On the flip side, the evidence is better, if still mixed, regarding the utility of early warning scores for deterioration on the inpatient side. Although such scores have been associated with improved inpatient mortality at Kaiser, mixed results have been published in this past year.⁴ On one hand, a retrospective evaluation f the commonly-deployed Epic Deterioration Index (EDI) showed a decrease in unplanned escalations in care without evidence of deleterious effects on mortality.⁵ Conversely, an evaluation comparing several early warning tools found little difference between the EDI and common bedside tools such as the MEWS and National Early Warning Score (NEWS/NEWS2).⁶ The true answer as to what value any electronic deterioration tool may or may not add to your institution will depend greatly on existing culture, workflow, and tools already in use.

Best Protection for Brains

Among many types of critical illness, evidence supports the noninferiority of “restrictive” transfusion strategies. The transfusion cut-offs vary, but it frequently appears reasonable to permit hemoglobin levels to drop, rather than proactively transfuse to prevent severe anemia. The brain is an entirely different organ system, however. The TRAIN trial evaluated whether a liberal or restrictive strategy was preferred in patients suffering traumatic brain injury, subarachnoid hemorrhage, or intracerebral hemorrhage.⁷ Considering the general sensitivity of the brain to hypoxemic and ischemic insults, it is not terribly surprising to find that liberal transfusion strategy was superior.

One of the paradoxical oddities of treatment for acute stroke is the reversal of anticoagulation to enable treatment with thrombolysis. In common practice, this usually involves using idarucizumab to bind dabigatran prior to further acute treatment. In a retrospective, registry-based study and systematic review, the authors reported generally favorable outcomes with few complications to this exceptional practice.⁸ However, the nature of these retrospective studies only serves to amplify the selection biases at work in current practice, demonstrating the baseline superior prognosis of those selected for treatment, rather than any value or safety of the treatment itself. Randomized, controlled trials remain necessary to ultimately determine the efficacy and safety of this practice.

Finally, in what serves as a bit of idle academic curiosity, a re-examination of the rate of intracranial hemorrhage after thrombolysis found starkly different statistics than those typically used in discussions with patients.⁹ In a re-analysis of the ENCHANTED study, the authors reported the incidence and outcomes of patients suffering asymptomatic intracranial hemorrhage (aICH), rather than just the typical symptomatic intracranial hemorrhage (sICH). The rates of sICH were consistent with the typical one percent to two percent generally quoted during informed consent, whereas rates of aICH were ten times this number. As expected, odds ratios for poor outcomes and death for those with sICH were in the 20-60 range, but aICH was certainly not benign, with odds ratios around two. Although there may be net chance of benefit to thrombolysis, the proportion harmed by treatment may approach 20 percent, vastly higher than appreciated.