Angeline
Your approach has promise if you can identify a city that implemented a diversion program and a comparison city that was similar beforehand but did not implement a diversion program. Simple versions of this approach just have one or two comparison locations, but more recent innovations create "synthetic controls". Synthetic controls look at the trend over time in a place that implemented a new program and compare that trend to infomation on a combination of places ('a synthetic control') that did not implement the new program. The goal is to combine the controls in such a way that they mimic the pre-policy change trends of the place that implemented the diversion program.
Kara Randolph recently had an approach like that to evaluate gun sales regulations, published in Am J Public Health, and Sanjay Basu a similar approach re the health effects of welfare reform, published in Am J Epi. You might see if these designs could be useful.
Maria