HMC Supplement Progress Report - OUTCOMES

February 2006

Supplement Title:                 Transbehavioral and Quality of Life Outcomes: Cross-Site Analyses and Resources

Supplement Chair/
Affiliation:                             Deborah J. Toobert, Ph.D., Oregon Research Institute

Purpose:                               To advance the science of behavior change and its evaluation by analyzing results using common                                              metrics and relating interventions to outcomes and maintenance based on epidemiologic risk and                                              patient functioning, and to advance the measurement and evaluation of behavioral intervention                                              outcomes, investigate important linkages among these outcomes, and identify intervention and                                              social-environmental factors influencing these outcomes.

Outcomes
Sharepoint Group:                Sharepoint Contact – Lisa Strycker
(In addition to HMCRC Members)

Progress to Date:  Our most important task to date has been to ensure that all participating sites are collecting the necessary measures for cross-site comparisons. We prepared and emailed a package of recommended instruments to all sites, and asked each member of the HMC Outcomes Supplement group to review these measures along with our "Common Constructs" table to confirm that an appropriate measure would be collected for each of the following at BASELINE:  1. Quality of Life. Since we plan to use the Healthy Days quality of life measure, all sites needed to add this at baseline. We provided a copy of the instrument to the sites and asked for verification that it had been added.  2. Diet. Those sites not focused on diet (Williams, Roll, Hughes) were asked if they had added the NCI fat and fruit-and-vegetable screeners at baseline. Dr. Williams determined that it made more sense to use his BCC sample for our supplement because he targeted diet in that study. Drs. Roll and Hughes agreed to add a diet measure. 
3. Physical Activity. With the exception of Drs. Elliot and Williams’ sites, all sites are collecting the CHAMPS. We clarified with Dr. Elliot the exact PA measure she is collecting at baseline, and determined with Geoffrey Williams that, in using his BCC sample, we will have an appropriate baseline PA measure.  4. Smoking. We clarified with each site exactly what smoking measure/s they were collecting at baseline.  5. Body Mass Index. All sites were already collecting BMI except Drs. Williams and Roll, who said they would collect height and weight (to compute BMI) for this project.  We also asked each site to provide an update of their geocoding status. With the exception of two sites, all sites are having their samples geocoded. We have been involved in adding variables to
the Boston environmental project which we also hope to use on this project. Geocoding is a problem for the Elliot group. Their intervention was randomized by fire station, and there are 12 to 24 stations within which firefighters trade station assignments. Also, the firefighters spend 1 day at the station, then 2 days at home. We determined that geocoding would not fruitful for the Elliot site, but we plan to use data from that site for analyses not involving geocoded data. Dr. Roll will not be able to geocode his sample because many participants are homeless.

Challenges to Date:  None

Next Steps:  As soon as all the baseline data are collected, Lisa Strycker will begin data management and analytic procedures, including: 1. Creating a data dictionary and coding rules that will be followed to transfer data in a common format. 2.  Secondary
data cleaning (initial cleaning will be performed by the individual sites) once the data are sent to ORI for analysis.  3.  Merging of the data sets from all sites, and merging of individual-level data with GIS data (e.g., U.S. Census variables such as racial composition, unemployment, poverty, vacant housing and community data, such as shopping, parks, traffic, restaurants).  4.  Conducting preliminary analyses to identify outliers and evaluate distributions, assess whether the data meet assumptions of the tests used, and transform data as needed.  5.  Conducting primary analyses. With more than 2,500 participants and an average of 500 per study, we will have good power to address all of our study questions. Analyses will be designed to evaluate the performance, characteristics, inter-relationships, intervention effects, and environmental context of the quality-of-life and transbehavioral outcomes.