The scoring scheme of the informant questionnaire on cognitive decline in the elderly needs revision: results of rasch analysis

Enhancing precision of the 16-item Informant Questionnaire on Cognitive Decline in the Elderly (IQCODE-16) using Rasch methodology

OBJECTIVE

This study aimed to investigate psychometric properties and enhance precision of the 16-item Informant Questionnaire on Cognitive Decline in the Elderly (IQCODE-16) up to interval-level scale using Rasch methodology.

DESIGN

Partial Credit Rasch model was applied to the IQCODE-16 scores using longitudinal data spanning 10 years of biennial follow-up.

SETTING

Community-dwelling older adults aged 70-90 years and their informants, living in Sydney, Australia, participated in the longitudinal Sydney Memory and Ageing Study (MAS).

PARTICIPANTS

The sample included 400 participants of the MAS aged 70 years and older, 109 out of those were diagnosed with dementia 10 years after the baseline assessment.

MEASUREMENTS

The IQCODE-16.

RESULTS

Initial analysis indicated excellent reliability of the IQCODE-16, Person Separation Index (PSI) = 0.92, but there were four misfitting items and local dependency issues. Combining locally dependent items into four super-items resulted in the best Rasch model fit with no misfitting or locally dependent items, strict unidimensionality, strong reliability, and invariance across person factors such as participants' diagnosis and relationship to their informants, as well as informants' age and sex. This permitted the generation of conversion algorithms to transform ordinal scores into interval data to enhance precision of measurement.

CONCLUSIONS

The IQCODE-16 demonstrated strong reliability and satisfied expectations of the unidimensional Rasch model after minor modifications. Ordinal-to-interval transformation tables published here can be used to increase accuracy of the IQCODE-16 without altering its current format. These findings could contribute to enhancement of precision in assessing clinical conditions such as cognitive decline in older people.

Establishing conversion of the 16-item Informant Questionnaire on Cognitive Decline in the Elderly scores into interval-level data across multiple samples using Rasch methodology

BACKGROUND

The 16-item Informant Questionnaire on Cognitive Decline in the Elderly (IQCODE-16) is a well-validated and widely-used measure of cognitive changes (CCs) among older adults. This study aimed to use Rasch methodology to establish psychometric properties of the IQCODE-16 and validate the existing ordinal-to-interval transformation algorithms across multiple large samples.

METHODS

A Partial Credit Rasch model was employed to examine psychometric properties of the IQCODE-16 using data (n = 918) from two longitudinal studies of participants aged 57-99 years: the Older Australian Twins Study (n = 450) and the Canberra Longitudinal Study (n = 468), and reusing the Sydney Memory and Ageing Study (MAS) sample (n = 400).

RESULTS

Initial analyses indicated good reliability for the IQCODE-16 (Person Separation Index range: 0.82-0.90). However, local dependency was identified between items, with several items showing misfit to the model. Replicating the existing Rasch solution could not reproduce the best Rasch model fit for all samples. Combining locally dependent items into three testlets resolved all misfit and local dependency issues and resulted in the best Rasch model fit for all samples with evidence of unidimensionality, strong reliability, and invariance across person factors. Accordingly, new ordinal-to-interval transformation algorithms were produced to convert the IQCODE-16 ordinal scores into interval data to improve the accuracy of its scores.

CONCLUSIONS

The findings of this study support the reliability and validity of the IQCODE-16 in measuring CCs among older adults. New ordinal-to-interval conversion tables generated using samples from multiple independent datasets are more generalizable and can be used to enhance the precision of the IQCODE-16 without changing its original format. An easy-to-use converter has been made available for clinical and research use.

Applying generalizability theory to examine assessments of subjective cognitive complaints: whose reports should we rely on - participant versus informant?

OBJECTIVES

This study aimed to apply the generalizability theory (G-theory) to investigate dynamic and enduring patterns of subjective cognitive complaints (SCC), and reliability of two widely used SCC assessment tools.

DESIGN

G-theory was applied to assessment scales using longitudinal measurement design with five assessments spanning 10 years of follow-up.

SETTING

Community-dwelling older adults aged 70-90 years and their informants, living in Sydney, Australia, participated in the longitudinal Sydney Memory and Ageing Study.

PARTICIPANTS

The sample included 232 participants aged 70 years and older, and 232 associated informants. Participants were predominantly White Europeans (97.8%). The sample of informants included 76 males (32.8%), 153 females (65.9%), and their age ranged from 27 to 86 years, with a mean age of 61.3 years (SD = 14.38).

MEASUREMENTS

The Memory Complaint Questionnaire (MAC-Q) and the Informant Questionnaire on Cognitive Decline in the Elderly (IQCODE).

RESULTS

The IQCODE demonstrated strong reliability in measuring enduring patterns of SCC with G = 0.86. Marginally acceptable reliability of the 6-item MAC-Q (G = 0.77-0.80) was optimized by removing one item resulting in G = 0.80-0.81. Most items of both assessments were measuring enduring SCC with exception of one dynamic MAC-Q item. The IQCODE significantly predicted global cognition scores and risk of dementia incident across all occasions, while MAC-Q scores were only significant predictors on some occasions.

CONCLUSIONS

While both informants' (IQCODE) and self-reported (MAC-Q) SCC scores were generalizable across sample population and occasions, self-reported (MAC-Q) scores may be less accurate in predicting cognitive ability and diagnosis of each individual.

Re-assessing the dimensional structure of the Informant Questionnaire on Cognitive Decline in the Elderly (IQCODE): empirical evidence for a shortened Brazilian version

BACKGROUND

The dimensional structure, effective number of item responses and item redundancies are controversial features of the Informant Questionnaire on Cognitive Decline in the Elderly (IQCODE) requiring more light. The aims of the present study are to revisit the dimensional structure and propose a shorter version of the instrument.

METHODS

The sample comprised 652 elderly and their informants, either attending a geriatric service of a public university clinic or enrolled in a health care provider database in Rio de Janeiro, Brazil. A Confirmatory Factor Analysis (CFA) first tested the originally proposed one-dimensional structure comprised of 26 items. This was followed by sequential Exploratory Structural Equation Model (ESEM) to evaluate alternative models, in particular a bi-dimensional solution. The identification of residual correlations (RC) lead to a shortened 20-item model, which was tested further via CFA.

RESULTS

The original model fitted poorly (RMSEA = 0.073; 90% CI: 0.069-0.077). Regarding the two-dimensional model, the exploratory procedure (ESEM) indicated several RCs and a lack of factor-based discriminant validity. The ensuing CFA on the one-dimensional model with freely estimated RCs showed an adequate fit (RMSEA = 0.051; 90% CI: 0.047-0.055). Addressing the identified RCs, the CFA on the abridged 20-item version also showed an adequate fit (RMSEA = 0.058; 90% CI: 0.053-0.064) and no further RCs.

CONCLUSION

A one-factor dimensional structure and a reduced version with 20 locally independent items were the most tenable solution. However, although promising, this simpler structure requires further examination before it may be fully supported and recommended.

An informant questionnaire for detecting Alzheimer's disease: are some items better than others?

A decline in everyday cognitive functioning is important for diagnosing dementia. Informant questionnaires, such as the informant questionnaire on cognitive decline in the elderly (IQCODE), are used to measure this. Previously, conflicting results on the IQCODEs ability to discriminate between Alzheimer's disease (AD), mild cognitive impairment (MCI), and cognitively healthy elderly were found. We aim to investigate whether specific groups of items are more useful than others in discriminating between these patient groups. Informants of 180 AD, 59 MCI, and 89 patients with subjective memory complaints (SMC) completed the IQCODE. To investigate the grouping of questionnaire items, we used a two-dimensional graded response model (GRM).The association between IQCODE, age, gender, education, and diagnosis was modeled using structural equation modeling. The GRM with two groups of items fitted better than the unidimensional model. However, the high correlation between the dimensions (r=.90) suggested unidimensionality. The structural model showed that the IQCODE was able to differentiate between all patient groups. The IQCODE can be considered as unidimensional and as a useful addition to diagnostic screening in a memory clinic setting, as it was able to distinguish between AD, MCI, and SMC and was not influenced by gender or education.

Measuring cognition in a geriatric outpatient clinic: Rasch analysis of the Montreal Cognitive Assessment

OBJECTIVE

To evaluate the psychometric properties of the Montreal Cognitive Assessment as a quantitative measure of cognitive ability.

DATA ANALYZED

A total of 222 cases extracted from a clinical database (57-91 years) of patients screened for cognitive impairment in outpatient geriatric assessment clinics.

DATA COLLECTED

Demographic information and individual item responses to Montreal Cognitive Assessment.

RESULTS

Comparison of the data with a unidimensional Rasch model indicated that the total score obtained by summing across all items yields a reliable (0.75) quantitative estimate of global cognitive ability. All items fit the model and together spanned a range of difficulty from -3.75 to +2.88 logits. Items were assessed for differential item functioning across such patient characteristics as age, education, and language spoken. We provide a table for converting Montreal Cognitive Assessment total scores onto a linearly scaled score, with guidelines for interpreting changes in Montreal Cognitive Assessment score in terms of their statistical significance.

CONCLUSIONS

The Montreal Cognitive Assessment can provide a reliable and valid quantitative estimate of cognitive ability in a geriatric cognitive disorders clinic setting.

Sensitivity of the informant questionnaire on cognitive decline: an application of item response theory

The Informant Questionnaire on Cognitive Decline (IQCODE) is a formal informant report instrument, originally developed by Jorm and Jacomb (1989; Psychological Medicine, 19(4), 1015). The goal of the present study was to evaluate the range of cognitive decline in which the IQCODE is most sensitive, using item response theory (IRT). Existing data (N = 740) from a sample of community-dwelling older adults was used for this purpose. A 2-parameter model estimating item difficulty and discrimination fit the data best. Additionally, the IQCODE provided the most psychometric information in the range of -0.5 < theta < 1.5, with peak information obtained at approximately theta = 0.4. Based on individuals' latent score (theta) estimates, items on the IQCODE are adequate for use as a screening tool for dementia. Results of the item calibration may be useful for targeted assessment needs, such as the development of short forms.

Validation of the Voice Handicap Index Using Rasch Analysis

BACKGROUND

The Voice Handicap Index is a tool for measuring the psychosocial consequences of voice disorders and consists of three dimensions. Previous psychometric evaluation of the VHI focused on the classic evaluation of reliability and validity, which is sample dependent. The authors used Rasch analysis to re-examine the dimensionality of the VHI and to produce item and scale statistics that are less sample dependent. In addition, they provide estimates of VHI item and person severities that are reported on the same logit unit scale, allowing a more straightforward interpretation of a VHI test score.

PATIENTS

Dysphonic patients (N = 530), who were referred for phoniatric examination, filled out the VHI.

RESULTS

Rasch analysis revealed two truly uni-dimensional constructs: the 20-item psychosocial scale and the 9-item physical-functional scale. Logit item severity measures ranged from -2.1 to +2.7. Person severity scores ranged from -4.4 to + 4.4 logits. The internal consistency of the reduced scales was similar to that of the original total VHI (0.95 and 0.84). The VHI consisted of two uni-dimensional constructs. Raw test scores can be transformed into logit unit severity measures, making the VHI more suitable for evaluating the effectiveness of voice-related therapy.