Challenges and recommendations for measuring time devoted to implementation and intervention activities in health...

There is little guidance for conducting health equity-focused economic evaluations of evidence-based practices in resource-constrained settings, particularly with respect to staff time use. Investigators must balance the need for low-touch, non-disruptive cost data collection with the need for data on providing services to priority subpopulations.

This investigation took place within a pilot study examining the implementation of a bundled screening intervention combining screening for social determinants of health and colorectal cancer at four federally qualified health centers (FQHCs) in the Boston metropolitan area. Methods for collecting data on personnel costs for implementation and intervention activities, including passive (automatic) and active (non-automatic, requiring staff time and effort) data collection, as well as three alternate wordings for self-reporting time-use, were evaluated qualitatively using data collected through interviews with FQHC staff (including clinicians, population health staff, and community health workers) and assessments of data completeness.

Passive data collection methods were simple to execute and resulted in no missing data, but missed implementation and intervention activities that took place outside planned meetings. Active cost data collection using spreadsheets was simple for users when applied to care processes already tracked in this fashion and yielded accurate time use data. However, for tasks where this was not typical, and when tasks were broken up over multiple sessions, spreadsheets were more challenging to use. Questions asking about time use for a typical rather than specific time period, and for typical patients, yielded the most reliable and actionable data. Still, even the best-performing question had substantial variability in time use estimates. Participants noted that patient characteristics of interest for equity-focused research, including language spoken, adverse social determinants of health, and issues related to poverty or mental health, all contributed significantly to this variability.

Passively collected time use data are the least burdensome and should be pursued in research efforts when possible, but should be accompanied by qualitative assessments to ensure the data are an accurate reflection of effort. When workflows are already tracked by active data collection, these are also strong data collection methods. Self-reported time use will be most accurate when questions inquire about "typical" tasks and specific types of patients.