NIOSH (who set the air filtration requirements for US respirators, eg: N95) completed a
study in 2020-12 that analyzed how much particulate makes its way through exhalation valves on non-elastomeric respirators. This was done with proper seals and a afaict all front-facing valves keep in mind.
This is interesting to me since they directly address the statements by some that respirators with valves are 'useless' in protecting others, and quote statements made that such respirators as 'good as wearing nothing at all'
They found them to be on average as good and sometimes better at mitigation (1-55% penetration) as common masks worn that lack exhalation valves (divided into three categories: surgical, procedural (which look virtually identical), cloth).
Cloth masks come out of this looking particularly bad, with 45-92% penetration through the fabric. Procedural masks had between 1-85% penetration, surgical between 2-17% (however they note that this doesn't account for regular fitment and in a prior NIOSH study when worn normally was between 7%-76% penetration).
They then tested the CDC's current recommendations for source control for valved respirators which is covering the valve with a surgical mask externally which they found is the least effective method of those they tested. Covering the valve with an ECG pad or surgical tape internally is highly effective, reducing the penetration to just 5% maximum.
They've yet to test elastomer respirators, which they say is something that would be useful to follow-up with, which I'd be interested in as there are models with downfacing exhalation valves, some of which are specifically marketed for medical use.