Thirty-One Days for a Drug She Was Already Taking

The fax machine in our office is nineteen years old and it is load-bearing. I mean that literally: if it fails, a category of work stops, because a meaningful share of American insurers still require prior authorisation requests to arrive by fax, and the alternative is a phone tree that averages twenty-six minutes before a human answers.

I spent thirty-one days last spring getting a patient re-approved for a medication she had been taking continuously for four years. Nothing about her condition had changed. Nothing about the drug had changed. Her employer had changed insurers, which meant the new insurer had no record of the previous approvals, which meant the drug required prior authorisation, which meant a form, which meant the fax.

I want to be careful here, because the easy version of this essay is that insurers are villains, and I do not think that is quite the claim. The claim is narrower and I think worse: prior authorisation has stopped being a clinical review and become a delay mechanism, and the delay is the point.

Here is the reasoning. A prior-authorisation requirement is defensible when it catches inappropriate prescribing — genuinely, some of it does, and I have had requests denied where the denial was correct and I had missed something. But the pattern of what gets flagged, in the cases that cross my desk, does not track clinical risk. It tracks cost. I cannot speak to any insurer's formulary as a whole and I am not going to pretend the sample on my desk is one. The expensive drug my patient had taken uneventfully for four years was flagged; the cheap one with a considerably worse side-effect profile was not. No reviewer looking at clinical risk would order it that way.

What the requirement does reliably produce is friction, and friction produces attrition. Some percentage of patients, faced with a month of delay, simply stop taking the drug. Some percentage of prescribers, faced with the form and the fax and the twenty-six minutes, write for the cheaper alternative instead. Both of those outcomes save the insurer money, and neither of them required anybody to deny anything. A denial can be appealed. An exhausted patient who gave up cannot be counted.

I have watched this work. She was sixty-one, she had been stable for four years, and by day nineteen she asked me whether she should just stop. She was not asking a clinical question. She was asking whether the fight was worth it, which is a question about her time and her dignity, and it was not a stupid question. The honest answer was that the fight was worth it and that I was sorry it was a fight.

That is the part I object to. Not that costs are controlled — they have to be, and anyone who says otherwise is not being serious about a system that consumes a fifth of the economy. I object to controlling them through a mechanism whose efficacy depends on people giving up. That is a design that works better the more tired the patient is, and worst for exactly the people with the least capacity to push: the ones working two jobs, the ones without a clinician's office making the calls for them, the ones for whom twenty-six minutes on hold means losing a shift.

There is a version of cost control that says no, out loud, with a reason attached, and can be argued with. I would take that trade. Tell me the drug is not covered and why, and I will either make the case or change the prescription. What I cannot argue with is a process that never says no and simply takes thirty-one days.

She restarted on day thirty-two. The approval, when it came, was for twelve months, at which point the whole thing begins again. The fax machine is still there. Somebody in our office checks its paper every morning, which is, in the richest country in the world, a clinical task.

Update, some months on. The twelve months ran out and the renewal went through in nine days, which I record because I would have reported the opposite outcome and because nine days is what this ought to look like. Nothing about the process changed. Her employer did not change insurers this year.