Time–motion analysis of external facilitation for implementing the Collaborative Chronic Care Model in general mental health clinics: Use of an interval-based data collection approach

Abstract

Background: Facilitation is an effective strategy to implement evidence-based practices, often involving external facilitators (EFs) bringing content expertise to implementation sites. Estimating time spent on multifaceted EF activities is complex. Furthermore, collecting continuous time–motion data for facilitation tasks is challenging. However, organizations need this information to allocate implementation resources to sites. Thus, our objectives were to conduct a time–motion analysis of external facilitation, and compare continuous versus noncontinuous approaches to collecting time–motion data. Methods: We analyzed EF time–motion data from six VA mental health clinics implementing the evidence-based Collaborative Chronic Care Model (CCM). We documented EF activities during pre-implementation (4–6 weeks) and implementation (12 months) phases. We collected continuous data during the pre-implementation phase, followed by data collection over a 2-week period (henceforth, “a two-week interval”) at each of three time points (beginning/middle/end) during the implementation phase. As a validity check, we assessed how closely interval data represented continuous data collected throughout implementation for two of the sites. Results: EFs spent 21.8 ± 4.5 h/site during pre-implementation off-site, then 27.5 ± 4.6 h/site site-visiting to initiate implementation. Based on the 2-week interval data, EFs spent 2.5 ± 0.8, 1.4 ± 0.6, and 1.2 ± 0.6 h/week toward the implementation’s beginning, middle, and end, respectively. Prevalent activities were preparation/planning, process monitoring, program adaptation, problem identification, and problem-solving. Across all activities, 73.6% of EF time involved email, phone, or video communication. For the two continuous data sites, computed weekly time averages toward the implementation’s beginning, middle, and end differed from the interval data’s averages by 1.0, 0.1, and 0.2 h, respectively. Activities inconsistently captured in the interval data included irregular assessment, stakeholder engagement, and network development. Conclusions: Time–motion analysis of CCM implementation showed initial higher-intensity EF involvement that tapered. The 2-week interval data collection approach, if accounting for its potential underestimation of irregular activities, may be promising/efficient for implementation studies collecting time–motion data.