Send letter to journal:
Re: Over-estimation of the prevalence of iron deficiency anaemia by screening with the HemoCue haemoglob

Dear Editor,

We note with interest the paper by Sherriff and colleagues about screening for anaemia in the ALSPAC study. Various small scale surveys, including our own, have revealed a prevalence of anaemia (defined as a haemoglobin <110g/l) of between 4-18% in White children, 17-28% in Asian children and 8-20% in Afro-Caribbean children of the age range 8-72 months [1]. Capillary blood samples obtained from a finger or heel prick were used in most of these surveys. Blood was collected either from the skin puncture directly into a microcuvette and measured by a portable HemoCue haemoglobinometer (HemoCue Ltd, Sheffield), or into specimen bottles or capillary tubes and analysed by the HemoCue (as was the case in the ALSPAC study) or by standard laboratory techniques. All junior paediatricians will be aware of the considerable squeezing of the heel or finger that may be required to fill a specimen bottle or capillary tube with blood and that this method is notoriously messy!

The HemoCue system has shown highly comparable results to the "gold standard" laboratory method of haemoglobin estimation for venous blood specimens from older children [2], especially when the venous blood collected into specimen bottles has been allowed to mix thoroughly prior to analysis. However few data exist in young children on the correlation between haemoglobin measurements obtained from venous specimens analysed in the laboratory and "direct" capillary blood specimens measured by the HemoCue. Nevertheless, this "direct" method of analysing one drop of blood obtained from a finger or heel prick has been widely used in a number of population-based early childhood anaemia screening programmes [3]. Now data from a small pilot study of methodologies being devised for a large scale randomised controlled trial of iron therapy in anaemic toddlers predominantly from Asian and Afro-Caribbean communities, has indicated that screening for anaemia by "direct" capillary sampling can significantly over-estimate the its true prevalence. 60 children aged 14 months attending their general practitioner for MMR immunisation were screening for anaemia by a thumb prick blood test performed by a trained practice nurse using an automated lancet. 17 children (28%) had a capillary haemoglobin <110 g/l according to the HemoCue (range 81-109 g/l). These children subsequently had venous blood drawn after the application of local anaesthetic cream. Only 4 were confirmed as being truly anaemic by laboratory analysis of venous blood ie 6.7% of the total screened. The capillary HemoCue results were all lower than the venous laboratory results with a mean discrepancy of 13 g/l (range 4-29 g/l). A similar discrepancy of 15 g/l was obtained in a study of 72 women attending for routine venous blood tests at an ante-natal clinic. The "direct" capillary HemoCue method therefore consistently under-estimated the true haemoglobin, presumably because of haemodilution by lymphatic fluid or sweat at the puncture site particularly if pressure has had to be applied to obtain an adequate specimen.

Clinically important discrepancies of greater than 10 g/l between repeat HemoCue haemoglobin measurements from single drops of blood have already been reported [4]. This variability can be reduced if a number of blood drops are mixed in an EDTA capillary tube prior to measurement [5]. This was the method used in the ALSPAC study. Subsequently, the reported prevalence of anaemia at 8 months in this study (23% with the conventional definition of anaemia) [6] is likely to be more accurate than if the single drop "direct" method had been used. However there is no verification of the accuracy of these estimates by comparison to venous blood test results.

The "direct" method of capillary blood sampling is then unacceptably inaccurate as a means of screening for anaemia especially if a HemoCue reading of <110 g/l is used to define anaemia. A lower instrument reading of 100 g/l might identify those who are truly anaemic with greater specificity, but 2 children from our series with venous haemoglobins less than 110 g/l would have then been missed. Our findings suggest it is essential to obtain venous blood to verify the capillary estimation of haemoglobin particularly if the single drop direct method has been used for screening for anaemia

Robert Moy
Institute of Child Health

Joe Kai
Sue Wilson
Department of Primary Care and General Practice

KK Cheng
Department of Public Health and Epidemiology

University of Birmingham Medical School

Referencs