Skip to main content
. 2014 Apr 11;9(4):e94321. doi: 10.1371/journal.pone.0094321

Table 3. Overall concordance of incident atrial fibrillation diagnosis based on Atherosclerosis Risk in Communities data and Centers for Medicare and Medicaid Services data.

All CMS* Inpatient (MedPAR) CMS Outpatient CMS
AF No AF Total AF No AF Total AF No AF Total
ARIC Cohort Follow-up AF 738 93 831 673 158 831 563 268 831
No AF 288 9015 9303 63 9240 9303 264 9039 9303
Total 1026 9108 10134 736 9398 10134 827 9307 10134
Kappa 95% confidence interval 0.77 (0.75–0.80) 0.85 (0.83–0.87) 0.65 (0.62–0.68)
% agreement 96 98 95
% positive agreement 66 75 51
% negative agreement 96 98 94

ARIC =  Atherosclerosis Risk in Communities.

CMS =  Centers for Medicare and Medicaid Services.

*All CMS includes MedPAR and outpatient claims.

Inpatient CMS includes MedPAR claims.

Outpatient CMS includes outpatient and carrier claims.

% agreement calculated as the number of participants with consistent classification of diagnosed AF from active ARIC cohort follow-up and surveillance of CMS divided by the total number of observations and converted to a percent.

% positive agreement calculated as the number of participants classified as having AF based on both active ARIC cohort follow-up and surveillance of CMS, conditional on being classified as having AF from at least one source, and converted to a percent.

% negative agreement calculated as the number of participants classified as not having AF based on both active ARIC cohort follow-up and surveillance of CMS, conditional on being classified as not having AF from at least one source, and converted to a percent.

Data are limited to participants enrolled in Medicare fee-for-service.