Objective: The present article evaluates the psychometric properties of the Portuguese-language versions of the CAGE, TWEAK and T-ACE questionnaires when used to identify alcohol misuse during pregnancy and compares these to historical data from the English-language versions. Also, construct validity of the Portuguese versions of those instruments was assessed.
Method: Data were collected from March to September 2000 in three large public maternity wards in Rio de Janeiro, Brazil. Sample size varied according to the analysis. An intra-observer reliability evaluation involved 95 replications carried out within a 24-48 hour period. Kappa was used as an estimator. Cronbach's alpha coefficient, whether or not each item from the analysis was excluded, and the item-rest point-biserial correlations addressed internal consistency (N = 786). For the appraisal of construct validity (n = 528), the relationship between alcohol misuse (evaluated by each instrument according to different cutoff points) and several putative underlying theoretically related dimensions (schooling, age, smoking, marital status, social support, illicit drug use, domestic violence and per capita household income) were evaluated.
Results: All kappas were above 0.70. The alpha coefficients for the Portuguese versions ranged from 0.48 to 0.68, a finding consistent with studies assessing the original instruments in English. Yet, irrespective of the instrument concerned a systematic increase was observed whenever the item on morning alcohol use ("Eye-opener") was removed from the analysis. Regarding construct validity, most tested hypotheses were corroborated, except for the CAGE using the cut-off point of 2.
Conclusions: This study confirms other authors' findings that all but the aforementioned CAGE may be recommended to identify heavy alcohol use during pregnancy in general as well as in Portuguese-speaking populations in particular.