The Fantastic Four: Quality Control Questions for the Clinical Lab

For two decades, I have been an ardent devotee of sigma metrics – a way of measuring quality and performance in the laboratory. In fact, my lab was the first in India to receive sigma certification from Westgard QC. Ever since, sigma has been one of the essential ingredients in my quality kitchen. I consider it one of the best statistical tools for measuring, monitoring, and improving the quality of testing in the clinical laboratory.

Even though I consider myself to have attained “sigma consciousness,” there is always more to learn – a fact that was brought home to me one fine day in Hyderabad. I was taking questions after a presentation when someone asked me a series of questions: the “fantastic four.”

1. What was the state of my lab’s analytical quality before we implemented sigma?
2. Can the quality kitchen thrive without sigma – even for a single day?
3. Have I ever used sigma as a tool for risk management or financial management?
4. Is sigma a mere publicity stunt? Only a few labs have attempted to bring sigma into routine quality control (QC) practice, but there are over 500 publications on sigma – most written by labs who don’t use it themselves.

These insightful questions prompted a deep dive into sigma metrics as a critical aspect of QC – an analysis I think will be useful for any lab using, or considering, the sigma approach.

Sigma metrics are derived from a mathematical calculation involving three components: total allowable error (TeA), bias, and standard deviation. On the sigma grading scale, an analyte with metrics above 6 exhibits “world-class performance,” whereas one with sigma less than 3 is considered “not fit for patient use.”

Sponsored Content

Genome Mapping Technology Case Study

Frances High explains how she used optical genome mapping to analyze structural variations in a large cohort of patients with congenital diaphragmatic hernia.

Learn more

And that brings us to the “ugly side” of sigma’s perception in the laboratory world. The tool is an industrial standard and the grading system is accordingly designed to improve productivity and reduce defects in industrial processes. Many laboratory medicine specialists may wonder how an industrial standard can fit into a clinical laboratory. If one of our tests shows a sigma of 1.5, are we reporting patient results with poor quality? This need to understand the intersection of sigma metrics and clinical utility drove me to answer the “fantastic four.”

What was the state of my lab’s analytical quality before we implemented sigma?
Sigma is a statistical quality measurement tool – nothing more. The past, present, and future of analytical quality lies not in the hands of sigma, but in the tight fist of analytical method design. So when an analyte’s sigma result is poor, the lab has only two viable options: to switch analytes or to lower its analytical goals. But does this serve the purpose of sigma – that is, to improve quality to meet clinical needs? My answer is no. It’s not analyte choice, but method selection, that is the main ingredient in our signature quality dish; sigma is added to the recipe at a later stage to test the strength of the chosen method.

Can the quality kitchen thrive without sigma – even for a single day?
In my opinion, no. I consider sigma the salt of my quality kitchen. Without salt, a kitchen cannot produce a quality dish – but too much or too little spoils the broth. You can’t blame the salt for that; it’s the fault of the chef who handles it! Similarly, you can’t wield sigma without first mastering the art of its application.

The first skill to learn is how to select an analytical goal. One size does not fit all! Various organizations have designed different sigma goals, but it’s the laboratory’s job to select the most ideal goal for any given analyte based on its clinical significance and needs. For example, you may want to select a stringent goal for creatinine, but a broader goal for AST, due to the differences in their clinical needs. Serum creatinine is the diagnostic and prognostic marker for chronic kidney disease and acute kidney injury – so even a fractional change affects treatment selection and outcome, meaning that creatinine needs a stringent analytical goal. AST, in contrast, is neither a diagnostic nor prognostic for liver disease – where ALT takes center stage – so a laboratory measurement error is deemed acceptable as long as its magnitude is within the analytical goal (total allowable error).

Have I ever used sigma as a tool for risk management or financial management?
Absolutely not! This question highlights the bitter truth of how sigma is used in clinical laboratory practice – but that’s not its true purpose. The metric should act as a whistleblower for a specific analyte’s performance quality. For example, if I discover that sigma for my potassium is 3.2, I know I need to search for the root cause of its below-par performance. Three key areas I will focus on are: i) my analytical goal, ii) my inaccuracy, or bias, and iii) my imprecision, or standard deviation. Once I’ve identified the cause of the low score, I can implement a set of QC rules – for instance, Westgard rules – to improve the performance of that particular analyte. Selecting and implementing appropriate QC rules based on the Ped (probability of error detection) and Pfr (probability of false rejection) paves the way to improving sigma… but at what cost?

The Westgard rules work on numbers (of QC rules, of QC runs, of QC failures…). When I try to fit these numbers into the science of risk management, I encounter a discrepancy – quality controls do not equal patient samples. Westgard sigma science focuses on QC; nowhere do patient results play a role. This prompts a question of my own: How can a lab claim to have improved its quality based on QC tools if it hasn’t taken into account medically unreliable results? Any tool whose aim is effective risk management must fit patient results into the puzzle.

And what of financial calculations? In practice, sigma’s value proposition is limited to quality improvement. The financial aspect of analytical process (in terms of cost incurred in QC reruns, calibrations, reagents and consumables, manpower, and even patient result recall costs in the event of QC failure) has been neither included nor explored in current sigma-oriented quality practice. As a result, the field as a whole is still in need of a process improvement tool that offers the golden trio: quality management, risk management, and financial management.

Is sigma a mere publicity stunt?
My one-word answer to this question is “sigma-phobia.”

Many laboratories experiment with sigma as an intriguing new “toy” – leading to a host of publications. But when it comes to the hardcore challenge of fully bringing sigma into the gamut of quality management systems, many professionals are hampered by worries about whether ISO 15189 has a place for sigma in quality management.

Many labs wrongly assume that, under ISO 15189:2012, sigma has no place in analytical processes. But look closer – the standard reads, “The laboratory shall design quality control procedures that verify the attainment of the intended quality of results (1).” To benefit clinical labs, the Clinical Laboratory Standards Institute has published their C24 guideline on Statistical Quality Control for Quantitative Measurement Procedures, which provides an evidence-based approach to adopting and implementing sigma metrics in quality practice. Organizations like these are paving the way for clinical laboratories to move toward stronger quality management processes.

The need of the hour is for labs to shed their sigma-phobia and move toward internal quality control practices aimed at maximizing patient safety and customer satisfaction. So I leave you with one final question: are you ready for sigma?