Article Text
Statistics from Altmetric.com
Treatment of very preterm babies in order to try and increase their blood pressure is a widespread, and extremely variable, practice. Some centres try to maintain a mean arterial blood pressure above 30 mm Hg, others try to achieve a blood pressure above the gestational age in weeks, and others tend to be permissive, only treating when the infant shows signs of poor perfusion.1
A systematic review that we performed a few years ago2 could find no reliable evidence comparing these approaches, nor any clear evidence that ‘hypotension’ is causative for short-term or long-term morbidities, nor could we find a clear threshold for intervention, nor any evidence that intervention improves outcomes.
Evidence which has accumulated over the last few years has addressed, but has not clearly answered these questions. Evidence from a recently published large German database study,3 showed an association, among infants of less than 1500 g birth weight and <32 weeks gestation, between a mean arterial blood pressure which was below the median at some point during the first 24 h, and an increase in mortality, and also in intraventricular haemorrhage (IVH) and bronchopulmonary dysplasia (BPD). Treatment with inotropes was strongly associated with an increased incidence of IVH and BPD. This and other published data are open to an interpretation that therapeutic interventions for low blood pressure could be a causative factor in the increased morbidity. However, in that new study the association between lower blood pressure and poorer outcomes was also seen in infants who did not receive treatment.
A study from Batton et al4 also provides new information. The authors have previously demonstrated marked variation in hypotension treatment in their cohort, and a surprising lack of efficacy of antihypotensive therapy,5 ,6 blood pressure increasing just as quickly among untreated infants as among those who received antihypotensive agents. It was already clear that fluid boluses, even though they are almost universally the first intervention for ‘hypotensive’ preterm infants, are largely ineffective, this group's data suggest that dopamine also may have limited efficacy.
In this new paper, the authors examined survival without neurological injury or developmental delay at 18–22 months in their cohort of infants less than 27 weeks gestation. They divided infants into 4 groups: those who received or did not receive antihypotensive therapy, and within each of those groups, those who had a slower than average rise in their blood pressure, and those with a faster increase. The primary outcome was not associated with a slower rise in blood pressure among untreated infants, but was strongly associated with hypotension treatment. Babies who received antihypotensive treatment, those with a good, and those with a poor response, were more likely to die, to have motor dysfunction, or to have developmental delay (in cognitive and language domains) than untreated infants. A potential mechanistic explanation is that the treated patients had a much higher incidence of brain injury on ultrasound (severe IVH or periventricular leukomalacia). These associations remained after adjusting for severity of illness measures.
It is obvious that 10% of very preterm infants at 3 h of age will have a blood pressure below the 10th centile for 3 hour-old very preterm infants. In contrast, approximately half of very preterm infants will have a blood pressure below their gestational age at some time during the first 24–72 h of age. Blood pressure changes dynamically, (and tends to progressively increase during the first 72 h of life), so many infants will dip below the postnatal age defined 10th centile at some point during the first 72 h. This is one of the differences between the data presented by Batton, and the previously mentioned data from the German Neonatal Network;3 the German Neonatal Network data compares the short-term outcomes in infants who had a single blood pressure (BP) less than the median low BP for their gestational age. Batton et al4 in contrast looked at dynamic changes in BP during the immediate postnatal period. The two approaches are valid but teach us different things about hypotension and its treatment in the newborn infant.
In the light of these new data we can pose again the following questions: Should we treat hypotension in the extremely preterm infant? At what BP should we intervene? What other clinical signs or methods of evaluating the circulation should be integrated into the treatment decision? Should measurement of blood flows or tissue oxygenation be routine in such infants? What should we treat with?
These questions are critically important and the answers remain deeply unsatisfying, the only scientifically valid answer to all of those questions is: ‘we do not know’.
There are many situations in neonatology where we have several unanswered but linked questions, where the answer to each question affects how to address the other. For example ‘should we treat hyperglycaemia with insulin’, ‘should we also reduce glucose intake’ and ‘at what level of hyperglycaemia should we intervene?’. The answer to each of these questions affects how we should address the others. This complex situation suggests two possible responses: we throw our hands in the air and stay frustrated about the complexities of life, or we arbitrarily choose a threshold or an intervention to test for one question so that eventually we will be able to answer the others.
For hypotension treatment I think the most reasonable approach is to test a currently recommended ‘best practice’ approach, such as treating at a BP less than the gestational age with a fluid bolus followed by dopamine, and compare that to a permissive approach of treating only when other signs of poor perfusion are found.7 At the same time, defining the different haemodynamic situations underlying hypotension, such as perinatal inflammation, pulmonary hypertension or primary cardiac dysfunction, will need to be integrated into such a trial, to better inform clinicians of responses to therapy.
We must, as a community, realise that the only way to answer many of the unanswered questions in neonatology is to put our prejudices aside and objectively evaluate the quality of the available evidence. For hypotension treatment, there are many who will say that they do not have ‘equipoise’; those who are convinced that not treating, in the face of poor autoregulation, risks cerebral ischaemia, and others who are convinced that treating the numbers is irrational and risks drug toxicity. But we should all be able to agree that, if reasonable people hold differing opinions, there is real uncertainty.8 Even if we have strong hunches or preferences for one treatment approach over another, the lack of certainty should enable us, indeed force us, to perform the research we need to reduce that uncertainty and improve care for our babies.
Observational data are not convincing for the benefit of one approach or another, and both are commonly used in current practice. Even neonatologists who currently have preferences for one approach or another have to be willing to put those preferences aside and test their hunches, within all the safeguards of a high quality randomised clinical trial. At the end of the trial one group of us or the other (or perhaps both) will find our prejudices destroyed. Unless you can be reasonably sure which group that will be, then such a trial is ethical; continuing to offer treatments that differ from one hospital to another, with no high-quality evidence to support one choice over another, is, in contrast, ethically questionable.
Depending on the results of such a trial, whether one group has better outcomes than the other or not, we will be able to design the next trial, and continue the remarkable progress that neonatology has made over the past 50 years.
Only by recognising uncertainty, and then designing and successfully performing trials to reduce it, will we be able to improve interventions and outcomes for extremely preterm infants. There really is no alternative.
Footnotes
Contributors KJB drafted the editorial, MJ and KJB edited and approved the final version.
Funding This work was funded by the grant for the HIP trial from the European Commission FP7 framework.
Competing interests None declared.
Provenance and peer review Commissioned; internally peer reviewed.