The application of Student’s t-test in internal quality control of clinical laboratory

Background

James O. Westgard advised that the relation between internal quality control (IQC) and statistical methods, such as t-test, should be seriously considered. The results of IQC fully fit in the requirements of the t-test. But there was no research on analyzing the results of IQC by t-test. So the study aimed to explore the value of t-test for IQC by analyzing a specific example.

Methods

The results of IQC about luteotropichormone (LH) were collected in July 2015 in two instruments from a clinical laboratory as a specific example. The pre-determined target value of IQC was evaluated by a one-sample t-test first. Then the consistency of two instruments was analyzed by comparing their results of IQC with two paired samples t-test.

Results

In middle LH level of No. 1 instrument, though there was no out-of-control IQC result, the difference between the mean of IQC results in July 2015 and the pre-determined target value was statistically significant (t = 2.311, p = 0.027). So it was earlier to find whether the pre-determined target value was inappropriate by using a one-sample t-test when no out-of-control results appeared in the traditional IQC chart. For high LH level, the difference between the IQC results of two instruments was statistically significant (t = 2.435, p = 0.020) in July 2015, so it was helpful to discover the inconsistency of two instruments by analyzing their IQC results with two paired samples t-test.

Conclusions

The one-sample t-test was helpful to guide the correction of pre-determined target value of IQC. Two paired samples t-test was beneficial to improve consistency between two instruments.