Verbal Fluency as a Rapid Screening Test for Cognitive Impairment in Early Parkinson’s Disease

Permission for the study was initially obtained from the local research ethics committee, and all participants signed an informed consent prior to inclusion.

Thirty-two patients who met the United Kingdom Parkinson’s Disease Society Brain Bank criteria, between Hoehn and Yahr stages I to II,¹⁰ were recruited for a broader ongoing study at the Institute of Cognitive Neurology (INECO). Twenty-two control subjects, matched by age, educational level, and premorbid intelligence were also included in the study. Control volunteers were obtained through word of mouth and were only included if they reported no history of neurological or psychiatric disorders. PD patients were grouped as either having (PDED N=15) on not having executive dysfunction (PDNED N=17) based on their performance in gold-standard executive tests (see below). It was considered that a patient had executive dysfunction if performance on at least one tests of the executive battery was more than 1.5 standard deviations under the mean of controls.

Demographic profile and total scores on tests of general cognitive status are summarized in Table 1. No significant difference was found for age (t (52) = 0.98; p=0.33) years of formal education (t (29) =-0.52; p=0.57) nor premorbid IQ (t (52) = -1.66; p=0.10) between PD patients and controls. Significant differences were found for age between PDED patients and controls (F2, 51=5, 54; p=0.028) and between the former and PDNED patients (F2, 51=5, 54; p=0.007).

Significant differences between groups were found on the verbal fluency test (F (2, 50) = 6.87, p<0.01, η² = 0.21). Although controls showed a mean of 18.09 (SD=5.07), PDNED presented a mean of 16.71 (SD=4.19) and PDED a mean of 11.60 (SD=3.96) words in this test. A post hoc analysis (Tukey honestly significant difference, MS=20.51, df=50) revealed that controls performed significantly better than both PDED (p<0.01) and PDNED (p<0.001). Moreover, both PD groups differed significantly between each other (p<0.01), with PDNED outperforming PDED. As expected, significant differences between groups were also found on most gold-standard executive tests. In this regard, significant differences were found in in the Trail Making Test Part B (F (2, 50) = 6.60; p<0.01, η² = 0.20). A post hoc analysis (Tukey honestly significant difference, MS=4383.6, df=50) showed that both controls (p<0.01) and PDNED (p<0.01) performed better than PDED, even if a significant effect of age (p<0.01) was observed on Trail Making Test Part B performance. Significant differences between groups were also found on the WCST (Wisconsin Card Sorting Test) (F (2, 50) = 29.46, p<0.01, η² = 0.50). A post hoc analysis (Tukey honestly significant difference, MS=0.76, df=50) revealed that PDNED (p<0.01) and controls (p<0.01) exhibited a higher performance than PDED. Unexpectedly, no significant differences were found between groups on the Digits Backward Test (F (2, 50) = 2.50, p=0.92, η² = 0.09). This lack of significant differences can be related to the fact that Digit Backward Test relies on verbal instead of visual working memory, the latter of which is most commonly decreased in early PD patients.¹⁵

Significant correlations were found between verbal fluency score, and the total number of categories abstracted on the WCST (r=0.52; p<0.001), and time to complete the TMT-B (r=−0.82; p<0.001). No significant correlations were found between the verbal fluency score and the Digits Backward Test (r=0.23; p<0.01)

A receiver operating characteristic curve analysis on the verbal fluency total score comparing healthy controls and PD patients (PDED and PDNED) generated a cut-off score of 14 points with sensitivity of 73.3% and specificity of 82.1% (Figure 1A).

With a verbal fluency score of 14, 73.3% of the PDED patients, 76.4% of the PDNED patients, and 86.3% of the controls were correctly classified. Area under the receiver operating characteristic curve was 0.85 [confidence interval (CI) 0.72–0.97; p<0.01]. Furthermore, when patients were divided into PDED and PDNED, a receiver operating characteristic curve analysis comparing both groups generated a cut-off score of 14 points, with a sensitivity of 73.3% and a specificity of 76.4% and an area under the curve of 0.81 (CI 0.66–0.96; p<0.01) (Figure 1B).

In the present study, we showed that the phonological verbal fluency task may be a good screening test for the detection of executive deficits in PD. Our findings show adequate concurrent validity, as revealed by the significant correlations that the test showed with executive tasks, and adequate discriminant validity, as evidenced by the test’s capability to significantly differentiate PDED patients from healthy controls and from PDNED patients. All the between-group differences remained significant even after controlling for the effect of age save an effect of age observed on the Trail Making Test B performance. This is not unexpected, as several studies^16–18 have shown that among adults processing speed is positively correlated with age.

Using a cut-off score of 14 points, the Verbal Fluency Task showed a sensitivity of 73.3% and a specificity of 76.4% in the detection of executive deficits in PD. These results suggest that this brief test is able to correctly classify the majority of PD patients with executive dysfunction and those without it.

Our results have important implications for clinicians working with patients with PD who cannot always access more extensive neuropsychological evaluations. The present investigation indicates that the phonological verbal fluency task can act as a sensitive and specific screening tool in the detection of executive deficits in this pathology. Even if an extensive evaluation of executive functions might be critical to obtain an accurate description of a patient’s neuropsychological profile, this 2-minute screening tool, combined with an adequate interview regarding the impact of cognitive dysfunction in patients’ lives, can be useful in the detection of patients who may benefit from more extensive cognitive evaluations and cognitively targeted interventions.

Although no single test can replace the value of a complete and focused history-taking or that of a comprehensive neuropsychological evaluation, the present study indicates that phonological verbal fluency can act as a reliable, brief, and easy-to administer tool for the screening of executive dysfunction in PD. In this regard, the phonological verbal fluency test could act as a simple tool to be used in everyday clinical practice to detect those patients with PD who will benefit from a more comprehensive neuropsychological battery, including a thorough cognitive history-taking and further test of executive and social functioning.