of the variability present in the acrophase, while rMEQ explained 62%. All in all, the performance of the rMEQ was reasonably good. This is probably related with the fact that rMEQ has been proven to be a very good measure of circadian typology (Adan & Almiral, 1991; Adan & Natale, 2002; Adan et al., 2012; Chelminski et al., 2000; Di Milia et al. 2013; Natale et al., 2006). As pointed out by Adan et al. in their review paper (Adan et al., 2012), many authors have reported an elevated degree of agreement between MEQ and rMEQ, with Pearson’s correlation coefficient ranging from satisfying to excellent (0.69–0.90). In our study, the demonstrated correlation was excellent (r = 0.95, p < 0.001), probably because our sample consisted of college students having a very restricted age range (mean age ± SD was 22.1 ±