Quality of HbA1c measurements in Australia – A Report from the RCPAQAP — ASN Events

Quality of HbA1c measurements in Australia – A Report from the RCPAQAP (#75)

Paul Williams 1 2 , Sabrina Koetsier 3 , Graham Jones 4
  1. Endocrinology Department, Royal Prince Alfred Hospital, Sydney, NSW, Australia
  2. University of Sydney, Sydney, NSW, Australia
  3. RCPAQAP, Adeliade 5000, SA, Australia
  4. SydPath, St. Vincents Hospital, Darlinghurst 2010, NSW, Australia

Introduction

In 2012 in preparation for the introduction of HbA1c for diagnosis of diabetes, RCPAQAP introduced a fresh whole blood HbA1c program to supplement  the existing program based on lyophilised samples. We report on the findings of this program.

Methods

Collections of whole blood were made from volunteer patients with a range of HbA1c concentrations between 4.8 and 11.3%. Aliquots were distributed to participating laboratories within 24-48h and results returned to in either NGSP% units or IFCC mmol/mol units for statistical analysis and reporting.  The data in this abstract is presented in NGSP % units.

Results

Participating laboratories increased from 111 in 2012 to 132 in 2015 by which time 52% reported in IFCC units. The total sample CVs (including within- and between-instrument, and between-method variation) ranged between 2.9 and 4.5% with an average of 3.5%. Bias compared with a certified reference method (RefMeth) was: All Laboratory Mean = 0.994 x RefMeth +0.06.  Performance within manufacturer groups was good with average sample CV and correlations for larger participants as follows:  BioRad (Cation Exchange) Ave CV 2.6%, Mean = 1.04 x RefMeth - 0.17 (+0.09 average bias when RefMeth = 6.50%) Roche (Integra): Ave CV = 2.7%, mean =0.974 x RefMeth + 0.20 (+0.03) Siemens(DCAs): Ave CV 2.8%, mean = 0.970 x RefMeth +0.172 (+0.00) Allere(afinion): Ave CV = 2.7%, mean =0.959 x RefMeth + 0.20* (-0.07).
The Afinion results in late 2013 showed a low bias of 0.4%. The company has subsequently corrected this issue (affected data removed from results above). A sample with haemoglobin F introduced significant bias in some methods reducing accuracy but not affecting precision. With an assigned target value of 8.1%, Biorad Cation Exchange gave 8.3± 0.16 and immunoassays gave the following lower results: : Siemens DPC (n=26)  gave 7.76±0.16:  Roche Cobas (N=16) gave 7.8±0.3for this sample.

Conclusions

The HbA1c results produced by routine laboratories in Australia are of a high quality. Different technologies can produce acceptable analytical performance when performed correctly although HbF samples can affect some methods. Laboratories should ensure selection of appropriate methods and use to maintain ongoing quality of results.