Table 3.

Variation Among Risk Stratification Approaches

	Automated (N = 5)	Manual (N = 5)	Hybrid (N = 5)
Description	Programming in the EHR or other database uses pre-selected criteria to assign patients a risk score.	Practice staff or clinicians review patients to generate a risk score, often based on pre-selected criteria.	Any mixture of clinical intuition, automated algorithm, and manual algorithm.
Practice Characteristics	Size: 2 Medium, 3 Large	Size: 3 Medium, 2 Large	Size: 2 Medium, 3 Large
	Location: 2 Rural, 2 Urban, 1 Suburban	Location: 5 Urban	Location: 2 Rural, 3 Urban
	Ownership: 2 Independent, 3 System	Ownership: 2 Independent, 3 System	Ownership: 1 Independent, 4 System
Workflow	Algorithm is programed into EHR or other database, and mapped to data sources	Criteria are developed to systematically assess risk	Criteria are developed to systematically assess risk, and data sources are identified
	Risk scores are generated	Empaneled patient lists are generated for each clinician	Algorithm is programed into EHR* or other database to generate risk scores
		Care team members review each patient	Care team members review each patient
Benefits	Efficient for large populations	Care team member(s) generates risk score	Care team member(s) generates risk score
	Automatically generates and updates scores	Can include information not reportable from EHR	Can include information not reportable from EHR
	Pre-packaged algorithms available	Technical expertise not required	Validated algorithm criteria available
	Risk scores generated within EHR or database
	Validated criteria
Challenges	Technical skills required	Time-intensive	Time-intensive
	Dependent upon consistent EHR documentation	Risk score manually entered into EHR or database	Technical skills required
	Difficulty including psychosocial criteria	Developing new criteria	Dependent upon consistent EHR documentation

EHR, electronic health record.