Ferrer, Robert Louis; Cruz, Inez; Burge, Sandra; Palmer, Ray (2014). 'Capabilities for healthy diet and physical activity and their relationship to behaviors and body mass index' Paper presented at the annual conference of the HDCA, 2-5 September 2014, Athens, Greece.

Introduction

Global shifts in diet and physical activity patterns are leading causes of preventable morbidity and mortality. Despite behaviors' complex genesis, clinical assessment and intervention has been grounded in a narrow evaluative space, largely limited to individuals' behavioral choices and motivations, with modest success rates. In recent papers, we made a theoretical(Ferrer & Carrasco, 2010) and empirical case(Ferrer, Cruz, Burge, Bayles, & Castilla, 2014) for using the Capability Approach to expand the evaluative space for health behavior interventions. Beginning with qualitative work in a disadvantaged community in South Texas, we identified important determinants of practical opportunities. From those findings we developed and tested the Capability Assessment for Diet and Activity (CADA), a structured questionnaire.

The purpose of this study is to attempt to confirm the CADA measurement model in a diverse primary care population and evaluate its performance as a predictor of diet, physical activity, and body mass index.

Methods

The study was approved by the Institutional Review Board at the University of Texas Health Science Center at San Antonio and other participating institutions. 

Participants were recruited from a research network of 8 primary care practices in Texas. Inclusion criteria included age 18 and older, speaking English or Spanish. The only exclusions were unwillingness to participate or cognitive impairment that precluded completing questionnaires. 

Measures: Practical opportunities for diet and physical activity were measured with a 35-item survey (CADA). Subscales for diet and activity cover neighborhood opportunities, convenience and cost, family support, non-family support, barriers, and time. As additional explanatory covariates, we measured health literacy (Newest Vital Sign instrument), locus of control for health behaviors, intention and perceived behavioral control (Theory of Planned Behavior), age, sex, educational attainment, race/ethnicity, and monthly income. We measured behavioral outcomes — diet and physical activity — using the Starting the Conversation - Diet instrument and the International Physical Activity Questionnaire, validated self-report measures that are feasible to administer in primary care settings. As a biological outcome, we calculated body-mass index (BMI) as kg/m2 from height and weight measured at the clinic visit.  

Approach: Confirmatory factor analyses with structural equation modeling sought to revalidate the CADA scales in a larger and more demographically diverse sample. Next, we evaluated construct validity of the capability scales as predictors of diet, physical activity, and BMI in models with and without additional explanatory variables. Robeyns's conceptual framework(Robeyns, 2005) guided the analysis, which used structural equation models to evaluate relationships between resources for healthy behaviors, conversion factors, and behavioral outcomes.

Results

Across the 8 sites, 746 patients were enrolled. The survey participation rate was 77%. The sample was predominantly female (67.7%), Hispanic (54.3%), and low income (50.2% earning less than 1500 USD/month). 

CFA models confirmed the factor structures, with good model fit for individual factors and higher-level factors for resources and conversion factors. 

Inverse gradients between income and health behavior capabilitiies were evident.

In subsequent models, the Dietary Conversion factor was significantly inversely related to BMI. A one standard deviation positive change was associated with a one-half standard deviation lower BMI. Both Activity factors (Conversion and Resources) were significantly independently and inversely related to BMI. A one standard deviation increase in these factors each contributed approximately a 1/5th decrease in BMI. 

CADA scales for diet and activity were also statistically significant predictors, respectively, of achieved diet and activity. For example, Activity Conversion Factor scores demonstrated a strong statistically significant relationship with weekly physical activity minutes, with both a direct effect and an indirect effect through intentions to engage in physical activity. Moving from low to high scores on the Activity Conversion Factor doubled the probability of achieving the recommended level of 150 activity minutes per week.

Study limitations include self-report measures of diet and activity and a cross-sectional design in which temporal relationships between capabilities and outcomes cannot be ascertained. For example, obesity could reduce practical opportunities for physical activity rather than the other way around.

Conclusions

 A mechanistic model guided by the Capability Approach confirmed theoretically predicted patterns of resources' and conversion factors' relationships to achieved diet and physical activity as well as BMI. Measuring capabilities for health behaviors is feasible and effects are robust even after accounting for participants' intentions to engage in healthy diet and activity. This framework enlarges the traditional evaluative space -- motivation and choice -- for health behavior interventions. More broadly, these data from a detailed health-related subset of overall capability help support the connection between theory and practical applications.