LAUSR

One of the notable features of being a clinician in movement disorders is how few tests we order compared to other specialties. Diagnosis of Parkinson’s disease (PD), in particular, remains firmly in the domain of the “art of medicine.” We use careful evaluation of symptoms, signs, and medication response to make a diagnosis, relying upon tests mainly for atypical or difficult cases. This situation can be disconcerting for patients and caregivers, who often would like further confirmation, and in this evolving field the question remains—how accurate are we? Estimating the accuracy of clinical diagnosis is not easy. The gold standard for diagnosing PD remains autopsy, but the vast majority of PD patients die without one. Studies using autopsy validation to measure our clinical accuracy have been uncommon and so are particularly precious. Previous autopsy validation studies have suggested a broad range of accuracy, depending on disease duration, age of the patient, and expertise of the diagnosticians. A 2016 meta-analysis study suggested an overall average accuracy of 81% in published series, reasonably good, but certainly with room for improvement. In this issue of Movement Disorders, the study by Virameteekul et al uses the Queen Square Brain Bank to assess diagnostic accuracy of PD. Accuracy was tested for both clinician diagnosis and the International Parkinson and Movement Disorders Society (MDS) diagnostic criteria for PD, which were assessed retrospectively. A total of 141 autopsy-diagnosed PD patients and 126 non-PD parkinsonism patients were included. The first goal was to evaluate the overall diagnostic accuracy of clinicians in diagnosing PD versus atypical parkinsonism. The main finding here was that diagnostic accuracy can be very good indeed. Applying the most optimal conditions (the final diagnosis before death made by a movement disorder clinician), accuracy was 97%. This notably high degree of accuracy is similar to a previous study from the same UK brain bank 20 years ago, which found among the highestaccuracy estimates published. It seems that the expert clinicians working with the Queen Square Brain Bank are particularly adept, and so results may not generalize to clinicians with less experience. Indeed, this study confirmed that expertise matters considerably; non-movement disorders specialists had a diagnostic accuracy of only 77%. Combining the two groups of clinicians (approximately 60% of patients were diagnosed by movement disorders specialists) resulted in an average final clinical diagnostic accuracy of 90%. Making a final diagnosis before death is obviously more accurate, as there has been time for atypical features to emerge and medication responsiveness to be measurable. So how about early diagnosis? This was also quite encouraging; 91.5% of movement disorders specialists’ diagnoses made in the first 5 years of disease were correct. Non-experts had 76% accuracy, with the combined group having 84% accuracy. Although both diagnostic sensitivity and specificity were good, sensitivity was generally lower than specificity for final diagnosis (those with PD were more likely to be misdiagnosed as non-PD than the other way around). This was particularly notable for the diagnosis of MSA; 20% (16/80) patients with a final clinical diagnosis of MSA were actually found to have PD on autopsy, whereas only 1 of 73 autopsy MSA cases was diagnosed clinically as PD. This represents an important clinical teaching point. It appears that the diagnosis of MSA is being overcalled late in the course of progressive PD. It should not be forgotten that many persons with PD have profound autonomic dysfunction and treatment-resistant motor features as the disease advances, which can mimic MSA. Perhaps we clinicians may need to bias ourselves somewhat toward retaining © 2023 International Parkinson and Movement Disorder Society.