Misclassification between competing event types, such as specific causes of death, produces bias in cumulative incidence functions. When comparing cumulative incidence functions under candidate interventions, such bias can lead to suboptimal decision making. Here, we present a semiparametric approach to estimate cumulative incidence functions under hypothetical interventions in settings with misclassification between two event types. We present variations of this approach informed by expert knowledge, external validation data, and internal validation data. Moreover, we propose a modified bootstrap approach to variance estimation that ensures appropriate confidence interval coverage probabilities. The proposed estimator is shown to perform well in finite samples via a series of simulation experiments and applied to estimate the cumulative incidence of cause-specific mortality under candidate clinical interventions.