Immunocompromised host

What the evidence shows

Steroids in Pneumocystis pneumonia are established in HIV and unproven outside it. Practice rests on a 333-patient trial that halved respiratory failure and death and found no benefit in mild disease — the severity threshold still used today (Bozzette 1990). A far smaller trial published in the same issue is more often quoted for its dramatic 75% versus 18% survival, but it enrolled 23 patients and was stopped early for benefit (Gagnon 1990). In HIV-negative immunocompromised patients the equivalent trial found 28-day mortality of 21.5% versus 32.4% with p=0.069 (PIC) — an 11 percentage point difference that is either a real effect in a trial too small to prove it, or nothing.

In allergic bronchopulmonary aspergillosis, itraconazole more than doubled the response rate in corticosteroid-dependent disease, 46% versus 19%, without added toxicity (Stevens 2000) — still the main randomised evidence in a disease that has very little.

Investigating respiratory failure in these patients is safer and less necessary than long assumed: bronchoalveolar lavage did not precipitate intubation, and non-invasive testing alone was non-inferior for identifying the cause (Azoulay 2010).