Glycated haemoglobin measurements from UK Biobank are different to those in linked primary care records: implications for combining biochemistry data from research studies and routine clinical care

Abstract

Biochemical tests of the same individual carried out on different test platforms are often not comparable due to bias in the assay method and calibration.1,2 Combining measurements from different sources, or diagnoses based on these measurements, is therefore not always valid. We highlight an example using glycated haemoglobin A1c (HbA1c) test results from two different sources in UK Biobank data: HbA1c measurements taken at baseline assessment using a single assay method (the Bio-Rad Variant II Turbo HPLC analyser3) and HbA1c measurements from linked UK primary care records, where the assay method was dependent on which National Health Service (NHS) laboratory the sample was processed in. We identified UK Biobank participants with no pre-existing or previous diagnosis of diabetes mellitus (any type), with a primary care HbA1c measurement ≤100 days before or after baseline assessment (n = 1039; a detailed method is provided in Supplementary Figure S1, available as Supplementary data at IJE online). In individuals without diabetes, HbA1c should be relatively stable within this short time frame. We found that UK Biobank baseline measurements were on average lower than primary care measurements with a mean difference of 2 mmol/mol (Figure 1), regardless of whether the primary care measurement was taken before or after the baseline assessment.