For Hospital Medicine, the Answer Should Always be: Randomize

I call them therapeutic fashions, things that we do because it's fashionable and popular. This week three of these were busted by...what else...the RCT

Three recent trials of (very) common interventions in hospitalized patients have all returned null results. As always, there are specific lessons about treating the three conditions and general lessons on evidence-based medicine.

ARISE FLUIDS

The Surviving Sepsis Campaign began 25 years ago after a 2001 small trial found that early goal-directed care reduced mortality in patients with sepsis. Goal-directed care was aggressive: making invasive measurements (central lines) and targeting these numbers with IV fluids, vasopressors and blood transfusion when needed. The mortality benefit was huge (16% in absolute terms). The invasive components of that trial were later reversed but early resuscitation with fluids remained.

Then came the tragic case of 12-year-old Rory Staunton from Queens who grazed his arm playing basketball and was sent home after an initial evaluation and later died of sepsis. This tragedy ushered in the mandating of early aggressive IV fluid therapy for sepsis via performance measures that had financial penalties.

Thing is: this is a lot of fluid, and not all sepsis patients are the same. Another way to manage the low blood pressure and under-perfusion of sepsis is with vasopressors—drugs that raise blood pressure.

The ARISE FLUIDs trial investigators randomized 1000 patients with early sepsis to the either high-dose IV fluids or early vasopressors with less IV fluid. The primary endpoint of days alive and out of the hospital was 76 in both groups. Adverse events were largely similar, except pulmonary edema which was 9x more likely in the early fluids arm (0.6% vs 5.0%).

SODa-BIC

Perhaps even more entrenched than the aggressive fluid strategy for sepsis is the notion of “fixing” metabolic acidosis with sodium bicarbonate. The sky is blue and hospital doctors give bicarb for acidosis. The HCO3 ion is a buffer to the excess acid.

A brief background: metabolic acidosis occurs because of low perfusion of tissues, usually from shock of one sort or another or kidney failure. Acidic blood is bad because it can cause kidney injury. You try to avoid metabolic acidosis, and when it occurs, treatment is directed at the underlying cause.

But. But. Acidosis is a number; and bicarb corrects that number. Yet it does nothing to fix the underlying cause. No matter, doctors give it; because they do. Previous trials have been inconclusive.

NEJM recently published the SODa-BIC trial where patients with metabolic acidosis (pH < 7.30) who were receiving vasopressors in the ICU were randomized to either a bicarb infusion or placebo.

The primary outcome of a major adverse kidney event (death, dialysis, or persistent kidney dysfunction) did not significantly differ (40.2% of the bicarb arm vs 39.4% in the placebo arm). Four patients (1.6%) in the sodium bicarbonate group had an adverse effect, as compared with none in the placebo group (P=0.06).

BIHCA

Another use of the acid buffer sodium bicarbonate is the treatment of patients who have in-hospital cardiac arrest. The thinking is that cardiac arrest causes massive under-perfusion of the organs and acidosis is a) likely and b) detrimental.

The background here differs from bicarb use in metabolic acidosis in the ICU. In ICU acidosis, there is at least some data to suggest benefit—albeit weak because of flawed trials. In the case of giving bicarb to cardiac arrest survivors, there is no data, and guidelines recommend against using it. No matter, habits die hard in Medicine; doctors still give the acid buffer.

JAMA published the BIHCA trial, which randomized 913 patients who had in-hospital cardiac arrest. One group got bicarb; the other a placebo. The primary outcome of return of spontaneous circulation did not differ (39% in the bicarb arm vs 37% with placebo). The risk ratio was 1.05 [95% CI, 0.88-1.24]; P = .62).

Comments

The specific lessons are easy: in sepsis, doctors can use judgement. In some cases, early vasopressors with less aggressive fluid may be appropriate. Routine use of super aggressive fluids in all patients should not be a performance measure. In the bicarb trials, it’s even easier: treat the underlying cause and resist the urge to give interventions to make lab values look better.

The larger evidence-based lessons though are far more important. Fluid resuscitation in sepsis and acid-base management join a long list of what I call therapeutic fashions that have been overturned by proper trials.

Examples include temperature management, oxygen targets, prevention of contrast-induced kidney injury, levosimendan in cardiac surgery and of course the older story of Swan-Ganz catheters.

Todays’ studies and this list should teach us that therapies that a) make sense and b) target physiology or c) make lab values look better should always be questioned.

The longer I practice, the more I realize that the surer we are of something, the more we should test it in trials. We are not humble enough about equipoise.

Empathy, compassion, and sitting down when taking a history; these things require no trial. Almost everything else, randomize or later look foolish.