A major study of electronic medical records by researchers at Florence A. Rothman Institute (FARI) introduced a new model for reporting laboratory test results, based on evidence of risk. This is the first application of “big data” techniques to aid physicians with diagnostic screening, according to researchers.
"Lab results for diagnostic screening are more meaningful when mortality risk is used to guide physicians. This leads to earlier diagnoses and improved patient outcomes," said Steven I. Rothman, co-author of the study and co-developer of Rothman Index, a measurement system being used at academic medical centers and residential care facilities to provide early warnings of clinical downturns. "Availability of data is not sufficient to improve health, a novel analytic method is required to hear the data speak," he added. "FARI’s next challenge is prodding laboratories to act: the lab tests are fine, but the lab reports ought to supply more accurate and useful information."
FARI’s study “A New Theory For Reference Intervals and Analyte Test Reporting Based on Clinical Risks Derived from Readily-Available EMR Data” received the Distinguished Abstract Award from the National Academy of Clinical Biochemistry and the Best Abstract Award for Informatics from the American Association for Clinical Chemistry.
G. Duncan Finlay, MD, president of FARI, explains, “We built a database of de-identified medical records from three hospitals in Florida, Texas, Connecticut, comparing the results of commonly tested blood chemicals (such as potassium) with risk of death.” According to Mayo Clinic, an estimated 60% to 70% of all decisions regarding a patient's diagnosis and treatment, hospital admission and discharge are based on laboratory test results.
“Surprisingly,” said Rothman, “most lab reference ranges are defined by a method developed in the 1800s, based on samples of 200 healthy people. In contrast, FARI analyzed 375,000 records to discover where the human body, as a chemical factory, prefers to operate – the zone of minimum risk is our replacement for the old reference ranges.”
"For example, potassium is not reported at risk unless the patient's test value is outside the central 95% of a lab’s population sample,” said lead author Alan B. Solinger, PhD, "The lab flags those outlying values as H or L (high, low) and physicians typically rule out a problem unless a test is flagged in the report." But when FARI researchers looked at risks associated with potassium values, patients in the upper third of the standard reference range had three times the risk of dying as patients in the lower portion of the range. "Heart-attack patients show this increased risk, as reported in Journal of the American Medical Association. However, FARI found this risk is not limited to patients with prior heart attack. Broadening the scope of study revealed that all patients have the same increased risk. We see significant risks within other reference ranges, which indicates a need for evidence-based reporting of lab tests," said Rothman.