Use of roster charts in the investigation and prosecution of nurses suspected of inflicting deliberate harm on patients

When there are no witnesses, no substantive material evidence, no plausible motive, and, moreover, nothing beyond speculation to indicate that a crime has even occurred, then the prosecution of suspected serial killer nurses will be difficult. It will lean heavily on statistical arguments. There are two pillars to these arguments; the first is to show, on the basis of a believed inexplicable spike of deaths and life-threatening events, that the best explanation is that of criminal activity. While identifying a cluster is an awkward statistical question, it is not the one we address here. Our focus is on the second pillar of the argument, one that aims to show the accused as the most likely culprit in the light of roster chart evidence. The basis of this evidence comes from what is judged to be an unusually high correlation of the accused's presence with the events when they took place. We show that such roster chart evidence is unreliable. A visual inspection of any chart will be misinterpreted, the reader coming to an erroneous conclusion. We consider a recent case and, via careful calculation, show that what was believed to be strong evidence of culpability is, in truth, anything but. The often heard statement "whenever something went wrong, the nurse was always there," has no basis in fact.