RT Journal Article SR Electronic T1 Preanalytic and analytic factors affecting the measurement of haemoglobin concentration: impact on global estimates of anaemia prevalence JF BMJ Global Health JO BMJ Global Health FD BMJ Publishing Group Ltd SP e005756 DO 10.1136/bmjgh-2021-005756 VO 6 IS 7 A1 Leila M Larson A1 Sabine Braat A1 Mohammed Imrul Hasan A1 Martin N Mwangi A1 Fernando Estepa A1 Sheikh Jamal Hossain A1 Danielle Clucas A1 Beverley-Ann Biggs A1 Kamija S Phiri A1 Jena Derakhshani Hamadani A1 Sant-Rayn Pasricha YR 2021 UL http://gh.bmj.com/content/6/7/e005756.abstract AB The accuracy of haemoglobin concentration measurements is crucial for deriving global anaemia prevalence estimates and monitoring anaemia reduction strategies. In this analysis, we examined and quantified the factors affecting preanalytic and analytic variation in haemoglobin concentrations. Using cross-sectional data from three field studies (in children, pregnant and nonpregnant women), we examined the difference in haemoglobin concentration between venous-drawn and capillary-drawn blood measured by HemoCue (ie, preanalytic) and modelled how the bias observed may affect anaemia prevalence estimates in population surveys and anaemia public health severity classification across countries. Using data from an international quality assurance programme, we examined differences due to instrumentation from 16 different haematology analyzers (ie, analytic). Results indicated that capillary and venous haemoglobin concentrations are not in agreement (bias +5.7 g/L (limits of agreement (LoA) −11.2, 22.6) in preschool age children; range from −28 g/L to +20 g/L in pregnant women; bias +8.8 g/L (LoA −5.2, 22.9) in non-pregnant women). The bias observed could introduce changes in population survey estimates of anaemia of up to −20.7 percentage points in children and −28.2 percentage points in non-pregnant women after venous adjustment. Analytic variation was minimal and unlikely to influence the diagnosis of anaemia. These findings suggest that global estimates of anaemia prevalence derived from capillary haemoglobin, as they often are, may be inaccurate and lead to erroneous public health severity classification, but that point-of-care, or other, instruments should not introduce variation if properly used.Data are available upon request. Data will be available upon appropriate request from the authors, following database lock, unblinding and publication of the parent clinical trials from which data from this study is derived.