I'd really be curious about distribution of the result they see. The folklore is definitely that that there's vary high variance in how people respond to SSRIs, and not recommending them because the average value is low is pretty irresponsible.