December 12, 2007 
In talking with a few PBMs, it is clear that they approve 90%+ of all prior authorization (PA) requests that come in. With that, I instinctively think of two questions:
- Why do it at all?
- If you’re going to do it, why have humans involved?
I haven’t seen the data on prior authorizations and how many people who hit a reject at the point-of-sale (POS) actually get a fill ever, but I imagine it is like step therapy (ST). With step therapy, only 50% of the people who hit the reject get a claim. (As I have mentioned before, 90% of those that don’t get a claim get samples, buy an OTC, pay cash, or find another solution.) But, the huge step therapy savings that plans and employers realize is not really about the movement to generics but about the lack of claims. Reduce your claims by 50% in a category and you save money.
So, I have to assume that PA has a similar role. The people that don’t really need it (or don’t understand the process) will either pay cash or find another solution, but they won’t have their doctor call the PBM for approval. (aka – The Sentinel Effect) Those that really need it (and understand the process) will call in and get approved.
So the next question is why are people doing this. I have heard from some pharmacists that it has to be humans so that physicians can’t figure out the approval algorithm and “game” the system. Somehow, I doubt that is what they are staying up at night trying to figure out. And, the agents taking the calls are following a very tight script anyways. I have argued for years that either a website or an inbound voice IVR that asks questions and based on answers determines the next question until the physician either fails the request or gets approval. Only exceptions would require a live person.
Enabling Healthy Decisions 
I think that the feedback you are getting from the PBM’s isn’t necessarily representative of the state of prior authorizations. I suspect that PA approval levels at PBMs are a function of the role the PBM plays in that area and of the clients served by the PBM in that area. Many large clients, such as health plans and large employers, manage their own clinical services, including prior authorizations, while small clients lack the resources to administer such programs.
Organizational politics and the demographics of the member base play a pivotal role in determining PA criteria. Labor groups tend to have very relaxed PA criteria, if any at all. Large non-union employers tend to have middle-of-the-road criteria. Health plans tend to have strict(er) criteria for their own insured member base, but not for their self-insured clientele.
My own experience is much different than that of your PBM contacts. A few years ago I worked in the Pharmacy department at Group Health Plan. I handled prior authorizations, among other things. The PA criteria GHP put in place for its insured business was pretty strict for all affected categories. For example, Lamisil (oral) and Sporanox required, #1 step therapy with griseofulvin, nystatin, or another generic antifungal, and #2 a fungal culture be done by a board certified dermatologist (and supplied to GHP for subsequent clinical review), and the fungal culture results indicate a fungus which is susceptible to the prescribed drug. No susceptible fungus, no Lamisil or Sporanox. And approvals only resulted in 12 weeks of therapy. Fungus isn’t gone in 12 weeks? Try something else.
PPI’s were just as strict. Step 1 was step therapy with, not one, but two failed generic H2 Antagonists. Step 2 was an endoscopy performed by a board certified gastroenterologist which returned a diagnosis of a condition for which the drug is approved, such as GERD or esophageal erosions. And the endoscopy report had to be supplied to GHP and reviewed by a Pharm.D. or an MD.
These steps resulted in as many as 90% of prior authorization requests being ultimately denied, depending on the therapeutic category. Some classes, in which step therapy could ultimately gain the patient approval (such as PPI’s), had approval rates of around 50%. I am sure the sentinel effect had an impact on the approval rates, but I think it had much less impact than the clinical protocol put in place.
I suspect that the PA’s being administered by the PBM’s are not tightly managed. Specifically, I don’t think any of the PBM’s are writing aggressive PA protocols and actively promoting the programs to their clients. We would see much lower approval rates if they were.
Prior authorizations can be a powerful cost containment tool, but the payor must have the guts to implement aggressive rules and a member base which will not mutiny, actively use politics (internal or otherwise) to fight the system, or use politics to circumvent the system.