MedicalResearch.com Interview with:
Gordon C S Smith, MD PhD DSc FMedSci
Professor & Head of Department,
Obstetrics and Gynaecology,
Medical Research: What were the key findings of the study?
Dr. Smith: We demonstrated that using ultrasound to scan all women in the last 3 months of pregnancy increased the detection of small babies. Our routine approach to antenatal care identified 1 in 5 small babies. Our research approach on the same patients, where all women were scanned in the last third of pregnancy, resulted in 3 in 5 small babies being detected.
We also found that, when we identified small babies, there was a scan measurement which helped us to distinguish which of those small babies were at increased risk of complications.
Medical Research: What’s the problem with small babies?
Dr. Smith: We have known for many years that small babies are at increased risk of subsequent complications, both in the short term and long term. For example, they are more likely than normal sized babies to have difficulties during labour due to shortage of oxygen.
We assume that this is because their growth has been restricted. This leads to them being less well prepared to face the stress of labour. We think that the poor growth usually occurs because of a problem in the way the placenta functions. The placenta, also known as the afterbirth, is the organ which allows the transfer of nutrients and oxygen to the baby. If the placenta isn’t functioning properly it is likely to impair the growth of the baby in the womb.
If we suspect during a pregnancy that the baby is small, it influences a number of aspects of the care of the mother. For example, we might deliver the baby earlier, we might recommend delivery in a high risk unit. It might also influence decisions about whether a mother has a caesarean delivery, or how the baby is monitored during labour.
Medical Research: How do we know whether a baby is small?
Dr. Smith: The main way that we determine the size of a baby in the womb is to perform an ultrasound scan. We measure the size of the baby’s head, abdomen and thigh bone and use a statistical model to estimate the weight. We then relate the estimated weight of the baby to the normal range for that stage of the pregnancy.
Like any other method of prediction, it isn’t perfect. The average error is about 8% of birth weight and errors of 15% or more are common. Another complicating factor is that many small babies are healthy. It is only a minority of small babies that have failed to grow properly and their small size reflects a problem.
Medical Research: Do all women get scanned to check whether the baby has grown OK?
Dr. Smith: In the UK and many other countries, women have routine scans at 12 weeks to date the pregnancy, and at 20 weeks to check for anomalies, such as heart defects or cleft lip. But they are not routinely scanned again in later pregnancy. Rather women are selected for late pregnancy scans if they are identified as high risk. For example, women who have pre-existing medical problems (such as diabetes), women who have had previous bad pregnancy outcomes, women who develop complications during the pregnancy, or if the midwife or doctor examining the woman thinks that the baby seems small. The last of these usually involves measuring the size of the bump with a tape measure. If a woman has none of these features, she is not usually scanned again after 20 weeks.
Medical Research: Surely it makes sense just to scan everyone?
Dr. Smith: Researchers have done large trials to address this question. A trial compares the outcomes in two groups of women who are given different treatments. In each of the trials, women were randomly allocated to one of two groups. In the first, everyone had at least one extra in late pregnancy. In the other group of women, they were only scanned if there was some specific reason to do so. The outcomes of the two groups were then compared. Combining all of these trials together in a single analysis, which included a total of 27,000 women, the outcomes seem to be similar whether women were routinely scanned or not. So research to date has not shown any value in scanning everyone.
Medical Research: If scanning doesn’t help, what was the point in doing the POP study?
Dr. Smith: Screening for a medical problem has two components. First, we use a test to check who is at a higher risk of having a bad outcome. Second, we intervene and do something to prevent the bad outcome in those women whom the test had picked up as high risk.
The problem with the existing research on late pregnancy scanning is that we do not know why screening didn’t work. It could be because scanning isn’t any good at picking up women who are high risk. Or it could be that the studies did not use an intervention which was effective in reducing the risk in the cases where there seemed to be a problem. Alternatively, although 27,000 sounds like a lot of women, it is impossible to say whether that was enough to get meaningful results.
Also, each of the different studies which were pooled in the analysis used different definitions of what constituted an abnormal scan. Many of the measurements used had not been thoroughly evaluated prior to testing them in the trial. So the previous research trialled a programme of screening using ultrasound without really knowing how well the screening test worked. The aims of our study were
(1) to find out how well universal scanning picked up small babies, and
(2) to find out which scan measurements, if any, were able to identify the small babies that were at increased risk of complications.
Medical Research: Who were the women studied?
Dr. Smith: The women eligible for the study were women attending the Rosie Hospital in Cambridge, England, who were having their first baby, and where their dating scan (done around 12 weeks) showed a single, viable pregnancy. About 8000 women were approached, and about 4,500 agreed to take part in the study. Of these, about 4000 attended for all of their scans, delivered at the hospital and were suitable for the analysis.
Medical Research: What was done for the purposes of research?
Dr. Smith: All the women in the study received normal antenatal care throughout their pregnancy. The research studied the effect of scanning everyone, which we called universal ultrasound. This involved additional scans and tests which are not part of normal care: at 20 weeks, the blood flow to the placenta was measured as part of the research study. Also, at 28 and 36 weeks, these blood flow measurements were repeated and the weight of the baby was estimated using ultrasound, again as part of the research.
Crucially, the results of the research scans were not revealed to the women or their doctors. Following the birth, the association between the research scans and the outcome was analysed statistically.
Medical Research: What if the research scan showed a major problem?
Dr. Smith: If the research scan showed serious problems in the pregnancy the results were revealed to the women and their doctors. To be consistent, we agreed before the study began which problems should be revealed. In total, between 150 and 200 women had their scan results revealed. In most of these cases, the result revealed was that the baby was not presenting head first at the time of the 36 week scan. In one or two cases, there were critical findings that resulted in major changes to the way the mother and baby were cared for.
Medical Research: Why did any of the results need to be blinded?
Dr. Smith: None of the measurements of the baby’s size and blood flow to the placenta were reported to the women or their doctors. This is because the whole purpose of the study was to determine how well these measurements might predict the outcome of the pregnancy, when compared with routine care. In order to compare routine care with a new test, the results of the new test have to be concealed. Revealing the results of the new test would inevitably influence the normal care. Hence, it wouldn’t be possible to say accurately what information the new test really added.
Medical Research: How many of the women ended up getting a scan requested for the purposes of their care?
Dr. Smith: All women got scanned in early pregnancy and had a detailed scan of the baby at 20 weeks, as they normally would. But only about 40% of the women studied ended up being referred for a scan after their 20 week scan, which was ordered by their doctors, based on their clinical need.
The women who were selected for extra scans tended to be older, they were more likely to be very underweight or overweight, they were more likely to have diabetes, and they were more likely to have had pregnancy complications. This is what we expected: currently, women tend only to be scanned if they are high risk or if a problem emerges.
Medical Research: How well did routine care and the research scans detect the small babies?
Dr. Smith: The routine antenatal care of the 4000 women in the study detected about 20% of small babies. The research scans performed on the same women detected 57% of small babies. Small was defined as being in the smallest 10% for the given stage of pregnancy.
For very small babies – that is in the smallest 3% of birth weight – routine care picked up about 32% whereas the research scanning picked up 77%.
Medical Research: So the universal scanning done for research was much better than the selective use of scanning done clinically?
Dr. Smith: It is not quite so simple as that.
The numbers went as follows:
With selective scans, there were 138 cases where the baby seemed to be small. In 69 cases the baby was confirmed to be small after the birth, but in another 69 cases it wasn’t.
With the research approach, where everyone was scanned, there were 562 cases where the baby seemed to be small. In 199 cases, it was but in 363 cases it wasn’t.
So although scanning everyone picked up 130 extra cases where the scan correctly identified a small baby, it also resulted in about another 290 cases where the scan indicated that the baby was small, when in fact the baby’s size was OK.
So the overall effect of scanning everyone was that for every 1 extra case where universal scanning correctly identified a baby as small, there were 2 additional cases where universal scanning diagnosed the baby as small, but got it wrong.
Medical Research: Was it possible to tell which small babies were at increased risk of complications?
Dr. Smith: When we looked at the outcomes of the 562 babies which seemed to be small on the POP study scan, we found that, overall, their risk of complications was 60% higher when compared with other babies. However, we felt that it was likely that many of the babies who appeared to be small were actually normal, either because the scan had got the size wrong, or the baby was small but healthy. So next, we studied a range of measurements which we had made during the pregnancy, which had previously been described as markers of poor fetal growth.
The first thing we did was to compare an estimated size of the baby which took into account the mother’s characteristics (height, weight and ethnic group) and we compared that to the standard measurement. Next, we examined a number of markers of fetal growth restriction and placental function. These were
(i) the blood flow to the placenta (on both the mother’s side and the baby’s side),
(ii) ratios of the different measurements of the baby, which had previously been thought to be associated with an abnormal growth patterns,
(iii) analysis of the rate of growth from 20 weeks onwards.
Medical Research: What worked?
Dr. Smith: First, adjusting the estimate of the size of the baby for the mother’s characteristics did not significantly improve prediction of complications following the birth. When we then looked at the markers of growth restriction and placental function, there was only one measurement which had strong evidence that it helped distinguish which small babies would experience complications. It was the abdominal circumference growth velocity.
If a baby was small and also had a low growth velocity it was at increased risk of complications compared with other babies.
However, if a baby was small but had a normal growth velocity, it was at the same risk of complications following birth as other babies.
Of the babies which were thought to be small at the POP study scan, about 30% have low abdominal circumference growth velocity, and in the other 70% it was normal.
So using this measurement, we could distinguish which small babies were at increased risk and which were healthy.
With all of the other measurements described to identify unhealthy babies, those which were small were still at increased risk of complications even when the ultrasound marker was in the normal range. For example, a measurement of flow in the umbilical artery is widely used in the assessment of the baby’s well being. However, in our study, small babies were still at increased risk of complications even when this measurement was normal.
Medical Research: What exactly is the abdominal circumference growth velocity?
Dr. Smith: Basically, this measurement indicates whether the baby was small from the beginning of pregnancy or whether it started off a normal size, and then stopped growing later on. I will explain this in a bit of detail. If you don’t think you need to know all of this skip ahead to the next section.
The technical approach is as follows. We take the measurement of the abdomen at 20 weeks. We then relate this to the average measurement using a thing called a z score. The z score is a number which indicates how much bigger or smaller a measurement was in comparison to the average measurement for that stage of pregnancy. The z score is positive when the baby is bigger than average and it is negative when the baby is smaller than average. If the baby is bang on the average, the z score is 0.
We calculated the z score at 20 weeks and then we calculated it again at the last scan before birth, usually at 36 weeks. If the z score got smaller, then the baby’s growth had slowed down. We regarded cases where the change in the z score was in the lowest 10% as abnormal (equivalent to a fall of 1.5).
Medical Research: Why measure the abdomen?
Dr. Smith: There is a lot of research that suggests the abdomen is the key measurement for assessing the growth of the baby. The measurement is made at the level of the liver. It also includes fat under the skin. Both the liver and the fat under the skin are places where a baby lays down energy stores. If the baby is short of nutrition in the womb – because, for example, the placenta isn’t working – it preserves the supply of nutrients to the critical organs, such as the brain and heart, at the expense of laying down energy stores. It makes a lot of sense that this measurement would be the one which really helps discriminate between healthy small babies, and those who are at high risk of complications during the birth.
Medical Research: So what happens next?
Dr. Smith: In the medium to long term, we are planning to conduct a trial to test whether pregnancy outcomes are improved if we screen all women using the combination of estimated fetal size plus the abdominal growth velocity, and intervening in cases where the baby shows the high risk pattern.
The results of the POPS study are critical in helping us design a trial, as we now know the combination of scan findings which identifies high risk babies.
In fact, knowing now what ultrasound really tells us in terms of predicting bad outcomes, it is not surprising that the previous trials of routine ultrasound in late pregnancy failed to show any benefit. Some of the measurements that were used in the previous trials turned out not to be very good at predicting bad outcomes.
The numbers of study participants is also important: our initial estimates are that a trial will need to recruit a lot more than 27,000 women. Also, the timing of scans is critical. We think that a scan at around 36 weeks is most likely to be useful, as at this stage there are a series of interventions that we can use to prevent bad outcomes if we identify a pregnancy as high risk. There are fewer options for intervening if we detect problems much earlier in the pregnancy.
Medical Research: But what do we do meantime? Shouldn’t we just be scanning everyone now?
Dr. Smith: The problem with this is the false positives that we mentioned earlier. If we scan everyone now, some women will be told they have a small baby when their baby is actually in the normal range. So, scanning everyone now would lead to a lot of unnecessary anxiety among the false positives. But as well as anxiety, this could also lead to harm. These women may end up being delivered early, or by caesarean. In those cases where the baby is actually healthy, these interventions may not just do no good, they may cause harm. We need a trial because it lets us know, overall, whether the good we might achieve in the cases where we get it right is greater than the harm we cause due to unnecessary intervention in the healthy pregnancies where we get it wrong.
Medical Research: Are there any immediate changes arising from the POP study?
Dr. Smith: The key thing that can change now is how we assess growth scans when they are done in late pregnancy. Each of the measurements that we had evaluated as markers of a high risk small baby were in clinical use. But we now know that the one measurement that really discriminates between healthy small and high risk small babies is the abdominal growth velocity. When we see the combination of a small baby with a low growth velocity, these babies will need careful monitoring, and consideration should be given to early delivery. Conversely, if a growth scan shows that the baby is small but the growth velocity of the abdomen is normal, there may be a case for a more hands off approach. The baby may need further assessment, but may not need early delivery. These decisions will be made by women with the advice of their doctor or midwife, who will be able to discuss the risks and benefits of different options.
So the information from the POP study is useful right now, as it clarifies how we should be interpreting growth scans of the baby in late pregnancy. However, this information can only be used where a pregnant woman has been scanned in later pregnancy i.e. in addition to the 20 week scan. A second scan after 20 weeks is needed to work out the growth velocity and weight.
Medical Research: Who funded this research? Is there more to come?
Dr. Smith: The bulk of the funding was from the UK Government, through the NIHR Cambridge Comprehensive Biomedical Research Centre.
GE, a scan manufacturer supplied two ultrasound machines for free, but had no role in the designing the study, or analysing or interpreting the results.
Sands, the UK stillbirth and neonatal death charity, paid the salary of the research statistician who performed the analysis for the Lancet paper.
This paper is the first report to come out from the POP study. However, we have further analyses in progress about other aspects of the scan data. And we also have a large collection of blood samples from the mothers taken at different stages during the pregnancy, as well as a collection of samples of the placenta obtained as soon as possible after the delivery. There are multiple studies analysing these samples and we expect further interesting findings to come out of this research.
Medical Research: What are the main conclusions?
Dr. Smith: We have shown that scanning all women in the last three months of pregnancy increases the detection of small babies by three fold. However, for every one additional new case of a small baby we correctly identify we also identify two new false positives.
When we find a small baby on scan, measurements of the growth of its abdomen helps identify whether the baby is truly growth restricted and at increased risk of complications.
A large trial to test these findings may help clarify whether we should be screening all first pregnancies using routine late pregnancy ultrasound.
Gordon C S Smith, MD PhD DSc FMedSci (2015). Should All Women Have Third Trimester Ultrasound?