T-score Maps

Item response theory can be used to identify the most likely response to a question based upon a person’s score on the symptom or outcome being measured. Here, responses expected for each item at different T-scores are shown for selected PROMIS® short forms.

PROMIS Depression 8a

In the image above, each item in the PROMIS Short Form v1.0 – Depression 8a is displayed along with response options. A T-score range for the measure is displayed at the top of the figure. To use the T-score Map, identify the score of interest at the top and find below the associated response for that score for an item. For example, persons with a T-score of 50 are most likely to respond “Never” to the item, “I felt worthless.” Persons with a T-score of 62 are most likely to respond “Sometimes.”

Adult Short Forms

Pediatric Short Forms

Parent Proxy Short Forms

Note that not all T-score maps will include the entire range of scores. Persons scoring below the lowest T-score are most likely to respond in the lowest category to all of the displayed items. Persons scoring above the highest T-score are most likely to respond in the highest response category to all of the displayed items.

Examples of Use

The T-score Maps help interpret PROMIS scores. For example, a clinician has an understanding of varying levels of a symptom. Mild depression may be characterized by periodic feelings of sadness whereas severe depression includes persistent feelings of hopelessness and helplessness. This clinical understanding can be mapped onto responses on specific PROMIS Depression items, and those responses can be mapped onto PROMIS T-scores. Thus, a T-score Map is a bridge between the numbers of the T-score metric to the meaning of being at a given score level.
Some clinicians have used T-score Maps to work with patients to set treatment goals. For example, a goal of reducing depressive symptoms by 5 points has no inherent meaning to patients. A T-score Map links that change to consequences that are relevant to patients. Instead of communicating to patients that the goal is to go from a PROMIS Depression score of 62 to 57, a clinician could explain the improvement as feeling helpless “rarely” instead of “sometimes”.


Item Response Theory (IRT) enabled the construction of the T-score Maps. IRT assumes that responses to a set of items or questions are related to an underlying “trait” (e.g., physical function). IRT models quantify how persons’ levels of a measured trait (e.g., high or low physical function) predict what responses they will give to an item (e.g., “With a little difficulty”). Therefore, to construct the T-score Maps, we wrote a program in the software, R, that calculates the most probable response for every item at every T-score. The program utilizes the item-level parameters associated with each item. For each item, every response option had a range of T-scores where it was the most probable response. That range is represented graphically by the band labeled with that response. In some cases, a response option was never the most probable response and therefore is not shown for that item.