Estimating the effects of exposure variables on outcomes has been a topic of high interest, particularly in the fields of medicine, epidemiology and social sciences. In some settings, however, these effects may be mediated by intermediate variables. Mediation analysis methods are used to estimate the direct effects of exposures, as well as indirect effects that may occur through mediating variables. We introduce a method to estimate the controlled direct effect of an exposure in recurrent event processes where measured and unmeasured mediator-outcome confounders may be present. Unlike traditional methods based on additive models, we focus on multiplicative models for event counts, which may include internal covariates representing a dependency on previous event occurrences. We present the results of simulation studies conducted to investigate the finite sample properties of the controlled direct effect estimator. Finally, we illustrate our method using a hospital readmission dataset.