Eh, this is alarmist nonsense that wants to get us to focus on the outliers than on the bulk of the users.
Sure, the per-person cost seems prone to causing outrage, but, and let's be honest about two things here:
1) the medicare budget in the state of Texas is FAR, FAR, FAR larger than $3m/year. So we're talking about a small portion of the overall cost, even if it's individually high
2) we're talking about the "high-risk" users here. we can debate over whether it's useful to focus on controlling/changing policy about them. But we can be virtually assured that we're not going to change their behavior. These are the same sort of extremely delinquent people in Marvin Wolfgang's "Delinquency in a birth cohort" study (where he found that 5% of his group committed ~95% of delinquent offenses).
3) Until you get down to the individual level in Texas, you have anecdotal experiences, and ZERO data whatsoever, about the conditions of these people. You assume their users, dealers, and people "working the system." And you know what? There's a good chance you're right. But we're a smart people in a smart(ish) society. Let's not base, modify, or otherwise suggest changes to our government policy based on what we *THINK* is the case. Until we know, for sure, who these people are, what their problems are, and if they're reselling pharmaceuticals, then we really shouldn't propose solutions to a problem we're only partly aware of.