229246 Defining the denominator in Maine's oral-health return-on-investment project: HEDIS or ETGs?

Monday, November 8, 2010

Kala E. Ladenheim, PhD, MSPH , Medical Care Development, Augusta, ME
Margaret I. Gradie, PhD , Medical Care Development, Augusta, ME
Kathleen E. Perkins, MPA , Director, Division of Health Improvement, Medical Care Development, Augusta, ME
Insurance claims are commonly mined to support care management decisions. Maine collects claims from all payers, and several groups are exploring how this data can support population health decisions. The Maine Oral Health return-on-investment (MeOHROI) project seeks to advance adult access to oral health services as a means of reducing the overall cost of health care, replicating work on oral-systemic connections for persons with diabetes using Maine private dental and medical claims. We serendipitously extracted two different sets of person-year records for privately insured persons with diabetes between 2005-2007, using two commonly used methodologies: the Episode-of-Treatment Group (ETG) approach, and a modified HEDIS approach to define persons with diabetes. Looking at the combined 106,544 person-year records, we anticipated that one would subsume the other. Instead, almost a third of the person-year records did not overlap at all, with the difference being split about 2:1 (10,096 HEDIS-only; 20,541 ETG-only; 43,668 both; 32,239 neither). To further explore the implications of the two approaches, we conducted planned analyses of periodontal care, costs, diabetes and complications of diabetes, four ways: ETG-definition; HEDIS-definition; both HEDIS and ETG; and either HEDIS or ETG. Some findings held across groups, but some differed not only in size but in the direction of the association. Our findings underscore the difficulty in relying on claims data to track chronic conditions for population health studies. A well-managed chronic condition may not appear in insurance records, particularly if it can be managed without drugs. Conversely, routine testing to rule out a condition can lead to over-counting. Our result underscores the need to get inside the black box of protocols used to extract health data from insurance claims when seeking to extrapolate to population health.

Learning Areas:
Biostatistics, economics
Chronic disease management and prevention

Learning Objectives:
Describe at least two approaches to extracting condition-related medical claims. Identify possible sources of error when using claims for population studies.

Keywords: Oral Health, Information Databases

Presenting author's disclosure statement:

Qualified on the content I am responsible for because: I am qualified to present because I oversee th eOral Health Prgram at my agency and am the principlal investigator on the project described in the poster.
Any relevant financial relationships? No

I agree to comply with the American Public Health Association Conflict of Interest and Commercial Support Guidelines, and to disclose to the participants any off-label or experimental uses of a commercial product or service discussed in my presentation.