Assistant Professor Institute of Pharmaceutical Sciences, Swiss Federal Institute of Technology, ETH Zurich, Zurich, Switzerland Zurich, Switzerland
Background: Zoledronic acid (ZOL, 5mg/100mL) is typically administered annually in outpatient infusion clinics to treat osteoporosis. ZOL is also indicated for cancer, but at a higher strength (4mg/5mL), and more frequent dosing (every 3-4 weeks). Although a more frequent dosing schedule may be used for cancer, there is no clinical reason for a patient with osteoporosis to receive consecutive dosing of ZOL within 30 days. Pharmacy dispensing (claims) data are commonly utilized to estimate drug exposure. In Canada, a pharmacy claim is created when the prescription is filled in the pharmacy, and patients pick-up their dose to bring with them to their infusion appointment. However, if a patient re-schedules their infusion appointment and does not pick-up their prescription until a future date, a second claim is created, and the original claim may remain and falsely overestimate exposure.
Objectives: Develop a data cleaning algorithm to better estimate when osteoporosis patients receive ZOL infusions, and exclude those with claims that do not coincide with the expected timing of ZOL administration.
Methods: We identified all ZOL claims for Ontarians aged >66 years from 2006/07 to 2020/08, and excluded patients with conditions that impact ZOL dosing (such as Paget’s and cancer). Among eligible patients receiving ZOL for osteoporosis, we applied the following: If a patient had 2 claims within 30 days, and the next claim was >335 days later, the second and third claims were kept and the first was removed as it was assumed the first appointment was rescheduled. However, if the third claim was < 335 days after the second claim, the patient was excluded. This algorithm was repeated for all subsequent claims.
Results: We identified 4,212 patients initiating ZOL (82% female, median age 75 [interquartile range (IQR) 70-81]. Of these, 2,268 (53.8%) had only a single claim. Among those that had more than one claim (n = 1944, 46%), prior to applying the algorithm, patients had a median time of 378 days [IQR 361-427] between claims. Notably, 54 patients had <=30 days between their first and second claim. After applying the algorithm, 432 (10.2%) patients were excluded. Among the 3,780 remaining in the cohort, the median time between claims was 390 days [IQR 369-461].
Conclusions: Careful attention to frequency in pharmacy claims for extended dosed drugs that require clinic appointment for administration is important to avoid overestimating exposure frequency.
