Parkinson's disease is a slowly emerging illness meaning that patients take several different drugs to manage their symptoms. Whilst some drugs, such as Levodopa, can control symptoms, no cures exist. Current therapies are also associated with significant side effects, including cognitive impairment (trouble remembering, learning new things, concentrating, or making decisions), somnolence (a state of drowsiness or strong desire to fall asleep) and dyskinesias (involuntary, erratic, writhing movements of the face, arms, legs or trunk). Hence there is a need to determine new ways of curing the disease. This study aims to determine if any non-Parkinson's disease drugs that the patient happens to be taking for other reasons have an unexpected positive impact on slowing the progression of Parkinson's disease. As CPRD Aurum does not directly measure Parkinson's disease progression, we will calculate a patient's Levodopa equivalent daily dose (LEDD) over time to measure disease progression and see if there are non-Parkinson's drugs that are associated with a slower rate of increase of LEDD.
We will examine non-PD drugs that have been prescribed to patients with PD for a non-PD reason to assess whether treatment is associated with a rate of change to the Levodopa equivalent daily dose (LEDD) and to assess any change in time from PD onset to a related clinical sequela. We will CPRD Aurum records for those patients with a PD diagnosis and PD prescription drugs. Records will be linked to A&E, admitted patient care and outpatient HES, and ONS death registration data. The population of interest comprises patients whose first medcodeid for PD or parkinsonism or a prodcodeid for an established PD symptomatic treatment is between 01/01/2000 and 31/12/2019. We will exclude those with a diagnosis code of Parkinson’s Plus or malignant neoplasm of the brain. Several statistical modelling approaches will be used to calculate outcomes. For longitudinal analysis of LEDD, we will adopt a multilevel model with two levels, repeated measures (L1) over time nested within patients (L2). The LEDD rate of change is measured before and after the prescription of the non-PD drug. We will use an analysis of covariance to test for differences between groups. In addition, we will use Cox regression to model the time (after diagnosis) from non-PD drug exposure (index) to the first event of falling, infection, delirium, or dementia. The Log Rank test will compare the non-PD drug exposure group to a control group. Subject to a sensitivity analysis, we will use age and year at index, BMI, smoking and alcohol status, ethnicity, and comorbidity diagnosis as covariates. Several methods to control confounds will be adopted, including confounding-by-indication and biases like lead time bias and those associated with socioeconomic and patient demographics. Every effort has been made to control for confounds and biases; for ones we do not, we recognise as limitations.
Primary Outcomes:
1. Rate of change of Parkinson’s Disease (PD) treatment (L-Dopa equivalent daily dose: LEDD) after disease onset.
Secondary Outcomes:
1. Time from PD onset to a significant potential related sequelae (fall, delirium, infection, death) for all cases of PD or other forms of parkinsonism e.g., multiple system atrophy (MSA), progressive supranuclear palsy (PSP).
2. Clinico-demographic features and healthcare resource use in the Parkinsonian population.
Cynthia Sandor - Chief Investigator - Imperial College London
Cynthia Sandor - Corresponding Applicant - Imperial College London
Anastasia Ilina - Collaborator - Imperial College London
Caleb Webber - Collaborator - Cardiff University
Mark Cunningham - Collaborator - Imperial College London
Payam Barnaghi - Collaborator - Imperial College London
Zameel Cader - Collaborator - University of Oxford
James Peach - Collaborator - Human Centric DD Ltd
HES Accident and Emergency;HES Admitted Patient Care;HES Outpatient;ONS Death Registration Data;Patient Level Index of Multiple Deprivation;Practice Level Index of Multiple Deprivation