Severe anaemia associated with Plasmodium falciparum infection in children: consequences for additional blood sampling for research

Hospitalized children in sub-Saharan Africa are frequently anaemic due to multiple causes, amongst which severe malaria caused by Plasmodium falciparum infection stands out [2, 3]. Concerns may arise when clinical or diagnostic studies are considered in this patient group, in particular when they involve sampling of additional blood. Indeed, as the oxygen binding capacity of blood would further decrease upon sampling of additional blood, the question of how much volume of blood can be sampled safely arises.

Several guidelines for safe blood withdrawal in children exist but they are mainly from institutions based in the USA and there is no general consensus. In a recent review on this subject, Howie concluded that beyond the neonatal period, sampling 3.8 % of total blood volume (equal to 3 ml/kg) over 24 h would be safe [1]. However, he also mentioned that caution should be taken in children with illnesses that impair the replenishment of blood volume or haemoglobin (Hb) [1]. It remains unclear how researchers should interpret this caution when planning studies that involve additional blood sampling in anaemic children.

When planning a diagnostic study on children suspected of severe malaria in sub-Saharan Africa, the authors decided, as part of the study preparation, to assess this 3.8 % safety guideline for its consequences through a desk-based simulation study. From a retrospective cohort of children hospitalized for fever in Burkina Faso, data about weight and Hb value were available; a large proportion of them were anaemic [4]. The estimated total blood volume of the children was corrected for the degree of anaemia. Next, the volumes of blood that were intended to be sampled as part of the planned study were matched to the 3.8 % safety guideline and the proportion of violations (exceeding) of the safety limit were calculated.

The distribution of the children’s Hb values are displayed per age group in Fig. 2. Table 2 shows the distribution of the TBVe and TBVf for best and worst case scenarios. For all age groups, TBVf was substantially lower than TBVe; differences were largest in the age groups 6–24 months and 2–6 years.

Overall, the 3.8 % safety guideline would have been violated in 5.3 % (35/666) and 11.4 % (76/666) of children (best and worst case scenario, respectively). This was most apparent in the age groups 2–6 months and 6–24 months (19 % (8/42) and 15.7 % (55/350) of children for the worst case scenario; see Fig. 3). The other age groups were less affected with, for the worst case scenario, violation of the safety limit in 8.8 % (12/204) and 1.4 % (1/70) of children aged 2–6 years and aged 6–12 years, respectively.

Calculations were also done with the median haemoglobin values found in previous studies in Tanzania [12] and Uganda [13] (Table 1). These median values are similar to the lower limit Hb reference values that were used for the present study. When the calculations were done with the Tanzanian median Hb reference values, the safety guideline would have been violated in 19/666 children (2.9 %). When the calculations were done with the Ugandan median Hb reference values, the safety guideline would have been violated in 32/666 children (4.8 %). Of note, if the safety limit was applied on the TBVe, the intended blood volumes would not have violated the 3.8 % safety guideline in any of the children.

As expected, violation of the 3.8 % safety guideline was related to the high proportion of anaemia in the cohort: 91.4 % (609/666) and 27.6 % (184/666) of children had anaemia and severe anaemia, respectively, with lowest Hb in children aged 6–24 months and 2–6 years (Fig. 2). Severe anaemia in turn was associated with recent or current falciparum malaria, which in turn coincided with the rainy season and shortly thereafter. When the worst-case scenario was considered, the proportion of children in whom the safety guideline would have been violated was 13.2 % (65/494) in the RDT-positive group versus 6.4 % (11/172) in the RDT-negative group (p = 0.016). For the best-case scenario, this difference was not statistically different.

Although in the present study severe anaemia was associated with a positive RDT, the RDT result by itself did not accurately predict possible violation of the safety guideline. For 429 out of 494 children (86.9 %) with a positive RDT, the safety guideline would not have been violated.

In this simulation study, it was demonstrated that, in children in a malaria-endemic setting, the safety guideline of 3.8 % of TBV drawn over 24 h would have been frequently violated when considering the TBVf and intended blood volumes of 6 and 10 mL as part of a planned diagnostic study. By contrast, using the estimated TBVe (i.e., not corrected for anaemia), the safety limit would have been violated in none of the children.

Violations were related to high proportions of severe anaemia, which was in turn associated with P. falciparum infection. These findings suggest that (at least in malaria-endemic settings) the 3.8 % safety guideline proposed by Howie should be modified, taking into account the TBVf rather than the TBVe. Howie indeed urged for caution when applying this guideline in children with impaired replenishment of blood volume or Hb [1].

Several limitations to the present approach should be considered. First, the 3.8 % safety guideline proposed by Howie was based on existing practices rather than on study evidence [1]. The author further only retrieved two studies that attempted to assess the impact of blood sampling for research in children; both studies, however, included only a few children and those included suffered from rare and specific diseases (children receiving chemotherapy and children with precocious puberty) [14, 15]. Meanwhile, no new studies on this issue has been published since the review by Howie.

Second, the 3.8 % safety guidelines was applied to the TBVf, which is only a proxy for the actual oxygen binding capacity. Moreover, the Hb reference values used might differ from reference values at the study site. Only a few studies have established haematological reference values in children in sub-Saharan Africa [12, 13, 16‐18]. They reported differences in reference Hb values compared to industrialized countries and among different geographical sites within sub-Saharan Africa [12‐16]. For this reason, calculations were also performed with the median Hb reference values as found in individual studies from Uganda and Tanzania (neighbouring countries of the intended study site). The results obtained with these calculations were similar to the results obtained in what was defined as the best-case scenario.

Intensity and seasonality of falciparum malaria may also be of influence; in the present Burkina Faso cohort, malaria transmission was seasonal and hyperendemic but actual Hb values and proportions of severe anaemia may be different in settings with different patterns and intensity of malaria transmission.

Finally, the clinical impact of the blood volume sampled could not be assessed. The design of the Burkina Faso study included drawing of blood cultures and EDTA-anti-coagulated blood at admission, but clinical and laboratory follow-up, including administration of blood transfusions, were not recorded [4].

Despite these limitations, it was shown that the blood volume safety guideline of 3.8 % may easily be violated in children in P. falciparum endemic areas in sub-Saharan Africa. Ethical committees and researchers should be wary of this and take appropriate precautions when considering drawing additional blood for research purposes. First, the volume of blood already sampled as part of standard (routine) patient care needs to be established, particularly since in most resource-limited settings micro-volume methods for clinical chemistry/haematology are not available. Second, before blood sampling by venipuncture, Hb can be determined on capillary blood, ideally by a bedside point-of-care method (such as HemoCue^®, HemoCue AB, Ängelhom, Sweden), or alternatively by a more cost-effective (but laboratory-bound) capillary haematocrit method. One might also consider assessing children for falciparum malaria before sampling. However, although in the present study severe anaemia was associated with a positive RDT, the RDT result by itself did not accurately predict possible violation of the safety guideline. Researchers should also consider alternative methods for sample collection. As an example, in case plasma is needed, it could be considered to collect a dried blood spot, which implies that a smaller amount of blood needs to be collected. Last and very important, research teams should be aware of the high proportions of severe anaemia among children in sub-Saharan Africa and training and supervision of clinical research staff should emphasize recognition of signs of severe anaemia and indications for blood transfusion. Children suspected of severe anaemia should either be excluded from additional sampling or, in view of the life-saving potential of some diagnostics, be sampled only when blood is available for transfusion. As the proposed guideline is based on limited evidence, future studies should attempt to assess the impact of blood sampling for research in children, especially in P. falciparum endemic areas.

The present analysis shows that the blood volume safety guideline of 3.8 % as proposed by Howie may easily be violated when considering the functional total blood volume in a cohort of febrile children in Burkina Faso. This was related to a high prevalence of (severe) anaemia in the study cohort, associated with current or recent falciparum malaria. When considering the sampling of additional blood for research in children in sub-Saharan Africa, researchers should carefully assess the volume of blood to be sampled and study protocols should include appropriate precautions with regard to patient safety.