摘要

Objective. To apply a continual improvement model to develop an algorithm for ordering laboratory tests to diagnose acute pancreatitis in a hospital emergency department.
Methods. Quasi-experimental study using the continual improvement model (plan, do, check, adjust cycles) in 2 consecutive phases in emergency patients: amylase and lipase results were used to diagnose acute pancreatitis in the first phase; in the second, only lipase level was first determined; amylase testing was then ordered only if the lipase level fell within a certain range. We collected demographic data, number amylase and lipase tests ordered and the findings, final diagnosis, and the results of a questionnaire to evaluate satisfaction with emergency care.
Results. The first phase included 517 patients, of whom 20 had acute pancreatitis. For amylase testing sensitivity was 0.70; specificity, 0.85; positive predictive value (PPV), 17; and negative predictive value (NPV), 0.31. For lipase testing these values were sensitivity, 0.85; specificity, 0.96; PPV, 21, and NPV, 0.16. When both tests were done, sensitivity was 0.85; specificity 0.99; PPV, 85; and NPV, 0.15. The second phase included data for 4815 patients, 118 of whom had acute pancreatitis. The measures of diagnostic yield for the new algorithm were sensitivity, 0.92; specificity, 0.98; PPV, 46; and NPV, 0.08].
Conclusion. This study demonstrates a process for developing a protocol to guide laboratory testing in acute pancreatitis in the hospital emergency department. The proposed sequence of testing for pancreatic enzyme levels can be effective for diagnosing acute pancreatitis in patients with abdominal pain.

  • 出版日期2017-4