Title: Allogeneic umbilical cord red blood cell transfusion for children with severe anaemia in a Kenyan hospital: an unmasked, single arm trial to assess safety, harm and efficacy. Oliver W. Hassall1,2, Johnstone Thitiri1, Greg Fegan1,3, Fauzat Hamid1, Salim Mwarumba1, Douglas Denje4, Kongo Wambua5, Kishor Mandaliya4,5, Prof. Kathryn Maitland1, 6, Prof. Imelda Bates2 1 Centre for Geographic Medicine Research (Coast), Kenya Medical Research Institute/ Wellcome Trust Research Programme, Kilifi 80108, Kenya 2 Liverpool School of Tropical Medicine, Liverpool L1 5QA, United Kingdom 3 Centre for Clinical Vaccinology & Tropical Medicine, University of Oxford, Oxford OX3 7LJ, United Kingdom. 4 Coast Provincial General Hospital, Mombasa 80100, Kenya 5 Regional Blood Transfusion Centre, Mombasa 80100, Kenya 6 Department of Paediatrics, Imperial College London, London SW7 2AZ, United Kingdom Corresponding author: Dr. Oliver Hassall (address as 2. above) ohassall@gmail.com + 44 (0) 7774 354716 1 Abstract Background Severe anaemia requiring an urgent blood transfusion is common in hospitalised children in sub-Saharan Africa but blood is frequently unavailable. Where conventional blood supplies are inadequate, allogeneic umbilical cord blood may be a feasible alternative. The aim of this trial was to assess the safety and efficacy of cord blood transfusion in children with severe anaemia. Methods Cord blood was donated at Coast Provincial General Hospital and screened for transfusion-transmitted infections and bacterial contamination. Red cells were produced by sedimentation during refrigerated storage. Children with severe anaemia but without signs of critical illness were recruited at Kilifi District Hospital and received a maximum of two group identical/compatible cord blood units. Participants were closely monitored for adverse events and followed up for one month. The primary outcome measure was the frequency and nature of adverse reactions associated with the transfusion. Secondary outcome measures were change in haemoglobin at 24 hours and one month after transfusion compared to pre-transfusion levels. The study has been completed. (Trial registration:ISRCTN66687527) Findings Fifty-five children received sedimented red cells from 74 cord blood donations. Ten children experienced 10 serious adverse events and 43 children experienced 94 non-serious adverse events. In none of these cases did an independent expert panel consider cord blood transfusion to be probably or certainly implicated (one-sided 97.5% confidence interval; 0 to 6.5%). The median rise in haemoglobin was 2.6g/dL (IQR 2.1g/dL to 3.1g/dL) 24 hours after transfusion, and 5.0g/dL (IQR 1.0g/dL to 6.8g/dL) 1 month after transfusion. Interpretation 2 The results demonstrated by this single arm trial, justify further studies comparing the safety and efficacy of cord blood transfusion and conventional adult-donated blood transfusion. Such trials should include operational analyses of the availability of cord blood and conventional blood. Challenges associated with cost, infrastructure and scale up also need exploring. Cord blood may be an important supplementary source of blood for transfusion in children in sub-Saharan Africa. Funding Wellcome Trust Training Fellowship (073604) 3 Introduction Background Sub-Saharan Africa has the highest risk of death in the first month of life and is one of the regions showing the least progress in reducing this high mortality rate.1 Severe anaemia is a major public health problem in sub-Saharan Africa and children aged less than 2 years are the most frequently affected. The prevalence of severe anaemia in hospitalised children is reported to range from 8-29% with case fatality rates of 8-18%.2 In children with severe, uncompensated anaemia, blood transfusion can substantially reduce mortality.3 Over 50% of deaths occur within 4 hours of admission and early intervention and the ready supply of safe blood are key components of the hospital treatment of severe anaemia in childhood.4,5 The supply of conventional blood for transfusion in sub-Saharan Africa is insufficient with only an estimated 52% of demand being met and a shortfall of at least 2 million units a year.6–8 Where the blood supply is limited and young children receive a significant number of blood transfusions, umbilical cord blood is a novel and potentially important source of blood for transfusion.9–11 Not only might cord blood provide increased numbers of small volume transfusions but also reduce the pressure on stocks of conventional, adult-donated blood thereby improving the supply of blood for emergency transfusions for other vulnerable groups. In sub-Saharan Africa, lack of blood for transfusion is implicated in 25% of maternal deaths due to haemorrhage.12 In order to test the feasibility of cord blood transfusion, we have established a cord blood donation programme on the labour ward at Coast General Provincial Hospital in Mombasa, Kenya. Previously we have demonstrated the acceptability to mothers of cord blood donation and transfusion; the feasibility of a two-stage informed consent process for cord blood donation; and the quality of variable volumes of whole cord blood stored in a fixed volume of anticoagulant-preservative solution.13,14 We have also shown that, for cord blood collected by our study team, rates of both bacterial contamination and seroreactivity for HIV, HBV, HCV and syphilis compare favourably to that of 4 conventional adult blood donated to the Regional Blood Transfusion Centre in Mombasa.15 Here we report the findings of, to our knowledge, the first clinical trial of allogeneic cord blood transfusion in children with severe anaemia. Objectives The primary objective of the study was to assess the frequency and nature of adverse reactions associated with umbilical cord red blood cell (UC-RBC) transfusion. The secondary objective was to assess the haematological efficacy of UC-RBC transfusion. Methods Trial design This was an unmasked, single arm trial designed to produce preliminary data on safety, harm and haematological efficacy of umbilical cord blood transfusion in children with severe anaemia. The protocol was reviewed and approved by the Kenya National Ethics Committee and the Research Ethics Committee of the Liverpool School of Tropical Medicine. The trial is registered as an International Standardised Randomised Controlled Trial, number ISRCTN66687527. Participants and study setting Participants were recruited from children aged less than 12 years admitted for paediatric in-patient care at Kilifi District Hospital (KDH), Kenya from 26th June 2007 to 20th May 2008. Eligibility criteria were designed to identify those children for whom a transfusion would provide clinical benefit based on WHO clinical guidelines but exclude those who were critically ill.16 Research staff from the KEMRI-Wellcome Trust research programme provide 24 hour clinical cover at KDH and at admission all children have a structured clinical assessment, including anthropometry and a standard set of laboratory investigations, including a haemoglobin concentration (Hb) estimation (Beckman Coulter, France), a blood film examination for malaria and a blood culture. Haemoglobin electrophoresis to detect haemoglobin S was done retrospectively for study children aged greater than 3 months of age. Full 5 investigation of the aetiology of severe anaemia was not part of the study protocol. Children were eligible for inclusion in the study if they had severe anaemia (Hb 10g/dL in children aged 3 months or less; Hb 4g/dL in children aged greater than 3 months) and the attending clinician requested a blood transfusion. Children with any of the following clinical features of critical illness were excluded: coma (Blantyre Coma Scale 2), prostration, shock, deep (acidotic) breathing, and hyperbilirubinaemia requiring exchange transfusion. In addition children were not eligible for the study if they had had a previous UC-RBC transfusion as part of this trial or were already enrolled in another intervention trial. A child was only enrolled in the study if sufficient cord blood was available and written consent was given by their caregiver. The intervention The intervention under investigation was the transfusion of umbilical cord red blood cells (UC-RBC). Cord blood was collected from placentas donated at Coast Provincial General Hospital in Mombasa and screened (for HIV, hepatitis B and C and syphilis) as described previously.15 Screened cord blood units were transported at 2-6C 50km by road to Kilifi, sedimented by storing vertically in racks at 2-6C and quarantined until screened for bacterial contamination. This was done by incubation of a 4ml sample in 40ml of brain heart infusion at 37C in the manner described previously.15,17 Incubation was for 48 hours and screening was by microscopic examination of a Gram stained smear. Volume, haemoglobin concentration and blood group of cord blood units were entered on an electronic database, which was used to ascertain whether sufficient cord blood was available as soon as a blood transfusion was requested for an eligible child. This was defined as at least 2.2g/kg of haemoglobin from a maximum of two group identical and/or blood group compatible cord blood units. Thus cord blood units were selected based on 6 estimated haemoglobin content rather than volume. In addition, no child was transfused more than 3.5ml/kg of CPDA-1. The hospital clinical laboratory used standard methods of blood grouping and crossmatching. In children without severe acute malnutrition (SAM) (defined as weight-for-height Z-score (WHZ) < -3), UC-RBC were transfused over 4 hours with no co-administration of furosemide and a maximum permitted volume of 20ml/kg. In children with SAM, UC-RBC were transfused over 3 hours with a maximum permitted volume of 10ml/kg; and 1mg/kg of furosemide administered intravenously at the start of the transfusion as per clinical guidelines.16 Outcomes The outcome measure to achieve the primary objective was the frequency and nature of adverse reactions occurring during, or within at least one month of, UC-RBC transfusion. Serious adverse reactions (SAR) were defined as any serious adverse event (SAE)1 that was judged probably or certainly related to the transfusion. Adverse reactions were defined as any adverse event (AE)2 judged probably or certainly related to the transfusion. The detection of adverse reactions was a two-stage process comprising the rigorous surveillance of adverse events (monitoring of harm) and an independent, expert judgement about their relationship to UC-RBC transfusion (assessment of imputability). Monitoring of harm (Figure 1) Monitoring of harm was by both passive and active surveillance. Children recruited to the study were admitted to a paediatric high dependency unit until 24 hours after the start of the transfusion. During the transfusion and for two hours afterwards children had continuous physiological monitoring. Temperature, pulse rate, respiration rate, oxygen saturation, and blood 1 Any untoward medical occurrence that is fatal, life-threatening, disabling, prolongs hospitalisation, or results in hospitalization.18 2 Any untoward medical occurrence.18 7 pressure were recorded before the start of the transfusion, 15 minutes after the start of a transfusion and every 30 minutes thereafter for the duration of the transfusion and for 2 hours subsequently. Two hours after the start of the cord blood transfusion, a blood sample was obtained for the estimation of serum potassium (Ilyte Ion Selective Electrode Analyser; Instrumentation Laboratory, US) and calcium (Selectra E; Vitalab, The Netherlands). A clinician reviewed every child and performed a study-specific structured clinical assessment designed to capture adverse events 2 hours after and 24 hours after the end of a transfusion, and at hospital discharge. For the rest of the child’s admission, monitoring of harm was by review of the daily clinical record kept by the attending clinicians. At hospital discharge, carers of children recruited to the study were given the cost of their fare home and the return fare back to the hospital and invited to bring the child to the hospital one month after the cord blood transfusion. They were encouraged to come back to the hospital before then if they had any concerns about their child. In addition, details of their homestead location were taken. Children who returned to the hospital had a structured clinical assessment. Those who did not attend were followed up at home by a fieldworker, who confirmed whether the child was alive and well by direct observation of the child and/or discussion with an adult family member. Carers were also encouraged to bring these children to the hospital for a full review. Assessment of imputability of adverse events The Principal Investigator (OH) and the Local Safety Monitor (LSM; an experienced consultant paediatrician) reviewed all serious adverse events and prepared a case summary, which was sent to the Safety Review Committee (SRC). The SRC comprised 3 paediatricians with extensive experience of the clinical care of children in sub-Saharan Africa and who were independent of the study. The SRC and the LSM came to a consensus decision regarding the probability that an SAE was caused by the transfusion 8 of UC-RBC and assigned it an imputability score based on an established 4point scale.19 All other (non-serious) adverse events were reviewed by a study clinician and the Principal Investigator. They were described according to an established adverse reaction nomenclature (COSTART: Coding Symbols for a Thesaurus of Adverse Reaction Terms)20 and the probability of a causative relationship with UC-RBC transfusion scored according to the same 4-point scale. A summary of these adverse events was reviewed by the LSM and SRC. Outcome measures for the secondary objective The outcome measure used to achieve the secondary objective was median change in haemoglobin concentration compared with pre-transfusion levels one day and one month after UC-RBC transfusion. A blood sample for haemoglobin estimation (Beckman Coulter, France) was taken 24 hours after the start of UC-RBC transfusion; unless a haemoglobin was requested for the clinical management of the child before this time in which case this result was used. A further blood sample for haemoglobin estimation was obtained from those children who returned for follow-up at one month. Sample size It was estimated that 100 children fulfilling the eligibility criteria for the trial would be admitted to KDH during a period of one year and that cord blood would be available and consent to transfuse given for 40-80% of these. Thus, during one year of study 40 to 80 children might be recruited to the trial. We intended to run the trial for one year and these numbers were set as a minimum and maximum sample size. The precision, as indicated by a confidence interval, of the frequency of adverse reactions (the primary outcome measure) at different event frequencies and sample sizes is shown in Table 1 of the appendix. Stopping rules The trial was to be stopped in the event of a Suspected Unexpected Serious Adverse Reaction (SUSAR), and not recommenced until a full review had 9 been undertaken by the SRC and their recommendations seen and approved by the ethics committees. In addition, in the event of an SAE the SRC advised whether they felt that the trial should continue with no change to the protocol, continue with a change to the protocol, or be stopped. Statistical methods Binary data were expressed as a percentage with 95% confidence intervals where appropriate. Where event frequencies were zero, a one-sided 97.5% confidence interval with a lower limit of zero was calculated. Continuous data were summarised by the median with range (minimum and maximum) and interquartile ranges (IQR). Observed differences in continuous data were compared for statistical significance using non-parametric statistics (Wilcoxon rank-sum). Role of the funding source OH was supported by a Wellcome Trust Training Fellowship (073604). The funder had in no role in design of the study; in the collection, analysis, and interpretation of data; in the writing of the report; and in the decision to submit the paper for publication. Results Participant flow and recruitment (Figure 1) There were 413 transfusion episodes to children over the period of the study and 87 children were considered eligible for the trial. In 24 cases, UC-RBC of sufficient haemoglobin content and/or blood group were not available and consent was declined for 6 children. Thus, 57 children were recruited to the study but 2 were withdrawn before UC-RBC transfusion. In one case, the laboratory made an error during compatibility testing and no further cord blood was available. In the second case, clinical review soon after recruitment demonstrated deep breathing (see exclusion criteria). Demographic and clinical characteristics of participants (Table 1) Fifty-five children received UC-RBC from 74 cord blood donations. Ages ranged from 2 days to 5 years and 8 months (median, 12 months) with 24 10 children aged 3 months or less. Weights of children ranged from 1.1kg to 14.5kg (median, 5.3kg). WHZ-scores ranged from -4.4 to -0.9 (median, 1.9) and 7 children had severe acute malnutrition (defined as a WHZ< -3). In those children aged 3 months or less, pre-transfusion haemoglobin ranged from 5.5g/dL to 10g/dL (median, 8.7g/dL). In those children aged greater than 3 months, pre-transfusion haemoglobin ranged from 1.9g/dL to 4.0g/d/L (median, 3.2g/dL). All children with SAM received 10ml/kg of UC-RBC; for those children without SAM, the median volume transfused was 13ml/kg (range, 10ml/kg to 20ml/kg). Numbers analysed In the event of an adverse event after a child had received UC-RBC from 2 units it is very unlikely that imputability could have been assigned to one of the two units, therefore the denominator for the primary outcome measure was the number of children transfused. Children who received subsequent conventional blood transfusions during the follow up period were included in the analysis of the primary outcome as these transfusions could themselves be evidence of harm related to UC-RBC transfusion. However, these children were not included in the analysis of haemoglobin change at 1 month as the subsequent transfusions would have confounded the effect of UC-RBC transfusion. Outcomes Of the 55 children who received UC-RBC transfusion, 10 experienced 10 serious adverse events (SAE) and 43 experienced 94 adverse events (AE) (Table 2). In no case (0/55) was UC-RBC transfusion considered probably or certainly implicated and thus the frequency of serious adverse reactions and adverse reactions was 0% (One-sided 97.5% confidence interval; 0 to 6.5%). There is, therefore, a 2.5% chance that the frequency of adverse reactions associated with UC-RBC transfusion exceeds 6.5%. The median change in haemoglobin on the day after UC-RBC transfusion (median 24 hours; IQR 17 – 24 hours) was a rise of 2.6 g/dL (IQR 2.1 – 3.1) (Table 3). In the 33 children who did not receive a further transfusion and for 11 whom a blood sample was obtained, the median change in haemoglobin at one month (median 29 days; IQR 28 – 35) was a rise of 5.0g/dL (Table 3). Four of the seven life-threatening SAE were new signs of critical illness (deep breathing and/or prostration) observed after recruitment at the pre-transfusion assessment and before the transfusion of UC-RBC. UC-RBC transfusion was therefore excluded as a potential cause of these SAE. Of the remaining 6 SAE, one was a death, 3 were judged life-threatening and 2 resulted in hospitalisation after discharge. These are described in detail below (see also Table X in the appendix): Study number: WG014 SAE severity: Fatal (Time after start of UC-RBC transfusion: approx. 7 days) Imputability level: Unlikely A 13-month old girl admitted with fever and cough and treated for pneumonia. Her admission Hb was 4.6g/dL but 5 days later this had dropped to 4g/dL and she received 100ml of UC-RBC, which raised her Hb to 6.5g/dL. Five days later, she was diagnosed with an acute lympoblastic leukaemia by microscopic examination of a peripheral blood film taken before UC-RBC transfusion but reported afterwards (21% lymphoblasts). She was referred to the local regional hospital for further management but left against medical advice and taken by bus to her mother’s marital home in Kisumu in western Kenya- a journey of approximately 850km. At discharge she had been noted to be ‘pale, sick-looking and febrile’. On follow-up in the community, the child’s grandfather reported that she died the day after arriving in Kisumu, approximately 1 week after UC-RBC transfusion. Study number: WG034 SAE severity: Life-threatening (26 hours) Imputability level: Possible A male infant with a birth weight of 1080g, delivered by emergency caesarean section for maternal ante-partum haemorrhage at approximately 28 weeks gestation. His initial Hb was 14.9g/dL and over the first 3 weeks of life he was treated for neonatal sepsis and jaundice. On day 23 of life, weighing 1120g, 12 he was noted to be pale and had an Hb of 8.8g/dL. He was transfused 20ml of UC-RBC, which increased his Hb to 15.5g/dL. Twenty-six hours after UC-RBC transfusion, he became unwell with diarrhoea, dehydration and a metabolic acidosis and was managed as probable neonatal sepsis. He made a full recovery after intervention with oxygen, broad-spectrum antibiotics and intravenous fluids. Study number: WG044 SAE severity: Life-threatening (7 days) Imputability level: Unlikely A pre-term infant boy born unexpectedly while his mother was walking to work on the family’s shamba and admitted on day 1 of life with an estimated gestation of 30 weeks and a weight of 1440g. His admission Hb was 9.4g/dL and he was transfused UC-RBC on day 2. Seven days after UC-RBC transfusion, he developed abdominal distension, respiratory distress and jaundice. This was managed as probable neonatal sepsis and he made a full recovery after intervention with broad-spectrum antibiotics, intravenous fluid therapy and phototherapy. Study number: WG049 SAE severity: Life-threatening (14 hours) Imputability level: Possible A female infant born at home and admitted on day of birth weighing 1200g, with an estimated gestation of 30 weeks and an Hb of 11.1g/dL. On day 9, weighing 1180g, she was noted to be pale, her Hb was 7.6g/dL and she was transfused 18ml UC-RBC raising her Hb to 11.0g/dL. Fourteen hours after UC-RBC transfusion she was febrile, tachypnoeic and having apnoeas. She was presumed septic and managed with broad-spectrum antibiotics, intravenous fluids, supplemental oxygen and aminophylline. A chest radiograph demonstrated right upper lobe consolidation consistent with a diagnosis of pneumonia. She made a full recovery. Study number: WG054 SAE severity: Hospitalisation (28 days) 13 Imputability level: Unlikely A 4-year old boy with known sickle cell disease admitted with respiratory distress and an Hb of 4.4g/dL. A blood transfusion was requested but he was ineligible for UC-RBC transfusion. No adult-donated blood was available until a replacement donor was found the following day, after which he was transfused 20ml/kg of whole blood resulting in a post-transfusion Hb of 8.3g/dL. Over the next 10 days he was treated for sepsis and his Hb fell to 3.6g/dL over this period. He was transfused 150ml of UC-RBC from two cord blood donations, his Hb rose to 6.2g/dL and he was discharged well with oral haematinics 3 days later. He returned mistakenly for study follow-up at 21 days after UC-RBC transfusion and at that time had an Hb of 6.6g/dL and was well. At his 28-day study follow-up assessment, he was found to have an Hb of 4.1g/dL, was admitted for a conventional blood transfusion and was discharged the following day with an Hb of 6.1g/dL. Study number: WG056 SAE severity: Hospitalisation (28 days) Imputability level: Possible A female infant born in hospital and admitted to paediatric ward at 19 hours of age with jaundice, low birth weight (2300g) and suspected prematurity, and treated for possible sepsis. At 5 days of age was noted to be pale, had an Hb of 9.7g/dL, and was transfused 25ml of UC-RBC which raised her Hb to 11.5g/dL. She was discharged well 3 days later fully breastfeeding and with a weight of 2120g. At 28-day follow-up, she was found to have an Hb of 6.2g/dL and was offered admission but her mother declined. She re-presented one week later and was admitted. Her Hb was 8.4g/dL and a blood transfusion was requested. No blood was available until a relative was found to donate, after which she received 20ml/kg whole blood and was discharged well on oral haematinics with an Hb of 14.1g/dL. Ancillary analyses In children aged less than 3 months, median change in haemoglobin at 1 month (median 29 days; IQR 28 – 36) was a rise of 0.5g/dL (IQR 0.2 – 1.2) compared with a rise of 6.1g/dL (IQR 5.3 – 8.1) in children aged greater than 14 3 months (median time to follow-up, 30 days; IQR 28 – 35) (Table 5). This difference is statistically significant (p<0.001). In those children aged greater than 3 months, the 7 children with severe acute malnutrition who received a maximum of 10mL/kg UC-RBC showed a median haemoglobin rise 1 day after transfusion of 2.1g/dL (IQR 2.0 – 2.9). In 23 children aged greater than 3 months without severe acute malnutrition who received a maximum of 15mL/kg the median rise in haemoglobin 1 day after transfusion was 2.6g/dL (IQR 2.2 – 3.1). The difference is not statistically significant (p=0.15, Wilcoxon rank sum). For the 5 children aged greater than 3 months with severe acute malnutrition for whom a result was available, the median rise in haemoglobin 1 month after UC-RBC transfusion was 8.1g/dL (IQR 7.8 – 8.2) compared to 5.9g/dL in the 15 children without severe acute malnutrition for whom there was a haemoglobin result. Discussion In this study 55 children with severe anaemia were transfused sedimented red blood cells from 74 umbilical cord blood donations. Of 10 serious adverse events and 94 adverse events, none were certainly or probably attributable to cord blood transfusion. Median increases in haemoglobin after transfusion were 2.6g/dL at 24 hours and 5.0g/dL at 29 days. These findings are consistent with the very few data concerning allogeneic cord blood transfusion that have been reported previously.9 Although we excluded children with signs of critical illness from the study at the time of recruitment, many children experienced adverse events which were unrelated to the cord blood transfusion. In four children, signs of critical illness not present at recruitment and which would have excluded them from the study were detected at the clinical assessment undertaken just prior to UC-RBC transfusion. To withdraw these children from the study at this stage and to secure and crossmatch adult-donated blood would have introduced an unacceptable delay in the management of critically ill children. This highlights the challenge of conducting studies focusing on safety and harm in hospitalised children in sub-Saharan Africa. Robust monitoring frameworks 15 are required to identify potential associations between the effects of the intervention and other confounding factors. A weakness of this study is that for those that did not attend the hospital for follow-up at 28 days there was no full clinical assessment. However, all these children were followed up in the community by a non-clinical fieldworker and the death of one child (WG014) was identified in this way. The rise in haemoglobin observed 1 day after UC-RBC transfusion seen in this study is consistent with estimations based upon the haemoglobin content of the transfused blood and the circulating volume of children: for a child with a circulating volume of 80mL/kg, the transfusion of 2.2g/kg of haemoglobin might be expected to raise the haemoglobin by 2.8g/dL. However, although cord blood units were selected for transfusion based upon an estimation of the unit haemoglobin content, it is not possible to ascertain from these data how much haemoglobin was actually issued and transfused to each child. The significant rise in haemoglobin 1 month after transfusion in children aged greater than three months is consistent with previous data from Kilifi and other sites in East Africa.3,4,21,22 Increases in haemoglobin over a similar time period are also observed in children with severe anaemia who do not receive a transfusion and survive.3,21,22 This highlights the importance of other therapies in the management of severe anaemia such as treatment of infection, antihelminthics, haematinics and diet. The relative importance of these will depend on the aetiology of the anaemia and this was not investigated here. The children aged less than 3 months in this study were likely to have a very different aetiology for their anaemia than the older children and many of them were presumed to have anaemia of prematurity. Several of these children required further transfusion and in those that did not the effect of a single UCRBC transfusion at one month was much more modest. Of note, however, is the number of young infants that were eligible for UC-RBC transfusion. This is a group of patients that carry a high burden of mortality in sub-Saharan Africa and who may potentially benefit substantially from more evidence about the role of transfusion in preventing the high death rates.1 These young children 16 might particularly benefit from the availability of cord blood for transfusion because they only require small volumes of blood for transfusion. The microbiological safety of cord blood provided by the donation programme that we have established at Coast Provincial General Hospital in Mombasa compares favourably to that of conventional blood from the same setting.15 Mothers who donate their infants’ umbilical cord blood are rigorously selected (including self-reporting of antenatal testing for syphilis and HIV), and aseptic cord blood collection undertaken by trained fieldworkers and not the midwives managing the deliveries.13,15 Furthermore, all cord blood donations in this study were screened for bacterial contamination. These rigorous techniques may be difficult to replicate outside of a research setting without considerable additional resources. Our findings suggest that further trials of umbilical cord blood transfusions are warranted but the challenges of conducting such trials and the barriers to potential scale up of such an intervention should not be underestimated. Attributing effects to the intervention is difficult in such a sick group of children. Despite this, further clinical trials should also include children with signs of critical illness who potentially have the most to gain from an improved blood supply. The infrastructure and training required to set up collection and administration of umbilical cord blood is complex and such trials would require meticulous monitoring during and after the transfusion. Poor haemovigilance systems in these settings means that very little is known about the harms associated with conventional blood transfusion which would be the comparator group in such trials.7 Other improvements and additions to the design of future trials include: better characterisation of anaemia aetiology and assessment of any correlation with benefits and harms of cord blood transfusion; immunological and genetic testing to compare rates of alloimmunisation and microchimerism; and operational analyses comparing the availability of cord blood and adultdonated blood for urgent transfusion in children, and the impact of cord blood 17 transfusion for children on the blood supply for adults requiring larger volumes of conventional blood. In this study, we have demonstrated that where demand for low volume transfusions for children is high and supplies of conventional blood are limited, umbilical cord blood may be a safe and efficacious supplementary source of blood for transfusion. Further trials comparing cord and conventional blood transfusion are merited. Conflicts of interest None of the authors have any conflicts of interest. Acknowledgement The authors would like to thank the following people without whom the study could not have been undertaken: Jay Berkley, Victor Bandika, Michael Boele van Hensbroek, Mike English, Trudie Lang, Kevin Marsh, Jennifer Othigo, Norbert Peshu, Sophie Uyoga, Tom Williams. Also many thanks to the Wazo Geni team, the midwives and mothers at Coast Provincial General Hospital, and the clinical staff at Kilifi District Hospital. We are grateful to the Wellcome Trust for funding the study. Role of the funding source The study funders had no role in study design; in the collection, analysis, and interpretation of data; in the writing of the report; and in the decision to submit the paper for publication. Research in context Systematic review A review of the literature relating to cord blood transfusion was conducted prior to this study. Allogeneic cord blood transfusion was first reported in the 1930’s before the advent of modern blood transfusion services. Subsequently, most research and clinical activity relating to cord blood transfusion has concerned autologous cord blood transfusion in pre-term neonates. A series 18 of about 200 mainly elderly patients with chronic and/or terminal disease has been transfused allogeneic cord blood in India. No adverse reactions were reported. To our knowledge, there has been no previous clinical trial of allogeneic cord blood transfusion in African children. Interpretation This trial demonstrates that there is a low probability of adverse events associated with the transfusion of sedimented red blood cells from umbilical cord blood donations (UC-RBC) to children with severe anaemia in subSaharan Africa. Haemoglobin recovery after UC-RBC transfusion is within expected limits. To our knowledge, this is the first time such a trial has been undertaken and umbilical cord blood may be a safe and efficacious supplementary source of blood for transfusion where demand for low volume transfusions for children is high and supplies of conventional blood are limited. Further work needs to be undertaken by clinical researchers to establish the safety and efficacy of cord blood transfusion compared to conventional blood transfusion. Additional research is also required on the operational aspects of cord blood collection including costs, impact on the conventional blood supply and scalability. 19 References 1. United Nations Children's Fund. Levels and Trends in Child Mortality. 2013. 2. WHO. The prevention and management of severe anaemia in children in malaria-endemic regions of Africa: A review of research. Geneva: WHO; 2001. 3. Lackritz EM, Hightower AW, Zucker JR, Ruebush TK, 2nd, Onudi CO, Steketee RW, et al. Longitudinal evaluation of severely anemic children in Kenya: the effect of transfusion on mortality and hematologic recovery. AIDS (London, England). 1997 Oct;11(12):1487-94. 4. English M, Ahmed M, Ngando C, Berkley J, Ross A. Blood transfusion for severe anaemia in children in a Kenyan hospital. Lancet. 2002 Feb 9;359(9305):494-5. 5. Lackritz EM, Campbell CC, Ruebush TK, 2nd, Hightower AW, Wakube W, Steketee RW, et al. Effect of blood transfusion on survival among children in a Kenyan hospital. Lancet. 1992 Aug 29;340(8818):524-8. 6. Allain JP, Owusu-Ofori S, Bates I. Blood transfusion in sub-Saharan Africa Transfusion Alternatives in Transfusion Medicine. 2004;6(1):16-23. 7. Tagny CT, Mbanya D, Tapko JB, Lefrere JJ. Blood safety in SubSaharan Africa: a multi-factorial problem. Transfusion. 2008 Jun;48(6):125661. 8. Tapko JB, Sam O, Diarra-Nama AJ. Status of Blood Safety in the WHO African Region: Report of the 2004 Survey WHO Regional Office for Africa; 2007. 9. Bhattacharya N. Placental umbilical cord whole blood transfusion: a safe and genuine blood substitute for patients of the under-resourced world at emergency. J Am Coll Surg. 2005 Apr;200(4):557-63. 10. Hassall O, Bedu-Addo G, Adarkwa M, Danso K, Bates I. Umbilical-cord blood for transfusion in children with severe anaemia in under-resourced countries. The Lancet. 2003;361:678-9. 11. Khodabux CM, Brand A. The use of cord blood for transfusion purposes: current status. Vox sanguinis. 2009 Nov;97(4):281-93. 12. Bates I, Chapotera GK, McKew S, van den Broek N. Maternal mortality in sub-Saharan Africa: the contribution of ineffective blood transfusion services. BJOG 2008; 115: 1331-9. 13. Hassall O, Maitland K, Fegan G, Thitiri J, Pole L, Mwakesi R, et al. The quality of stored umbilical cord and adult-donated whole blood in Mombasa, Kenya. Transfusion. 2010 Nov 13;50(3):611-6. 14. Hassall O, Ngina L, Kongo W, Othigo J, Mandaliya K, Maitland K, et al. The acceptability to women in Mombasa, Kenya, of the donation and transfusion of umbilical cord blood for severe anaemia in young children. Vox sanguinis. 2008 Feb;94(2):125-31. 15. Hassall OW, Thitiri J, Fegan G, Pole L, Mwarumba S, Denje D, et al. The microbiologic safety of umbilical cord blood transfusion for children with severe anemia in Mombasa, Kenya. Transfusion. 2012 Jul;52(7):1542-51. 16. WHO. Pocket book of hospital care for children: guidelines for the management of common illnesses with limited resources. Geneva: WHO; 2005. 20 17. Hassall O, Maitland K, Pole L, Mwarumba S, Denje D, Wambua K, et al. Bacterial contamination of pediatric whole blood transfusions in a Kenyan hospital. Transfusion. 2009 Aug 4;49(12):2594-8. 18. EMEA. ICH Guideline for Good Clinical Practice. London: The European Agency for the Evaluation of Medicinal Products; 2002. 19. MHRA. UK Blood Safety and Quality Regulations. Implementation of the EU Blood Safety Directive. Background and Guidance on reporting Serious Adverse Events & Serious Adverse Reactions. London: MHRA. 20. FDA. COSTART: Coding Symbols for Thesaurus of Adverse Reaction Terms. Rockville, MD: Public Health Service, FDA; 1995. 21. Akech SO, Hassall O, Pamba A, Idro R, Williams TN, Newton CR, et al. Survival and haematological recovery of children with severe malaria transfused in accordance to WHO guidelines in Kilifi, Kenya. Malaria journal. 2008;7:256. 22. Holzer BR, Egger M, Teuscher T, Koch S, Mboya DM, Davey Smith G. Childhood anemia in Africa: to transfuse or not transfuse? Acta Trop. 1993;55:47-51. 21 Figure 1 Trial participant flow Transfusions to children 413 Age <= 3m 105 Hb > 10g/dL 12 Age > 3m 308 Hb <= 10g/dL 93 Hb <= 4g/dL 114 Hb > 4g/dL 294 Children eligible by Hb 207 Children ineligible by clinical criteria 120 Children eligible by clinical criteria 87 Cord blood unavailable Declined consent 24 6 Children enrolled 57 Withdrawn before transfusion 2 Children transfused 55 Outpatient follow-up Community follow-up 44 11 22 Table 1 Selected admission characteristics of trial participants (SS= homozygous for sickle cell genotype; MPs= malaria parasites; WHZ= weightfor-height Z-score; Hb= pre-transfusion haemoglobin) Aged <= 3 Aged > 3 months months N 24 31 55 Male 12 16 28 SS - 6 6 MPs 0 6 6 WHZ< -3 - 7 7 Pre-term 14 - 14 Weight (kg)* 1.6 8.6 8.7 3.2 (5.5 to 10.0) (1.9 to 4.0) Hb (g/dL)* All 5.3 (1.1 to 14.5) - * Median (range) 23 Table 2 Serious adverse events and adverse events experienced by children receiving umbilical cord red cell transfusion. Serious Adverse Events Adverse Events 10 10 94 43 Timing Before transfusion Transfusion + 24h 24h to 28d follow-up 4 1 5 4 12 78 Indicator of severity Fatal Life-threatening Hospitalisation 1 7 2 N/A N/A N/A Imputability level* Not assessable Excluded (0) Unlikely (0) Possible (1) Likely/Probable (2) Certain (3) 0 4 3 3 0 0 0 15 76 3 0 0 N No. of children * Numbers in parentheses refer to 4-point imputability score described in the text 24 Table 4 Description of Serious Adverse Events occurring after umbilical cord red cell transfusion ID No. Age* Sex Description (time after CBT start) Indicator of severity Comments Imputability level (score) 14 1y 1m F Diagnosis of an acute leukaemia- probably acute lymphoblastic leukaemia (N/A) Fatal Unlikely (0) 44 2d M Abdominal distension, respiratory distress, jaundice (7d) Life-threatening 54 4y M Anaemia (28d) Hospitalisation 34 49 24d 9d M F Diarrhoea, dehydration and metabolic acidosis (26h) Life-threatening Sepsis, peumonia, apnoeas (14h) Life-threatening 56 5d F Anaemia (28d) Hospitalisation Diagnosis by microscopic examination of a peripheral blood film; film taken before UC-RBC transfusion but reported afterwards. Child was referred to regional hospital but discharged against medical advice and taken by bus to mother’s marital home in western Kenya (Kisumu). Child was noted to be ‘pale, sicklooking and febrile’ at discharge Child reported by grandfather to have died the day after arriving in Kisumu, approximately 1 week after UC-RBC transfusion. Pre-term infant (estimated gestation, 30 weeks; birth weight, 1440g) Probable neonatal sepsis Full recovery after intervention with broad-spectrum antibiotics, nil by mouth, intravenous fluid and phototherapy. Sickle cell disease Discharged well 3 days after UCRBC transfusion (Hb 6.2) with haematinics Returned early (21d) for follow-up (mistakenly) (Hb 6.6) Hb 4.1 at 28d follow-up; admitted for conventional blood transfusion; discharged next day (Hb 6.1). Pre-term infant (estimated gestation, 28 weeks; birth weight, 1080g) Probable neonatal sepsis Full recovery after intervention with oxygen, broad-spectrum antibiotics and intravenous fluid. Pre-term infant (estimated gestation, 30 weeks; birth weight, 1200g) Radiologically confirmed pneumonia Full recovery after intervention with broad-spectrum antibiotics, oxygen, aminophylline and blood transfusion. Low birth weight (probable prematurity), neonatal jaundice and possible sepsis Discharged well 3 days after cord blood transfusion (Hb 11.5) Hb 6.2 at 28d follow-up; offered admission, declined but represented 1 week later; admitted; Hb 8.4 and transfused conventional blood once donor found (initially no blood available) Discharged next day with haematinics * At enrolment to the study 25 Unlikely (0) Unlikely (0) Possible (1) Possible (1) Possible (1) Table 5 Haemoglobin concentrations (median, IQR) of children receiving UCRBC stratified by age N Aged <= 3 Hb months Hb change N Aged > 3 months Hb Hb change N Pretransfusion 1 day 1 month 24 24 13 8.7 (7.8 – 9.2) 11.4 (10.5 – 12.4) +2.7 (2.2 – 3.8) 9.2 (8.4 – 9.9) +0.5 (0.2 – 1.2) 31 31 20 3.2 (2.7 – 3.8) 5.8 (5.3 – 6.2) +2.6 (2.1 – 3.1) 9.3 (8.3 – 11.0) +6.1 (5.3 – 8.2) 55 54* 33§ - +2.6 (2.1 – 3.1) +5.0 (1.0 – 6.8) - - All Hb change * Excludes 1 child for whom consent was declined for a blood test § Excludes 22 children: 10 received further transfusions, 11 were followed up in the community, 1 was not bled in error. 26