Lower Mortality, No Increased Hip Fracture Risk: Anticholinergic Use in Parkinson’s Patients – A BMC Geriatrics Study

Data Source

This study uses claims data from Taiwan’s National Health Insurance. This database includes healthcare data for over 99% of the population. It has information on demographics, clinical visits, diagnostic codes, prescription orders, and related costs. Diagnoses use the International Classification of Disease codes (ICD-9 and ICD-10). The institutional review board at E-Da Hospital approved this study.

Subject Selection

Parkinson’s Disease Cohort

We enrolled patients aged 50 or older diagnosed with Parkinson’s disease (PD) between January 1, 2000, and December 31, 2015. These patients had no prior PD diagnosis in the past two years. We identified PD using ICD codes: ICD-9 code 332 and ICD-10 code G20. Inclusion criteria were:

1. One or more inpatient admissions for PD, or
2. Three or more outpatient visits for PD within six months.

This method improved diagnostic accuracy. The index date was set one year after the first diagnosis to allow for treatment to start.

Exclusion criteria included:

1. Patients with secondary PD, coded as ICD-9-CM 332.1 or ICD-10-CM G21.
2. Patients with specified mental diseases like senile organic psychosis, schizophrenia, or affective psychoses.
3. Patients with cerebrovascular disease before PD diagnosis.

We divided the PD cohort into two groups:

• PD with Antiparkinsonism Anticholinergic Agents (AAs): Patients who had a continuous prescription of AAs for at least 90 days within one year before the index date.
• PD without AAs: Patients who never received AAs during the study period.

We matched both cohorts in a 1:1 ratio based on age, gender, index date, and comorbidity propensity scores.

Non-Parkinson’s Disease Cohort

The control group included participants without a PD diagnosis and no prescriptions for AAs during the study. We matched them in a 4:1 ratio with the PD cohort based on age, gender, index date, and co-morbidity scores.

Covariates

Statistical Analysis

We first analyzed baseline characteristics and comorbidities of all groups. Traditional survival analysis often focuses on single events, which may overlook other risks. Our study used the Fine and Gray regression hazards model to assess death alongside other outcomes, yielding subdistribution hazards (sHR). We calculated p-values using Gray’s test.

We compared the PD with AAs cohort and the PD without AAs cohort against the non-PD cohort to evaluate hip fracture risks. We adjusted for age, gender, index date, and comorbidities.

We also compared risks of developing hip fractures between the two PD cohorts. The study accounted for time-dependent variables to include the effects of time and drug dosage. Additionally, we used a Cox proportional hazards regression model to determine hazard ratios for death among the cohorts.

We applied the Kaplan-Meier method to analyze hip fracture incidence and used the log-rank test to compare differences among cohorts. Data management and sHR calculations were conducted using SAS software. The Fine and Gray model was executed using the PHREG package.