Describe Scoring & Calculation Algorithm
Most quality measures produce rates; however, there are other scoring methods such as categorical value, continuous variable (CV), count, ratio, and composite/scales. Measure information must include a description of the scoring type.
The measure developer should describe the type of scoring, accompanied by an explanation of how to interpret the score, such as
- Higher score indicates better quality; improvement noted as increase in rate
- Lower score indicates better quality; improvement noted as decrease in median value
- Score within a defined interval (or a “passing score” over or under a certain threshold) indicates better quality
Develop the Calculation Algorithm
The calculation algorithm—sometimes referred to as the performance calculation, measure logic, or measure flow—is an ordered sequence of data element retrieval and aggregation through which the measure identifies the numerator and denominator events or CV values. At the measure specification stage, the calculation algorithm is not necessarily an equation or a body of computer programming code, but instead is a depiction of the path from the raw data to the result. The measure developer must describe how to combine and use data collected to produce measure results. The calculation algorithm can be a graphical representation (e.g., flowchart), text description, or combination of the two. A calculation algorithm is a required item in the CMS consensus-based entity Intent to Submit and measure submission. Revisions and updates continue to the calculation algorithm through to the measure implementation stage where it is known as the implementation algorithm.
Development of the calculation algorithm should be based on the written description of the measure. The measure description must contain enough information to develop the algorithm. The measure developer needs to check the calculation algorithm for consistency with measure text, as the calculation algorithm will serve as the basis for development of computer programming to produce measure results. The measure developer should account for each scenario and ensure there is a logical end point for each scenario. They should establish this through alpha testing and preliminary feasibility assessments. These assessments will inform beta testing and also minimize implementation and reporting burden on the measured entity. See Measure Testing for more information on responsibilities during this process.