Can a screening questionnaire accurately identify mild Parkinsonian signs?

Abstract

Background: Mild parkinsonian signs (MPS) are early features that, when present, increase the risk of neurodegenerative disease and mortality. Current methods to identify MPS are limited to neurological examination. Our objective was to assess the ability of a 9-item Parkinson's Disease Screening Questionnaire (PDSQ), which has high sensitivity in the detection of overt Parkinson's disease (PD), to detect MPS.

Methods: Measures including the PDSQ, Unified Parkinson's Disease Rating Scale and University of Pennsylvania Smell Identification Test were administered to 267 participants without neurodegenerative disease. Two published definitions of MPS were used to classify cases.

Results: PDSQ scores were higher for cases compared to controls (p < 0.001 for the first case definition and 0.07 for the second). However, the questionnaire had low sensitivity (47 and 59%) and specificity (62 and 63%) in the detection of MPS. Adding factors such as age, gender and smell test score to the questionnaire in a predictive model only marginally improved the test characteristics.

Conclusion: The results show the screening questionnaire does not accurately identify MPS. More accurate tests are needed to improve the detection of this early syndrome which can lead to motor disability, neurodegenerative disease and mortality.

Figure 1

Receiver Operating Characteristic (ROC) curves for Case Definition #1 (panel a) and Case Definition #2 (panel b). ROC curves are shown for the PDSQ alone (open arrow) and for output from a logistic regression model that includes the PDSQ and the UPSIT, age and gender (closed arrow). There were no significant differences between the PDSQ and the model (p>0.05 for both case definitions). The diagonal line below the ROC curves identifies the ROC area that would be achieved by chance alone.