Differences between capillary and venous blood counts in children—A data mining approach

Capillary sampling of blood counts is a well‐established alternative to venipuncture in paediatrics. However, the sampling method has to be considered when interpreting test results, as measurements differ. Ethical and practical considerations prevent simultaneous venous and capillary sample acquisition in comprehensive paediatric cohorts that span all ages for the purpose of a direct method comparison, resulting in uncertainty regarding the interpretation of capillary test results.


| INTRODUCTION
Basic hematologic tests are among the most commonly performed laboratory tests, with diagnostic and therapeutic implications for many diseases and traits during all periods of life. 1 The blood samples necessary for analysis are most commonly obtained using venipuncture.
However, in paediatric laboratory medicine, capillary blood sampling is well established as a complement or alternative to venous sampling for the analysis of haematological and biochemical analytes.
Capillary blood sampling is performed most often in paediatric patient groups where laboratory testing is performed frequently (e.g. paediatric haematology/oncology) or where difficult venous access would otherwise oppose sample acquisition (e.g. pre-term neonates and children with chronic diseases). 2 Benefits of capillary sampling include a reduction of children's stress and pain, more rapid sample acquisition, and reduced need for operator training. 3  are significantly higher when measured in capillary blood, while the platelet count (PLT) is reported to be lower (Table 1). Measurements in red cell indices (mean corpuscular haemoglobin, MCH, mean corpuscular haemoglobin concentration, MCHC, and mean corpuscular volume, MCV) also differ, however, the reported deltas between sampling techniques are inconsistent between studies. Different physiological mechanisms have been proposed to explain the differences between capillary and venous blood counts 7,10,15,17,20 : leakage of plasma out of capillaries, arterioles, and venules results in a relative hemoconcentration in capillary blood, 15 "enrichment" of larger cells (i.e. leukocytes and red cells) due to laminar flow in capillaries with "displacement" of platelets to the vessel wall, 10,17,20 and release of active substances (e.g. thromboplastin) due to samplinginduced tissue injury with consecutive changes in cell activity and concentration. 21 This demonstrates the importance of differentiating between capillary and venous peripheral blood to avoid misinterpretation of test results. 17,21 While optimum guidance of clinical decisions would theoretically require separate decision limits for venous and capillary samples, this is not feasible in clinical practice. Instead, addition or subtraction terms (i.e. "deltas") for capillary test results are favoured, which would allow "conversion" of capillary test results to venous test results. 14 However, the determination of these "delta values" is challenging: while blood count test results in the majority of individuals are within age-specific reference limits (which also need to be adjusted according to sampling method), clinical decision limits are most often especially important outside the range of physiological test results (e.g. when to perform red cell or platelet transfusions in hemato-oncological patients). 16 Conversion deltas therefore need to be established both in the physiological and the pathological range. Specific to children, age-dependent differences in body composition require inclusion of patients from all paediatric age groups in such studies. 22 While a variety of studies cover paediatric age groups, none covers children from birth to 18 years (Table 1). A major challenge when setting up these studies in children is the fact that both capillary and venous blood sampling are associated with pain, blood loss and physical injury, albeit minimal. Acquisition of an additional sample, that is not required for patient care (or acquisition of both a venous and a capillary sample in the case of healthy children) to determine "delta values" is therefore ethically challenging and has limited the number of samples analysed in previous studies.
To address these challenges, both large multicenter studies and data mining approaches have been proposed. 22,23 While multicenter studies reduce hospital-dependent bias and result in a diverse patient group, the ethical challenges specified above remain. Conversely, data mining approaches use data obtained during patient care to gain clinically and scientifically relevant insights. 24 This bypasses ethical obstacles to (repetitive) blood sampling in children, and the large amount of laboratory test results stored in hospital and laboratory information systems allows data-mining studies to access large datasets.
In this study, we examined capillary and venous blood counts from two large paediatric tertiary care centers to systematically calculate the deltas between venous and capillary sampling methods in children from birth to adulthood and in both physiological and pathological ranges.  Bellamy et al. [ To this end, we performed graphical and statistical evaluations using Python (https://www.python.org/) and R (https://www.r-project.org/).

| RESULTS
The differences between capillary and venous test results are shown in Table 2 The associations between venous test results and capillary test results are shown in Figure 1A and Figure S1 A. Although the correlation between venous and capillary test results is strong, a substantial number of measurements differ with a clinically relevant extent between both methods. Importantly, the deviation from a hypothetical perfect correlation is asymmetrical, indicating a systematic shift due to methodological differences.
A detailed analysis of the time interval between capillary and venous measurements is shown in Figures 1B and 2, and in

| Influence of age
To analyse the effect of age on capillary-venous deltas, we examined differences separately for each year-based age group ( Figure 4, Figure S4). This analysis shows relatively stable differences for all age groups, although notable differences in the first  (Table 1). 16,21 We used blood counts performed during routine paediatric clinical care to determine the systematic differences between capillary and venous haematology test results (Table 2). This approach enabled the inclusion of children of all ages and analysis of a wide range of physiological and pathological test results, in contrast to the majority of published studies, which focus on specific age ranges in healthy children or children with certain diseases.
Our results show higher haemoglobin, haematocrit, and MCV values in capillary blood counts (+6.5 g/L, +2.38%, and +2 fl) than in venous samples, while the platelet count is lower in capillary samples (À7 Â 10 9 /L). For the remaining analytes, the differences between capillary and venous samples are so small that they should not affect clinical decision making in most cases. Comparison of the calculated differences to EFLM specifications for Desirable Total Error (Table 2) confirms this. Additionally, our findings for haemoglobin, haematocrit, and platelet count are in line with results from previous studies, 3,19 while results for red cell indices except MCV, white cell count and RDW are inconsistent in-between previous studies and between previous studies and our report ( Table 1). The latter finding therefore supports our conclusion that differences in sampling methods in red When examining the effect of children's ages on differences in test results, our study offers reassuring findings: for most analytes, the impact of age is negligible in comparison to the overall differences ( Figure 4, Figure S4). and ΔRDW) are due to the sampling method, while ΔWBC and ΔRBC represent intrinsic and extrinsic changes rather than sampling method differences. 10,19 In conclusion, we used a data mining approach to investigate the differences between capillary and venous blood counts in children of all ages both in physiological and pathological ranges. The dataset examined is therefore the largest of its kind, containing 15 218 paired samples from 5794 patients, and allows guidance on the interpretation of capillary test results for the complete paediatric age range and different ranges of test results.

SUPPORTING INFORMATION
Additional supporting information may be found in the online version of the article at the publisher's website.