Score Cut Point Methods

Empirically established cut points or thresholds can be derived that communicate the severity level or clinical meaningfulness of a score.

A single cut point may be used to define an “acceptable state” on the symptom or outcome of interest. Multiple cut points can be used to carve the full measurement continuum into interpretative levels. Shi and colleagues (2019) provide a review of methods for establishing cut points. All begin with an anchor such as interference with daily functioning, another symptom PRO, or a global rating of health/function. Statistical methods for identifying cut points include multivariate analysis of variance (MANOVA), receiver operating characteristic (ROC) curve analysis, and regression modeling. Learn more>>

Defining Patient Acceptable Symptom State

One approach to cut scores is to identify what has been called the Patient Acceptable Symptom State (PASS). As Tubach and colleagues expressed it, “Feeling good rather than feeling better matters more to patients.” Using the PASS approach, a cut-score is defined for the threshold between what patients consider acceptable and unacceptable states. The PASS cut score typically is defined based on responses to a single-item global scale. A widely used standard for the PASS is the 75th percentile of the subsample of persons who report “important improvement” based on global impression of change. Another approach is to ask patients if they would consider it “acceptable” to remain in their current state for the next few years.

Standard Setting Using Bookmarking

It is not uncommon for researchers to establish score thresholds based on scores on single items (e.g., 0-10 scales), but deriving cut points for multiple-item scales can be challenging. Recently, researchers have adopted a method applied in educational testing to identify thresholds for levels of academic outcomes (e.g., math proficiency levels). This method, called Bookmarking, has been used to establish thresholds for severity levels (no problems, mild problems, moderate problems, severe problems) in multiple patient populations. See Cook, Cella, & Reeve (2019)>>

Bingham et al (2021) used Bookmarking methods to estimate meaningful change for making treatment decisions for an individual patient. Individuals with rheumatoid arthritis (RA) and clinicians treating RA provided input on PROMIS Pain Interference and Fatigue. Learn more>>

Linking to Existing Cut Points

The PROsetta Stone^® tables link scores on common legacy measures to the PROMIS metric. Known cut-scores from a legacy measure can be translated into its corresponding score on the PROMIS metric. For example, a score of 10 on the PHQ-9 is associated with a PROMIS depression score of 59.9. One can now utilize PROMIS scores 60 and above in the same way that the PHQ cut-score of 10 is used.

HealthMeasures Score Cut Points

Learn more about score cut points for: