Control variables in mediation analysis?


I want to conduct a mediation analysis, which will investigate whether the change in depressive symptoms during treatment (time) is mediated by the change in OCD symptoms. Currently, I have time as my predictor, OCD symptoms as my mediator and depressive symptoms (at time t+1) as my outcome. However, I need to control for the impact of prior levels of depression on later levels of depression. Am I right in thinking that in order to control for the impact of prior depressive symptoms on my outcome variable, I need to include depressive symptoms at an earlier time point as a parallel mediator (see figure attached). Or is there another way of controlling for variables in mediation analysis?