Abstract:
|
Cox`s proportional hazards model is one of the most widely used statistical models. Its popularity is partially due to its compact outcome measure, the hazard ratio. However, recently concerns regarding the causal interpretation of hazard ratios and advantages of alternatives, such as Aalen`s additive model, have gained attention. Despite theoretical reasonings in favour of Aalen`s model, experiences in real life studies are still scarce and possible benefits of its use unknown to applied researchers. To evaluate gains and difficulties as well as perception and acceptability when applying Aalen models in combination with Cox models, we chose 5 statistically challenging case studies from hemostaseology and nephrology with intrinsically causal research questions for re-analysis. We involved clinical collaborators to first, assess the current knowledge on interpretation of results from Cox and Aalen models, second to (re-)define their research question in causal terms, third, to get feedback on analysis plans and corresponding results and fourth, to gauge whether they would be prepared to present the results at clinical scientific meetings. We will present results from this evaluation
|