Secondary Logo

Journal Logo

Institutional members access full text with Ovid®

Cross-sectional analysis of cognitive function in the MACS with multivariate normative comparisons

Wang, Zhenga; Molsberry, Samantha A.c; Cheng, Yua,b; Kingsley, Lawrenced,e; Levine, Andrew J.f; Martin, Eileeng; Munro, Cynthia A.h,i; Ragin, Annj; Rubin, Leah H.i,k; Sacktor, Nedi; Seaberg, Eric C.k; Becker, James T.l,m,n for the Neuropsychology Working Group of the Multicenter AIDS Cohort Study

doi: 10.1097/QAD.0000000000002312

Background: Prevalence estimates of cognitive impairment in HIV disease vary widely. Here we used multivariate normative comparison (MNC) to identify individuals with impaired cognition, and to compare the results with those using the Frascati and Gisslén criteria.

Methods: This project used data collected before October 2014 from bisexual/gay men from the Multicenter AIDS Cohort Study. 2,904 men (mean age 39.7yr, 52.7% seropositive) had complete data in six cognitive domains at their first neuropsychological evaluation. T-scores were computed for each domain and the MNC was applied to detect impairment among seronegative and seropositive groups.

Results: The MNC classified 6.26% of seronegative men as being impaired using a predetermined 5% false discovery rate (FDR). By contrast, the Frascati and the Gisslén criteria identified 24.54% and 11.36% of seronegative men as impaired. For seropositive men, the percent impairment was 7.45%, 25.73%, and 11.69%, respectively, by the MNC, Frascati and Gisslén criteria. When we used seronegative men without medical comorbidities as the control group, the MNC, the Frascati and the Gisslén criteria identified 5.05%, 27.07% and 4.21% of the seronegative men, and 4.34%, 30.95%, and 4.48% of the seropositive men as having cognitive impairment. For each method, serostatus was not associated with cognitive impairment.

Conclusions: The MNC controls the FDR and therefore avoids the low specificity that characterizes the Frascati and Gisslén criteria. More research is needed to evaluate the sensitivity of the MNC method in a seropositive population that may be sicker and older than the current study sample and that includes women.

aDepartments of Statistics

band Biostatistics

cUniversity of Pittsburgh; Population Health Sciences Program, Graduate School of Arts and Sciences, Harvard University

dDepartment of Epidemiology

eand Infectious Diseases and Microbiology

fGraduate School of Public Health, University of Pittsburgh; Department of Neurology, David Geffen School of Medicine, UCLA

gDepartment of Psychiatry, Rush University School of Medicine

hDepartments of Psychiatry

iand Neurology

jThe Johns Hopkins University School of Medicine; Department of Radiology, Northwestern University

kDepartment of Epidemiology, Bloomberg School of Public Health, The Johns Hopkins University

lDepartments of Psychiatry


nand Psychology, The University of Pittsburgh.

Correspondence to James T. Becker, Department of Psychiatry, Neurology, and Psychology, 3501 Forbes Avenue, Suite 830, University of Pittsburgh, Pittsburgh PA 15213. Tel: +412 246 6970; e-mail:

Received 30 October, 2018

Revised 26 April, 2019

Accepted 3 June, 2019

Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal's Website (

Copyright © 2019 Wolters Kluwer Health, Inc.