---

Suppose you see a patient with hepatorenal syndrome and acute kidney injury. What to do? You give pressors—the evidence base is good.

But what about albumin?

You look at the ADQI/ICA recommendations. It discusses two trials, Maiwall et al. 2022 and Philips et al. 2021, and from these trials it concludes:

“The use of albumin was associated with a significantly greater improvement in haemodynamics in the short term, [but] … did not improve renal outcomes or need for RRT compared to crystalloids.”

Let’s take a look at these trials.

First, Maiwall et al. 2022 state the following.

“Albumin was superior to plasmalyte in achieving MAP > 65 mmHg at 3 hours (62% vs. 22%; p <0.001). [A] reduced need for dialysis (48% vs. 62%; p = 0.16), and a longer time to initiation of dialysis (in hours) (68.13±47.79 vs. 99.7± 63.4; p = 0.06) were seen with albumin.“

Note the lack of significance of the dialysis outcomes. However, the dialysis outcomes are secondary. The trial’s primary outcome, for which it is powered, is mean arterial pressure.

Similarly, the primary outcome in Philips et al. 2021 is the reversal of hypotension, and the trial is powered for this as well. Secondary outcomes include urine output, which is not found to be different in the albumin and saline groups (p=0.291).

Hence, both trials were powered for mean arterial pressure, not renal outcomes. Mean arterial pressure, a physiologic signal, likely does not fluctuate much, and therefore, in between group comparisons, has more signal than outcomes like the time to initiation of dialysis, which may depend on external factors like the hospital workflow. Urine output is also a physiologic signal, but it might be noisier than something like mean arterial pressure. Detecting a difference between the treated and untreated group might be easier for a less noisy outcome. Mean arterial pressure, therefore, might require a relatively small sample to detect an effect, whereas time to dialysis, and other renal outcomes, might require a larger one.

Hence, when interpreting potentially noisier secondary outcomes in these trials, one has to be careful not to assume that “absence of evidence is evidence of absence.” Overall, it seems, the trials did not imply that albumin did not improve renal outcomes (as stated in the consensus guidelines). Instead, the trials were inconclusive.

The ADQI/ICA recommendations then go on to give a nice biological rationale for using albumin. This rationale might be dismissed, if one misinterprets the trials as evidence of absence. In general, although the wording could be improved in the ADQI/ICU recommendations, the authors luckily avoid dismissing this rationale, as does UpToDate.