Development of a computerized adaptive test to assess health-related quality of life in adults with asthma

Abstract

Objective: The purpose of this research was to calibrate an item bank for a computerized adaptive test (CAT) of asthma impact on health-related quality of life (HRQOL), test CAT versions of varying lengths, conduct preliminary validity testing, and evaluate item bank readability.

Methods: Asthma Impact Survey (AIS) bank items that passed focus group, cognitive testing, and clinical and psychometric reviews were administered to adults with varied levels of asthma control. Adults self-reporting asthma (N = 1106) completed an Internet survey including 88 AIS items, the Asthma Control Test, and other HRQOL outcome measures. Data were analyzed using classical and modern psychometric methods, real-data CAT simulations, and known groups validity testing.

Results: A bi-factor model with a general factor (asthma impact) and several group factors (cognitive function, fatigue, mental health, physical function, role function, sexual function, self-consciousness/stigma, sleep, and social function) was tested. Loadings on the general factor were above 0.5 and were substantially larger than group factor loadings, and fit statistics were acceptable. Item functioning for most items and fit to the model was acceptable. CAT simulations demonstrated several options for administration and stopping rules. AIS distinguished between respondents with differing levels of asthma control.

Conclusions: The new 50-item AIS item bank demonstrated favorable psychometric characteristics, preliminary evidence of validity, and accessibility at moderate reading levels. Developing item banks for CAT can improve the precise, efficient, and comprehensive monitoring of asthma outcomes and may facilitate patient-centered care.

Figure 1

Item Characteristic Curves (ICCs) a. AIS Item with Properly Functioning ICC b. AIS Item with Poorly Functioning ICC c. AIS Item Pre- and Post- Collapsed Response Options

Figure 1

Item Characteristic Curves (ICCs) a. AIS Item with Properly Functioning ICC b. AIS Item with Poorly Functioning ICC c. AIS Item Pre- and Post- Collapsed Response Options

Figure 1

Item Characteristic Curves (ICCs) a. AIS Item with Properly Functioning ICC b. AIS Item with Poorly Functioning ICC c. AIS Item Pre- and Post- Collapsed Response Options

Figure 2

Information Function for AIS Item Versions with Similar Content

Figure 3

Correlation between Full AIS Item Bank and CAT Versions (N=1106)

Figure 4

CAT Data Simulations Note: Theta corresponds to the IRT score before rescaling to a standardized metric (mean of 50 and standard deviation of 10 in the developmental sample).