Posted by Sten Westgard, MS
Fresh on the heels of a study about error rates in a Romanian hospital, now we've got a new study about pre-analytical error rates at a Chinese hospital:
Corrected reports in laboratory medicine in a Chinese University hospital for 3 years, Liu X, Jiang Y, Zeng R, Zaho H, Clin Chem Lab Med 2014;52(4):e57-e59
Want to guess how many errors this lab experienced out of 1.1 million test reports in 2012?
-----126.
They only corrected 126 test reports out of more than 1.1 million. Sounds pretty good, right?
They further broke down the type of errors that caused those corrected test reports, down to the role of the person who committed the error (clinician, phlebotomist, and also technologist).
Here are a few of the numbers:
| Error | 2012 Frequency per 10,000 |
Sigma |
| Clinician errors | ||
| Entered wrong patient info | 2 | 5.1 |
| Added test order | 0 | >6 |
| Wrong test order | 1 | 5.3 |
| Phlebotomist | ||
| Patient ID error | 5 | 4.8 |
| Specimen quality problem | 1 | 5.3 |
| Detection Technologist | ||
| Mistook (Wrong) specimen | 0 | >6 |
| Inappropriate specimen processing | 1 | 5.3 |
| Reagent quality problem | 0 | >6 |
| tested incorrect analyte | 0 | >6 |
| Missed test | 7 | 4.7 |
| Manual entry error | 6 | 4.8 |
| Corrected test value | 63 | 4 |
| Missed notification of critical value |
22 | 4.4 |
These are pretty high Sigma-metrics. Some processes appear to be defect-free. Impressive if these accurately reflect what's going on with the laboratory system. Where we lack some information is what happens to an error that gets caught before the test is released, so no corrected error report is necessary? If that were added in, we would see a more comprehensive picture of error rates.