Attributing Resource Use Measures

Resource use measures attribute the monetary value of the care provided as part of an episode of illness, care of a population, and event to a measured entity, in combination with quality or health outcome performance. The measure developer can attribute care to a single measured entity or multiple measured entities. Single attribution identifies the decision-maker, for example the primary care physician, and holds this individual responsible for all care rendered. Multiple attribution acknowledges cases where there are multiple decision makers who are playing a role in the care rendered.

As with all measures, the measure developer should consider the attribution model early in the measure development process and determine the attribution model based on the type of patient, clinical circumstance, and the measured entity providing the care. For more information, see the Attribution Model section of the Develop Specifications page.

Peer Group Identification and Assignment

Unlike quality measures, which normally compare performance to an agreed-upon standard (e.g., providing flu vaccinations to a percentage of eligible patients) and direction for improvement (e.g., higher or lower score is better), preferred resource use amounts often are not standardized, and it is not always clear whether higher or lower resource use is preferable. Instead, measure developers should apply resource use measures to compare one measured entity’s performance to the average performance of their peers. For this reason, it is essential to identify an appropriate peer group for comparison.

Calculating Comparisons

The observed-to-expected (O/E) ratio compares the value for each resource use measure attributed to a measured entity (i.e., observed amount) and divides it by the average resource use within the identified peer group (i.e., expected amount—the amount of resource use expected if the measured entity was performing at the mean). Measure developers may use more sophisticated statistical approaches (e.g., multilevel regression).

Setting Thresholds

After estimating the value of a resource use measure, to provide more context for the values, the measure developer should determine whether to apply thresholds or remove outliers. Outliers can be the result of inappropriate treatment, rare or extremely complicated cases, or coding error. Users of the measure results often do not discard outliers, but instead examine them separately. Measure developers should document these actions so users can understand the full context.

Providing Detailed Feedback 

After completion of the analytic steps, users of resource use measures must decide which analytic results to report publicly or include in provider feedback.

Reporting With Descriptive Statistics 

It is critical to choose the right statistics when reporting resource use measure results. Factors influencing this choice include whether to use the results for public reporting or for feedback to providers. Well-crafted, descriptive analytic results can provide the detailed information necessary to make feedback actionable for all interested parties. However, it is important to balance detailed reporting with the possibility of information overload.