1166. Utility of CMS Claims Data for Disease Surveillance
Session: Poster Abstract Session: Public Health
Friday, October 9, 2015
Room: Poster Hall
Background: To evaluate the utility of CMS Medicare and Medicaid claims data for disease surveillance using  influenza (flu)

Methods: We compared CMS claims data, including Medicare outpatient (OTP), physician office visits (Carrier) and Medicaid (MAX) data, with the National Respiratory and Enteric Virus Surveillance System (NREVSS) data from CDC as the “gold standard” for flu activities during 2008 - 2010 for 10 U.S. states, one from each HHS  region. Correlations between weekly test order counts (flu_test) and weekly diagnosis counts (flu_dx) were calculated for each data source. We correlated the CMS flu_test and flu_dx results to the NREVSS weekly positive test counts (NREVSS flu_dx). We developed linear regression models with flu_dx in NREVSS as the dependent variable using the CMS flu_dx and flu_test  data, and compared the adjusted R2 values resulting from each model.  Analyses were performed separately for each state and for the pooled 10-state data.

Results: The correlation coefficients (CC) between flu_dx and flu_test within each data source ranged from 0.61 to 0.93 (P<0.01 for all CCs). The CCs between NREVSS flu_dx and CMS flu_dx and between NREVSS flu_dx and CMS flu_test ranged from 0.46 to 0.79 (P <0.01 for all CCs) with significant variation among states (0.26 to 0.85).  When using flu_test to estimate flu_dx, R2 was 0.78 for NREVSS, 0.90 for OTP, 0.57 for Carrier and 0.82 for MAX. When using flu_test combined with flu_dx from the CMS data to estimate flu_dx in NREVSS, the R2 was 0.55 for OTP (range from 0.53 to 0.77 for the 10 states); 0.57 (0.46 to 0.84) for Carrier; and 0.67 (035 to 0.76) for MAX data. Among various models, a combination of CMS flu_dx and CMS flu_test best correlated with the NREVSS flu_dx results

Conclusion:

We demonstrated that flu test order counts highly correlated with flu diagnosis counts in the selected states and that CMS claims data had the potential for flu activity surveillance. The models’ fit varied depending on the data source (OTP, Carrier, or MAX) and states.  Based on our findings using flu data as a model, we believe CMS claims data has broader potential for use in disease surveillance, although state-specific factors may affect its utility.  Future studies are needed to explore this potential for other diseases.

Lin Fan, PhD, Centers for Disease Control and Prevention, Atlanta, GA, Rex Astles, PhD, Centers for Disease Control and Prevention, atlanta, GA and Howard Burkom, PhD, Applied Physics Laboratory, johns hopkins University, baltimore, MD

Disclosures:

L. Fan, None

R. Astles, None

H. Burkom, None

See more of: Public Health
See more of: Poster Abstract Session

Findings in the abstracts are embargoed until 12:01 a.m. PDT, Wednesday Oct. 7th with the exception of research findings presented at the IDWeek press conferences.