Posted by Sten Westgard, MS
Two interesting abstracts and papers concerning HbA1c came out recently. As laboratories switch from fasting blood glucose to HbA1c to diagnosis diabetes, the importance of the method performance of HbA1c methods is becoming critical.
But what's more important? Bias or CV?
The first answer comes from an abstract from the IFCC 2014 WorldLab conference in Istanbul:
HbA1c as the diagnostic criterion for diabetes reduces incidence and prevalence of DM2 by 25% but strongly depending on analytical quality, Poster 469, I. Brandslund, AA Nielsen, PH Petersen, A Green, CK Christensen, H. Chistensen.
They found that "Analytical variation of HbA1c had a much greater impact on the number of patients with diabetes than bias. At a bias of 0%, an increase of CVanalytical from 2.7% to 3.7% increased the number of diabetes cases by 90%, and an increase in bias to +5% by 40%." They concluded that Cv needs to be less than 2.8% and bias also needs to be less than 2.8% for optimal diagnostic use of HbA1c.
The second paper was recently published in Clinical Chimica Acta:
Influence of analytical bias and imprecision on the number of false positive results using Guideline-Driven Medical Decision Limits, Petersen PH, Klee GG, Clin Chim Acta 430C(2014)1-8.
Their findings were somewhat different: "A sharp cut-off is also a challenge to analytical performance in the medical laboratory. However, the recommendation of the need for two independent positive results to confirm the decision reduces the influence of analytical imprecision considerably, with CV values up to 4-5% having negligible effects. Regarding analytical bias, on the other hand, high levels of trueness are required as a positive bias of 2% results in a doubling of false positive in the two examples of HbA1c and cholesterol. A small analytical bias of no more than +/- 1% seems to be needed as long as sharp decision limits are used in "Guideline-Driven Medical Decision Limits"
So, as long as we're able to double the number of tests we use on patients to diagnose HbA1c, we don't have to worry about method imprecision. This is easy to say, but hard to contemplate: In 2006, Medicare Part B had 24,208,700 glucose tests. If we replace glucose testing with HbA1c testing, does our healthcare system really have the ability to absorb the cost of running 24 million more tests? It might be easier to seek out more precise methods.
But both studies definitely agree that bias is a big issue. If we're going to use 6.5% HbA1c as the diagnostic decision, every bit of bias in the method will change the number of patients diagnosed either as false positives or false negatives. It will be imperative for laboratories to choose test systems that align closely with the reference methods - and minimize bias whenever it's found.
Comments