OBJECTIVES: The IQCODE is an informant-rated scale of estimated cognitive decline. This study examined the scalability of the 16-item form of the IQ-CODE and its validity to predict performance on a comprehensive cognitive battery. METHODS: The medical files of 169 psychogeriatric day care attendants contained an IQCODE. Mokken's nonparametric item response model for polytomous items was used to assess the unidimensionality of the IQ-CODE. Tests of orientation, episodic memory and executive functioning (Behavioral Dyscontrol Scale, Category Fluency, and Expanded Mental Control Test) were administered in order to examine predictive validity. RESULTS: Hi-coefficients of scalability for the 16 items of the IQCODE ranged from 0.33 to 0.50. Loevinger's H was 0.40 for he full scale, which makes it a moderately strong, unidimensional scale. A 'strong' scale (H = 0.54) was found by selecting a subset of seven items, representing the latent construct of problem solving ability. The correlations of the total IQCODE score with neuropsychological tests differed significantly from zero but the coefficients were quite small, ranging from -0.12 to -0.25. A nonparametric optimal discriminant analysis (ODA) was used to derive a cutoff score for the IQCODE that would maximize the classification accuracy in differentiating patients with comparatively high or low levels of cognitive functioning. Given an effect strength for sensitivity (ESS) of 21%, the optimal IQCODE cutoff score (3.75) did not reliably discriminate between different levels of cognitive ability. More than half of the patients (56%) who were predicted to belong to a group of low cognitive ability, in fact performed at a higher level. Conversely, overestimation of cognitive ability occurred less often (20%). CONCLUSIONS: In this sample of psychogeriatric day care attendants the IQCODE tended to underestimate the patient's cognitive scores. In order to understand deficits or find remaining competencies, direct assessment of cognitive abilities may be preferred to informant based measures.
OBJECTIVES: The IQCODE is an informant-rated scale of estimated cognitive decline. This study examined the scalability of the 16-item form of the IQ-CODE and its validity to predict performance on a comprehensive cognitive battery. METHODS: The medical files of 169 psychogeriatric day care attendants contained an IQCODE. Mokken's nonparametric item response model for polytomous items was used to assess the unidimensionality of the IQ-CODE. Tests of orientation, episodic memory and executive functioning (Behavioral Dyscontrol Scale, Category Fluency, and Expanded Mental Control Test) were administered in order to examine predictive validity. RESULTS: Hi-coefficients of scalability for the 16 items of the IQCODE ranged from 0.33 to 0.50. Loevinger's H was 0.40 for he full scale, which makes it a moderately strong, unidimensional scale. A 'strong' scale (H = 0.54) was found by selecting a subset of seven items, representing the latent construct of problem solving ability. The correlations of the total IQCODE score with neuropsychological tests differed significantly from zero but the coefficients were quite small, ranging from -0.12 to -0.25. A nonparametric optimal discriminant analysis (ODA) was used to derive a cutoff score for the IQCODE that would maximize the classification accuracy in differentiating patients with comparatively high or low levels of cognitive functioning. Given an effect strength for sensitivity (ESS) of 21%, the optimal IQCODE cutoff score (3.75) did not reliably discriminate between different levels of cognitive ability. More than half of the patients (56%) who were predicted to belong to a group of low cognitive ability, in fact performed at a higher level. Conversely, overestimation of cognitive ability occurred less often (20%). CONCLUSIONS: In this sample of psychogeriatric day care attendants the IQCODE tended to underestimate the patient's cognitive scores. In order to understand deficits or find remaining competencies, direct assessment of cognitive abilities may be preferred to informant based measures.
Authors: Jennifer K Burton; Patricia Fearon; Anna H Noel-Storr; Rupert McShane; David J Stott; Terry J Quinn Journal: Cochrane Database Syst Rev Date: 2021-07-19
Authors: Christina P Boyle; Cyrus A Raji; Kirk I Erickson; Oscar L Lopez; James T Becker; H Michael Gach; Lewis H Kuller; William Longstreth; Owen T Carmichael; Brandalyn C Riedel; Paul M Thompson Journal: Hum Brain Mapp Date: 2020-09-10 Impact factor: 5.038