Validation of a computer-adaptive test to evaluate generic health-related quality of life.

BACKGROUND: Health Related Quality of Life (HRQoL) is a relevant variable in the evaluation of health outcomes. Questionnaires based on Classical Test Theory typically require a large number of items to evaluate HRQoL. Computer Adaptive Testing (CAT) can be used to reduce tests length while maintaining and, in some cases, improving accuracy. This study aimed at validating a CAT based on Item Response Theory (IRT) for evaluation of generic HRQoL: the CAT-Health instrument. METHODS: Cross-sectional study of subjects aged over 18 attending Primary Care Centres for any reason. CAT-Health was administered along with the SF-12 Health Survey. Age, gender and a checklist of chronic conditions were also collected. CAT-Health was evaluated considering: 1) feasibility: completion time and test length; 2) content range coverage, Item Exposure Rate (IER) and test precision; and 3) construct validity: differences in the CAT-Health scores according to clinical variables and correlations between both questionnaires. RESULTS: 396 subjects answered CAT-Health and SF-12, 67.2% females, mean age (SD) 48.6 (17.7) years. 36.9% did not report any chronic condition. Median completion time for CAT-Health was 81 seconds (IQ range = 59-118) and it increased with age (p < 0.001). The median number of items administered was 8 (IQ range = 6-10). Neither ceiling nor floor effects were found for the score. None of the items in the pool had an IER of 100% and it was over 5% for 27.1% of the items. Test Information Function (TIF) peaked between levels -1 and 0 of HRQoL. Statistically significant differences were observed in the CAT-Health scores according to the number and type of conditions. CONCLUSIONS: Although domain-specific CATs exist for various areas of HRQoL, CAT-Health is one of the first IRT-based CATs designed to evaluate generic HRQoL and it has proven feasible, valid and efficient, when administered to a broad sample of individuals attending primary care settings.

Desarrollo del CAT-Health, primer test adaptativo informatizado para la evaluación de la calidad de vida relacionada con la salud en España / Development of the CAT-Health, the first adaptive computerized test (CAT) for the evaluation health-related quality of life in Spain

Fundamento y objetivo: Desarrollar un test adaptativo informatizado (TAI) para la medición genérica del estado de salud percibido por el paciente. Material y método: A partir de cuestionarios de calidad de vida relacionada con la salud (CVRS) validados en España, un panel de expertos elaboró un banco de ítems con respuestas tipo Likert de 5 alternativas, que se aplicó a una muestra piloto y posteriormente a otra muestra general de pacientes en 7 centros de salud de Asturias. A la muestra general se le ofrecieron 2 modalidades de respuesta: a) en papel; b) en pantalla táctil. Se evaluó el cumplimiento de los supuestos requeridos por los modelos psicométricos de la teoría de la respuesta al ítem (TRI) (discriminación, fiabilidad y validez) mediante análisis factorial y otros métodos. Se analizó la función de información y se realizó un estudio de simulación. Resultados: Se creó un banco de 140 ítems. Comprobado el cumplimiento de los supuestos de la TRI mediante análisis factorial, se realizó una selección de ítems mediante criterio estadístico y aplicación de la TRI: se eliminaron 35, lo que dio como resultado un test de 105 ítems. Al aplicarlos a la muestra general, se estimaron nuevamente los parámetros de los ítems, se comprobó el supuesto de invariancia de los parámetros y se eliminaron 9 ítems más; esto dio como resultado una versión definitiva de 96 ítems. Conclusiones: El TAI desarrollado resulta apto para evaluar genéricamente y de forma rápida y fiable la CVRS. Este test supondrá un avance muy importante para la medición de los resultados reportados por los pacientes en la práctica clínica

Objectives: To develop a Computerized Adaptive Test to obtain a generic measure of the patients perceived health state. Methods: An experts panel created a pool of five-alternative Likert format items, chosen among Health Related Quality of Life questionnaires validated in our country. That pool was applied to a pilot sample and later to another general sample of patients belonging to 7 Primary Health Centers in Asturias. Two response options were offered to the general sample: 1) on paper 2) on a touch panel. The Item Response Theory psychometric properties (discrimination, reliability and validity) were evaluated by means of the factorial analysis and other methods. The information function was analyzed and a study of simulation was developed. Results: A pool of 140 items was created. The requirements for the application of the IRT were verified by means of the factorial analysis. A selection of items was realized following the statistical criterion and the application of the IRT. The elimination of 35 articles resulted in a test composed by 105 items. Once the questionnaire was applied to the general sample, the parameters of the items were estimated again. The parameters invariance was verified once more and other 9 items were eliminated. The result was a definitive version of 96 items. Conclusions: The Computerized Adaptive Test developed proves to be a fast and reliable tool to evaluate the Health Related Quality of Life generically. This test may imply a very important advance for the measurement of the results reported by the patients in the clinical practice

[Development of the CAT-Health, the first adaptive computerized test (CAT) for the evaluation health-related quality of life in Spain]. / Desarrollo del CAT-Health, primer test adaptativo informatizado para la evaluación de la calidad de vida relacionada con la salud en España.

OBJECTIVES: To develop a Computerized Adaptive Test to obtain a generic measure of the patients' perceived health state. METHODS: An experts' panel created a pool of five-alternative Likert format items, chosen among Health Related Quality of Life questionnaires validated in our country. That pool was applied to a pilot sample and later to another general sample of patients belonging to 7 Primary Health Centers in Asturias. Two response options were offered to the general sample: 1) on paper 2) on a touch panel. The Item Response Theory psychometric properties (discrimination, reliability and validity) were evaluated by means of the factorial analysis and other methods. The information function was analyzed and a study of simulation was developed. RESULTS: A pool of 140 items was created. The requirements for the application of the IRT were verified by means of the factorial analysis. A selection of items was realized following the statistical criterion and the application of the IRT. The elimination of 35 articles resulted in a test composed by 105 items. Once the questionnaire was applied to the general sample, the parameters of the items were estimated again. The parameters invariance was verified once more and other 9 items were eliminated. The result was a definitive version of 96 items. CONCLUSIONS: The Computerized Adaptive Test developed proves to be a fast and reliable tool to evaluate the Health Related Quality of Life generically. This test may imply a very important advance for the measurement of the results reported by the patients in the clinical practice.