Using latent class analysis to examine usage patterns of website features for pregnant women

Background: Online interventions have emerged as a popular strategy to promote healthy behaviors. However, there is little agreement about how best to measure involvement in the intervention and what demographic characteristics might be associated with involvement. The objectives of this study were to: 1) identify patterns of website usage in an online intervention to reduce pregnancy weight gain and 2) determine if demographic characteristics differed across these patterns.

Methods: We report on a cohort of racially and socioeconomically diverse pregnant women ages 18-35 enrolled in a randomized controlled trial to prevent excessive gestational weight gain. Women were randomized to receive: website access with articles, resources, and blogs (control, n=557) or the same website with an additional weight tracker and goal setting tools (intervention, n=1112). Latent-class analysis was used to identify patterns of usage and chi-squared analysis was used to examine associations with demographic characteristics including: income, BMI, race, and age.

Results: Four distinct patterns of usage emerged in our intervention and control arms: non-users (21% and 18%, respectively); minimal users (34% and 34%); viewers-of-content (34% and 34%); and super-users (11% and 14%). Overall patterns of website usage were similar between the intervention and control website, but the demographics of those patterns varied. For instance, among super-users there were a higher proportion of overweight/obese women in the intervention arm compared to the control arm (p=0.09).

Discussion: These emergent patterns illuminate how website features cluster among a diverse study population and could guide the development of future interventions.