Message passing vs memory manipulation

My app consists of different decoupled modules and usually the output of one module is the input of the next module. However for a specific use case I want to collect data across all modules and print it out in the end. I see two options:

1. Message passing: Passing a container object through the different modules which fill it with data. This is how I would usually do it since the hardcoded dependencies make the code very clear and understandable. However, refactoring really becomes annoying, since I often need to change the arguments and return-values of the call hierarchy. Also, I'm not sure if just passing a "container/collector object" and filling it without explicitly returning it is a good design approach.
2. Memory manipulation: Injecting the container object at all places where I need to collect data data. Inject the object also at the place where I need to retrieve the data. Allows for quick refactoring since I just need to inject the collector/container object at the specific places. However, it feels like the code becomes hard to understand, since i don't have a call hierarchy to understand where the information is coming from.
   

Are there reasons to prefer one option over the other in my scenario or in general?