Study Finds Medicare Claims Data Can Produce Good Surrogates for Measuring Outcomes

An interesting new study -- abstracted here -- found that applying new statistical models to Medicare claims data produced estimates of heart failure mortality "that are very good surrogates" for those generated by models using patients' medical records.  If indeed what we're witnessing, in effect, is a crosswalking of claims data to patient records, then the new models would represent an important step forward in quality measurement.  More broadly, the study seems another indication that big administrative data sets may be where reimbursers and regulators will increasingly graze to get the answers they need (see "The Coming Data Convergence . . . Straws in the Wind").