The CAHPS Consortium recommends transforming the 0-10 scale into three groups for both practical and statistical reasons. From a practical standpoint, displaying the data in three separate groups allows users to make easy comparisons among the health plans, medical groups, hospitals or other units being assessed.
From a statistical standpoint, collapsing the data in this way maximizes the differences between groups relative to the within-group variance. Thus, health plans or providers in the "10-group" are different from those in the "9-8-group"; however, plans with a score of "6" are not necessarily different from those with a score of "5."
A final issue concerning how the three groups are defined is that sponsors may be reluctant to restrict the top group to the single score of "10," even if the decision has been statistically derived. This practical consideration lends support to defining the three groups as "10-9," "8-7," and "6-0," the grouping that the National Committee for Quality Assurance uses for HEDIS® reporting on health plans.
On the other hand, in some situations, especially when the mean rating is very high, the "10," "9-8," and "7-0" categories may yield stronger between-plan (or between-provider) discrimination and more informative bar chart displays. The CAHPS Analysis Program has an option that allows the user to choose between these two classifications.
For information about various ways the CAHPS 0-10 scale has been collapsed, read: Damiano PC, Elliott M, Tyler MC, Hays RD. Differential use of the CAHPS 0-10 global rating scale by Medicaid and commercial populations. Health Serv Outcomes Res Methodol. 2004 Dec;5(3-4):193-205.