When does a doctor recommend ECT? Unfortunately, the answer depends much more on who the doctor is than what a clearly defined hierarchy of treatment modalities might suggest.
Generally, ECT is considered to be efficacious, but with numerous limitations. It must be done in a hospital setting - an anesthesiologist is required, to prescribe a neuromuscular blocker to paralyze your muscles, and a sedative to render you both unconscious and to ensure total amnesia for the procedure - and must be administered three times a week. (In a later maintenance phase, these can be once a week.) There are a number of risks: death, seizures, memory loss, fractures or muscle damage, cardiac problems, and more.
Despite the risks, however, ECT is widely considered to be a great option, especially for severe cases. The United Kingdom ECT Review Group reviewed the available data and found that it was generally effective, and possibly even more efficacious than medication. The APA Guidelines encourages it for patients with psychotic features or history of “medication intolerance.”
The caveat here is that the reviewed studies, and, indeed, most of the published studies, are open label, and not controlled trials. Furthermore, the controlled trials themselves are often subject to selection bias; patients often have failed other medications before they get to ECT, making ECT appear to be more efficacious than the previously failed medications, even though this is not logically necessary. For example, if one fails bupropion but does well on sertraline, it doesn’t mean sertraline is stronger or more efficacious than bupropion.
A recent study (http://www.medscape.com/viewarticle/567804_print) using the largest sample up to this point of patients with available data on prior treatment history found that prior medication failure (or “medication resistance”) did not predict how well ECT would keep the patient in remission. What does this mean? It could mean either that ECT isn’t as powerful as previously thought, in other words prior treatment failure doesn’t mean ECT will work, or it can mean the opposite: that prior treatment failure isn’t a reason not to use ECT.
It’s a subtle distinction, but one that matters. On the one hand, one should not be discouraged from trying it just because nothing else has worked. But, on the other hand, one shouldn’t be pushed into trying ECT simply because most of the available treatments have failed. My particular bias on this issue is that the risks generally outweigh the benefits, and a history of two or three or even ten medication failures may not (in my opinion) justify it. Other things, such as severity of illness, time available (ECT seems to work faster, so it maybe better for more urgent cases), and safety concerns would be much more important to me.
Like anything else, this is an individual decision: a doctor can advise you, but you should learn as much about the procedure as you can on your own.