Validation of Transitional Shift Patterns Among Women with Multiple Sclerosis

Validation of Transitional Shift Patterns Among Women with Multiple Sclerosis: Application of Growth Mixture Modeling

Kin-Kit Li, Bradley J. Cardinal, and Gianni F. Maddalozzo

Abstract

Developmental studies (i.e., transition among stages over time) using the Transtheoretical Model (TTM) of exercise behavior change have been limited. Furthermore, when they have occurred, the analytic strategies employed usually fail to consider the individuals’ developmental trajectories and the possible heterogeneity of the trajectory patterns. The purpose of this study was to incorporate these characteristics and examine the validity of the five transitional shift patterns (i.e., stable active, stable sedentary, activity adopters, activity relapsers, and perpetual preparation) proposed by Cardinal, Engels, and Smouter (2001) empirically using a growth mixture modeling technique. It was hypothesized that a model with five trajectory patterns would fit the data the best. Participants for this study were 105 women with multiple sclerosis (M age = 50.9) living in Oregon. Questionnaires including TTM variables measuring exercise behavior change were delivered to the participants three times with one year apart without any intervention. The response rate was 79%, with 78 participants completed all the TTM items. Missing values were handled using the maximum likelihood estimation provided by M-plus (final n = 103). The stages of change (SOC) items across time were used as the dependent variables. First, the intercept and slope of the unrestricted model were tested. Second, models with different numbers of trajectory patterns were compared. The model fit of the unrestricted model with no class classification was satisfactory (χ2 = 4.14, p = .04; CFI = .96). Apart from that, the model fit of the model with six classes (BIC = 841.85; AIC = 786.52; entropy = .92; Lo-Mendell-Rubin LRT ratio = 53.77, p = .02) was significantly better than the mode with five classes (BIC = 885.17; AIC = 837.75). Although the results did not support the proposed five-class model, the six trajectory patterns matched well with the transitional shift patterns, with two classes representing the pattern of activity adopters at the same time. This was the first study to incorporate individuals’ developmental trajectories and the heterogeneity of the trajectory patterns into SOC. The preliminary results partially supported the transitional shift patterns of SOC. Practically the results suggest that exercise promotion strategies can be focused on individuals with different developmental trajectory patterns. Future study should use a larger sample size, which will provide a platform for examining the determinants and outcomes of the different patterns.

Supported by National Multiple Sclerosis Society Grant No. PP1026.