Since the HAQ was created, the science of item response theory (IRT) has developed, a science based on mathematical models that focus on the individual items rather than the total test or instrument. Also called “latent trait theory,” IRT is based on the idea that a latent trait is at the core of an illness and that this trait cannot be directly measured but can be indirectly assessed by individual items.1 Physical function is an example of a latent trait, in that “we know what it is and if you ask us, we can define it, sort of. We know it when we see it,” Dr. Fries says.
With IRT, the levels of a latent trait will depend on the person’s item-level responses rather than the score on the total instrument or test. An item’s properties ultimately characterize an individual and become an estimate of his or her unique functional status.2 Unlike fixed-length legacy tests, such as the HAQ, that were developed by classic test theory, an instrument created using IRT does not require that a patient answer a predetermined number of questions or items. Instead, the patient’s trait levels can be estimated with any subset of items appropriate to that patient from a pool of items—a process that requires much less time to complete.
The PROMIS initiative is developing self-reported measures of not only physical functioning but also of other domains, including fatigue, pain, emotional distress, and social health. Item banks, now available at www.nihpromis.org, can be used to assess pain behavior, fatigue, sleep disturbance, anger, depressive symptoms, and other domains that apply to patients with rheumatologic and other chronic conditions.
Assess More Symptoms Quickly
PROMIS’s intent is to develop, evaluate, and standardize item banks to measure patient-reported outcomes across various medical conditions, Dr. Khanna says. “You are able to assess many more symptoms or aspects of disease—whether physical, mental, or social functioning—and assess them using a very few number of items. An item bank such as physical function may have 124 items, but, on average, people just need to complete seven or eight items.”
A major goal is that the reliable and valid item banks can be administered as computer adaptive testing (CAT), which means that the computer program will select the most informative questions for a particular patient based on each previous response, thus using a minimum number of questions that the patient needs to answer. A patient gets the first question and then a second question, unique to that patient, that is based on the response to the first question; a third question follows that is based on the answers to the first two questions.
