Clinical Analytics Attributed Physician
NOTE: To help you get started, download our Physician Attribution FAQs sheet.
Label Charges and Procedures with Ordering Physician
Many hospital cases have multiple physicians working with them, including operating physicians, consulting physicians, anesthesiologists, and others. The Ordering Physician function in Clinical Analytics labels procedures and charges with the physician that ordered it. This way, the primary physician doesn't get assigned charges he or she was not responsible for, and secondary physicians who may have been ordering frivolous charges/procedures are easier to track. This Ordering Physician field is populated from the data your facility sends to Clinical Analytics.
Overview and definition
The purpose of Physician Attribution is to determine, from patient encounter claims records at a hospital, which physician made the largest contribution to the outcome. This problem is important because physicians are evaluated for quality and efficiency of care based on medical records. If physicians are misattributed, the overall quality of this evaluation process may suffer, ultimately leading to worse outcomes for both the patient and the hospital.
Typically, attribution is currently performed in an inefficient and possibly ad-hoc manner. We wish to provide our clients with a system for attribution that is easy for them to use and consistent with best practices. Ideally, our attribution system provides useful information for the evaluation processes performed by our clients and ultimately lead to better outcomes.
Below, we give an overview of the development of our Physician Attribution algorithm, including the theory behind it and how we are testing it. The basis of the algorithm is attribution by cost where clinical logic is used to weigh certain charges more heavily. Based on our testing, this provides excellent results. The functionality includes a transparent report of the algorithm results and the ability to override the Clinical Analytics Attributed Physician if necessary. Note that Ordering Physician fields must be populated for us to run the algorithm on your server.
Attribution by Cost
In Clinical Analytics, we have access to all of the ICD-9/10 diagnosis and procedure codes and all of the charges for each encounter. We run the MS-DRG grouper and the APR-DRG grouper on the diagnosis and procedure lists. This is the clinical data we have to use to determine the Primary Physician.
Every charge is expected to have a Revenue Code, a charge amount, a cost and an ordering physician. Some hospitals also provide CPT codes and/or HCPCS codes with relevant charges. In order to attribute care to a primary physician based on charges, there are two main approaches: Attribution by Quantity and Attribution by Cost. Attribution by Quantity means using the number of distinct lines of charge as the main guide to choosing the primary physician. Attribution by Cost means that the combined cost from the charges is used instead of the number of distinct charge lines.
We chose Attribution by Cost for three main reasons:
- Our purpose is to identify the physician who made the largest contribution to the outcome of the encounter. Cost tends to be a better proxy measure for contribution than quantity
- In an inpatient encounter, Attribution by Quantity can be distorted by repetitive charges for minor pharmaceuticals and supplies
- The physician with the highest total cost in an encounter is also the one primarily responsible for the use of the hospital’s resources to treat the patient
When possible, we measure cost using the cost field on the charges, rather than the charge amount, because certain types of charges tend to be heavily marked up. This gives a more accurate measurement of the actual cost and, indirectly, the importance of the care provided by each physician. Our method is robust enough that we can get use charge amount instead, however, if a particular hospital’s cost data is not available or not satisfactory.
The diagram below offers a high-level overview of the general attribution methodology. There are additional conditions for certain types of encounters, but most attribution cases follow a process similar to this.
Clinical logic using APR-DRG and other information
For certain types of encounters, Clinical Analytics can identify the most important charges and use this information to make a more accurate attribution. Clinical Analytics uses the 3M APR-DRG Grouper to classify all encounters. The 3M APR-DRG grouper assigns each encounter to a Diagnosis Related Group (DRG) from the list of diagnoses and procedures. The algorithm to assign a DRG is built on clinical logic and well supported statistically. Based on the encounter’s DRG, the cost from certain charges may be prioritized over others.
For example, in any procedural DRG, the Principal Procedure is the most important thing that took place during the encounter. By looking for charges from the operating room (by Revenue Code) on the same day as the Principal Procedure, Clinical Analytics can accurately determine the physician responsible for the Principal Procedure and hence the encounter. If an encounter is grouped into a DRG indicating the insertion of a Pacemaker, the algorithm can identify the charge for the Pacemaker by revenue code and assign the primary physician accordingly. These additional rules make the Clinical Analytics Attributed Physician even more accurate.
Transparency and override capability
Although the algorithm to assign Clinical Analytics Attributed Physician is very accurate, some very complicated encounters may require further review. Axiom facilitates this process with a transparent display of the results and a convenient override capability. In an encounter’s clinical case summary, the Clinical Analytics Attributed Physician is listed along with a quick explanation of how the Clinical Analytics Attributed Physician was chosen. In addition, the physicians with the highest total cost, the highest total charge amount and the most total orders on the encounter are also listed. If necessary, a hospital administrator can override the Clinical Analytics Attributed Physician to any other physician involved in the encounter. Any overrides are logged by the software and can be used to improve the algorithm in the future.
Testing and results
The Clinical Analytics Physician Attribution algorithm has been tested and verified on client data. By running the algorithm and examining the results on a variety of encounters, the method has been improved in multiple iterations. In the most recent tests, we have seen great results. Beta testing will soon begin with selected client facilities.
Submit Ordering Physician data
TIP: Clinical Analytics can determine Attributed Physicians only if facilities submit complete Ordering Physician data for all charges and procedures. For more information about data loads, see Data management, or Email data operations.
