Causal identification of an intervention’s effect is challenging in the presence of interactions between subjects. For example, in infectious disease, an individual’s outcome depends on others’ treatments and the time when they get infected. Although many proposed estimands claim to successfully identify the causal effect of a vaccine on susceptibility or infectiousness, they either are defined in restrictive settings or fail to recover the true effect under some scenarios. In this paper, we construct an innovative causal identification strategy non-parametrically in a conceptual partnership model, and propose estimands for a vaccine’s effect that is more closely linked to its biological effect. The setting considered makes no assumptions on the functional form of infection risk or the way that this risk changes due to exposure to an infectious partner or vaccination. Systematic comparisons with existing quantities are illustrated using a realistic simulation of an HIV vaccine trial. Further insights regarding evaluation of the unbiasedness of an estimand in a variety of experimental designs are also discussed.