A controversial policy put forward by one of the nations largest insurers could have dramatic effects is adopted more broadly.

A 27-year-old woman presented to the emergency room with severe right sided abdominal pain of acute onset. After a workup which included a CT scan, it was found that the pain was due to an ovarian cyst and she was discharged. Good news, right?

Not exactly.

The woman, who was insured through Anthem, was charged more than $12,000 for the visit under a program in which the insurer would retroactively deny coverage for an emergency visit that was deemed inappropriate based on the ultimate diagnosis.  The true story, profiled on vox.com, generated national headlines and a lot of scrutiny of the major insurer.

But the program has continued and even expanded, albeit with some changes.

Now this paper, appearing in JAMA Network Open tells us what we can expect when programs like this become the norm.

Researchers used data from the National Hospital Ambulatory Medical Care Survey ED subsample to examine over 28,000 emergency room visits. They then asked a simple question. If the Anthem policy was applied, how many of these visits would be denied?

Let’s go through how the policy works.

First, Anthem screens diagnosis codes from the ER visit – not presenting symptoms. The list of trigger codes is secret, but a leak of the list for Missouri forms the basis for this analysis. So if you go to the ED and are diagnosed with, for example, Acute Bronchitis or Edema or Dysmenorrhea or Esophageal reflux to name a few, your claim may be denied.

Just about 50% of ED visits in the analysis have a diagnosis that would trigger an evaluation for inappropriate use.

But Anthem has stated that certain procedures or situations legitimize the ED visit – if the patient receives IV fluid or gets an ECG for example. You can see the other factors here.

Applying that screen drops the number of denied claims down to 15.7%, about 1 in 6.

Should a patient roll the dice? Well, it would be useful if there were certain symptoms that patients would know are not covered but remember – this is based on diagnosis not symptoms. In fact the presenting symptom overlap between visits that could be denied and visits that wouldn’t be denied was around 90%. Among those symptoms? Chest pain, shortness of breath, vertigo, and headache.

Is that headache a true emergency, a “worst headache of life”, or just a bit of stress on the job?  A $20,000 bill could hang in the balance.

If you do end up with one of those diagnoses, what happens?

In a statement to me (they did not respond to my request for a direct interview), a spokesperson for Anthem said:

“Anthem would request a statement from the consumer as to why they went to the ED as well as hospital records from the visit. A medical director would review the medical records and the consumer’s statement using the prudent layperson standard… before making a determination on the claim”.

That “prudent layperson” language is important. Federal law mandates a “prudent layperson” standard for coverage of ED visits – if a prudent layperson would have gone to the ER, insurance should pay for it. Is an Anthem-paid medical director really the one to judge by the “prudent layperson” standard? It seems rather ridiculous to me.

There’s a bit of an insidious effect of a policy like this though, and that may be the ultimate goal here.

 A Missouri report found that roughly 6% of ED claims were denied by Anthem under this policy – a fair bit less than 1 in 6, but 73% of those were overturned on appeal.  Between that and the negative press Anthem is receiving, it hardly seems that they are saving much money this way.

But maybe the goal is the press. Maybe the goal isn’t really to retroactively charge patients who show up in the ED, but to scare people away from the ED in the first place. This mailer, sent to Anthem members in Missouri supports that theory.

And the really insidious part about that strategy is – we may never know how many people are harmed by it.

This post was first published on medscape.com.