´óÏóÊÓƵ

• Claims data or
• General ledger data spanning periods before, during, and after transformation

Analysis Methods:

• Descriptive analysis (trend analysis)
• Econometric modeling (linear regression, difference-in-difference method)

• Does not require burdensome data collection
• Analyses can be adjusted for patient case-mix, demographics, and other variables, enhancing the applicability of results beyond study settings

Drawbacks/Challenges:

• Claims data do not necessarily reflect costs of primary care transformation efforts incurred by clinics
• Changes in cost trends can be caused by factors other than PCMH transformation efforts

Keys to Success:

• Use linear regression to control for patient case-mix, demographics, and other variables
• Consider fixed or random effects models to address unobserved variation
• Consider a difference-in-difference method to compare differences in costs in transformed vs. untransformed clinics before and after the transformation effort
• Use a comparable control group to address attribution challenges; consider using propensity score matching to identify a comparable control group

Analysis Methods:
Basic accounting methods (multiply number of units by unit costs and add together).
If cost estimates are produced for multiple clinics, results can be stratified to show how costs vary by geography, clinic type, transformation components implemented, and other characteristics.

• Can be used to account for uncompensated time spent and opportunity costs in addition to outlays
• Does not require statistical analysis software or skills

Drawbacks/Challenges:

• Burdensome data collection
• During the maintenance period, staff cannot distinguish between time spent on PCMH activities and their regular jobs
• Potential for recall bias in retrospective studies

Keys to Success:

• Ample time and adequate staff for data collection
• Pre-established rapport with clinics providing data
• Web-based data collection tools for data collection efforts across large numbers of clinics