Page 40 - Zero for nine: Reducing alcohol use during pregnancy via health counselling and Internet-based computer-tailored feedback

Chapter 2
average number of standard drinks per day and his average number of drinking days per week were computed by four questions assessing how often the partner used alcohol during weekdays (0 = never; 4 = four days) and weekend days (0 = never; 3 = three days) in the previous four weeks, and how many standard glasses (equivalent to 10 grams of alcohol) he drank on these respective occasions. The partner’s support was measured with one item on a 5-point scale (I support my spouse in not drinking alcoholic beverages during her pregnancy; 1 = no, definitely not; 5 = yes, definitely).
Data Analysis
Data were analyzed using SPSS 20.0. To assess correlates of prenatal alcohol use, independent sample t-tests and chi-square tests were used to detect differences between pregnant women who did and did not use alcohol. In addition, to assess the relationship between the partner’s influence as perceived by the pregnant woman and as reported by her partner, Pearson’s correlation analyses for pairs of continuous variables and Spearman’s correlation analyses for pairs of an ordinal and a continuous/dichotomous variable were conducted (Field, 2005). Finally, hierarchical logistic regression analyses were conducted. To assess the relative impact of perceived and reported partner influence next to other cognitive factors as derived from the I-Change Model in explaining alcohol use among pregnant women, we tested three models: (a) a model with all I-Change items, including partner influence as perceived by the pregnant woman; (b) a model with all I-Change items in which the pregnant woman’s perception of the partner’s influence was replaced by her partner’s report of this influence; (c) a model with all I-Change items with both the pregnant woman’s perception and her partner’s report of his influence included. The items were entered in two steps. In the first step, all items were included except partner influence; the second step included perceived and/or reported partner influence. Because of the exploratory nature of this analysis, we used a backward approach (Field, 2005), whereby in both steps, variables were excluded from the model if the probability was greater than .10 based on the likelihood ratio statistic (LR).
38