Objective: To examine diagnoses and symptoms in progressive supranuclear palsy (PSP) patients in the 5 years before PSP diagnosis and to develop a model to predict PSP diagnosis among patients diagnosed with Parkinson’s disease (PD).

Background: PSP is a rare neurodegenerative disorder diagnosed ~3 years from symptom onset and often preceded by a PD diagnosis. Early diagnosis of PSP may improve prognosis.

Method: PSP cases and age and gender-matched controls were identified between 2012-2017 from a US health insurance claims database, Truven Health MarketScan®, of >80M patients. PSP was defined as >2 ICD-10 G23.1, or >1 ICD-9 333.0 and >1 ICD-10 G23.1, occurring at least 27 days apart. Index date was the date of the first diagnostic code for PSP. The prevalence of selected diagnostic codes related to potential PSP symptoms or misdiagnosed disorders was examined up to 5 years pre-index date in cases and controls. Prevalence ratios (PR) and 95% confidence intervals were computed.

Results: Among 590 PSP patients with >=30 days of follow-up time pre-index date, 55% had a prior PD diagnosis. In the 3-4 years pre-index date, among 188 PSP cases, the most frequent diagnoses were gait abnormalities (29%), pain in joints (24%), pain in limb (18%), fatigue (16%), cerebrovascular disease (16%) and urinary disorder (16%). Several diagnoses were significantly more prevalent in PSP than controls including dysphagia (PR=34.3), Alzheimer’s disease (PR=15.2) and dementia (PR=8.3). In the 4-5 years pre-index date, among 60 PSP cases, pain in joints (23%), gait abnormalities (18%), fatigue (17%), cerebrovascular disease (15%) and anxiety/phobia (12%) were most common. Prevalence was significantly higher in PSP than controls for memory loss (PR=12.1), cerebrovascular disease (PR=4.5) and gait abnormalities (PR=3.7).

Conclusion: We identified diagnoses potentially predictive of PSP as far as 5 years before diagnosis, many of which were significantly more frequent in PSP than controls. In future analyses, we will use these diagnoses, along with other data, to develop an algorithm for predicting PSP diagnosis among patients diagnosed with PD using diagnostic codes, medications, and terms in medical record notes in a US electronic health record (EHR) database, the de-identified Optum® EHR database (2007-2018). This model, based on real world data, may aid with earlier detection of PSP in clinical practice.

« Back to 2019 International Congress

MDS Abstracts - https://www.mdsabstracts.org/abstract/development-of-a-predictive-model-for-progressive-supranuclear-palsy-using-real-world-data/