Design and treatment protocol
We conducted a multicenter open randomized Phase II trial, with the aim of providing preliminary results regarding the safety of higher volume transfusions and their feasibility and acceptability to clinicians and transfusion services. Eligible children from two clinical centers in Eastern Uganda (Mbale and Soroti Regional Referral Hospitals) were randomly assigned on admission to hospital (ratio 1:1) to receive either: (1) 20 ml/kg whole blood transfusion (Tx20) (alternatively 10 ml/kg of packed red blood cells (standard of care)); or (2) 30 ml/kg whole blood transfusion (Tx30) (or 15 ml/kg packed red blood cells). Sites were instructed to transfuse blood over three to four hours, as recommended by WHO.
Study population
Screening procedure
We aimed to enroll 160 children, >60-days- and <12-years-old, with severe anemia at admission to the pediatric ward. Dedicated trial clinicians and nurses were employed to conduct the study. Potentially eligible children with clinical evidence of pallor were identified and registered in the eligibility screening log. A rapid bedside test by HemOcue (Ängelholm, Sweden) and bedside examination determined hemoglobin level and severity. Children were eligible if they had severe anemia (hemoglobin <6 g/dl) at the time of admission to hospital), no previous transfusion during the course of the current illness and a guardian or parent willing/able to provide consent (Figure 1: Trial Flow). Complicated severe anemia was defined as children with hemoglobin 4 to 6 g/dl in conjunction with markers of clinical severity (reduced conscious level or respiratory distress) or profound anemia (hemoglobin <4 g/dl). Children with malignancy, surgery, acute trauma or acute severe malnutrition were excluded from the study.
Outcome measures
The primary outcome was correction of severe anemia (to hemoglobin >6 g/dl) at 24 hours. Secondary outcomes included: (1) meeting criteria for additional transfusion (development of profound anemia (hemoglobin <4 g/dl) or hemoglobin 4 to 6 g/dl with new markers of severity (impaired consciousness or respiratory distress)) from eight hours post randomization; (2) serious adverse events defined according to Good Clinical Practice [16] and including suspected pulmonary edema (bilateral basal crepitations with hypoxemia (oxygen saturations <90%); biventricular heart failure (severe tachycardia (<12-months-old: >180 beats per minute (bpm); 12-months to 5-years-old: >160 bpm; >5-years-old: >140 bpm) plus an increasing liver size) or suspected transfusion reaction; (3) mortality through 48 hours and 28 days post-admission; and (4) redevelopment of severe anemia (hemoglobin <6 g/dl) or mortality at 28 days post-admission.
Study procedures
Consent
Prospective written, informed consent was obtained from parents or guardians of the children. The information sheet was in their usual language and was read aloud to those unable to read. Parents and guardians were encouraged to ask questions about the trial prior to signing the consent form. In cases where prior written consent from parents or guardians could not be obtained because of severity of the illness a provision was approved for verbal assent from a legal surrogate followed by delayed informed consent as soon as practicable.
The Mbale Research Ethics Committee, Mbale, Uganda approved the protocol. The trial was registered, prior to enrollment, with ClinicalTrials.Gov identifer: NCT01461590 (registered 26 October 2011).
Randomization
Randomization was stratified by clinical center. The treatment allocation (Tx30 or Tx20) was kept in numbered, sealed opaque envelopes, each signed across the seal. The cards were numbered consecutively and opened in numerical order. The randomization list and envelopes were prepared before the trial by a statistician at the Kilifi Clinical Trials Facility and the list was not available to the investigators.
Sample size
The study aimed to generate pilot safety and efficacy data on a higher transfusion volume (30 ml/kg) in children with severe anemia. Numbers required to address the trial objectives were therefore balanced against exposing children to a therapeutic intervention (dose) for which there are limited data. The overall sample size of 160 children (approximately 80 having signs of severity as defined above) randomized to 20 versus 30 ml/kg provided at least 80% power to detect major (20% to 25%) increases in the proportions experiencing the primary and secondary outcomes.
Clinical monitoring and study assessments
Following consent and randomization, lactate (LactatePro®), glucose, malaria status (by blood film and Optimal® rapid diagnostic test (RDT)) and cross match were performed. Following national guidelines, HIV testing was conducted after completion of admission procedures, with pre- and post-test counseling done in accordance with routine practice. Blood was collected at admission into ethylenediaminetetraacetic acid (EDTA), stored at -80°C and typed by PCR for the hemoglobinopathies sickle cell anemia (HbSS), sickle cell trait (HbAS) and the common African variant of α-thalassemia and for the red cell enzymopathy G6PD deficiency at the end of the study, as described in detail previously [17], using DNA extracted using Qiagen DNA blood mini kits (Qiagen, Crawley, UK).
Transfusions were given in standard infusion sets incorporating a graded, filtered burette. Since the volume of the burette was only 150 ml, each transfusion, depending on volume, was given as consecutive aliquots (150 ml maximum/aliquot) until the transfusion was complete. For analysis, a separate transfusion was defined by a gap of 30 minutes or more between consecutive aliquots.
All children were reassessed at 30 minutes, 1 hour, 90 minutes, and 2, 4, 8, 16, 24 and 48 hours for consciousness level, vital signs (heart rate, oxygen saturation, respiratory rate, axillary temperature, blood pressure) and for adverse events. Hemoglobin was monitored 8-hourly on the day of admission and daily thereafter. Glucose and lactate were reassessed at 8, 16 and 24 hours, with lactate assessed again at 48 hours. At follow up (day 28) hemoglobin and malaria parasite status were reassessed.
Serious adverse event reports were sent to the Clinical Trials Facility, Kilifi, Kenya within two days and were also monitored against source documents by visiting monitors. An independent clinician removed all references to the randomized arm prior to review by the Endpoint Review Committee (ERC), which included an independent chair (JE), one independent clinician (IB), one clinician involved in trial management but not patient enrollment (KM) and one clinician not involved in the day-to-day running of the trial (DMG). The ERC had access to clinical narratives, bedside vital observations, serial laboratory and bedside blood tests and concomitant treatments. They adjudicated (blind to randomized arm) on whether fatal and non-fatal events could be related to transfusion or the volume transfused, and the main cause of death.
Further management
An additional transfusion was permitted after eight hours (at the time of the first protocol hemoglobin reassessment) for children who still had either (1) hemoglobin <4 g/dl or (2) hemoglobin 4 to 6 g/dl and a sign of severity (respiratory distress or impaired consciousness). If a child required maintenance fluids, 5% dextrose was given at 3 to 4 ml/kg per hour until the child was able to drink. All children received standard treatments recommended by national guidelines, depending on their illness, including parenteral anti-malarials, antibiotics and/or antipyretics, anticonvulsants, oxygen (for oxygen saturations <90%) and glucose for hypoglycemia. Use of diuretics during blood transfusion was discouraged and reserved for children developing new signs suggestive of pulmonary edema or biventricular heart failure (defined as respiratory distress plus oxygen saturation <90%, bilateral basal crepitations, severe tachycardia and increasing liver size following transfusion).
Statistical analysis
All analyses followed intention-to-treat and all statistical tests were two-sided. For the primary endpoint (correction of severe anemia at 24 hours), the arms were compared using Cox proportional hazards regression for the cause-specific hazard of a hemoglobin >6 g/dl before death, and the relative difference estimated by cause-specific hazard ratios. The cumulative incidence of severe anemia before death was also estimated; comparison of the sub-distribution hazard corresponding to the cumulative incidence between arms [18] gave similar results to the cause-specific hazards (not shown). Secondary outcomes were compared between arms using risk ratios and Fisher’s exact test. Other continuous characteristics (for example, volume and rates of the initial transfusion) were compared between arms using two sample t-tests assuming equal variance in the arms; categorical characteristics (for example, number of transfusions per child) were compared between arms using Fisher’s exact test. Change in vital signs, hemoglobin, glucose and lactate from baseline were compared between the arms using two sample t-tests at scheduled assessments assuming equal variance in the arms, and across all time points (global tests of difference) using generalized estimating equations (normal distribution, independent correlation structure).