Using pared-mean methods to set data-driven benchmarks
Using pared-mean method to set data driven benchmarks
The pared mean method determines "top performance". This is defined as the best outcome accomplished for at least 10 percent of the population. The explanation is taken from an article by Allison J., Kiefe C.I. and N. W. Weissman, and uses an example of mammography screenings.
How is the pared-mean calculated?
- Rank order providers or other units of analysis (e.g., health departments, jurisdictions) in descending order of adherence. In this example, metropolitan statistical areas were ranked according to average mammography rates.
- Order providers in descending sequence until you have a subset that equals or exceeds 10 percent of all patients in the survey. In this example it was 10 percent of women over the age of 50 in the survey.
- Calculate the benchmark based on the subset of units analyzed, dividing the total number of patients in the subset receiving therecommended intervention (e.g., mammography screenings by the population).
In the example of the mammography screenings, a benchmark rate of 71 percent was found, exceeding the Healthy People 2000 target of 60 percent.
Data sources to use for the pared-mean method include vital statistics and the Behavioral Risk Factor Surveillance System.
" This method is not feasible for all Healthy People objectives. Data may not be available for some objectives, or the nature of the objective may not lend itself to using the pared-mean method. For example, access to preventive care should be available for 100 percent of the population, regardless of what the data show."