Evaluation of biochemistry tests with different process sigma levels via patient-based quality control implementation

Introduction: Clinical laboratories are responsible for ensuring the quality and accuracy of reported test results. The total testing process is divided into three phases, and the performance of the analytical phase is evaluated with internal quality control and external quality assessment programs. However, these quality management tools only cover daily or monthly programs. There is a need for monitoring the continuity of quality control between quality control intervals. Objective: This study aimed to calculate the process sigma values of some biochemical tests conducted at the Clinical Biochemistry Laboratory of KTU Faculty of Medicine Farabi Hospital and to establish and optimize patient-based quality control application rules for these parameters. Additionally, it aimed to assess the supportiveness, applicability, and usefulness of these methods as complementary to traditional internal quality control practices, as well as to determine the relationship between process sigma values of the tests and algorithms. Materials and Methods: Considering the total allowable error limits of CLIA 2024, process sigma values were calculated for clinical chemistry parameters from the year 2021. Patient data from 2021 was obtained from the Laboratory Information System (LIS) and sorted according to the analysis date. Moving average and moving median algorithms with different series sizes, as well as exponentially weighted moving average algorithms with different coefficients, were applied to patient data to establish control limits. Various percentage biases were applied to the datasets to conduct power function and bias simulation analyses, determining the most suitable algorithms. The established control limits were then applied to patient data from the year 2022, and the number of alarms per thousand was calculated, examining their relationship with process sigma. Results: In general, it has been observed that the probability of generating alarms increases as the sample size or coefficient decreases, but alarms occur in later patient counts. Patient-based quality control algorithms may achieve TEa targets in albumin, iron, calcium, magnesium, potassium, sodium, and total protein tests; however, they fail to achieve these targets in ALT, BUN, inorganic phosphate, GGT, glucose, creatinine, LDH, triglyceride, and uric acid tests. Different algorithms have been selected for tests in the positive and negative bias directions. The proximity of test distribution to normality plays a significant role in determining the most suitable algorithm. Despite the similarity of process sigma levels between 2021 and 2022, it has been found that control limits do not fit with some tests. Conclusion and Recommendations: The most important step in the implementation of patient-based quality control algorithms is determining the control limits. Exclusion criteria, truncation limits and transformation can be applied for tests to fit a normal distribution. No relationship was found between process sigma levels and the power of algorithms. It is recommended that patient-based quality control algorithms be applied not only to tests with process sigma levels <4 but to all tests. Keywords: Patient-based quality control applications, process sigma level, analytical performance