It's not obviously flawed. EPA is a non-proprietary equation, albeit calculated slightly differently by different stat aggregators, and it's well understood and used by everyone from teams analytics depts to gamblers. What is unclear is whether it's predictive or not. But it is not obviously flawed.
It doesn't surprise me tbh. For one, the difference in EP between a drop and a completion is always going to be higher than you might expect for a team like the Bills w/fewer attempts than league average, and when you consider propensity to run in red zone vs throw, drops in areas that would allow for FGs instead of punts etc I don't see anything that doesn't jive with what I'm seeing in terms of dropped passes