Causes of errors in diabetic urine testing by hospital personnel

Errors in urine testing were studied using 119 health professionals who tested simulated urine specimens from six hypothetical patients. A record of the hypothetical patients' test results for 2 previous days accompanied each specimen. Five of the six records were loaded high or low to investigate testers' biases as a result of their knowledge of previous values. Testers demonstrated an aversion to extreme readings. At the lower concentration, 83% of erroneous readings were higher than the correct value. At the higher concentration, 70% of errors were below the true value. The results also revealed that readings were influenced by knowledge of past records. A higher reading for the sample with the biased-high record occurred 50% more often than a lower reading. Differences in error were nonsignificant for different educational levels and past experiences of the testers. Urine testing accuracy appears to be affected by a tendency toward avoidance of extremes and knowledge of previous results.