The HIV dementia scale: Predictive power in mild dementia and HAART

Background: HIV-associated dementia (HIV-D) is a subcortical dementia consisting of cognitive and motor symptoms that ultimately affects as many as 20% of patients with AIDS and is associated with significant morbidity and mortality. With the advent of highly active antiretroviral therapy (HAART), the use of sensitive and efficient screening tests for HIV-D continue to be needed for identifying individuals who develop this disorder. Objective: The objective of this study was to compare the HIV Dementia Scale (HDS) with comprehensive neuropsychological procedures in detecting both minor cognitive and motor disorder (MCMD) and HIV-D in a population of patients with varying durations of HAART. Methods: Forty-six HIV-seropositive patients completed both the HDS and a battery of neuropsychological tests as they enrolled in a MRI study. Each person was also assigned a MSK score based on clinical neurological examination. HDS score of <= 10 were considered cognitively impaired and scores > 10 were considered cognitively unimpaired. Two separate sensitivity analyses were performed. Global Z scores (NPZ8) averaged from eight individual neuropsychological subtests were compared to the HDS score for each subject. An NPZ8 score -2.0 standard deviations (S.D.) below the mean was used to define HIV-D. Additionally, HIV-D, defined as -2.0 S.D. below the mean on one test or - 1.0 S.D. below the mean on two or more tests from the NPZ8, were also compared to the HDS. Finally, performance on these cognitive measures was used to predict duration of HAART in this sample. Results: Using the average NPZ8 score based on American Academy of Neurology consensus criteria yielded a test sensitivity of 30%, a specificity of 0%, a positive predictive value of 0%, and a negative predictive value of 58% when compared to clinical MSK ratings. Comparison of the number of impaired tests with MSK severity yielded a test sensitivity of 43%, a specificity of 91%, a positive predictive value of 83%, and a negative predictive value of 61%. HDS scores were less efficient in predicting the presence of subtle and mild HIV-D in this sample. Conclusion: While the HDS is a useful bedside test that a physician may quickly administer to HIV seropositive patients to assist in diagnosing suspected cases of frank HIV-D, the HDS, as a screen, is not as accurate in detecting HIV-D as a more thorough neuropsychological examination. With an increasing prevalence of HIV-D and minor cognitive/motor disorder (MCMD) following the introduction of HAART, the development of more sensitive bedside measures is essential in order to identify individuals with these disorders and monitor treatment regimens. (C) 2007 Elsevier B.V. All rights reserved.