Reference Populations

A reference population is a specific, relevant group of people (e.g., general population, clinical population).

• During the development of a HealthMeasure, a large sample from the reference population was thoroughly assessed with the measure. That sample’s average score was 50.
• Thus, if a person completes a HealthMeasure today and gets a score of 50, we can say that their score is the same score as the average for the reference population.

Knowing the reference population for a measure helps you interpret a score because you know who you are being compared to.

Overview of HealthMeasures Reference Populations

For most HealthMeasures, the reference population is the general population. PROMIS® adult measures were developed with a large sample of individuals who represented the 2000 U.S. General Census (the general population). PROMIS pediatric and parent proxy “GenPop” measures were developed with a large sample of individuals who represented the 2020 U.S. General Census (the general population).

There are some exceptions that you would expect. Specifically, measures that are likely only relevant to people with certain health conditions often use a clinical sample for the reference population. For example, the PROMIS Dyspnea measures have people with Chronic Obstructive Pulmonary Disease (COPD) as the reference population. The Smoking measures use daily smokers, and the Self-Efficacy for Managing Chronic Conditions measures use a clinical sample as the reference population.

The tables below show the reference populations for PROMIS, Neuro-QoL™, ASCQ-Me®, and NIH Toolbox® measures.

• A T-score is a standardized score, like z-scores and IQ scores. All standardized scores have a “middle” score; it is zero for z-scores, 100 for IQ scores, and 50 for T-scores.
• All T-scores have a mean of 50 and standard deviation of 10. This middle score is the mean of a large sample (i.e., the centering sample) that is representative of a relevant population (i.e., the reference population).

When you know that a T-score of 50 is the mean for a specific population, it is easier to interpret a score quickly. You can compare someone with a HealthMeasure score to other people.

For example, if a person has a PROMIS Fatigue score of 60, we know that they are one standard deviation worse than the general population in terms of fatigue. That is because the reference population for PROMIS Fatigue is the general population. If a person has a PROMIS Nicotine Dependence score of 60, we know that they are one standard deviation worse (more dependent) than daily smokers because the reference population is daily smokers.

During measure development, T-scores are derived through multiple steps:

• Measures are administered to a large representative sample (the centering sample) that represents the reference population.
• Item response theory (IRT) is used to analyze the responses and create initial scores that estimate each respondent’s level of the measured domain.
• These initial scores are then transformed to the T-score metric using a linear transformation that sets the reference population’s average to 50 and standard deviation to 10.
• This transformation ensures that T-scores have a consistent meaning; a score of 50 always represents the average for the reference population.
• Once this process is complete, the measure can generate T-scores for any individual. Those T-scores are directly comparable to the reference population.