Diagnostic Accuracy of New and Old Cognitive Screening Tools for HIV–Associated Neurocognitive Disorders

M Trunfio; D Vai; C Montrucchio; C Alcantarini; A Livelli; MC Tettoni; G Orofino; S Audagnotto; D Imperiale; S Bonora; G Di Perri; A Calcagno


HIV Medicine. 2018;19(7):455-464. 

In This Article


Study Design and Participants

We carried out a pilot prospective monocentric cross–sectional study, approved by the Ethics Committee (Comitato Interaziendale di Orbassano, PROspettico sui predittori di DecadImento Neurocognitivo in pazienti HIV–positivi, PRODIN study, protocol number 103). Data were gathered consecutively from patients' visits at our out–patient infectious diseases clinic, Amedeo di Savoia Hospital (Turin), from March 2010 to March 2017. The enrolled subjects systematically underwent clinical examination and screening for neurocognitive impairment. Patients presenting with an altered IHDS score (≤ 10) or complaining of neurocognitive symptoms were referred for a comprehensive neurocognitive battery. Eligibility criteria were: age ≥ 18 years and sufficient basic physical, cognitive and linguistic skills to undergo testing. Exclusion criteria were: current or previous neoplastic, vascular, traumatic, neurodegenerative or infective central nervous system (CNS) disorders, alcohol or drug abuse in the last 6 months, major psychiatric disorders, and a Beck Depression Inventory II and/or Hamilton Anxiety Rating Scale score ≥ 30. Study results were reported according to the Standards for Reporting Diagnostic accuracy studies (STARD) guidelines, 2015.

Index Tests

The four screening tests were administered during the visits by trained physicians aware of the clinical history of the patients, but prior to the administration of the full neurocognitive battery used as a reference standard. FAB requires approximately 10 min to be administered.[18] The six subtests provide a thorough assessment of the following frontal lobe functions: conceptualization, mental flexibility, motor programming, sensitivity to interference, inhibitory control and environmental autonomy, as previously described.[18] A brief summary of each of the six tasks of FAB and the scoring system is shown in Figure S1; for further details see Dubois et al.[18] and supplementary Figure S1. Raw scores were adjusted for age and education and a final score ≤ 14 was deemed to be abnormal, as previously validated in an Italian HIV–negative normative population.[23] CDT requires approximately 5 min to be administered and evaluates visuoconstructive and visuospatial skills, symbolic and conceptual representation, hemiattention, semantic memory and executive functions.[20,24] A raw score ≤ 3 was considered abnormal according to the modified version of CDT described by Shulman in an HIV–negative population affected by dementia.[20,24] We chose Shulman's method as it showed the highest area under the receiver operating characteristic curve (AUROC) and accuracy in predicting dementias.[21] IHDS requires approximately 5 min to be administered and includes measures of psychomotor speed, processing and short–term memory.[25] A score ≤ 10 was considered abnormal, as already validated for an HIV–positive population.[25] 3QT was derived from the brief test reported by Simioni et al.[11] which represents a self–assessment of attention, memory and processing speed. An affirmative answer to at least one of the three questions was deemed abnormal, as already suggested.[11]

Figure S1.

Frontal Assessment Battery (FAB): sub-tests and scoring system

Reference Standard

A comprehensive neurocognitive battery was designed to evaluate the cognitive domains recommended to be assessed by the Frascati criteria[1] and used to diagnose HAND as a reference standard test. It was composed of: Trail Making Test A for processing speed/reaction time, Trail Making Test B and the Stroop Test for executive functioning, Digit Spans and Digit Symbol for attention/working memory, the Corsi Test and Disyllabic Words Serial Repetition Test for visuo–spatial and verbal short–term memory, respectively, the Free and Cued Selective Reminding Test and the Story Recall and Rey Complex Figure Delayed Recall Test for verbal long–term memory and learning, a phonemic verbal fluency test for language skills, the Grooved Pegboard Test for Dominant/Nondominant Hand for motor skills and the Rey Complex Figure Copy Test for visuoconstruction. All tests were administered by a trained neuropsychologist who did not know the index test results and they were scored in accordance with published manuals. Raw scores were converted to age– and education–adjusted normative T–scores. HAND diagnosis and determination of severity were performed according to the Frascati criteria.[1] Daily functioning impairment was assessed using the Instrumental Activities of Daily Living (IADL).


For each index and reference test category, subjects were coded as not impaired or neurocognitively impaired according to the established cut–off points. We then evaluated HAND prediction accuracy by cross–tabulation of the index test results with the reference standard results; this cross–tabulation was used to calculate the SE, SP, positive predictive value (PPV) and negative predictive value (NPV) for each screening. In addition, the odds ratio (OR), likelihood ratio for a positive and a negative result (LR+ and LR–, respectively), correct classification rate (CCR) and Youden's Index (Y–I) were calculated to evaluate the clinical utility of the index tests. ROC curves were obtained to assess the AUROC and the variability in screening accuracy according to different exploratory cut–off points. For the cross–tabulations, two–tailed Pearson's chi–squared test or Fisher's exact test were used. The significance level was set at 0.05 to judge the strength of evidence for hypothesis testing. All the analyses were performed using stata/mp15 (StataCorp LLC, College Station, Texas, USA).