Development of the prenatal health inventory of behaviors (PHI-B)

Health promoting behaviors have a critical influence on outcomes during pregnancy, at delivery, and on life-long health in women. Despite the significance of these behaviors, scientifically sound and comprehensive measures of health behaviors during pregnancy and preconception are lacking. The purpose of this study was to develop a comprehensive, psychometrically sound inventory to measure women’s health behaviors associated with the preconception and pregnancy time periods. The study was divided into four phases. Phase I focused on conceptualization of prenatal health behavior utilizing Rubin’s construct of maternal tasks and an extensive literature review. Critical domains and definitions of healthy behavior associated with pregnancy were generated. During Phase II items representing these domains were developed with help of perinatal experts and construction of the inventory took place. Phase III consisted of a pilot study. Individual pregnant women, as well as two groups of pregnant women, completed the inventory and critiqued it for readability, redundancy, format, cultural sensitivity, and completion time. Based on their comments and item analysis, the inventory, renamed the Prenatal Health Inventory of Behaviors (PHI-B), consisted of 45 items. Phase IV was the main study to test reliability and validity of the PHI-B in a multi-ethnic sample of 333 pregnant women. During this phase the PHI-B demonstrated satisfactory internal consistency reliability for the scale as a whole (coefficient alpha = .82), but coefficient alphas for the theorized domain subscales ranged from poor to adequate. Preliminary factor analysis and a content validity index failed to support the theorized domain subscales. Further factor analysis revealed a five-factor solution, somewhat dissimilar from the original theorized domains of health behavior. These five factors were named Surveillance of Health, Good Health Practices, Ensuring Safety, Balancing Health, and Adherence. Evidence of test-retest reliability for these factor-analytic based subscales was found. Evidence of concurrent validity was provided by obtaining Pearson correlations between scores of well-being, depression, and physical discomforts, while evidence of convergent and divergent validity of the inventory was found with a general measure of health behaviors and a measure of social desirability, respectively. Assessing factor-analytic subscales for hypothesis-based construct validity revealed interesting associations, particularly for stress management and depressive symptoms.