Jollis and colleagues [1] draw conclusions regarding claims versus database discordance that extend beyond their data.

First, claims records and their corresponding comorbid diagnostic elements selected on the basis of associated coronary arteriography are not representative of all claims exclusive of coronary arteriography and their associated morbidities. One key reason is that approximately 75% of the indexed claims are accounted for by only three diagnostic-related groups (DRGs). Review of 2 fiscal years of Medicare claims representing 83 503 inpatient admissions in New Hampshire, of which 3173 included coronary arteriography, showed that the three main DRGs and their relative percentages of all medical and surgical cardiac patients (Medical Disease Category 5) were as follows: DRG 112, percutaneous procedures (25.6%); DRG 124, circulatory disorders other than acute myocardial infarction with cardiac catheterization and complex diagnosis (31.1%); and DRG 125, circulatory disorders other than acute myocardial infarction with cardiac catheterization without complex diagnosis (18.6%). Each different grouping has a distinct risk-related morbidity profile and expected outcome The relative mean hazard score representing the risk for mortality within 1 year of hospitalization for patients grouped in DRG 125 is significantly lower than that associated with DRG 124. Seventy-five percent of all principal diagnoses for DRG 125 are for chest pain, unspecified angina, or coronary atherosclerosis, whereas 70% of the principal diagnoses for DRG 124 are more evolved disease categories for the intermediate coronary syndrome and heart failure. Patients with lower hazard scores are younger and are more likely to be women. They also have fewer comorbid illnesses.

Second, Jollis and colleagues' record selection method puts them at risk for underestimating comorbid conditions [4], especially for patients hospitalized more than once in the same analysis period for later and more invasive surgical treatment that might include presurgical arteriography. For example, patients hospitalized for bypass grafting are much more likely to have had comorbid historical acute myocardial infarction and other comorbid cardiovascular diseases during these later hospitalizations.

Finally, the use of claims data from only one institution suggests that their undercoding of comorbid conditions is unique to their institution. The authors admit that their institution was placed on intensified review for 7 of 16 quarters by the local peer review organization (PRO). One wonders what problematic institutional practices may have occurred within their treating hospital to require intensified review and possibly differentiate this hospital from other hospitals.

Jollis and colleagues show nothing of enhanced analytical or clinical utility with the claimed riches of their data.