Abstract Body: Introduction: INTEREACT 4 demonstrated lowering blood pressure in the prehospital setting improves outcomes for patients with intracerebral hemorrhage (ICH) but worsens outcomes for acute cerebral ischemia (ACI). Consequently, hyperacute antihypertensive therapy could potentially aid prehospital patients identified as likely having ICH by clinical scales and helmet technologies. The diagnostic performance characteristics needed to yield net benefit have not been well-delineated.
Methods: We modeled net treatment benefit and harm using magnitude of beneficial and adverse effects of BP lowering in INTERACT 4 with a 2-stage algorithm. In stage 1, we continuously varied prehospital diagnostic test sensitivity, specificity, disease prevalence, with output of positive predictive values (PPVs) for ICH. In stage 2, PPVs were converted to net treatment effect using INTERACT 4 outcomes.
Results: The results of the stage 1 PPV module are shown in Figures 1 and 2, with continuous variation of specificity (1) and sensitivity (2) at different biologically plausible levels of ICH prevalence for Western and Asian populations. PPV is more strongly influenced by test specificity than test sensitivity. At sensitivity 80% and specificity of 80%, as prevalence varied from 20%, 30%, 40%, 50%, PPVs were 50%, 63%, 73%, and 80%. The results of the stage 2 mRS outcomes module are shown in Figure 3 for the endpoint of functional independence (mRS 0-2). With PPVs below 30%, treating test-positive patients with BP lowering resulted in net decrease in mRS 0-2 outcome but with PPVs above 30%, treating test positive patients results in net increase in mRS 0-2 outcome, with benefit magnitude increasing as PPV increased. To achieve the minimally clinically important difference (MCID) in mRS 0-2 increase of value to patients of 1.3%, test PPV is 39% or higher. To achieve the MCID that would be practice-changing of 5%, test PPV is 67% or higher. At test PPV of 90%, the mRS 0-2 increase would be 8.2%.
Conclusion: This model provides a framework for developing prehospital ICH recognition instruments by profiling BP-lowering treatment effects associated with test performance and biologically plausible disease prevalence. Prehospital tests that yield PPVs for ICH of 67% or higher would be associated with treatment benefit magnitudes traditionally considered sufficient to change practice.