Physicians Should Administer Low-Dose Corticosteroids Selectively to Septic Patients until an Ongoing Trial Is Completed

Most physicians prefer administering therapies that have a physiologic rationale and proven benefit. We urgently need beneficial therapies for sepsis severe enough to cause shock, which is responsible for approximately 400 000 intensive care unit admissions and 200 000 deaths each year in the United States (1). In this issue, a meta-analysis by Minneci and colleagues (2) from the National Institutes of Health (NIH) concludes that low but not high doses of corticosteroids reduce mortality from septic shock and that we should administer these agents to all patients who have vasopressor-dependent sepsis. How the investigators reached these conclusions and why I agree with the first but not the second of them are the subjects of this editorial.

The adrenal glands release cortisol during various stresses, of which septic shock is one of the most profound. Both high and low plasma levels of the endogenously released hormone are associated with increased mortality, presumably because high levels reflect severe stress and low levels an inability to respond to stress by releasing cortisol (3). The beneficial actions of cortisol are due to its anti-inflammatory effects (including inhibition of cytokine production and prevention of circulating inflammatory cell migration into tissues) and cardiovascular effects (including attenuation of nitric oxide synthase induction and enhancement of the vasoconstrictive response to catecholamines) (4).

Between 1959 and 1986, investigators administered short courses of high (also called “pharmacologic”) doses of corticosteroids (for example, methylprednisolone, 30 mg/kg of body weight, in 4 intravenous doses over 24 hours) to almost 1300 septic patients in 9 clinical trials. One rationale for these trials was to interrupt the inflammatory cascade hypothesized to perpetuate sepsis (5). Unfortunately, no large randomized trial demonstrated that corticosteroids improved mortality. Such improvement was seen only in a single-center study with both prospective and retrospective enrollment. Furthermore, 2 …