Vitamin C and Sepsis: Resist the Urge to be Duped.

My Facebook feed has been buzzing with news of a breakthrough in the treatment of sepsis–a deadly condition caused by bacterial infection.

Do not be fooled. Please. Resist the urge to be duped.

A recent study published in the prestigious journal Chest concluded that a combination of high-dose vitamin C, anti-inflammatory corticosteroids and thiamine given to patients with sepsis “may prove to be effective.”

NPR covered the story — the headline exuded hope: Doctor Turns Up Possible Treatment For Deadly Sepsis.

The excitement surrounding this preliminary and inconclusive report is instructive on how our brains are susceptible to medical hype.

This story is important because I believe one of the reasons for widespread overtreatment in the US is our collective (doctors and patients alike) susceptibility to extravagant publicity, which, in most cases, exaggerates the benefits of an idea or treatment.

First the background on the sepsis and vitamin C story.

Millions of people die from sepsis. Death occurs because the bacteria release toxins that damage our organs. Many smart people are looking for better treatments.

Thus far, the only treatment of sepsis is to give antibiotics to kill the bacteria and use supportive measures to see people through the period of organ failure–ventilators for failing lungs, dialysis for failing kidneys and medicines to increase the falling blood pressure. Too often, the organs don’t ever recover and the person dies.

Everyone wants to lower the death rate from sepsis. Our wanting sets the stage to be duped.

Now to the “study:” (I put study in quotes because it wasn’t really a study; keep reading and I will explain.)

In the report, a group of doctors at one hospital added the vitamin C, steroid, and thiamine combo to the normal supportive care in 47 patients with sepsis. Four of the 47 patients (8.5%) died.

Then they went back into their medical records and found 47 similar patients with sepsis who did not receive the vitamin C combo. Of this group, 15 of 47 (40%) patients died.

The authors performed statistical tests and found the difference between the 8.5% and 40% death rate to be significant. Sounds good, right?

There are two serious limitations to this paper:

First, this was a comparison of two very small groups of patients from one hospital. When you flip a fair coin only a few times, you can get a string of either heads or tails. As you increase the number of coin flips, the heads/tails ratio gets closer to 1:1. When small studies find large differences, you don’t know whether it was due to a play of chance.

Second, this was not a normal medical study. A normal way to test whether an unknown treatment works is to give one group of patients the treatment and another group a placebo. The two groups should be randomly assigned and the investigators should not know the treatment assignments.

The researchers in this case did not directly compare the vitamin C-combo to a placebo control. They did not randomly assign patients to one group or another. And they were not blinded to the treatment.

So we don’t know whether the death rates differed because of the vitamin C combo or many other factors.

Yes. This is an intriguing observation. It would be cool if something as simple as vitamin C worked.

But at this point, it’s only a theory. The authors spend many words in the full paper explaining ways in which vitamin C combined with steroids and thiamine might work, but they call for additional study:

We believe that the results of our study provide sufficient information for the design of an adequately powered, high quality pragmatic trial to confirm the findings of our study.

Buried near the end of the NPR article is a warning from a sepsis researcher:

A result can look really exciting when you do it on a group in one hospital with one set of clinicians, and then when you try to validate with a larger group in multiple centers — thus far we’ve been unsuccessful with anything.

The way medical science is supposed to work is that interesting observations like this lead to proper experimental studies. Without such proper studies, we can’t tell whether our brains are duping us into seeing signal when there is only noise.

The take-home here is to be on the look out for hype. When it comes to medical science, always read past the headlines, be aware of your own biases and don’t expect rapid progress.

JMM