Studying Critical Values

Abstract

No study to date has used laboratory critical values to evaluate variations in patient adverse events. We retrospectively analyzed a database of critical values to determine their distribution by hospital unit over time. The data were drawn from the Ohio State University Medical Center Information Warehouse (Columbus) for a 58-month period. Critical values were plotted over time on statistical control charts and analyzed for unusual peaks in monthly occurrence rates. Chart review of individual patient results yielded several predictor variables for the unusual peaks. Of these, occurrence of patient adverse events was the most relevant independent predictor variable for a month with an unusual number of critical values vs a normal month. This result epidemiologically confirms the basic premise of critical value reporting and suggests that the control-chart method of this type could be a new statistical tool to compare clinical activity of different hospital locations at different times.