In sub-Saharan Africa severe anemia in children, defined as a hemoglobin less than either 5 g/dl or 6 g/dl, remains a leading cause of hospital admission [1] and a major factor in the 800,000 malaria deaths/year [2]. Young children and women account for more than three quarters of the blood transfusions in sub-Saharan Africa - most given as emergency life-saving treatments [3]. Despite high demand, blood supply is inadequate to meet the needs. On average only 2.3 units of blood are donated per 1,000 population in sub Saharan Africa, compared with 8.1 and 36.7 in medium and high-income countries [4]. Furthermore, most nationally funded blood services depend almost entirely on whole blood for transfusion, since the model of exclusive component preparation has technical, financial and possibly serious negative consequences and can only be feasibly implemented if substantially supported by external aid [3].

Owing to these resource-limitations, guidelines have been developed by the World Health Organization (WHO) that encourage the rational use of blood transfusion to treat severe anemia in order to prevent overuse and to reduce the risk of transfusion-transmitted infection [5, 6]. Nevertheless, evidence is lacking regarding which hemoglobin threshold criteria, volume and timing of transfusion intervention are associated with optimal survival outcomes. Consequently adherence to the current guidelines is poor [7, 8]. Children with severe anemia have high rates of in-hospital mortality (9% to 10%) [9], suggesting that the current recommendations are not optimal. An inadequate supply of blood to treat emergencies has previously been highlighted as a key factor driving poor outcomes, with 63% of the early deaths (<6 hours) occurring while awaiting transfusion [10]. A further consideration is whether the volume of transfusion is sufficient to correct the anemia. Currently, WHO recommends 20 ml/kg of whole blood (or 10 ml/kg packed cells) for all levels of anemia <4 g/dl, irrespective of initial hemoglobin or <6 g/dl if complicated by life-threatening features [5]. Few data are available on volumes received. One prospective study of pediatric transfusions in Siaya, Kenya, reported the mean volume of transfusion as 25 to 26 ml/kg whole blood [7]; however, 14% of transfusions were <15 ml/kg. Others following WHO guidelines (20 ml/kg) have shown only a modest hemoglobin rise of mean 2.5 to 3.3 g/dl [10–12] with approximately 25% remaining severely anemic (<5 g/dL) [10] following initial transfusion. Unpublished data from the Fluid Expansion as Supportive Therapy (FEAST) trial [13] indicates that of 1,422 children transfused, 322 (23%) received two or more transfusions, the proportion being greater (212/612, 35%) in those with hemoglobin <4 g/dl at enrollment. Multiple transfusions not only incur additional resource utilization but expose children to added risks of infection, transfusion reaction and adverse events. Using standard formulae to calculate the volume required [14], the current doses prescribed under-treat children with profound anemia by nearly 30% [15].

Larger initial transfusion volumes have not been systematically evaluated. We therefore investigated whether a greater initial volume of whole blood (30 ml/kg) compared to the standard recommendation (20 ml/kg whole blood) safely treated severe anemia with respect to a superior hematological correction and reduced the need for extra transfusions.

We evaluated the safety and efficacy of a higher volume of initial transfusion (30 ml/kg) than currently recommended (20 ml/kg) in a controlled trial in 160 children presenting to two hospitals in Eastern Uganda with severe anemia with respect to hematological recovery, mortality, adverse events and the need for additional transfusion. More than half of the children had a febrile illness, 60% had evidence of current or recent Plasmodium falciparum malaria, and 50% and 30% had respiratory distress and/or severe lactic acidosis, respectively. Seven (4%) children died before 28 days following admission; six fatalities occurred prior to discharge. Children randomized to 30 ml/kg (Tx30) had a superior hemoglobin recovery at 24 hours (the primary outcome) and through to 28 days (global P <0.0001); the observed data suggest that the number of children meeting the criteria for repeated transfusion was lower (5% versus 15%, P = 0.06) and there was no indication that the higher initial volume resulted in an increase in adverse or fatal events compared to those in the Tx20 arm. The combined endpoint of re-development of severe anemia and survival at 28 days following admission also favored the arm receiving 30 ml/kg, but the fact that this was designed as a pilot safety trial meant that it was not powered to detect differences of this magnitude and did not reach statistical significance (P = 0.25). Of the eight adverse events adjudicated by the ERC, one was probably or definitely related to transfusion (non-fatal allergic reaction); the others were seven fatalities that either occurred in hospital and were unrelated (four) or possibly-related (three) to transfusion, or occurred four days after discharge.

This is the first trial to examine a higher initial volume of blood transfusion for treatment of severe life-threatening anemia in African children. Enrollment in the trial was pragmatic, with few exclusion criteria, at the point of admission to hospital. Most transfusions were started within 75 to 130 minutes of enrollment, indicating an efficient transfusion service, which may have underpinned the low aggregate in-hospital mortality (4%) that is much lower than published case-series and prospective studies in comparable study populations in Africa [10, 11]. Children were managed in busy emergency rooms and pediatric wards. Children received an additional standard bundle of care, suggesting that implementation of such bundles, as well as urgent transfusion, could have also contributed to lower early mortality than in most of sub-Saharan Africa. Only one transfusion used packed cells rather than whole blood. The absence of adverse events related to volume overload provides reassuring endorsement of the relative safety of whole blood for pediatric transfusions in severely ill African children. Our study challenges the strong United States President’s Emergency Plan for AIDS Relief (PEPFAR) recommendation (made on the basis of patient safety) for component preparation, including packed cells [3]. It also supports the recent questions around implementation of these specific PEPFAR requirements for transfusion services, which are both costly and not evidence-based.

The trial was conducted at two hospitals in Eastern Uganda, in an area of intense all-year malaria transmission and where severe anemia as a cause of hospital admission is a major public health problem. Despite this, only 60% of the participants in the trial had evidence of malaria; and four of the six inpatient fatalities were in children with non-malarial febrile illnesses, indicating that the study has external validity for pragmatic management of children with severe anemia, including anemia with a non-malaria etiology. The study findings remain pertinent since, in many places, malaria has remained at the same or increased levels [24] and thus severe anemia remains a major cause of hospitalization in sub-Saharan Africa. The high frequency of children with sickle cell anemia has been noted previously in other hospital studies of severe anemia in malaria-endemic Africa [19]. The imbalance of the proportions with convulsion at admission most likely occurred due to chance (since the randomization was masked) and can occur in trials involving small numbers. Since convulsions are a risk factor for poor outcome we do not believe that these resulted in a lower mortality in the Tx30 arm [25]. Similarly, the higher in-hospital fatality rate in the 20 ml/kg transfusion arm (representing standard of care) was also consistent with chance, owing to the small size of the trial. Of note, virtually all transfusions in the trial were whole blood rather than packed cells, reflecting the difficulties local transfusion services have in preparing packed cells [3], and general lack of availability in populations similar to those studied here.

This trial primarily demonstrated that a higher initial volume of transfusion (30 ml/kg) could be feasibly and safely implemented resulting in improvements in early hematological correction and global outcome, but also suggested a reduced need for repeat transfusion prescription compared to the usual standard of care (20 ml/kg) recommended by WHO guidelines. Since the WHO transfusion guidelines have not been systematically evaluated, this has resulted in variation in practice across African countries. Incomplete hematological response in children with severe anemia may underpin the poor outcomes including relapse, readmission [26] and death [9, 26]. A policy to increase initial transfusion volume may also result in substantial cost savings, averting the overuse of scarce resources and decreasing the safety risk of further transfusion to the child; however, this cannot be recommended as the standard of care until tested in a larger trial. Further evaluation in a definitive randomized controlled trial examining efficacy and cost-effectiveness is therefore warranted.