“…Yet, for many services (for example, advanced imaging), claims will only reliably identify the provider who is paid to perform the service, while the provider who decides to order the service (and the parameters of that decision) is of greater interest to policymakers. Claims also normally do not contain all the clinical data, 4 Manual abstraction is costly [3][4][5][6] and timeintensive 4,7 Sample sizes are likely to be small Medical records often lack detail and may omit decisions that occurred [3][4][5][6]8 and/or include decisions that never occurred 8 Only some EHR data fields can be automatically extracted (e.g., checkboxes, drop-down menus) 5 May be challenging to sufficiently control for differences in providers' patient case mixtures 3,4 Physicians who permit access to medical records may not be representative May be challenging to sufficiently control for differences in providers' patient case mixtures, 4 and bias in coding of claims may distort results (e.g., upcoding) 11 Often not possible to attribute a given service to the provider responsible for placing the order (e.g., an imaging study is likely to be attributed to a radiologist, rather than the primary care physician who submitted the order)…”