Implementation Outcomes of a Health Information Technology Program For Vulnerable Diabetes Patients

This study evaluated the Self-Management Automated Real-Time Telephone Support (SMART-Steps) program, which was developed through a previous Agency for Healthcare Research and Quality (AHRQ) grant (R18 HS 017261). SMART-Steps used an automated telephone self-management (ATSM) support system to provide monitoring and education for diabetic patients enrolled in the San Francisco Health Plan (SFHP). The focus of this more recent project was on examining the fidelity of the implementation of SMART-Steps to better-inform tailoring of health information technology interventions, particularly those focused on diverse populations and multiple languages.

The fidelity analysis focused on two primary areas:  health systems integration and moderating factors. The former included population-based data linkage to determine patient eligibility; electronic exchange of health information to deliver the intervention to patients; and electronic integration of health information to identify patients requiring a health coach callback or a callback for a medication or lab trigger. The moderating factors included representativeness of participants (reach); quality of intervention delivery in health coach callbacks (responsiveness); and consistency of delivery of the intervention over time.         

The specific aims of this project were to:

• Estimate the proportion of patients identified as SMART-Steps-eligible who were ineligible, and describe reasons for ineligibility.
• Determine if SMART-Steps patients received ATSM calls with intended frequency (weekly), content (questions/language), and duration (27 weeks).
• Estimate the frequency with which electronic exchange for out-of-range triggers (from ATSM and SFHP clinical registry/pharmacy claims) resulted in a documented callback, in a sample of patients stratified by language.
• Compare SMART-Steps-enrolled to -eligible patients for clinic, age, language, sex, hemoglobin A1c, insulin use, blood pressure, cholesterol, and prior medication non-adherence.
• Describe the quality of intervention delivery from care managers’ callbacks, including frequency of supplemental self-management support, call duration, adherence to protocols, and creation of patient action plans, for a diverse sample of patient triggers.
• Over the course of SMART-Steps implementation, identify differences in average length of callbacks, proportion of callbacks made for triggers, and whether wait-list patients (vs. not) had differential ATSM engagement.
• Summarize fidelity assessment findings, adaptations, and implications for real-world ATSM implementation and related health IT interventions into a guide, with SFHP partnership.        

The assessment was conducted between 2009 and 2011 among 252 patients with type 2 diabetes. Fidelity to health systems integration was high. Eligibility data linkages were successful, with 76 percent of potential participants determined to be eligible. Ninety-six percent of patients received correctly sequenced ATSM calls; 70 percent of those requiring a callback received one from a health coach. There was a high level of variation in callbacks by type of patient problem and study language, warranting consideration for implementing health IT interventions in diverse populations.