Marijuana isn't ready for the prescription pad
In November, the citizens of my home state approved a medical marijuana law. The very next day, I started getting calls from patients (often not may own) asking how they could get it. I'm not fan of draconian laws that imprison people for getting stoned, but when it comes to medical interventions (rather than legal ones) I have an informed opinion. The new law allows Michigan residents to grow weed for their own consumption if they have approval. The law does not allow doctors to prescribe marijuana, rather it allows them to certify that the patient has a condition designated by statue as qualifying them for the medical marijuana program.
When I prescribe a pharmacologic intervention, I usually have some data to back up my decision. My most commonly prescribed medications, such as metformin, ACE inhibitors, beta blockers, statins, and aspirin, have clear dosing options and have clear outcome data that support their use. Marijuana is not a clearly science-based treatment.
That is not to say it isn't medically useful. There is a great deal of anecdotal data for its use in a variety of conditions, and there is scientific plausibility underlying this data. There are also data supporting the concept of cannabis dependence, and there is scientific plausibility to support the idea that smoking anything is probably bad for you. In other words, the available clinical data do not give a doctor a clear way to evaluate the risk/benefit ratio of pot.
In some circumstances, the decision is a bit more clear. In hospice patients or other patients with end-stage diseases, there is probably little harm in using cannabis, although we don't have a lot of data here either.
With marijuana, we have a drug that is not standardized, and has no clear indications for which it has been well-studied. There is no other drug whose use I would recommend on such scant data. There may be considerable promise in cannabis and its derivatives, but until the government allows more study, I'm not writing it.
Addendum
Astute commenters have asked why there is even a problem here---if all I need to do is fill out a certification form with a diagnosis, then what's the problem? There isn't even a prescription to write!
But if I fill out certification forms for medical marijuana, I am de facto recommending it as a therapy. My ethical responsibilities are the same as they would be had I written an actual prescription.