Bulletin of the World Health Organization

Blood sample volumes in child health research: review of safe limits

Stephen RC Howie a

a. Bacterial Diseases Programme, Medical Research Council (UK) Laboratories, PO Box 273, Banjul, Gambia.

Correspondence to Stephen RC Howie (e-mail: showie@mrc.gm).

(Submitted: 28 May 2010 – Revised version received: 03 September 2010 – Accepted: 05 September 2010 – Published online: 10 September 2010.)

Bulletin of the World Health Organization 2011;89:46-53. doi: 10.2471/BLT.10.080010


Safeguarding the interests of research participants is an ethical imperative that is conveyed in all accepted research guidelines and standards, and most prominently in the Declaration of Helsinki.1 Because children are vulnerable, we have a special obligation to ensure their safety when they become research subjects. Blood sampling in children who participate in medical research is an issue of concern to parents, researchers and research ethics committees alike.

The Gambia Government–Medical Research Council Joint Ethics Committee, a long-established entity2 of which the author is a member, frequently discusses blood sample volumes when considering research projects for approval. The committee decided to reassess its policy on safe limits for paediatric blood samples and asked the author to review the subject to assist its deliberations.

The concept of “minimal risk” is useful for the protection of children who take part in clinical research and was the guiding principle for the review. A full discussion of minimal risk, which figures prominently in research regulation in the United States of America,3 is beyond the scope of this paper, but it is defined, in essence, as a low risk outweighed or balanced by potential benefits to the individual, society or both. Regulation in the United States, as elsewhere, recognizes that in certain situations consideration should be given to research that carries a greater than minimal risk to the participant, but stronger justification would be required in terms of the resulting benefit and far more stringent case-by-case assessment by the responsible ethics committee or institutional review board.

The objective of this review is to determine paediatric blood sample volume limits that are consistent with physiological “minimal risk.” The review does not address other considerations that might influence local policy in this area, such as volume acceptability (to parents and society, for instance) or the most appropriate means of collecting blood samples.


Data for this review were identified by searching Medline, EMBASE, the Cochrane Library,4 Clinical Evidence,5 BloodMed,6 the web sites of the American Academy of Paediatrics7 and the Royal College of Paediatrics and Child Health,8 Google,9 paediatric texts (Nelson,10 Rudolph11) and the reference lists of relevant articles for all the literature published up to September 2009 on safe blood draw volume limits in persons less than 18 years of age (the United Nations definition of a child). The Medline search was undertaken in two parts: (i) To identify the literature on the adverse effects of blood sampling, the following search strategy and terms were used: (blood specimen collection) AND (infant OR child, preschool OR child OR adolescent OR paediatrics) AND (anaemia OR adverse effect); (ii) to identify the literature on recommendations for blood sampling in clinical research, the following search strategy and terms were used: (blood specimen collection) AND (infant OR child, preschool OR child OR adolescent OR paediatrics) AND (guidelines OR practice guidelines). A similar strategy was followed for the EMBASE search using equivalent EMTREE terms. No language restrictions were placed on the Medline and EMBASE searches. The titles and abstracts of the references returned by these searches were screened and the full manuscripts of all references relevant to the review question were obtained.

The review was originally undertaken in October 2006 and updated in August and September 2009. In addition, on 17 October 2006 a draft of the review report was sent to 32 health professionals with relevant expertise, both within and outside the Medical Research Council’s Gambian unit. Feedback was received from 14 of them and the report was revised accordingly.

Ethics approval: approval to undertake and publish this review was granted by the Gambia Government–Medical Research Council Joint Ethics Committee (L2009.E10).


The references identified through the literature search are shown in Table 1. No review of the evidence surrounding the question of safe blood sampling volume limits was identified in Medline, the Cochrane Library or any other source. Several relevant sources addressing the underlying physiology and pathology of the effects of blood sampling were identified, together with several guidelines that are summarized in Table 2. The data collected are presented in sections on basic physiology and pathology, the empirical evidence available and existing guidelines. Key recommendations are provided in Box 1.

Box 1. Key recommendations regarding blood sampling in children involved in clinical research

  • Existing guidelines for blood sample volume limits (ranging from 1–5% of total blood volume within 24 hours and up to 10% of total blood volume over 8 weeks) are consistent with the limited evidence available on “minimal risk” to children.
  • Lower limits for sick children are advisable, and a maximum of 3 ml/kg post-neonatally within 24 hours (3.8% of total blood volume), in line with United States policy, seems to be a reasonable guideline, although each study must be judged on its own merits and greater caution may be needed in children with illnesses that impair the replenishment of blood volume or haemoglobin.
  • The blood volumes needed for clinical care and research should be carefully assessed and a clear justification should be provided to institutional review boards and ethics committees, which they can then assess and monitor.
  • Effective coordination of clinical and research activities is needed to minimize the burden to the child.
  • Collection procedures should minimize discomfort for the child as much as practically possible.
  • Blood collection volumes and procedures should be acceptable to the community in which the research is conducted.
  • More research on the impact of blood sampling is needed to better assess appropriate blood volume limits in particular patient groups.

Basic physiology and pathology

Total blood volume

A person’s total blood volume (TBV) is related to body weight. The TBV of a child is around 75–80 ml/kg and is higher in the neonatal period (from 85 ml/kg it rises to a peak of 105 ml/kg by the end of the first month and then drops progressively over ensuing months).18 Thus, the TBV of a 3.5-kg 2-week-old will be about 350 ml while that of a 10-kg 15-month-old will be about 800 ml.

Impact of blood sampling on the circulation

A large acute loss of blood volume may compromise the circulation, though the blood volumes required for this to occur are generally much higher than those taken in child health research. According to the British Committee for Standards in Haematology Transfusion Task Force Guidelines for Transfusion,19 in children having blood collected for possible future autologous transfusion it is safe to draw up to 12% of TBV (around 10 ml/kg) at each donation, provided the child is in a stable condition and has a normal blood haemoglobin (Hb) level.

Replenishing lost red cells and Hb

Red cells are usually replenished fairly rapidly, since the marrow can acutely double or triple its usual red cell replacement rate of 1% per day if necessary.30,31 In patients with iron deficiency anaemia, blood Hb rises by up to 0.2–0.3 g/dl per day in adults on iron supplements31,32 and possibly higher in children who have received iron supplements for a few days.21 In a study of girls with central precocious puberty, Hb levels had recovered after 3 months of iron supplementation following a maximum draw of 12% of TBV, though how quickly recovery was achieved within those 3 months is not known.14 Slower red cell replenishment is seen in patients with aplastic anaemia, renal failure and a variety of chronic conditions in which erythropoiesis is inhibited.33

Empirical evidence

Impact of blood sampling on Hb in children

Two studies that attempted to measure the impact of blood sampling for research in children were identified. The first was a retrospective review of 11 children (aged 1–19 years) undergoing chemotherapy.15 In a single chemotherapy cycle (out of a total of six cycles given to each patient), between 1% and 15% of TBV was drawn for research purposes over a 3-day period, in addition to the blood samples needed for clinical care. The differences between pre-treatment Hb level and the lowest Hb level after treatment during “research” and “non-research” cycles were compared. The data were inconclusive. Overall Hb levels dropped by an additional 1.5 g/dl on average when research samples were taken, but unexpectedly no relationship was found between the change in Hb in an individual patient and the volume of blood drawn (as a percentage of TBV). The data showed that among children who lost around 5% or less of TBV over 3 days no meaningful change in Hb occurred (an average 0.1 g/dl increase, n = 9 children), while among those who lost well over 5% of TBV there was an appreciable drop in Hb (a drop of 2.75 g/dl on average, n = 2 children). These data need to be interpreted cautiously as the numbers were small and the study was open to confounders, such as differences in blood transfusion practice among physicians.

The second study prospectively followed 34 girls aged 1.6 to 9.2 years who underwent repeated blood sampling as part of an intervention study for central precocious puberty in the United States.14 Participants had 6–10 ml/kg of blood (7.5–12.5% of TBV) drawn over a 24-hour period every 3 to 6 months for up to 3 years. After each blood draw patients were given an equivalent volume of normal saline intravenously plus 1–6 mg/kg daily of elemental iron for 6 weeks. At first sampling, the average acute drop in blood Hb was 1.2 g/dl, and the maximum drop was 2.3 g/dl. Subsequent pre-sampling Hb levels, measured every 3 months, averaged 12.5 g/dl and did not differ significantly from baseline levels. In the author’s view, the need for volume replacement and iron therapy along with the appreciable drop in Hb the participants experienced, albeit temporarily, places this volume of sampling in a category of greater than minimal risk. Nevertheless, the study provides direct evidence that supports the British Committee for Standards in Haematology Transfusion Task Force Guidelines for Transfusion policy in terms of the amount of blood that can be safely collected for autologous blood transfusion (10 ml/kg, or around 12.5% of TBV) and that is useful for ethics committees or institutional review boards attempting to quantify the risks and benefits associated with sampling in the “greater than minimal risk” category.

Neonates are one group in which the impact of blood sampling on Hb levels is particularly concerning. Three studies in neonates that were indirectly relevant to the question posed in the review were identified. The first was a study from a developed country in which 99 very premature neonates (gestational age 24–32 weeks) had an average of 13.6 ml/kg of blood (equivalent to around 13% of TBV) drawn for testing during the first month after birth, and 19% of the 99 neonates received at least one blood transfusion.12 This study did not quantify the impact of blood sampling on Hb levels and did not attempt to separate the impact of blood loss through sampling from the effect of neonatal illness, but the findings do suggest that this level of sampling was often associated with the need for transfusion, which takes it out of the minimal risk category. This conclusion is consistent with the findings of a similar study of 253 premature infants (gestational age 25–32 weeks) that showed a strong correlation between blood sample volume and the need for transfusion.16 A median of 18 ml/kg (18% of TBV) of blood was taken and 53% of the infants required transfusion during their neonatal intensive care admission (median admission duration: 12 days; range: 0–135 days). A third study followed 50 very-low-birth-weight infants whose Hb was measured on admission and on day 8, along with the volume of blood taken for sampling.13 The average blood volume collected was 4.5% of TBV and the average drop in Hb was 3.4 g/dl (from a mean of 19.8 g/dl to 16.4 g/dl). A proportion of this drop would be expected to result from the physiological fall in Hb that is expected in the neonatal period, but the contribution of this was not assessed. However, multiple regression analysis (which included severity of illness and blood volume in the regression model, but not birth weight or gestational age) indicated that illness severity had a greater impact on the drop in Hb than blood sampling losses. The fact that blood sampling can exacerbate the effect of illness on Hb levels should be taken into account when determining appropriate blood sample volumes in sick children.

Existing guidelines

Indirect guidance from transfusion guidelines

The British Committee for Standards in Haematology endorses the following formula for calculating the volume of packed red cells that should be given to a patient for a top-up transfusion:19

Volume to transfuse = (desired Hb – actual Hb) × weight × 3

where Hb is measured in g/dl and weight in kg. So for example, to raise the Hb from 6.5 g/dl to 7.5 g/dl in an anaemic 10-kg 18-month-old one would need to give 30 ml (3 ml/kg, or 4% of TBV) of packed red cells. Packed red cells have a Hb level around twice as high as a normal child’s blood (haematocrit 50–80% versus around 33–40%)20 so, conversely, the taking of 30 ml as a sample from a normal child would be presumed to drop the Hb by half as much as the giving of 30 ml of packed red cells would raise it (and the drop may be less than half as much if the child were anaemic given that the absolute loss of Hb would be less). Therefore a sample of 3 ml/kg (4% of TBV) taken on one occasion in the same 10-kg 18-month-old with a Hb of 7.5 g/dl would be expected to reduce the Hb to no less than 7.0 g/dl. This loss would normally be replenished within days.

Existing policies on blood draw volumes in paediatric research

Nine policies or guidelines directly relevant to this report were identified and their contents are summarized in Table 2, along with the Gambian Government–Medical Research Council Joint Ethics Committee policy.17,2229 Overall the policies and guidelines identified specify blood volume limits between 1% and 5% of TBV on a single draw (or over 24 hours) and of up to 10% of TBV over 8 weeks.


This review has shown that while several guidelines are available for reference, there is a limited amount of direct evidence on which to base them. It is apparent that safety guidelines should be based on the size of the child instead of on his or her age. Beyond this, the evidence concerning the impact of blood sampling on a child’s Hb or general well-being comes from specific contexts from which general guidelines should be cautiously drawn. Nevertheless, on the whole the available data are consistent with the conclusion that all guidelines for blood draw volumes identified in this review are within the limits of “minimal risk” to study participants. Notably, no reports of adverse events associated with the use of any of these guidelines were identified in the course of this review, although no data are available to indicate how well such guidelines are being followed. In any case, better evidence, based on larger numbers of study subjects, a wider range of study contexts and more robust study designs (randomized trials ideally), is clearly needed before firmer conclusions can be reached. Good studies conducted in important participant groups from which no data are currently available, such as children in resource-poor countries, would be particularly useful.

While the literature search for this review was comprehensive, some relevant data may not have been identified, particularly in the grey literature. Such a limitation is potentially more important in an area such as this, which has a very small body of evidence.

Ethics committees and institutional review boards should consider several factors when deciding how to apply general volume guidelines for blood sampling. A lower blood volume limit for sick children, i.e. children with either acute or chronic disease, is advisable, and a maximum of 3 ml/kg within 24 hours (3.8% of TBV) post-neonatally is in line with United States policy applying to instances where expedited institutional review board chair’s approval for a research project could be given.17 However, a more cautious approach will be needed in the presence of particular illnesses, such as those associated with a low blood Hb level or blood volume depletion. Renal failure, bone marrow dysfunction, a variety of chronic illnesses and certain treatments can inhibit erythropoiesis and shorten red cell survival, thereby reducing the body’s ability to replenish the blood taken through sampling. Chemotherapy is an example of such a treatment, which makes the data supplied by Cole and colleagues particularly relevant.15 Illnesses associated with depletion of blood volume and Hb, especially when this is rapid (such as severe malaria or disseminated intravascular coagulation from severe sepsis), also call for greater caution. Baseline anaemia itself requires caution, particularly if the Hb is very low or is associated with haemodynamic compromise. In settings where environmental factors such as endemic malaria or widespread iron deficiency tend to lower average Hb levels and may slow the recovery of Hb levels after sampling, these factors should also be taken into account.

The limits discussed here are total blood sample volumes that include what is required for both clinical care and for research purposes exclusively. It is very important that an ethics committee or institutional review board have a full picture of the sampling demands that will be placed on a child in a particular study, including those for routine clinical care, to best determine what limits should be placed on volumes.

To put these limits in perspective, most paediatric research that calls for blood sampling would not require blood volumes anywhere near the higher limits seen in the guidelines reviewed. Nonetheless, it is important to decide where to set the limits for minimal risk. Ethics committees and institutional review boards may consider approving studies involving larger volumes of blood than those associated with minimal risk, but only after fully weighing the risks and benefits and the well-being of the child participant. Pharmacokinetic research is an area with comparatively large sampling requirements and a dearth of paediatric data, partly because of the large sampling requirements. As the regulatory environment pushes sampling requirements further, sound paediatric data may become even scarcer, placing an onus on institutional review boards and regulators to walk a fine line between over-caution and under-caution.34,35

One issue not addressed in this review but that should be addressed in local policies is the means by which blood is obtained, for instance, the acceptable number of needle pricks. Policy applied in this area may, in many studies, have a greater impact on a child’s overall well-being than the question of blood volume, to the point of restricting the blood volume obtainable to far less than is allowed by physically safe limits. Blood drawing by experienced phlebotomy staff and the use of local anaesthetic creams can make the experience less traumatic for the child. It is very important that the blood required for clinical care and for research be collected at the same time whenever possible to minimize discomfort for the child, and this requires effective coordination. In some settings communication between health-care and research staff may not be good and research samples may be taken without regard for the sample volumes already obtained for clinical care. It is up to researchers to show ethics committees and institutional review boards that systems are in place to ensure that the procedures followed and the combined blood volumes obtained are within the limits established by agreed guidelines. Having been satisfied on this point at the approval stage, ethics committees and institutional review boards can assess compliance by requiring that blood sampling be included in monitoring and progress reports on the study.

Most policies identified by this review set blood drawing limits well below the boundaries of what is likely to represent “minimal risk,” and relaxing those limits might enhance the potential benefits of research for children overall. However, while good health policy takes account of physiology, it must also take account of acceptability within the population it serves. Data regarding the limits of physical safety can make an important contribution to locally-formulated policy but are not sufficient in isolation to determine that policy. The outcome of the policy reappraisal that prompted this review is an example of this. After the review, the Gambian ethics committee felt reassured that its policy on blood drawing was well within physically safe limits, based on the available evidence, but did not change its guideline except to allow requests for greater volumes to be judged on a case-by-case basis. A conservative guideline was deemed appropriate in a setting where much of the population does not consider blood a “renewable resource.”27 Responsible bodies should determine how to apply the available safety information in their local context in a manner that both protects the immediate well-being of children who participate in clinical research and also promotes the future well-being of all children.


I thank the Gambia Government–Medical Research Council Joint Ethics Committee for initiating this review and for appraising the report on which this paper is based. I am also grateful to others who reviewed the report or who provided additional information: Climent Casals, Bouke DeJong, Godwin Enwere, Brian Greenwood, Philip Hill, Ross Howie, Paul Knight, Kathryn Maitland, Jenny Mueller, Kim Mulholland, Onike Rodrigues, Martin Ota, Michael Walther, Hilton Whittle, and the KEMRI–Wellcome Trust Research Programme in Kilifi, Kenya. The responsibility for this review and its conclusions, including any errors, are the author’s.


The UK Medical Research Council funded this study.

Competing interests:

None declared.