P1 "Validation of consumption information on alcohol use within the standardized M-CIDI interview"

Dipl.-Psych. Sören Kuitunen-Paul

Supervisior: Prof. Dr. Hans-Ulrich Wittchen (LMU München, TU Dresden)

submitted: 13. Dezember 2018

Background. Alcohol consumption is both a prevalent behavior (Rehm et al., 2003) and potentially harmful, causally increasing the risk of DSM-5 Alcohol Use Disorder (AUD) development, morbidity, and mortality (Parry, Patra, & Rehm, 2011; Rehm, Marmet, et al., 2013; Shield, Rylett, & Rehm, 2016). Knowledge of alcohol consumption is crucial for the assessment and therapy of AUD and related consequences (Kuitunen-Paul, Manthey, et al., 2017; Rehm, Marmet, et al., 2013). Unfortunately, consumption self- reports are often biased (Greenfield & Kerr, 2008; Kip, Spies, et al., 2008; Tourangeau & Yan, 2007) so that the validity of consumption estimates on individual and population levels is generally diminished.

Aims. This thesis examines assessment material characteristics of a widely used instrument, the Munich Composite International Diagnostic Interview (M-CIDI) (Lachner et al., 1998; Wittchen & Pfister, 1997), as a possible source for bias. This includes:

(a) real-life beverage assignment across presented drink groups; (b) number and relevance of presented drink groups;
(c) standard drink conversion rules;
(d) conversion of categorical drinking frequency.

Methods. Overall, we analyzed M-CIDI consumption reports of N = 6,420 persons from population, community, and inpatient samples coming from a survey, an epidemiological and two experimental studies (AF-CIDI, APC, DEGS-MH1, LeAD). Additionally, k = 23,569 alcoholic beverages sold in Germany between 2010 and 2016 were categorized according to M-CIDI drink groups. Statistical methods included correlation analyses, MANOVAs for single and repeated measures, as well as multivariate / ordinal / logistic regressions.

Results. Main findings regarding each aim were:

(a) Alcoholic beverages were frequently misclassified (20 %), especially when they were rarely consumed or when they were not listed in the M-CIDI supplementarysheet. Resulting differences in alcohol-by-volumes estimates were comparably low (Kuitunen-Paul et al., under review).

• (b)  The M‐CIDI supplement sheet displays more categories than comparable

  instruments (AUDIT, TLFB, WHO‐CIDI). Beer, wine, and spirits represent the most

  prevalent drink categories (Kuitunen-Paul, Rehm, et al., 2017).

• (c)  Traditional M-CIDI standard drink conversion factors were inconsistent for different pouring sizes of the same drink and, to a smaller extent, across drink

  categories (Kuitunen-Paul, Rehm, et al., 2017).

• (d)  Converting categorical frequencies did not cover the full range of drinking

  frequencies and leads to a maximal coverage of 44 % regarding official alcohol sales numbers (Kuitunen-Paul et al., in preparation).

Conclusions: We found clear evidence that the M-CIDI alcohol consumption assessment results in meaningful self-reports despite interviewee-based biases, however, a number of revisions to the supplementary sheet are warranted. Therefore, we presented suggestions including:

(a),(b) reduced number of presented drink groups and vessel sizes;

(c) recalculation and unification of standard drink conversion factors, including the option to report the remembered amount and alcohol-by-volume of consumed beverages;

(d) explicit conversion factors for each frequency category.

Future studies might not only try to replicate the presented findings, but rather assess further influences on self-reports, e.g., cognitive biases, reference period, social desirability, stigma avoidance, or acute alcohol effects. While this thesis helps to define standards for interview and questionnaire assessments, the future of self-reporting might include momentary assessment or biomarkers rather than retrospective self- reports.

Publications:

Kuitunen-Paul, S., Kuitunen, P. T., Kadrić, F., Lachenmeier, D. W., Čolić, J., Leonhardt, L., & Scheffel, C. (submitted). Assignment of alcoholic beverages in the Munich Composite International Diagnostic Interview (M-CIDI): An online survey among German students and non-students.

Kuitunen-Paul, S., Rehm, J., Lachenmeier, D. W., Kadrić, F., Kuitunen, P. T., Wittchen, H.-U., & Manthey, J. (2017). Assessment of alcoholic standard drinks using the Munich composite international diagnostic interview (M-CIDI): An evaluation and subsequent revision. International Journal of Methods in Psychiatric Research, 26 (3), e1563. doi:10.1002/mpr.1563

Kuitunen-Paul, S., Höfler, M., Zimmermann, U. S., Sommer, C., Smolka, M. N., Sebold, M., . . . Wittchen, H.-U. (in preparation). A number is a number is a number: Modelling conversion factors for categorical drinking frequencies in epidemiological research.