Outcomes Framework - Types of Scales

Numerical Scales

Most scales in common use are numerical in that when scored the final result is a number. Examples include the PHQ-9, which is scored between 0 and 27, and the GAD-7, which is scored between 0 and 21.

When evaluating numerical scales, it is possible to construct Success and Inclusion Criteria corresponding to the baseline and final scores. Examples include:

• Success when a particular score is attained, e.g. scoring < 5 on the PHQ-9
• Criteria only includes patients where the baseline score is as specified, e.g. counting patients only where their baseline PHQ-9 score was over 15.

In many circumstances, using properties calculated from the raw score - for example, a clinical cutoff, or a Treatment Episode Response indicator - is simpler to understand than using raw scores.

However, many scales, particularly for less common diagnoses, do not support these characteristics and you may have to use the raw score directly.

Categorical Scales

Some scales have no numerical score. Examples include most variants of the Columbia Suicide Severity Rating Scale (C-SSRS), the CATS, and parts of the Vanderbilt. These scales only provide a clinical severity readout - e.g. “High Risk”, “Clinical”.

Because these scales have no numerical score, you cannot do many common types of analysis on them. You cannot measure ‘a 5 point change’. Instead, you can do analysis on the basis of the change in clinical categories. For example, you could measure ‘% of patients who started with a high-risk score on the C-SSRS who are now in the low-risk group’.

Unscored Scales

Some scales are purely administrative, and support neither a numerical score, nor a clinical cutoff. These scales are not suitable for outcomes metrics and are therefore not supported.

Information on how to select scale can be found here.