Within-Person Across-Neuropsychological Test Variability and Incident Dementia
Total Page:16
File Type:pdf, Size:1020Kb
PRELIMINARY COMMUNICATION Within-Person Across-Neuropsychological Test Variability and Incident Dementia Roee Holtzer, PhD Context Neuropsychological tests are used to predict and diagnose dementia. How- Joe Verghese, MD ever, to our knowledge, no studies to date have examined whether within-person across- Cuiling Wang, PhD neuropsychological test variability predicts dementia. Objective To examine whether within-person across-neuropsychological test vari- Charles B. Hall, PhD ability predicts future dementia. Richard B. Lipton, MD Design The Einstein Aging Study (EAS) is a population-based longitudinal study of aging and dementia located in Bronx County, New York. We used Cox proportional EVELOPING STRATEGIES TO hazards models using age as the time scale to estimate hazard ratios (HRs) for perfor- improve the prediction mance on individual neuropsychological tests (Free and Cued Selective Reminding Test, and diagnosis of dementia Digit Symbol Substitution subtest of the Wechsler Adult Intelligence Scale Revised, has paramount therapeutic and the Vocabulary subtest of the Wechsler Adult Intelligence Scale Revised) and for Dand public health implications.1 Neu- within-person across-neuropsychological test variability as predictors of incident de- ropsychological tests are used to mentia. Analyses were stratified by sex, and controlled for education and medical ill- evaluate the diagnosis of dementia,2,3 ness. the transitional stages that precede it, Setting and Participants A total of 1797 participants (age Ն70 years) enrolled in such as mild cognitive impairment,4,5 the EAS between October 1993 and December 2007. Participants seen for the base- and cognitive changes over time.6 line visit only (n=750), prevalent dementia cases (n=72), and those with missing fol- Level of performance on tests of low-up information (n=78) were excluded. A total of 897 individuals were included memory7,8 and executive function9,10 in this investigation. Participants had follow-up visits every 12 to 18 months. has been reported to predict future Main Outcome Measure Incident dementia. dementia, and memory declines Results Sixty-one cases of incident dementia were identified during follow-up (mean more rapidly up to 7 years prior to [SD], 3.3 [2.4] years), of which 26 were in the highest quartile of within-person across- diagnosis of dementia.11,12 neuropsychological test variability. Adjusting for sex, education, and medical illness, vari- When neuropsychological tests are ability was associated with incident dementia (HR for 1-point difference in variability, used for diagnostic purposes, an 3.93 [95% confidence interval {CI}, 2.04-7.56]). The association persisted even after individual’s level of performance on adjusting for level of performance on individual neuropsychological tests (HR for 1-point difference in variability, 2.10 [95% CI, 1.04-4.23]). Comparing Cox models using neu- specific tests is measured against ropsychological tests with and without within-person across-neuropsychological test vari- healthy normative samples to deter- ablity showed that the former improved the prediction of dementia. Sensitivity in a model mine cognitive impairment.13,14 How- predicting dementia at 1 year also improved when neuropsychological test variability ever, this approach does not take was included. into account intraindividual variabil- Conclusions In this population, within-person across-neuropsychological test vari- ity in cognitive function. The tax- ability was associated with development of incident dementia independent of neuro- onomy of intraindividual variability, psychological test performance. This finding needs to be confirmed in future studies. considered as inconsistency in cogni- JAMA. 2008;300(7):823-830 www.jama.com tive performance within a person, includes the following definitions: task administered on 1 occasion or Author Affiliations: Departments of Neurology (Drs Holtzer, Verghese, and Lipton) and Epidemiology and (1) variability on the same cognitive over short periods (cross-sectional Population Health (Drs Wang, Hall, and Lipton), Ferkauf task across multiple assessments over studies show that this form of vari- Graduate School of Psychology (Dr Holtzer), and Insti- 16 tute for Aging Research (Dr Lipton), Albert Einstein Col- a long period (such variability may ability is increased in aging and lege of Medicine, Yeshiva University, Bronx, New York. be high in healthy individuals across dementia,17,18 but has shown variable Corresponding Author: Roee Holtzer, PhD, Albert Ein- 15 stein College of Medicine, Yeshiva University, Rousso the life span ); (2) variability on results in individuals with mild Bldg, Room 306, 1165 Morris Park Ave, Bronx, NY repeated trials of a single cognitive cognitive impairment19,20); and (3) 10461 ([email protected]). ©2008 American Medical Association. All rights reserved. (Reprinted) JAMA, August 20, 2008—Vol 300, No. 7 823 Downloaded From: https://jamanetwork.com/ on 10/01/2021 INCIDENT DEMENTIA AND NEUROPSYCHOLOGICAL TEST VARIABILITY variability in performance across care-eligible individuals were first con- view at baseline and the in-person neuropsychological tests adminis- tacted by letter, then by telephone, interviews in the EAS, but the scores tered in a single session (such vari- explaining the purpose of the study. The are not used when assigning dementia ability is expected in the normal telephone interview included verbal diagnosis in consensus case confer- population,21,22 including older consent, a brief medical history ques- ences.24) Second, impairments on adults without dementia,23 reflecting tionnaire, and telephone-based cogni- basic activities on the Lawton-Brody the individual’s relative cognitive tive screening tests.24 Following the in- activities of daily living,29 secondary to strengths and weaknesses). This terview, an age-stratified sample of cognitive impairment, had to be docu- study focused on within-person individuals who matched on a com- mented using previously described across-neuropsychological test vari- puterized randomization procedure was procedures.30 Incident dementia using ability because it can be estimated invited for further evaluation at the the above algorithmic procedures was directly from standardized clinical medical center. This procedure was used as the main outcome measure in neuropsychological procedures. implemented to ensure that the indi- the primary analyses reported herein. Increased within-person across- viduals included in the EAS repre- Agreement between the algorithmic neuropsychological test variability has sented the Bronx population at that age and clinical diagnosis of dementia was been related to poor cognitive func- stratum, and that those who agreed to 96% in this cohort. tion in normal older adults in cross- participate were not different from non- sectional studies.23 Accordingly, we responders in terms of key demo- Clinical Diagnosis of Dementia examined whether increased within- graphic characteristics. Written in- The EAS participants suspected to have person across-neuropsychological test formed consent was obtained at clinic dementia received a complete diagnos- variability was associated with future visits according to study protocols and tic workup as previously described for dementia after controlling for level of approved by the Committee on Clini- this cohort.31 Diagnoses of dementia performance on individual neuropsy- cal Investigation (institutional review were assigned according to the crite- chological tests. We also assessed board of the Albert Einstein College of ria of the Diagnostic and Statistical whether variability improved the sen- Medicine). Manual of Mental Disorders, Fourth Edi- sitivity and specificity of a model pre- A total of 1797 participants were en- tion,32 at case conferences attended by dicting development of dementia 1 year rolled in the EAS between October 1993 at least 1 study neurologist, a neuro- later. and December 2007. Participants seen psychologist, and a geriatric nurse or for the baseline visit only (n=750), social worker. Alzheimer disease was METHODS prevalent dementia cases (n=72), and diagnosed according to the criteria for Study Population those with missing information (n=78) probable disease detailed by the Na- Participants were enrolled in the Ein- were excluded. Hence, a total of 897 tional Institute of Neurological and stein Aging Study (EAS), a longitudi- participants were included in this Communicative Disorders and Stroke nal study of aging and dementia lo- investigation. Participants had fol- and the Alzheimer Disease and Re- cated at the Albert Einstein College of low-up visits every 12 to 18 months, at lated Disorders Association.3 The State Medicine in Bronx County, New York. which they underwent detailed neuro- of California Alzheimer Disease Diag- The study design, recruitment, and fol- logical and neuropsychological evalu- nostic and Treatment Centers criteria low-up methods have been previously ations. was used to assign diagnoses of prob- described.24,25 Briefly, the EAS used tele- able, possible, or mixed-vascular de- phone-based screening procedures to Algorithmic Diagnosis of Dementia mentia.33 In individuals diagnosed with recruit and follow-up a community- To address the issue of diagnostic cir- dementia, neuroimaging was used to based cohort since 1993. The primary cularity, we derived an algorithmic help allocate the diagnosis of probable aim of the EAS was to identify risk fac- diagnosis of dementia that was inde- Alzheimer disease or probable