Statistician, postdoctoral research fellow Leibniz Institute for Prevention Research and Epidemiology - BIPS, Germany
Background: In 2016, a medical scandal was revealed in Germany: a pharmacist had diluted chemotherapies over several years. On the day of arrest, ≥50% of the 117 preparations planned to be dispensed on that day contained no cytostatic drug or only 10-20% of the prescribed dose. The full extent and the duration of this fraud is still unknown. It is also not clear which and how many patients were affected, but patients receiving chemotherapy from this pharmacy can be identified in claims data to assess whether they had a worse prognosis as compared to other patients.
Objectives: Exemplified by breast and non-solid cancers we aimed to compare I) the prognosis of patients receiving cytostatic drugs from the pharmacy that illegally diluted chemotherapy (PharmaBottrop) with those receiving treatment from other pharmacies and II) the course of treatment between both groups.
Methods: Using claims data (years 2004-2019) from five German statutory health insurance providers, we selected patients with non-metastatic breast cancer or with non-solid cancers receiving their first parenteral chemotherapy from the PharmaBottrop. These patients were matched to ≤4 similar cancer patients receiving their first chemotherapy from another pharmacy (controls). Matching factors included age, year of diagnosis and cancer subtype. Start of follow-up was the day of dispensation of the first chemotherapy. Endpoint for breast cancer was relapse, and for non-solid cancer it was death. Hazard ratios were estimated using Cox regression models. Further, cancer treatment including the number of days on which patients received chemotherapy was assessed in groups.
Results: With respect to breast cancer, we included 255 patients from the PharmaBottrop and 979 patients as controls. In the group PharmaBottrop and in controls, respectively, 79 patients (31.0%) and 310 (31.7%) had a breast cancer relapse (hazard ratio: 1.03, 95% confidence interval 0.79-1.34). The proportion who died was 11.4% in the group PharmaBottrop and 14.2% in controls. With respect to non-solid cancers, we included 149 patients in the group PharmaBottrop of whom 54 (36.2%) died during follow-up; of the 581 included controls 246 (42.3%) died. The corresponding hazard ratio was 0.85 (0.61-1.19). The time until death was similar in both groups. Both for breast cancer and for non-solid cancers, the median number of days with a chemotherapy dispensation was one third higher in patients in the group PharmaBottrop than in controls (p < 0.05).
Conclusions: Our study suggests that the prognosis of patients treated in the PharmaBottrop was not worse than for other patients, possibly because physicians recognized the non-response to treatment and compensated it by a longer treatment phase.
