A guest post by William Paolo (@paolomd1), the Program Director of Emergency Medicine at SUNY Upstate.
Spontaneous intracerebral hemorrhage (ICH) continues to carry a great deal of morbidity and mortality, with little hyperacute emergent patient interventions available to the treating physician. For these reasons multiple avenues to alter patient prognosis have been explored, from continuous ICP monitoring to decompressive craniectomies, with little demonstrable patient benefit. Elevations in blood pressure (BP) in all forms of measurement (SBP, DBP, MAP) have long been associated with worse outcomes in individuals with spontaneous ICH. At issue is the nature of the association between ICH and a rise in BP, as it is not entirely clear that the subsequent elevation is directly responsible for causative harm rather than correlated poor outcomes. As BP is a simple empirical data point, multiple studies have attempted to define the relationship between poor outcomes and BP to determine if acute BP lowering will result in net patient benefit. It should be noted that the theoretical benefit of BP lowering in the face of ICH is counterbalanced by the potential for harm if elevations of MAP are in fact adaptive to allow for adequate cerebral and penumbral perfusion in the face of rising ICP.
In June of 2013 the New England Journal of Medicine published the INTERACT2 study in which individuals with spontaneous ICH were randomized into intensive BP lowering versus current guidelines based lowering. The study assessed the proportion with poor outcomes (a composite of death and major disability) as its primary outcome, and a secondary outcome consisting of multiple measures of quality of life and death. INTERACT2 found no difference in the primary outcome and discovered no difference between the hematoma growth across cohorts. Only after conducting an ordinal analysis did the authors report a better functional primary outcome in the intensive BP lowering group, a result not fully supported by a principal analysis of the study’s raw findings.
Continuing this work, The Lancet Neurology has published a post-hoc analysis of the INTERACT2 database in which it is hypothesized that variability in BP is predictive of poor outcomes in addition to total BP elevation. The authors reanalyzed the INTERACT2 raw data looking for the association between variation in BP and patient outcomes. Given multiple statistical assumptions and parsing of the initial data, the study determines that there is indeed an association between BP variation and poor outcomes. The study concludes that the findings are in fact relevant to practice and should cause the provider not only to lower the BP acutely in ICH, but to aim for a smooth reduction in BP with little variability. Unfortunately it is difficult to endorse the study’s findings given its multiple limitations and assumptions. The entire INTERACT2 database was examined as a single cohort, lumping in together both treatment arms into a single evaluative group. The authors posit that this demonstrates the robustness of the association, however it may in fact prove the opposite. While it may be true that variability is predictive independent of treatment, it may also be true, given the grouped analysis, that treatment does not affect outcomes, and that BP across treatment groups predicts poor outcomes without the ability to positively intervene to change net patient results. Further, the study is a post-hoc analysis in which a hypothesis is generated prior to the generation of data allowing only the demonstration of association. No matter how robust the findings, clinical practice changes should not be recommended based on a study whose design can only generate new hypotheses from their discovered correlations to be confirmed with prospective evaluations. Unfortunately, the debate will continue regarding BP management in ICH, as this study does not alter clinical practice nor change the negative findings of the INTERACT2 trial.
“Blood pressure variability and outcome after acute intracerebral haemorrhage: a post-hoc analysis of INTERACT2, a randomised controlled trial.”