Ultrasound has an extraordinary safety record. It has been used in obstetrics for almost four decades with no proven harmful effects. In this issue of the Journal there is an updated review of the epidemiological literature1. The authors searched the literature extensively and analyzed the data according to Cochrane review guidelines. The results are reassuring. Apart from an unexplained weak association between ultrasound and non-right handedness in boys, there are no indications of deleterious effects from obstetric ultrasound. The authors conclude that exposure to diagnostic ultrasound during pregnancy appears to be safe. So why is safety of ultrasound an issue; or is it an issue?
We do not know that modern ultrasound devices are safe
Most of the available epidemiological evidence on ultrasound safety is derived from B-mode scanners in use before the mid 1990s. There are hardly any epidemiological data on the use of color flow or pulsed wave Doppler, and today's scanners can produce 10–15 times higher output levels than did these earlier scanners2. If biological effects of ultrasound are dose-dependent, this updated review of the epidemiological literature is not helpful at all for present-day ultrasound operators and pregnant women.
Ultrasound operators do not know how to use the real-time display of safety information on the screen
The American Institute of Ultrasound in Medicine (AIUM) and National Electrical Manufacturers Association (NEMA) introduced the ‘output display standard’ (ODS) in the early 1990s. ODS implies the use of biophysical indicators, such as the mechanical index (MI) and the thermal index (TI), for real-time display of safety information during scanning. The Food and Drug Administration (FDA) in the USA adopted the ODS and issued regulations demanding that the ODS information be provided by the manufacturers in all commercially available devices on the ultrasound market after 1992. In practice, this transformed the responsibility of the safe use of ultrasound from the manufacturer to the operator of the machine. The machines still have upper limits for energy output (intensity less than 720 mW/cm2), but it is the responsibility of the ultrasound operator to consider the output displays (MI and TI) and to scan with output levels according to the ALARA principle (as low as reasonably achievable).
Ten years after the introduction of ODS, Karel Maršál surveyed the knowledge among ultrasound users of some safety aspects of diagnostic ultrasound3. A questionnaire was distributed to 145 doctors, 22 sonographers and 32 midwives from nine European countries. All of them were using diagnostic ultrasound on a daily or weekly basis. The results of this study were depressing. About one third knew the meaning of MI and TI, and only 28% knew where to find the safety indices on the screen of their own machine. More alarmingly, only 43 (22%) of 199 respondents knew how to adjust the energy output on their machine3. Theoretically the ODS may well be an excellent concept, but that is not much help if not even the ultrasound experts know where to find the output displays and how to turn down the output levels on their own machines. It is fair to say that the ODS has failed to provide a basis for safe scanning—at least when applied to obstetric examinations.
Doppler is used in the first trimester in normal pregnancies
According to the European Committee for Medical Ultrasound Safety (ECMUS) safety statement4, pulsed Doppler ultrasound should only be used in the first trimester under careful control of exposure levels and exposure times. The introduction of the 11 to 13 + 6-week scan to screen for fetal chromosomal anomalies has challenged this safety statement. Pulsed Doppler across the fetal tricuspid valves and in the ductus venosus can be used to refine risk assessments for Down syndrome and other trisomies. This may not be a problem if Doppler is used sequentially, that is, after serum screening and/or measurements of nuchal translucency thickness have revealed a high risk. It may be a problem, however, if pulsed Doppler is used routinely in all pregnancies or for extensive time periods in normal pregnancies for the purpose of training and qualifying for accreditation.
If Doppler were to have an adverse fetal effect, we could hypothesize that it would most likely do so early in gestation, when there is more rapid cell division and when the fetal blood flow is less well developed and therefore less likely to dissipate heat derived from Doppler examination. Further, these examinations are at the level of the ductus venosus or fetal heart, very close to a bone (i.e. the spine)/soft tissue interface, where a heating effect would be greatest. The main reason for advocating a restrictive or precautionary use of Doppler ultrasound in early gestation is not the fact that we know that Doppler ultrasound can cause harm, because we do not. Restrictive use is advocated because we do not know that Doppler ultrasound is safe, and because of the fact that the first trimester is a particularly vulnerable period of fetal life.
Journal policy on publishing reports of first-trimester Doppler ultrasound research is not followed
In 1999, the Editors of the Journal took a strong position regarding the publication of research papers using first-trimester Doppler5. The Journal policy was to accept papers on color and pulsed wave Doppler in the first trimester only if several requirements had been fulfilled, including the use of ODS and the ALARA principle, explicit publication of machine settings and exposure times, and mandatory obtaining of informed patient consent and ethical review committee approval5.
A search in PubMed (March 2009) on the following key words: ‘Doppler’, ‘first trimester’ and ‘Ultrasound Obstet Gynecol’, gave a total of 126 papers. Among 21 papers published in the Journal in 2007 and 2008, 11 were studies using uterine artery Doppler, seven were studies using fetal Doppler at 10–14 weeks, two were Doppler studies after 14 weeks and one paper was a case report. The papers on uterine artery Doppler prediction of pre-eclampsia and fetal growth restriction are probably non-controversial from a safety point of view, because the fetus is not insonated during the Doppler examination. However, among the seven papers on fetal Doppler at 10–14 weeks6–12, only one6 appeared explicitly to fulfil the requirements listed by the Journal Editors in 19995. (We don't know that the others didn't, we just know that they did not obviously do so.) We believe that this exemplifies how easy it is to forget the safety issue when writing and reviewing research papers. It may be time to reinforce Journal policy regarding research papers involving Doppler in the first trimester.
There is a possible link between experimental and epidemiological evidence on ultrasound and handedness
The association between ultrasound and non-right handedness in boys is discussed in this issue of the Journal1. Yet, who cares about this unexplained weak association? Being left-handed is not a problem. Barack Obama is left-handed. He is doing fine, as did the other four left-handed US presidents in the last 75 years.
In general, left-handers are no different from right-handers. This does not preclude that sinistrality can be associated with pathological conditions13, which can best be explained by a very small group in whom left-handedness is caused by early brain damage. Yet even the prevalence of 39% (5 of 13) left-handers among US presidents in the last 75 years, compared with 10% in the general population, is not considered a sign of brain damage by satirical comedians or enemies of the US. Thus, the weak association between ultrasound exposure during pregnancy and non-right handedness in boys is interesting, but not alarming. However, a study from 2006 of fetal mouse brains demonstrated that exposure from a commercially available ultrasound device was capable of producing disturbed neuronal migration14, although the exposure times were extensive (up to 420 min) and were not comparable to common obstetric practice. Also, the current understanding of the biological mechanisms behind left-handedness is contradictory to a possible effect of disturbed neuronal migration because of ultrasound exposure15, 16. Nevertheless, a possible link between disturbed neuronal migration in mouse brains after exposure to modern ultrasound devices and epidemiological evidence of non-right handedness in boys after exposure to old ultrasound devices, relates directly back to the fundamental problem: we do not know that modern devices are safe.
Is ‘souvenir scanning’ a problem for the future of ultrasound?
It could be argued that keepsake or souvenir scanning (‘ultrasound for fun’) is none of our business. The medical profession do not advocate it, and people should be able to do whatever they want with their money. On the other hand, as professionals involved in ultrasound, we must stand up for the unborn babies and the future of ultrasound if we think souvenir scanning is wrong. According to official statements from most ultrasound societies, we do think it is wrong17. The World Federation for Ultrasound in Medicine and Biology (WFUMB) symposium on safety of non-medical use of ultrasound at the International Society of Ultrasound in Obstetrics and Gynecology (ISUOG) 2007 world conference in Florence reinforced this position.
Communicating our concern about souvenir scanning is difficult. This is because the medical profession is ‘talking with two tongues’. On the one hand we say that ultrasound is perfectly safe in a medical setting, and we will be happy to sell you a picture from the scan. On the other hand we say that ultrasound for fun may harm your baby, so you shouldn't do it. How can the public understand this double communication?
We are not spin doctors trained in difficult double communication, but we do believe that the answer is to be frank and report the lack of knowledge on safety of modern ultrasound devices. We do not believe that ultrasound is harmful, but we do not know for sure. We do believe in the benefits of medical ultrasound, and that the benefits outweigh the risks (if any). Souvenir scanning has no medical benefit, and cannot outweigh any possible risk. This is why it is hard to justify souvenir scanning. Having said this, the current vogue for three-dimensional scanning is not normally associated with significantly higher power outputs or length of examination than those of two-dimensional ultrasound, so the balance between safety and the profession being proscriptive must be carefully weighed.
Where do we go from here?
We will have to live with uncertainty regarding ultrasound safety in the years to come. There is no such thing as zero risk, and absence of evidence of harm is not equivalent to evidence of absence of harm. More research is welcomed, but the time has passed when randomized controlled trials with ‘ultrasound free’ control arms could be done. As professionals involved in ultrasound, we must regulate ourselves sensibly—or else someone else is likely to. We owe it to our patients, to unborn babies and to the future of ultrasound.
There is a possible way forward. We should start recording the output exposure levels and exposure times during all scans. This was suggested in a paper in the Journal in 200018. In that paper, this was accomplished by going through hours of videotapes manually. This could, however, be done automatically by software in the ultrasound devices, and be compiled as part of an ultrasound report printout from the machine. However, we would need the manufacturers to provide this service in all scanners (Figure 1).
The potential problems are clear—fear of litigation because of stored records with exposure levels and exposure times. Still, this is probably the only way to create large databases of ultrasound exposure levels and exposure times for future epidemiological research, and it is far better for us—the practitioners—to regulate ourselves sensibly than to have it forced upon us. That is the danger if we, as sonographers and sonologists, fail to act now.