COMPUTERIZED DIAGNOSTIC TESTING

How item banks and their application can influence measurement practice in rehabilitation medicine: a PROMIS fatigue item bank example

OBJECTIVE

To illustrate how measurement practices can be advanced by using as an example the fatigue item bank (FIB) and its applications (short forms and computerized adaptive testing [CAT]) that were developed through the National Institutes of Health Patient Reported Outcomes Measurement Information System (PROMIS) Cooperative Group.

DESIGN

Psychometric analysis of data collected by an Internet survey company using item response theory-related techniques.

SETTING

A U.S. general population representative sample collected through the Internet.

PARTICIPANTS

Respondents used for dimensionality evaluation of the PROMIS FIB (N=603) and item calibrations (N=14,931).

INTERVENTIONS

Not applicable.

MAIN OUTCOME MEASURES

Fatigue items (112) developed by the PROMIS fatigue domain working group, 13-item Functional Assessment of Chronic Illness Therapy-Fatigue, and 4-item Medical Outcomes Study 36-Item Short Form Health Survey Vitality scale.

RESULTS

The PROMIS FIB version 1, which consists of 95 items, showed acceptable psychometric properties. CAT showed consistently better precision than short forms. However, all 3 short forms showed good precision for most participants in that more than 95% of the sample could be measured precisely with reliability greater than 0.9.

CONCLUSIONS

Measurement practice can be advanced by using a psychometrically sound measurement tool and its applications. This example shows that CAT and short forms derived from the PROMIS FIB can reliably estimate fatigue reported by the U.S. general population. Evaluation in clinical populations is warranted before the item bank can be used for clinical trials.

Measuring fatigue for children with cancer: development and validation of the pediatric Functional Assessment of Chronic Illness Therapy-Fatigue (pedsFACIT-F)

Fatigue is a major concern for cancer patients of all ages. The lack of an appropriate assessment tool has impeded our understanding of its prevalence and significance, specifically in the pediatric cancer population. This paper documents the reliability and validity of the pediatric Functional Assessment of Chronic Illness Therapy-Fatigue (pedsFACIT-F) in a comprehensive manner. The 11-item PedsFACIT-F was developed via literature review, feedback from patient/parent/clinician, and a face-to-face consensus meeting. Its reliability and validity were examined on the basis of data from 159 pediatric patients with cancer via classical test theory and Rasch analysis. Results showed that the pedsFACIT-F demonstrated good internal consistency (Cronbach alpha), acceptable item-total correlations, and met the unidimensionality assumption set by confirmatory factor analysis. All items had acceptable fit statistics in the Rasch analysis and demonstrated stable measurement properties by age, sex, and cancer type. Scores on the pedsFACIT-F significantly discriminated between patients with and without anemia and among patients with different functional status; clinically relevant minimally important differences were estimated accordingly. The pedsFACIT-F was significantly correlated to the PedsQL Multidimensional Fatigue Scale. In conclusion, the pedsFACIT-F demonstrates satisfactory reliability and validity and can be a useful tool in clinical trials and other research.

An item response theory-based pain item bank can enhance measurement precision

Cancer-related pain is often under-recognized and undertreated. This is partly due to the lack of appropriate assessments, which need to be comprehensive and precise yet easily integrated into clinics. Computerized adaptive testing (CAT) can enable precise-yet-brief assessments by only selecting the most informative items from a calibrated item bank. The purpose of this study was to create such a bank. The sample included 400 cancer patients who were asked to complete 61 pain-related items. Data were analyzed using factor analysis and the Rasch model. The final bank consisted of 43 items which satisfied the measurement requirement of factor analysis and the Rasch model, demonstrated high internal consistency and reasonable item-total correlations, and discriminated patients with differing degrees of pain. We conclude that this bank demonstrates good psychometric properties, is sensitive to pain reported by patients, and can be used as the foundation for a CAT pain-testing platform for use in clinical practice.