Great example Emily. I'm sure you know Alyssa Mooney, who is actually studying this question for her PhD dissertation.
There are a lot of challenges w/ the methods. One issue is that the group of people who are arrested might change as a result of the policy, ie policy may change arrest patterns based on their knowledge of the policy. Finding a comparison state or states is an appealing option, although you really need states with good surveillance data on the relevant population to track their health outcomes. Since members of the relevant population (low level users of illegal substances) are difficult to identify, it is a challenge.
Hannah Cooper (now at Duke I think) had a similar question regarding the opposite type of policy, NYC Mayor Giuliani's aggressive drug policing policies, which she hypothesized led people to engage in unsafe practices.
I think the basic approach is to find an indicator outcome that you consider closely tied to the policy and assess how it changes with the introduction of the policy. Hannah used as I recall something like ED visits related to skin infections common among IV drug users.
Maria