Algorithm-based decision rules to safely reduce laboratory test ordering

PURPOSE: Our study develops decision rules to define appropriate intervals at which repeat tests might be indicated for commonly ordered laboratory tests for hospitalized patients. METHODS: The final data set includes 5,632 adult patients admitted to the University of Virginia Hospital between July 1995 and December 1999. These patients had a hospital length of stay of five days or more and had results recorded for three routinely ordered laboratory tests for each of the first five days of their hospitalization. We use the serum potassium test to illustrate our algorithmbased decision rule methodology. RESULTS: Our decision rule begins with testing on the first two days of hospitalization and allows for repeat testing after observation of any non-normal values. The results show that the algorithm-based decision rule would lead to a 34% reduction for serum potassium tests for the first five days of hospitalization. Only one out of the 5,632 patients in our sample had a critical value that occurred only on a non-test day and, thus, was missed by the algorithm. CONCLUSIONS: The algorithm results are encouraging. We demonstrate that the number of tests can be reduced while missing critical values in only a small fraction of patients. Testing algorithms such as these can be used to reduce laboratory test ordering without compromising the quality of patient care.