Objective One legitimate aim of the pharmaceutical industry is maximizing profit. In an ageing population this can for example be achieved with preventive therapies for older patients. Atrial fibrillation is an age-dependent risk state indicating oral anticoagulation (OA) for stroke prevention. Vitamin-K-antagonists (VKA) have long been the mainstay of OA. Probably in most countries, warfarin was used for OA. Since the approval of direct oral anticoagulants (DOACs) they are increasingly replacing VKAs, although their benefit, in comparison to VKA, is limited and the direct drug costs are much higher. In Germany almost exclusively phenprocoumon is used as VKA. Phenprocoumon differs in pharmacological properties, like the half-life. Also, time in therapeutic range is much better in Germany than in the RCTs with warfarin. Still, there is no single RCT comparing DOACs and phenprocoumon. Nevertheless, the more expensive DOACs largely replaced phenprocoumon in the treatment of patients with atrial fibrillation (AF). We therefore analysed the benefits and harms of DOACs and phenprocoumon for patients with atrial fibrillation in a real-world setting in Germany.

Methods In a retrospective observational cohort study claims data from a group of small to medium-sized health insurance companies covering in- and outpatient care from 2015 to 2019 were analysed. The dataset included sociodemographic information of the patients, their diagnoses and diagnostic and therapeutic procedures from in-and outpatient care. Patients with AF and a first prescription of phenprocoumon or DOAC in standard dose were analysed regarding the outcomes thromboembolic events, major bleeding and death during a twelve months follow-up period. Excluded were patients with a VKA or DOAC prescription in the twelve months before observation, patients with more than one kind of DOAC/VKA, different doses of DOAC or DOAC in reduced dose at index date and patients with less than twelve months of follow-up time (in the case of survival), with undefined age or sex or warfarin as VKA. Cox regression with adjusted cause specific hazard ratios (HR) and comparison after propensity score matching (PSM) were used to analyse the data.

Results In total, 71,961 patients were analysed with Cox regression, out of whom 20,179 received phenprocoumon and 51,782 DOAC in standard dose. The analysis widely showed similarity between phenprocoumon and DOACs regarding effectiveness and safety. There were only three statistically significant differences after Bonferroni correction: a lower bleeding risk with composite DOACs and apixaban (HR [95% CI] = 0.67 [0.59, 0.76], and 0.54 [0.46, 0.63] respectively) and a higher risk of death with rivaroxaban (1.21 [1.10, 2.34]). The analysis after PSM showed almost similar results.

Conclusion The small superiority or noninferiority of DOACs over warfarin seen in the RCTs might not translate into relevant advantages of DOACs over phenprocoumon. To confirm this hypothesis an RCT with phenprocoumon would be needed and could help reduce overspending. The introduction of DOACs instead of phenprocoumon, without adequate evidence from RCTs, is an example of how factors other than evidence are likely to drive the introduction of low-value-added pharmaceutical innovations, thereby threatening equitable health care.