ABSTRACT
INTRODUCTION: Greater transit use is associated with higher levels of physical activity, which is associated with lower health risks and better health outcomes. However, there is scant evidence about whether health care costs differ based on level of transit ridership. METHODS: A sample (n=947) of members of Kaiser Permanente in the Portland, Oregon area were surveyed in 2015 about their typical use of various modes of travel including transit. Electronic medical record-derived health care costs were obtained among these members for the prior three years. Analysis examined proportional costs between High transit users (3+ days/week), Low transit users (1-2 days/week), and Non-users adjusting for age and sex, and then individually (base models) and together for demographic and health status variables. RESULTS: In separate base models across individual covariates, High transit users had lower total health care costs (59-69% of Non-user's costs) and medication costs (31-37% of Non-users' costs) than Non-users. Low transit users also had lower total health care (69%-76% of Non-users' costs) and medication costs (43-57% transit of Non-user's costs) than Non-users. High transit users' outpatient costs were also lower (77-82% of Non-users). In fully-adjusted models, total health care and medication costs were lower among High transit users' (67% and 39%) and Low transit users' (75% and 48%) compared to Non-users, but outpatient costs did not differ by transit use. CONCLUSIONS: Findings have implications for the potential cost benefit of encouraging and supporting more transit use, although controlled longitudinal and experimental evidence is needed to confirm findings and understand mechanisms.
ABSTRACT
In a randomized controlled trial, we found that a cognitive behavioral program (CBP) was significantly more effective than usual care (UC) in preventing the onset of depressive episodes, although not everyone benefitted from the CBP intervention. The present paper explored this heterogeneity of response. Participants were 316 adolescents (M age = 14.8, SD = 1.4) at risk for depression due to having had a prior depressive episode or having current subsyndromal depressive symptoms and having a parent with a history of depression. Using a recursive partitioning approach to baseline characteristics, we (Weersing et al. 2016) previously had identified distinct risk clusters within conditions that predicted depressive episodes through the end of the continuation phase (month 9). The present study used the same risk clusters that had been derived in the CBP group through month 9 to reclassify the UC group and then to examine group differences in depression through month 33. We found that in this overall very high-risk sample, the CBP program was superior to UC among youth in the low-risk cluster (n = 33), characterized by higher functioning, lower anxiety, and parents not depressed at baseline, but not in the middle (n = 95) and high-risk (n = 25) clusters. Across conditions, significantly more depression-free days were found for youth in the low-risk cluster (M = 951.9, SD = 138.8) as compared to youth in the high-risk cluster (M = 800.5, SD = 226.7). Identification of moderators, based on purely prognostic indices, allows for more efficient use of resources and suggests possible prevention targets so as to increase the power of the intervention.
