Essential Tremor Diagnosis in a Rural Neurology EHR Cohort: Comparison and Validation in a Random Sample Using Both 1998 and 2018 MDS Criteria

Background : Essential Tremor is one of the most common movement disorders with new diagnostic criteria. We want to evaluate the sensitivity, specificity, and positive and negative predictive value of Essential Tremor diagnosed by neurologists using the Movement Disorder Society 1998 and 2018 criteria, with at least one or two visits with the same diagnosis. Methods : This retrospective cohort study was conducted within the Geisinger Health System in Northeast – Central Pennsylvania, United States. We identified all patients with a neurologist's diagnosis of Essential Tremor and without Parkinson's Disease. We divided the patients into two groups: Essential Tremor versus Essential Tremor with a differential diagnosis if the patient had an International Classification of Diseases code suggesting a differential diagnosis for Essential Tremor. A random sample was selected for manual data collection. We built a code in R software to determine the classes (true and false positives, true and false negatives) following the 1998 and 2018 criteria, with at least one or two visits. Then, we calculated the accuracy metrics with 95% confidence intervals, shown in the brackets. Results : The sample included 447 with Essential Tremor and 137 with Essential Tremor with a differential diagnosis. The findings from at least one visit using the 1998 criteria show sensitivity of 96% (94-98%), specificity 60% (53-67%), positive predictive value 82% (78-86%), and negative predictive value 88% (82-94%), while the 2018 show sensitivity 96% (94-98%), specificity 38% (32-44%), positive predictive value 53% (48-58%), and negative predictive value 93% (88-98%). The outcomes for at least two visits show similar sensitivity and negative predictive value, and higher specificity of 71% (61-81%) or positive predictive value of 85% (80-90%) for 1998 criteria and 51% specificity (43-59%), or 65% positive predictive value (57-71%) for the 2018 criteria. Conclusions : Sensitivity and negative predictive value remain high regardless of criteria or number of visits. The 1998 consensus is more specific and has a higher positive predictive value. Both criteria' specificity and positive predictive value improve with two or more visits.