Ultrasound: Weighing the Propaganda Against the Facts
Editor’s note: This article first appeared in Midwifery Today, Issue 51, Autumn 1999.
Subscribe to Midwifery Today Magazine
The use of ultrasound in antenatal care is big business, and in any big business marketing is all-important. As a result of decades of enthusiastic marketing, women believe they can ensure the well-being of their babies by reporting for an early ultrasound scan and that early detection of a problem is beneficial for these babies. That is not necessarily so, and there are a number of studies which show that early detection can be harmful.
In response to women’s desire for information about the implications of routine ultrasound examinations, Jean Robinson and I wrote the book Ultrasound? Unsound, in which we reviewed the research evidence and drew attention to some of the hazards (Beech and Robinson, 1996). But since then more evidence has accumulated. For example:
It is ironic that women who have had previous miscarriages often have additional ultrasound examinations in order to “reassure” them that their baby is developing properly. Few are told of the risks of miscarriage or premature labour or birth.
Obstetricians in Michigan (Lorenz et al., 1990) studied fifty-seven women who were at risk of giving birth prematurely. Half were given a weekly ultrasound examination; the rest had pelvic examinations. Preterm labour was more than doubled in the ultrasound group—52 percent—compared with 25 percent in the controls. Although the numbers were small the difference was unlikely to have emerged by chance.
A large randomised controlled trial from Helsinki (Saari-Kemppainen et al., 1990) randomly divided over 9000 women into a group who were scanned at sixteen to twenty weeks compared with those who were not. It revealed 20 miscarriages after 16–20 weeks in the screened group and none in the controls.
A later study in London (Davies et al., 1993) randomised 2475 women to routine Doppler ultrasound examination of the umbilical and uterine arteries at nineteen to twenty-two weeks and thirty-two weeks compared with women who received standard care without Doppler ultrasound. There were 16 perinatal deaths of normally formed infants in the Doppler group compared with four in the standard care group.
It is not only pregnant patients who are at risk, however. Physiotherapists use ultrasound to treat a number of conditions. A study done in Helsinki (Taskinen et al., 1990) found that if the physiotherapist was pregnant, handling ultrasound equipment for at least twenty hours a week significantly increased the risk of spontaneous abortion. Also, the risk of spontaneous abortions occurring after the tenth week was significantly increased for deep heat therapies given for more than five hours a week and ultrasound more than ten hours a week.
Diagnosis of placenta praevia
The Saari-Kemppainen study also revealed the lack of value in early diagnosis of placenta praevia. Of the 4000 women who were scanned at 16 to 20 weeks, 250 were diagnosed as having placenta praevia. When it came to delivery, there were only four. Interestingly, in the unscanned group there were also four women found at delivery to have this condition. All the women were given caesarean sections and there was no difference in outcomes between the babies. Indeed, there are no studies which demonstrate that early detection of placenta praevia improves the outcome for either the mother or the baby. The researchers did not investigate the possible effects on the 246 women who presumably spent their pregnancies worrying about having to undergo a caesarean section and the possibility of a sudden haemorrhage.
Since the publication of Ultrasound? Unsound further studies have raised questions about the value of routine ultrasound scanning.
Babies with serious defects
Almost all babies receive a dose of ultrasound, but even at the best centres wide variations occur in detection rates for babies with major heart abnormalities. Both national and international detection rates differ widely in published studies (which are usually undertaken in centres of excellence), but the majority of mothers will be exposed to older machines in ordinary hospitals and clinics. The skill of the operators will vary (everybody has to learn sometime), but even with the best machines and the best operators misdiagnoses occur. A study from Oslo (Skari et al., 1998) looked at how many babies born with serious defects had been diagnosed by antenatal scans, and whether the early diagnosis made any difference to the outcomes. Women in Norway have a scan at 17–21 weeks done by trained midwives, who refer to obstetricians if an abnormality is suspected.
In nineteen months, thirty-six babies were referred from a population of 2.5 million. They had diaphragmatic hernias, abdominal wall defects, bladder extrophy or meningomyelocele. Only thirteen of the thirty-six defects had been detected before birth (36 percent). They found that only two of eight congenital diaphragmatic hernias were picked up on ultrasound, half the cases of abdominal wall defects (six out of twelve), 38 percent of the meningomyelocele (five out of thirteen) and none of the three cases of bladder extroversion. The mothers had an average of five scans (from one to fourteen); those in whose cases abnormality was detected had an average of seven.
Three out of the 13 babies diagnosed antenatally died. There was one death in the 23 undiagnosed. All 13 babies with antenatal diagnosis were delivered by caesarean. Nineteen of the 23 undiagnosed babies had an uncomplicated vaginal delivery. The diagnosed babies had lower birth weight and two weeks shorter gestation. Although the babies with pre-diagnosed abdominal wall defects received surgery more quickly (4 hours versus 13 hours), the outcomes were the same in both groups. Although small, this is an important study.
Pregnant women often automatically assume that antenatal detection of serious problems in the baby means that lives will be saved or illness reduced. Knowing about the problem in advance did not benefit these babies; more of them died. They got delivered sooner, when they were smaller, a choice that could have long-term effects. All 12babies with abdominal wall defects survived. But for the six detected on the scan, their length of hospital stay was longer and they spent longer on ventilators, though the numbers are too small to be significant. They were operated on sooner (4 hours rather than13 hours) but the outcomes were the same.
One of the promises held out by antenatal scanning is that obstetricians will be able to identify the baby with problems and do something to help it. A German study from Wiesbaden hospital (Jahn et al., 1998) found that out of 2378 pregnancies only 58 of 183 growth retarded babies were diagnosed before birth. Forty-five fetuses were wrongly diagnosed as being growth retarded when they were not. Only 28 of the 72 severely growth-retarded babies were detected before birth despite the mothers having an average of 4.7 scans.
The babies diagnosed as small were much more likely to be delivered by caesarean—44.3 percent compared with 17.4 percent for babies who were not small for dates. If the baby actually had intrauterine growth retardation (IUGR) the section rate varied hugely according to whether it was diagnosed before birth (74.1 percent sectioned) or not (30.4 percent).
So what difference did diagnosis make to the outcome for the baby? Preterm delivery was five times more frequent in those whose IUGR was diagnosed before birth than those who were not. The average diagnosed pregnancy was two to three weeks shorter than the undiagnosed one. The admission rate to intensive care was three times higher for the diagnosed babies.
The long-term emotional impact
The effects of screening on both parents can be profound. For example, women waiting for the results of tests try not to love the baby in case they have to part with it. The medical literature has little to say about the human costs of misdiagnosis unless the baby was mistakenly aborted, and even then it tends to focus on legal action. However, a letter in the British Medical Journal revealed how a diagnosis of a minor anomaly can have serious long-term implications for the family:
A couple was referred for amniocentesis during the wife’s second pregnancy on the grounds of maternal age, thirty-five years, and anxiety. Their three-year-old son played happily during the consultation. When his wife and son had left the room after the procedure the husband confided that they had opted for amniocentesis to avoid having another “brain damaged” child. On questioning it became apparent that an ultrasound examination before their son’s birth had shown a choroid plexus cyst. Despite having a healthy child, the husband remained convinced that this cyst could cause his son to be disabled (Mason and Baillie, 1997).
Evaluating the risks
When ultrasound was first developed researchers suggested that “the possibility of hazard should be kept under constant review“ (Donald, 1980), and they said that it would never be used on babies under three months. However, as soon as vaginal probe ultrasound was developed, which could get good pictures in early pregnancies (and get nearer to the baby giving it a bigger dose), this initial caution was ignored.
Research by Lieberskind revealed “the persistence of abnormal behaviour . . . in cells exposed to a single dose diagnostic ultrasound ten generations after insonation.“ She concluded, “If germ cells were… involved, the effects might not become apparent until the next generation“ (Lieberskind, 1979). When asked what problems should be looked for in human studies, she suggested: “Subtle ones. I’d look for possible behavioural changes, in reflexes, IQ,, attention span“ (Bolsen, 1982).
Because ultrasound has been developed rapidly without proper evaluation it is extremely difficult to prove that ultrasound exposure causes subtle effects. After all, it took over ten years to prove that the gross abnormalities found in some newborn babies were caused by thalidomide. However, there are a number of ultrasound studies which raise serious questions that still have to be addressed.
The first evidence we saw of possible damage to humans came in 1984 when American obstetricians published a follow-up study of children, aged seven to twelve years born in three different hospitals in Florida and Denver, who had been exposed to ultrasound in the womb (Stark et al., 1984). Compared with a control group of children who had not been exposed they were more likely to have dyslexia and to have been admitted to hospital during their childhood, but no other differences were found.
In 1993 a study in Calgary, Alberta which examined the antenatal records of seventy-two children with delayed speech of unknown cause were compared with those of 142 controls who were similar in sex, date of birth and birth order within the family. The children were similar in social class, birth weight and length of pregnancy. The children with speech problems were twice as likely as controls to have been exposed to ultrasound in the womb. Sixty-one percent of cases and only 37 percent of controls had had at least one exposure.
A Norwegian study (Salvesen, 1993) showed an increase in left-handedness, but no increase in dyslexia. While the increase in left-handedness was not large, it does suggest that ultrasound has an effect on the development of the brain. It should be noted, however, that the scanners used in this study emitted very low doses of ultrasound–lower than exposures from many machines nowadays—the women had only two exposures, and it was real time, not Doppler, a more powerful form of ultrasound.
Assessing the risks
“Present day ultrasonic diagnostic machines use such small levels of energy that they would appear to be safe, but the possibility must never be lost sight of that there may be safety threshold levels possibly different for different tissues, and that with the development of more powerful and sophisticated apparatus these may yet be transgressed“ (Donald, 1979).
Donald’s foresight was remarkable. The machines in use today are far more powerful than the machines used a decade or more ago, and new variants are being developed all the time.
There has been inadequate research into the potential long-term effects. Measuring the outcome of any intervention in pregnancy is very complicated because there are so many things to look at. Intelligence, personality, growth, sight, hearing, susceptibility to infection, allergies and subsequent fertility are but a few issues which, if affected, could have serious long-term implications, quite apart from the numbers of babies who have a false positive or false negative diagnosis. Because a baby grows rapidly, exposing it to ultrasound at eight weeks can have different effects than exposure at, for example, 10, 18 or 24 weeks (this is one of the reasons the effects of potential exposure are so difficult to study). Women are now exposed to so many different types of ultrasound: Doppler scans, real-time imaging, triple scans, external fetal heart-rate monitors, handheld fetal monitors. Unlike drugs, whereby every new drug must be tested, the rapid development of each new variation of ultrasound machine has not been accompanied by similar careful evaluation by controlled, large-scale trials.
Despite decades of ultrasonic investigation, no one can demonstrate whether ultrasound exposure has an adverse effect at a particular gestation, whether the effects are cumulative or whether it is related to the output of a particular machine or the length of the examination. How many exposures are too many? What is the mechanism by which growth is affected? A large-scale study (Newnham et al., 1991) showed decreased birth weight, although a later study suggested the babies soon make up the deficit. It should not be forgotten, however, that numerous studies on rats, mice and monkeys over the years have found reduced fetal weight in babies that had ultrasound in the womb compared with controls. Nor should it be forgotten that in the monkey studies (Tarantal et al., 1993) the ultrasound babies sat or lay around the bottom of the cage, whereas the little control monkeys were up to the usual monkey tricks. Long-term follow up of the monkeys has not been reported. Do they reproduce as successfully as the controls? And, as Jean Robinson has noted: “Monkeys do not learn to read, write, multiply, sing opera, or play the violin.“ Human children do, and perhaps we should consider seriously whether the huge increases in children with dyslexia and learning difficulties are a direct result of ultrasound exposure in the womb. Furthermore, when a woman is scanned her baby’s ovaries are also scanned. So if the woman had seven scans during her pregnancy, when her pregnant daughter eventually presents years later at the antenatal clinic, her developing baby will already have had seven scans. Do women really know what they consent to when they rush to hospital to have their first ultrasound scan, then trustingly agree to further scans?
Beverley A Lawrence Beech, honourary chair of the Association for Improvements in the Maternity Services (AIMS), is a freelance writer and lecturer and lives in the United Kingdom.
- Beech, B., and J. Robinson. (1996). Ultrasound? Unsound. London: Association for Improvements in the Maternity Services (AIMS).
- Bolsen, B. (1982). Question of risk still hovers over routine prenatal use of ultrasound. JAMA, 247: 2195-97.
- Donald, I. (1979). Practical Obstetric Problems. (5th ed). London: Lloyd-Luke, Medical Books Ltd.
- Donald, I. (1980). Sonar—Its present status in medicine. In A. Jurjak (Ed), Progress in Medical Ultrasound, 1: 001–04. Amsterdam: Excerpta Medica.
- Jahn, A., et al. (1998). Routine screening for intrauterine growth retardation in Germany; low sensitivity and questionable benefit for diagnosed cases. Acta Ob Gyn Scand, 77: 643–89.
- Lorenz, R.P., et al. (1990, June). Randomised prospective trial comparing ultrasonography and pelvic examination for preterm labor surveillance. Am. J. Obstet. Gynecol 1603–10.
- Mason, G. and C. Baillie. (1997). Counselling should be provided before parents are told of the presence of ultrasonographic ‘soft markers’ of fetal abnormality (Letter). BMJ 315: 180–81.
- Newnham, J.P., et al. (1991). Effects of frequent ultrasound during pregnancy: a randomised controlled trial. The Lancet, 342: 887–90.
- Saari-Kemppainen et al. (1990). Ultrasound screening and perinatal mortality: controlled trial of systematic one-stage screening in pregnancy. The Lancet, 336: 387–91.
- Salvesen, K.A., et al. (1992). Routine ultrasonography in utero and school performance at age 8–9 years. The Lancet, 339.
- Skari, H., et al. (1998). Consequences of prenatal ultrasound diagnosis: a preliminary report on neonates with congenital malformations. Acta. Ob Gyn Scand, 177: 635–42.
- Tarantal, A.F., et al. (1993). Evaluation of the bioeffects of prenatal ultrasound exposure in the Cynomolgus Macaque (Macaca fascicularis). Chapter III in Developmental and Mematologic Studies, Teratology 47: 159–70.
- Taskinen, H., et al. (1990). Effects of ultrasound, shortwaves, and physical exertion on pregnancy outcome in physiotherapists. Journal of Epidemiology and Community Health 44: 196–201.
- Understanding Obstetric Ultrasound (2nd edition) by Jean Proud
- Midwifery Today Issue 50 Theme: Homebirth; from this issue, read online Ultrasound: More Harm Than Good? by Marsden Wagner