Chapter 1: ULTRASOUND BIOEFFECTS AND SAFETY: WHAT THE PRACTITIONER SHOULD KNOW

"Is this safe for my baby?" Ultrasound practitioners hear this question almost every day in clinical practice. The answer generally given is: "Of course. Ultrasound is not x-rays, it is not invasive, it has been used for close to fifty years and is perfectly safe." While this answer may, in fact, contain some correct facts (ultrasound is not x-rays), the concept of perfect safety is not scientifically valid, and furthermore, the level of knowledge regarding potential effects of ultrasound in tissues is, by and large, very low among clinicians. Ultrasound in obstetrics is convenient, painless, and results are available immediately. The belief exists that is does not pose any risk to the pregnant patient or her fetus. However, ultrasound is a form of energy and, as such, has effects in biological tissues (bioeffects). The physical mechanisms responsible for these effects are thermal or nonthermal (mechanical). The nonthermal mechanisms can further be separated into acoustic cavitation (inertial and noninertial) and noncavitational mechanisms, ie, acoustic radiation force (time-averaged force exerted by the ultrasound beam), acoustic radiation torque (producing in the insonated tissue a tendency to rotate or spin), and acoustic streaming (circulatory flow). It is the role of science to show whether any of these bioeffects may be harmful. The question has been debated since the introduction of ultrasound in clinical obstetrics, particularly as it relates to the fetal nervous system^2,3 and continues to be discussed currently.^{4,5,6,7,8, and 9}

This chapter presents basic notions of acoustics and physics as they relate to ultrasound, examines some literature on bioeffects and the safety of ultrasound, reviews statements of various ultrasound organizations, and affords a practical approach to limit the potential risks to the fetus of exposure to diagnostic ultrasound (DUS).

A detailed description of ultrasound physics can be found in various publications.^{10,11, and 12} However, certain properties of ultrasound are very important when trying to understand safety and bioeffects. Equally important are tissue characteristics, such as attenuation coefficient. A basic knowledge of instrument controls ("knobology") is essential not only for appropriate clinical usage, but it is imperative to avoid potential harm.

The Ultrasound Wave

Sound is a mechanical vibratory form of energy. It propagates through a medium by means of the motions of the particles in the medium, under the influence of the alternating positive and negative components of the wave. Megapascal (MPa) is the unit for pressure. Ultrasound instrumentation can generate peak pressures of 5 MPa and above. This is in comparison to the atmospheric pressure, which is 0.1 MPa. Several other characteristics define the ultrasound beam. The ultrasonic wave progresses in the insonated tissue at a velocity that is related to the sound characteristics as well as the tissue characteristics. For practical purposes, the average speed of sound propagation in biological tissues is estimated at 1540 ms/sec. Frequency is the number of cycles per second, measured in hertz (Hz). The limits of human hearing spans from approximately 20 to 20,000 Hz. Diagnostic ultrasound is, generally, 2 to 10 million Hz (megahertz, MHz). Wavelength is the distance between 2 corresponding points on a particular wave. It is inversely proportional to the frequency. Equipment resolution (the shortest distance between 2 objects or parts of an object to be represented by 2 separate echoes) depends on the wavelength: axial resolution ranges between 2 and 4 wavelengths. Hence, the shorter the wavelength (ie, the higher the frequency), the better the resolution (the distance between 2 points is smaller). The trade-off is that the higher the frequency (better resolution), the lower the penetration of the beam through a given tissue (Figure 1-1).

Diagnostic ultrasound is pulsed, ie, pulses of acoustic energy separated by "silent" gaps. The number of pulses occurring in 1 second is the pulse repetition frequency (PRF) and is controlled by the instrument in B-mode. In Doppler mode, it can be changed by the end user. Another important parameter is the duty factor: this is the fraction of time that the pulsed ultrasound is on. With an increase in PRF, the duty factor increases. The pulse amplitude reflects pressure and is the maximum variation from the baseline, expressed in megapascals (MPa). Since the ultrasound wave is sinusoidal, there are periods of positive and negative pressure. When the ultrasound wave exerts pressure on the resisting insonated tissue, work is produced. The ability of the wave to do work is its energy (in joules). The rate at which the energy is transformed from one form to another is the power (in watts or milliwatts). Intensity represents the rate at which energy passes through area unit. Average intensity of a beam is expressed by the beam power (in milliwatts, mW), divided by the cross-sectional area of the beam (in cm²) and is, therefore, expressed in mW/cm². As stated earlier, DUS is performed with a pulsed wave. The intensity is proportional to the square of the instantaneous ultrasound wave pressure. There are pulses of energy intermingled with periods where no energy is emitted. Depending on the time and location of the measurement, several parameters can be described in relation to time or space: temporal peak intensity (the greatest intensity), average intensity over time, ie, including "silent" time between pulses (temporal-average intensity), maximal intensity at a particular location (spatial-peak intensity), as well as average-spatial intensity. By combining time and space, 6 intensities can be described: spatial average–temporal average (I_SATA), spatial average–pulse average (I_SAPA), spatial average–temporal peak (I_SATP), spatial peak–temporal average (I_SPTA), spatial peak–pulse average (I_SPPA), and spatial peak–temporal peak (I_SPTP). The most practical, and commonly referred to, is the I_SPTA.

The maximal permitted value varies by clinical application. This had been determined in 1976 by the US Food and Drug Administration (FDA),¹³ but was modified in 1986.¹⁴ The most recent definition dates from 1992.¹⁵ These values are shown in Table 1-1. One can observe from the table that, for fetal imaging, the I_SPTA has been allowed to increase by a factor of almost 16-fold from 1976 and almost 8-fold from 1986 to 1992, yet, all epidemiological information available regarding fetal effects predates 1992. A remarkable fact is that intensity for ophthalmic examination has not changed from the original 17 mW/cm², a value approximately 42.5 times lower than the present allowed value for fetal scanning.

Tissue Characteristics

When the ultrasound wave travels through a medium, its intensity diminishes with distance.¹⁶ In completely homogeneous, idealized materials, the signal amplitude would be reduced only because the wave is spreading. Biologic tissues, however, are different and induce further weakening by absorption and scattering (an effect called attenuation) and by reflection. Many models have been described to help calculate attenuation, particularly in obstetrical scanning,¹⁷ but the most commonly used model uses an average attenuation of 0.3 dB/cm/MHz.¹⁸ It is important to note that the attenuation increases logarithmically with frequency and distance traveled. Technically, many measurements of acoustic power are performed in water, which has almost no attenuation. To apply these calculations to tissues, values are multiplied by this factor, an action called derating.¹⁹ Absorption is the sound energy being converted to other forms of energy, and scattering is the sound being reflected in directions other than its original direction of propagation. Since attenuation is proportional to the square of sound frequency, it becomes evident why higher frequency transducers have less penetration (but better resolution; see Figure 1-1). One needs, therefore, to be closer to the organ of interest, such as through transesophageal or, in obstetrics and gynecology, transvaginal scanning. Another possibility is increasing the power of the instrument, resulting in improved resolution, as depicted by the red arrow in Figure 1-1. This is seemingly simple, but instrument outputs are regulated in the United States (see The Output Display Standard section). Another important parameter is acoustic impedance, which can be described as the opposition to transmission of the ultrasound wave. It is proportional to the velocity of sound in the tissue (estimated at 1540 ms/sec) and to the tissue density.

Instrument Outputs

Although some publications of various instrument outputs are available,^{20,21, and 22} these are generally quickly outdated, since manufacturers introduce new commercial machines to the market (or modify existing ones) at a rate too fast for immediate objective evaluation. From a clinical standpoint, there is no easy way to verify the actual output of the instrument in use. In addition to different instruments, each attached transducer will generate a specific output, further complicated by the different modes that may be applied.²³ When comparing modes, the I_SPTA increases from B-mode (34 mW/cm², average) to M-mode to color Doppler to spectral Doppler (1180 mW/cm²). Average values of the temporal averaged intensity are 1 W/cm² in Doppler mode but can reach 10 W/cm².²³ Therefore, caution should be exercised when applying Doppler mode, particularly in the first trimester. Color Doppler, while having higher intensities than B-mode, is still much lower than spectral Doppler. This is mainly due to the mode of operation—sequences of pulses, scanned through the area of interest ("box"). Most measurements are obtained from manufacturers' manuals, having been derived in laboratory conditions. Real-life conditions may be different.²⁴ Furthermore, machine controls can alter the output. If one keeps in mind that, for instance, the degree of temperature elevation is proportional to the product of the amplitude of the sound wave by the pulse length and the PRF, it becomes immediately evident why any change (augmentation) in these properties can add to the risk of elevating the temperature, a potential mechanism for bioeffects (see Thermal Effects). The 3 important parameters under end-user control are the scanning (or operating) mode, including transducer choice; the system setup and output control; and the dwell time.

1. Scanning mode: B-mode carries the lowest risk, and spectral Doppler carries the highest (with M-mode and color Doppler in between). High pulse repetition frequencies are used in pulsed Doppler techniques, generating greater temporal average intensities and powers than B- or M-mode, and hence greater heating potential. An additional risk is that since, in spectral Doppler, the beam needs to be held in relatively constant position over the vessel of interest, there may be a further increase in temporal average intensity. Naturally, transducer choice is of great consequence since it will determine frequency, penetration, resolution, and field of view.

2. System setup: starting or default output power and, particularly, mode (B-mode, Doppler etc.) control changes. A subtler element is fine tuning performed by the examiner to optimize the image and influence output but with no visible effect (except if one follows thermal index [TI] and/or mechanical index [MI] displays). Controls that regularize output include focal depth (usually with greatest power at deeper focus but occasionally, on some machines, with highest power in the near field); increasing frame rate; and limiting the field of view, for instance, by high-resolution magnification or certain zooms (Figure 1-2). In Doppler mode, changing sample volume and/or velocity range (all done to optimize received signals) changes output. A very important control is receiver gain. It often has similar effects to the above controls on the recorded image but none on the output of the outgoing beam, and is therefore completely safe to manipulate. In other words, the receiver gain should be maximized before output is increased. In addition, over the years, output of instruments has increased.²²

3. Dwell time is directly under the control of the examiner. Interestingly, dwell time is not taken into account in the calculation of the safety indices nor, in general, until now, reported in clinical or experimental studies. However, one needs to remember that it takes only 1 pulse to induce cavitation, and about a minute to raise temperature to its peak. Directly related with dwell time is examiner experience: knowledge of anatomy, bioeffects, instrument controls, and scanning techniques. It can be safely assumed that the more experienced the examiner, the less scanning time will be needed.

A standardized method of providing the end user a parameter related to acoustic output and expressing potential for bioeffects is clearly needed; hence, the generation of the Output Display Standard, based on the 2 most likely interactions of ultrasound with tissues: thermal and nonthermal or mechanical.²⁵

Normal core human body temperature is generally accepted to be 37°C with a diurnal variation of ±0.5°C to 1.0°C, although 36.8°C ± 0.4°C (95% confidence interval) may be closer to the actual mean for large populations.²⁶ During the entire gestation, temperature of the human embryo/fetus is higher than maternal core body temperature²⁷ and gradually rises until the final trimester (near term). The fetal temperature generally exceeds that of the mother by 0.5°C.²⁸ Thermally induced teratogenesis has been demonstrated in many animal studies, as well as several controlled human studies.²⁹ While elevated maternal temperature in early gestation has been associated with an increased incidence of congenital anomalies,³⁰ the majority of these studies do not involve ultrasound-induced temperature elevation.

Edwards and others have demonstrated that hyperthermia is teratogenic for numerous animal species, including humans,³¹ and suggested a 1.5°C temperature elevation above the normal value as a universal threshold.³² Some scientists believe that there are, indeed, temperature thresholds for hyperthermia-induced birth defects, hence the ALARA (as low as reasonably achievable) principle. However, there is some evidence that any positive temperature differential for any period of time has some effect. In other words, that there may be no thermal threshold for hyperthermia-induced birth defects.³³ From careful thermal dose determinations derived from published literature in this area, it may be that hyperthermia-induced birth defects are induced in accordance with an Arrhenius relation for chemical rate effects, and thus have no threshold.³⁴ Any temperature increment for any period of time has some effect. Likewise, the higher the temperature differential or the longer the temperature increment, the greater the likelihood of producing an effect. Gestational age is a vital factor: milder exposure during the preimplantation period can have similar consequences to more severe exposures during embryonic and fetal development and can result in prenatal death and abortion or a wide range of structural and functional defects.

The organ at greatest risk is the central nervous system (CNS) due to a lack of compensatory growth of undamaged neuroblasts. In experimental animals the most common defects are of the neural tube, microphthalmia, cataract, and microencephaly, with associated functional and behavioral problems.³¹ Defects of craniofacial development including clefts,³⁵ the axial and appendicular skeleton,³⁶ the body wall, teeth, and heart³⁷ are also commonly found. Hyperthermia in utero (due to maternal influenza) has recently been described as a risk factor for subsequent childhood psychological/behavioral disturbances³⁸ and, more particularly, schizophrenia.³⁹ Nearly all these defects have been found in human epidemiological studies following maternal fever or hyperthermia during pregnancy. It should be emphasized that these investigations have not involved ultrasound-induced hyperthermia effects. Yet, there are data on the effects of hyperthermia and measurements of in vivo temperature induced by pulsed ultrasound, but not in human beings.^{40,41,42, and 43} These data have been widely reviewed.^{31,34,44,45, and 46} However, there is a serious lack of data that examine the effects of ultrasound while rigorously excluding other confounding factors. Two widely accepted facts are that ultrasound has the potential to elevate the temperature of the tissues being scanned,^{47,48,49, and 50} and elevated maternal temperature, whether from illness or exposure to heat, can produce teratologic effects.^{30,31,34,51,52, and 53} The major question is whether DUS can induce a harmful rise in temperature in the fetus.^{54,55, and 56} Some believe that this temperature rise is, in fact, a major mechanism for ultrasound bioeffects.^29,34 Temperature elevation in the insonated tissue can be calculated and estimated fairly accurately if the field is sufficiently well characterized.^57,58 For prolonged exposures, temperature elevations of up to 5°C have been obtained.⁵⁴ Temperature change in insonated tissues depends on the balance between heat production and heat loss. A particular tissue property that strongly influences the amount of heat transported is local perfusion, which very clearly diminishes the risk, if present. Similar experimental conditions caused a 30% to 40% lower maximal temperature increase in live versus dead sheep fetuses exposed in the near field,⁴² while in guinea pig fetuses exposed at the focus the difference was approximately 10%.⁴³ These findings were estimated to be secondary to vascular perfusion in live animals. A significant cooling effect of vascular perfusion was observed only when the guinea pig fetuses reached the stage of late gestation near term, when the cerebral vessels were well developed. In the midterm, there was no significant difference when guinea pig fetal brains were exposed, alive (perfused) or postmortem (nonperfused), in the focal region of the ultrasound beam.⁴³

In early pregnancy, under 6 weeks gestation, there appears to be minimal maternal-fetal circulation, that is, minimal fetal perfusion, which may potentially reduce heat dispersion.⁵⁹ The lack of perfusion is one reason why the spatial peak-temporal average intensity (I_SPTA) for ophthalmic applications has been kept very low, in fact much lower than peripheral, vascular, cardiovascular, and even obstetric scanning, despite the general increase in acoustic power that was allowed after 1992 (see Table 1-1). There are some similarities in physical characteristics between the early, first-trimester embryo and the eye. Neither is perfused; they can be of similar size; and protein is present (in an increasing proportion in the fetus). At about weeks 4 to 5, the gestational sac is about the size of the eye (2.5 cm in diameter), and by week 8 it is around 8 cm in diameter. This may allow whole-body fetal scanning (and possibly temperature increase), a concept that is generally ignored in the literature dealing with thermal effects of ultrasound. So is, often, the issue of transducer heating, which may be particularly relevant in the first trimester, if performing endovaginal scanning.^60,61 There are additional concerns in early gestation because of the lack of or the minimal perfusion. Only at about weeks 10 to 11 does the embryonic circulation actually linkup with the maternal circulation.⁶² There may thus be some underestimation of the actual DUS-induced temperature in early gestation, mainly because of the absence of perfusion. The perfusion issue is in addition to modifications of tissue temperature due to ambient maternal and fetal temperatures. Furthermore, motions (even very small) of the examiner's hand as well as the patient's breathing and body movements (in the case of obstetric ultrasound, both the mother and the fetus) tend to spread through the region being heated. However, for spectral (pulsed) Doppler studies, it is necessary to have the transducer as steady as possible. This is because, in general, blood vessels are small in comparison to the general organ or body size being scanned with B-mode imaging, and hand movements while performing Doppler studies will have more undesired effects on the resulting image. As described earlier, the intensity (I_SPTA) and acoustic power associated with Doppler ultrasound are the highest of all the general-use categories. Ziskin⁶³ reported that among 15,973 Doppler ultrasound examinations, the average duration was 27 minutes (and the longest 4 hours!).

There is a mathematical/physical relation between temperature elevation and several beam characteristics. The elevation is proportional to the product of the wave amplitude, length of the pulse, and PRF. Hence, manipulating any of these via instrument controls will alter the in-situ conditions. It is clear that temperature increases of 1°C are easily reached in routine scanning.⁶⁴ Elevation of up to 1.5°C were obtained in the first trimester and up to 4°C in the second and third trimesters, particularly with the use of pulsed Doppler.⁶⁵ There is a large body of literature on heat shock proteins (HSPs), the production of which is triggered by a core temperature increase and the function of which is to protect against hazardous effects of elevated temperature as well as to induce some thermotolerance, ie, the ability to withstand higher elevations than in the past, with no harmful results.⁶⁶ While their production is activated by whole-body temperature elevation, and may be speculated in ultrasound-induced thermal effects, it has not been shown to actually occur during experimental (or clinical) insonation.

Ultrasound bioeffects also occur through mechanical mechanisms.^67,68 These are interactions between the ultrasound wave and the tissue that do not cause a significant degree of temperature increase (less than 1°C above physiologic temperature). These include acoustic cavitation as well as radiation torque and force, and acoustic streaming secondary to propagation of the ultrasound waves. While included in this category, some effects are, in fact, the result of the mechanical interaction but are actually physical (shock wave) or chemical (release of free radicals) effects. Table 1-2⁹ summarizes nonthermal effects described in the literature in laboratory or animal experiments—and not in humans—which may be pertinent to fetal ultrasound.

Investigations with laboratory animals clearly indicate that nonthermal interactions of ultrasound fields with tissues can produce biological effects in vivo.⁶⁸ It is interesting to note that chemical effects of ultrasound were described more than 80 years ago!⁶⁹ Cavitation seems to be the major factor in mechanical effects⁷⁰ as it has been demonstrated to occur in living tissues under ultrasound insonation.^71,72 Two types of cavitation can be described—stable and inertial (previously defined as transient)—both of which need the presence of gas bubbles to occur. Stable cavitation indicates vibrations or small backward and forward movements with possible resulting microstreaming. Inertial cavitation indicates expansion and reduction in volume, secondary to alternating positive and negative pressures generated by the ultrasound wave. Expansion in growth is less with each cycle until collapse occurs with production of very high pressure (hundreds of atmospheres) and very elevated temperature (thousands of degrees), but on such a small area (less than 100 nanometers) and for such a brief time (few tens of nanoseconds) that it will not be felt and is very hard to measure (adiabatic reaction—occurring without the gain or loss of heat) but can produce microstreaming—a phenomenon that has been described also with no clear involvement of bubbles,^{73,74, and 75} or even release of free radicals.^76,77

Acoustic streaming is easily demonstrated by watching ultrasound-induced movements of solid-matter-containing fluids in insonated cavities . Radiation torque refers to the induction, in objects found in the acoustic field, of rotation or of the tendency to rotate. Biological effects of ultrasound in animals such as local intestinal,⁷⁸ renal,⁷⁹ and pulmonary⁸⁰ hemorrhages have been attributed to mechanical effects, although cavitation could not always be implicated. Furthermore, since gas bubbles do not seem to be present in fetal lungs or bowels (where effects have been described in neonates or adult animals), the risk from mechanical effect secondary to cavitation appears to be minimal.⁸¹ There are several other effects that do not appear to involve cavitation such as tactile sensation of the ultrasound wave, auditory response, cell aggregation, and cell membrane alteration. Hemolysis has also been reported.⁸² It seems, however, that the presence of some cavitation nuclei is necessary for hemolysis to occur. At present, there is no clear clinical indication for the use of ultrasound contrast agents (a source of cavitation nuclei, when injected into the body before ultrasound examination) in fetal ultrasound, and to date, no studies have specifically investigated the interaction of ultrasound and microbubble contrast agents in fetal tissues in vivo. Nevertheless, it should be noted that in the presence of such contrast agents, fetal red blood cells are more susceptible to lysis from ultrasound exposure in vitro.⁸³

Additionally, fetal stimulation caused by pulsed ultrasound insonation has been described, with no apparent relation to cavitation.⁸⁴ This effect may be secondary to radiation forces associated with ultrasound exposures. These forces were suspected at the earliest stages of ultrasound research⁸⁵ and are known to possibly stimulate auditory,⁸⁶ sensory,⁸⁷ and cardiac tissues.⁸⁸ No harmful effects of DUS, secondary to nonthermal mechanisms, have been reported in human fetuses. A very intriguing nonthermal effect of ultrasound is acceleration of bone fractures healing in animals and humans.⁸⁹ Because of these known effects of ultrasound in living tissues and the fact that pressures involved with Doppler propagation are much higher than B-mode, caution is further recommended, based on scientific evidence of potential effect, particularly in the first trimester.⁹⁰

In 1992, the FDA yielded to pressure from ultrasound clinical users as well as manufacturers to increase the power output of instruments. The rationale for this request was that higher outputs would generate better images, and thus improve diagnostic accuracy. To allow clinical users of ultrasound to use their instruments at higher powers than originally intended and to reflect the 2 major potential biological consequences of ultrasound (mechanical and thermal, see pp. 5-7), the American Institute of Ultrasound in Medicine (AIUM), the National Electrical Manufacturers' Association (NEMA), and the FDA (with representatives from the Canadian Health Protection Branch, the National Council on Radiation Protection and Measurements,⁹¹ and 14 other medical organizations²⁹) developed a standard related to the potential for ultrasound bioeffects. The full name was the Standard for Real-Time Display of Thermal and Mechanical Indices on Diagnostic Ultrasound Equipment, generally known as the Output Display Standard or ODS.¹⁵ The importance of this document and what it describes is that it represents historically the first attempt at providing to the end user quantitative safety-related information. One important result is that the end users are able to see how manipulation of the instrument controls during an examination causes alterations in the output and thus on the exposure. As a consequence, for fetal imaging the output, as expressed by the I_SPTA, went from a previous value of 92 mW/cm² to 720 mW/cm² (see Table 1-1).

To allow the output to reach such levels, the manufacturers were requested to display, on screen and in real-time, 2 types of indices with the intent of making the user aware of the potential for bioeffects, as described earlier. These indices are the thermal index (TI), to provide some indication of potential temperature increase, and the mechanical index (MI), to provide indication of potential for nonthermal (ie, mechanical) effects^15,29,92 (Figure 1-3). The TI is the ratio of total acoustic power to the acoustic power estimated to be required to increase tissue temperature by a maximum of 1°C. It is an estimate of the maximal temperature rise at a given exposure. There are 3 variants: for soft tissue (TIS), to be used mostly in early pregnancy when ossification is low; for bones (TIB), to be used when the ultrasound beam impinges on bone at or near the beam focus, such as late second and third trimesters of pregnancy; and for transcranial studies (TIC) when the transducer is essentially against bone, mostly for examinations in adult patients. These indices were required to be displayed if equal to or over 0.4. It needs to be made very clear that TI does not represent an actual or an assumed temperature increase. It bears some correlation with temperature rise in degrees Celsius but in no way allows an estimate or a guess as to what that temperature change actually is in the tissue.⁹² The MI represents the potential for nonthermal damage in tissues but is not based on actual in-situ measurements. It is a theoretical formulation of the ratio of the pressure to the square root of the ultrasound frequency (hence, the higher the frequency, the lesser risk of mechanical effect).

Both the TI and MI can and should be followed as an indication of change in output during the clinical examination. A clear extension of the above statements is that education of the end user is a major part in the implementation of the indices. Attempts have been made to educate the end users,⁹³ but, unfortunately, this aspect of the ODS does not seem to have succeeded as end users' knowledge of bioeffects, safety, and output indices is found lacking.^94,95 Furthermore, several assumptions were made, which prompts some questions on the clinical value of these indices. Maybe the most significant (from a clinical aspect) is the choice of the homogeneous attenuation path model (defined as the H³ model), with an attenuation coefficient of 0.3 dB/cm/MHz. The reason to employ models of that nature is the impossibility, for obvious reasons, to perform certain measurements in pregnant women. This coefficient may be an overestimation of the attenuation in many clinical scenarios, a situation that would underestimate the actual exposure. In National Council on Radiation Protection and Measurements (NCRP) report number 140,²⁹ there is an entire chapter (Chapter 9) indicating conditions where both indices may be inaccurate, eg, long fluid path (full bladder, amniotic fluid, ascites, or hydrocephalus) or path through increased amounts of soft tissue such as obese patients. Because of these uncertainties, the accuracy of the TI and MI may be within a factor of 2 or even 6.⁹⁶ For example, an on-screen TI of 1 may correspond to an actual value of 0.5 or 2 if the error factor is 2, but possibly 0.33 or 6, if the error factor is 6 (as previously stated, these are not actual temperature indications). A further disturbing and confusing element is that outputs reported by manufacturers are not necessarily equivalent to those calculated in the laboratory.⁹⁷

Risk means the chance or the possibility of loss or bad consequence. It refers to the possibility, with a certain degree of probability, of damage to health, environment, and objects, in combination with the nature and magnitude of the damage.⁹⁸ These are the 3 important characteristics of risk: probability of occurrence, and nature and magnitude of harm. It has been, specifically, applied to the use of medical instruments.⁹⁹ A complicating factor that makes definition and classification difficult is that the concept of risk means various things to different people. Age, background, education, morals, religion, and many other traits will direct this evaluation and not only the absolute possible result of the activity, putting the participant at risk. For instance, in bungee jumping, rupture of the elastic cord and subsequent death may be, indisputably, the worst possible outcome but different people evaluate this and make decisions that are not necessarily based on this absolute result. Furthermore, the reason to take a possible risk has to be investigated.

Two approaches are possible in risk evaluation: how much harm is acceptable to obtain the desired results (risk-benefit ratio) or how much harm can be avoided by withholding the action or modifying it (the precautionary principle). The risk-benefit principle is what is almost universally used in medicine to justify a medical diagnostic procedure (such as ultrasound) or a therapeutic intervention. If the benefit to be obtained from the procedure in terms of diagnosis (ultrasound) or intervention (a newly discovered and not yet commercialized cancer or AIDS drug, for instance) is deemed to be sufficient, then even if this diagnostic or interventional procedure carries some risks (recognized or presumed to be possible), the benefit overrides these risks, assuming the subject understands those risks and is willing to take them. The precautionary principle (PP) is a diametrically opposed ethical, political, and economic approach stating that if a certain action may cause severe damage to the public, in the absence of a scientific consensus that harm would not ensue, the burden of proof falls on those who would advocate taking that action.¹⁰⁰ This principle is much less familiar to the medical field, although "first do no harm" is its direct application, but it may be extremely relevant when considering safety and risks of a procedure, such as prenatal ultrasound. The concept originated in the 19th century when John Snow, a London, UK, physician, determined that cholera was due to the extensive, common use of an unclean water supply and recommended closing of this source of water, although it was the sole one in a large vicinity.¹⁰¹ This may have been the first epidemiological analysis of a disease. Although the beginning of the PP was medical, it became a social idea in Germany in the 1930s as Vorsorge, "forecaring." This later became the Vorsorgeprinzip, the forecaring or precautionary principle, in West German environmental law in the 1970s.¹⁰² The idea was adopted by decision and policy makers but, remarkably, much more extensively in Europe than in the United States. Some key concepts in the original formulation were environmental harm to a population and responsibility: "When an activity raises threats of harm to human health or the environment, precautionary measures should be taken even if some cause and effect relationships are not fully established scientifically. In this context the proponent of an activity, rather than the public, should bear the burden of proof" (the Wingspread Statement on the Precautionary Principle¹⁰³). From environmental research it spread to toxicology and was first applied only recently in the United States to a clinical medical field.¹⁰⁴ However, several medical mishaps clearly belong to the history of the development of the PP—from the diethylstilbestrol debacle¹⁰⁵ to the thalidomide tragedy.¹⁰⁶ While referring mostly to environmental issues, such as global warming, the PP can certainly be extended to other medical activities (such as diagnostic ultrasound) and be applied to individuals (such as fetuses). The simple enunciation of the principle, particularly in reference to diagnostic ultrasound in general and entertainment ultrasound in particular, is that even if a particular action or procedure has not been proved to be harmful, it's better to avoid it so as not to take the risk until safety is established through clear, scientific evidence, popularly expressed as "better safe than sorry."¹⁰⁷ This is also the basis of the Hippocratic Oath, which includes the recommendation to first do no harm. A major difference with the risk-benefit principle is that proponents of the PP believe that public action is necessary if there is any evidence of likely or substantial harm, however limited but plausible, and the burden of proof is shifted from showing the presence of risk to demonstrating its absence.¹⁰⁸ As such, epidemiologic research on chronic diseases and the use of surrogates for human studies (eg, animal research or tissue cultures) have been shown to be uncertain.¹⁰⁹ There are several variations of the PP, but all have some common key elements:

1. There must be scientific uncertainty about nature of harm, probability, magnitude, and causality (fulfilled by DUS).

2. Mere speculation is not enough to invoke the PP. Scientific analysis must have triggered the process (also fulfilled by DUS).

3. Per definition, the PP deals with procedures with probability of unclear outcome, in that it is different from prevention or from risk-benefit assessment where some clear knowledge or precise suspicion exists, and where decision may be made to go ahead despite this risk by, for instance, taking additional measures to attempt and limit the danger. Clearly, the ALARA principle is the exact application of this element^110,111 (fulfilled by DUS).

4. In general, the PP applies to unacceptable ("serious," "irreversible," "global") high levels of risk to large populations, present or future, local or distant¹¹² (may not be the case for DUS).

5. One needs to intervene (not observe or procrastinate) before damage has been demonstrated (eg, "do not perform DUS").

6. The intervention must be proportional to the possible risk: indicating DUS may be acceptable but not nonclinical use of DUS. A level of "zero risk" is probably never attainable.

Those who support the PP make the following very strong argument for precaution: serious damage may be caused if one uses a risk-based approach. A well-known example is what constitutes toxic levels of lead in paint. As early as 1897, it was known that lead may be toxic, but at first the upper limit of safety for children was assumed to be 60 μg/dL of blood, and this had terrible results. The "safe" level was reduced over the years to 40, then 20, then 10, which it is today, although some scientists feel that even 2 μ/dL may pose some risk.¹¹³ The basic conclusion of risk analysis with the PP is that measures against a possible risk should be taken (such as exposure avoidance) even if the available evidence is weak (or maybe absent) regarding the existence of that risk as a scientifically established fact.¹¹⁴ In many European countries this "stop first then study" approach (a clear application of the PP) has been adopted (particularly for chemicals). The exact opposite is often true in the United States where something, once introduced, has to be proven harmful by science before being removed or forbidden. A major goal of the PP is to help delineate (preferably quantitatively) the possibility that some exposure is hazardous, even in cases where this is not established beyond reasonable doubt.¹¹⁵ The classical statistical approach to hypothesis testing is unhelpful, because lack of significance can be due to either uninformative data or genuine lack of effect (type II error).¹¹⁶

There are many critics of the PP because of the risk of exaggeration in caution and slowing down of scientific progress.^117,118 A major issue is that the PP relies very heavily on a single conjecture: prevention is better than cure. There is no scientific evidence for this. Furthermore, it may be true that often it is better to be "safe than sorry" and the primum non nocere (first do no harm) principle is a direct application of this, but preventative measures can be long lasting and possibly incapacitating, whereas cures can be targeted and effective.¹¹⁷ What is more, no moral opinion is formed of people when treating them, but if the main focus is upon precaution, then it can be deemed morally wrong not to take preventative measures. The whole precaution approach is imbued with what may appear to many as an excessively moralistic tone.¹¹⁹ Furthermore, the probability of a problem occurring that one tries to avoid has to be high (which does not apply, as far as we know, to ultrasound) and preventative measures have to be effective. Hence this approach may be adopted with some restrictions and this is, in fact, exactly what ALARA recommends.¹¹¹ Most scientists and professional organizations have recommended such a practice in clinical obstetrical ultrasound.^{120,121, and 122}

The first descriptions of ultrasound as an imaging mode date from the 19th century.¹²³ The French engineer Paul Langevin designed an ultrasound machine, using Pierre Curie's principle of the piezoelectric effect. During World War I, he attempted to use this instrument to detect submarines through echo location (hence the later coined term SONAR: SOund Navigation And Ranging). He also demonstrated that the waves produced by his machine could kill small animals in an insonated water bath, and could cause pain to his assistants when they were required to plunge their hands in the water bath in the path of the beam. Other bioeffects observed included the searing of skin when touching a resonant quartz bar, and explosive atomization (!) of fluid drops from the end of the rod. Since that time, the question of effects and safety has been on the minds of researchers⁸⁵ and has given rise to literature too extensive to review in detail.^{2,3,6,46,120,124,125,126,127,128,129,130,131,132,133,134,135, and 136} Initially, cell suspensions and cell and tissue cultures were employed and many reports described clear effects of the ultrasound waves on these, mostly secondary to cavitational and other nonthermal mechanisms, such as cell aggregation,¹³⁷ membrane damage,¹³⁸ and cell lysis,¹³⁹ among others. Plants were another extensively studied organism for effects of ultrasound,¹⁴⁰ particularly the Elodea leaf, since internal gas channels are present.¹⁴¹ Insects have been exposed to ultrasound with significant effects, such as death of eggs and larvae as well as abnormal development, presumably secondary to the presence of gas-filled channels.¹⁴² Additionally, alterations at the chromosomal and even DNA levels have been described.¹⁴³ The above effects have been reviewed extensively elsewhere,^5,29 and while of major scientific and historical importance, are not of major relevance to clinical exposure of human fetuses.

Animal Research

Effects of ultrasound were demonstrated in animals more than 80 years ago.⁸⁵ Since then, multiple studies have been performed with ultrasound on a wide variety of species. Studies of gross effects on the brain and liver of cats were first performed with well-defined lesions and demyelination in the brain¹⁴⁴ and tissue damage in the liver,¹⁴⁵ resulting from ultrasound exposure of a few seconds at 1 and 3 MHz, respectively. Other observed effects include limb paralysis as a result of spinal cord injury in the rat,^146,147 as well as lesions in the liver, kidney, and testicles of rabbits.¹⁴⁸ While some effects are likely due to mechanical influences, very high temperature elevations (much higher than anything reachable with diagnostic ultrasound) have also been observed and may be more directly involved with the tissue damage. Effects in muscles have been obtained, but with outputs much higher than those usually generated in clinical studies,¹⁴⁹ and so have intestinal⁷⁸ and lung¹⁵⁰ hemorrhages, also at acoustic pressures well above those generated by ultrasound fields. These are helpful in understanding the mechanisms involved with possible bioeffects of DUS. It should also be noted that some similar effects have also been demonstrated with acoustic fields much closer to clinically pertinent ones, in particular lung and intestinal hemorrhage.⁷⁸ Several major clinical end-points for bioeffects that could have direct relevance to human studies include fetal growth and birth weight, effects on brain and CNS function, and change in hematological function, and will be considered in more detail. Decreased birth weight after prenatal exposure to ultrasound has been reported in the monkey^151,152 and the mouse,^153,154 but not convincingly in the rat.¹⁵⁵ Therefore, clear species differences seem to exist,¹⁵⁶ making it difficult to generalize, and even more difficult to extrapolate to humans.

Tarantal and Hendrickx¹⁵¹ evaluated 30 pregnancies in monkeys, half of which were exposed to ultrasound. The scanned fetuses had lower birth weights and were shorter than the control group. No significant differences were noted between the groups with regard to the rate of abortions, major malformations, or stillbirths. Moreover, all showed catch-up growth when examined at 3 months of age.¹⁵¹ It should be noted that in-situ intensities were higher than what is considered routine in clinical obstetrical imaging in humans. Hande and Devi¹⁵⁷ evaluated the effect of prenatal exposure to diagnostic ultrasound on the development of mice. Swiss albino mice were exposed to diagnostic ultrasound for 10 minutes on day 3.5 (preimplantation period), 6.5 (early organogenesis period), or 11.5 (late organogenesis period) of gestation. Sham-exposed controls were maintained for comparison. Fetuses were dissected out on the 18th day of gestation, and changes in total mortality, body weight, body length, head length, brain weight, sex ratio, and microphthalmia were recorded. Exposure on day 3.5 of gestation resulted in a small increase in the resorption rate and a significant reduction in fetal body weight. Low fetal weight and an increase in the incidence of intrauterine growth-restriction were produced by exposure on day 6.5 postcoitus.¹⁵⁷

Others have also demonstrated restricted growth in newborns after in-utero exposure to DUS.¹⁵⁸ Subtler findings have also been described. Pregnant Swiss albino mice were exposed to diagnostic ultrasound (3.5 MHz, 65 mW, I_SPTP = 1 W/cm², I_SATA = 240 W/cm²) for 10, 20, or 30 minutes on day 14.5 (fetal period) of gestation.¹⁵⁹ Sham-exposed controls were studied for comparison. There were significant alterations in behavior in the exposed groups as revealed by decreased locomotor and exploratory activity, and an increase in the number of trials needed for learning. No changes were observed in physiological reflexes and postnatal survival. The authors concluded that ultrasound exposure during the early fetal period can impair brain function in the adult mouse.¹⁵⁹ Likewise, Hande et al¹⁶⁰ found that anxiolytic activity and latency in learning were more noticeable in ultrasound-treated animals. The authors exposed pregnant Swiss mice to diagnostic levels of ultrasound (3.5 MHz, maximum acoustic output: I_SPTP = 1 W/cm² and I_SATA = 240 mW/cm², acoustic power = 65 mW) for 10 minutes on postcoital day 11.5 or 14.5. At 3 and 6 months postpartum, offspring were subjected to behavioral tests. The effect was more pronounced in the 14.5 days postcoital group than in the 11.5 days group. They concluded that exposure to diagnostic ultrasound during late organogenesis period or early fetal period in mice may cause changes in postnatal behavior.¹⁶⁰ Temperature elevations were induced by ultrasound in guinea pig fetal brains.⁴³ In fact, mean temperature increases of 4.9°C close to parietal bone and 1.2°C in the midbrain were recorded after 2-minute exposures, albeit at exposure conditions higher than what is usually employed in clinical examinations.⁴³ This greatest temperature rise recorded close to the skull correlated with both gestational age and progression in bone development.⁴⁰ The skull bone becomes progressively thicker and denser between 30 and 60 days' gestational age (normal gestation for guinea pigs is 66 to 68 days). After only 2 minutes of insonation with an I_SPTA of 2.9 W/cm² (about 4 times higher than currently permitted by the FDA for diagnostic use), mean maximum temperature increases varied from 1.2°C at 30 days to 5.2°C at 60 days. It is important to note that most of the heating (80% of the mean maximum temperature increase) occurred within 40 seconds. The rate of heating is relevant to the safety of clinical examinations in which the dwell time may be an important factor. Because maximal ultrasound-induced temperature increase occurs in the fetal brain near bone, worst-case heating will occur later in pregnancy, when the ultrasound beam impinges on bone, and less will occur earlier in pregnancy, when bone is less mineralized. However, milder insults early in gestation may be as significant (or more) than more severe ones in later stages.

Neurons of the cerebral neocortex in mammals, including humans, are generated during fetal life in the brain proliferative zones and then migrate to their final destinations by following an inside-to-outside sequence. Recently, Ang et al¹⁶¹ evaluated the effect of ultrasound waves on neuronal positioning within the embryonic cerebral cortex in mice. Neurons generated at embryonic day 16 and destined for the superficial cortical layers were chemically labeled in over 335 animals. A small but statistically significant number of neurons failed to acquire their proper position and remained scattered within inappropriate cortical layers and/or in the subjacent white matter when exposed to ultrasound for a total of 30 minutes or longer during the period of their migration. The magnitude of dispersion of labeled neurons was variable but systematically increased with duration of exposure to ultrasound (although not linearly, with some extended exposure yielding less effect than lower ones). These investigators concluded that further research in larger and slower-developing brains of nonhuman primates and continued scrutiny of unnecessarily long prenatal ultrasound exposure is warranted. It is unclear whether a relatively small misplacement in a relatively small number of cells that retain their origin cell class is of any clinical significance. It is also important to note that there are several major differences between the experimental setup of Ang et al¹⁶¹ and the clinical use of ultrasound in humans.⁶ The most noticeable difference was the length of exposure of up to 7 hours in the setup of Ang et al. No real mechanistic explanation was given for the findings, and furthermore, there was no real dose effect with high effects at the penultimate high dose, but less so at the highest dose. Moreover, scans were performed over a small period of several days. The experimental setup was such that embryos received whole-brain exposure to the beam, which is rare in humans. In addition, brains of mice are much smaller than those in humans, and develop over days. This should not completely deter from the study, but encourages caution. It should be noted that some have described a complete lack of effects of prenatal ultrasound exposure on postnatal development and growth¹⁶² or behavior.¹⁶³ Another recently published study is worth considering.¹⁶⁴ Chick brains were exposed, in ovo, on day 19 of a 21-day incubation period to B-mode (5 or 10 min), or to pulsed Doppler (1, 2, 3, 4, or 5 min) ultrasound. After hatching, learning and memory function were assessed at day 2 post-hatch. B-mode exposure did not affect memory function. However, significant memory impairment occurred following 4 and 5 min of pulsed Doppler exposure. Short-, intermediate-, and long-term memory was equally impaired, suggesting an inability to learn. Chicks were also unable to learn with a second training session. In this study, exposure to pulsed Doppler ultrasound adversely affected cognitive function in chicks. Although some methodological issues exist and extrapolation to humans is unwarranted, these findings justify further investigations.

The hematological system is the second major system to be investigated for ultrasound effects. The following have been assessed: hemolysis, coagulation factors and platelets, and leukocyte production and function.¹⁶⁵ Increased hemolysis has been demonstrated for ultrasound in (human) fetal cells as compared to adult cells, but only in the presence of ultrasound contrast agents, with human cells being less fragile than certain tested animals.^83,166 Other alterations have been described in the hemolytic system¹⁶⁷ but appear to be of minimal, if any, clinical significance.

Human Research and Epidemiology

In 2005, the American Institute of Ultrasound in Medicine (AIUM) published the following statement: "Based on the epidemiological data available and on current knowledge of interactive mechanisms, there is insufficient justification to warrant a conclusion of a causal relationship between diagnostic ultrasound and recognized adverse effects in humans. Some studies have reported effects of exposure to diagnostic ultrasound during pregnancy, such as low birth weight, delayed speech, dyslexia, and non–right-handedness. Other studies have not demonstrated such effects. The epidemiological evidence is based on exposure conditions prior to 1992, the year in which acoustic limits of ultrasound machines were substantially increased for fetal/obstetrical applications."¹⁶⁸ Applied to ultrasound, epidemiology is the study of effects on human populations as a result of ultrasound scanning and, in the case of obstetrical ultrasound, this should include the pregnant patient as well as her infant. Laboratory animal experiments under similar diagnostic exposure levels have shown some effects from ultrasound, under certain conditions. Effects have also been reported in humans, but a definitive statement regarding risk should, ideally, include direct analysis of the effects in human populations. Several epidemiological studies have been published.^4,46,169 For an extensive discussion, including elements of statistics, see Chapter 12 in NCRP report number 140,²⁹ an extensive review by Newnham,¹³² and AIUM document, Conclusions Regarding Epidemiology for Obstetric Ultrasound.^170,171. Relevant details will be summarized.

Several biological end points have been analyzed in the human fetus/neonate in an attempt to determine whether prenatal exposure to diagnostic ultrasound had observable effects: intrauterine growth restriction (IUGR) and low birth weight, delayed speech, dyslexia, neurological and mental development or behavioral issues, and, more recently, non–right-handedness. Occasional studies report an association between diagnostic ultrasound and some specific abnormalities such as lower birth weight,¹⁶⁹ delayed speech,¹⁷² dyslexia,¹⁷³ and non–right-handedness.^174,175 With the exception of low birth weight (also demonstrated in monkeys,¹⁶⁶) these findings have never been duplicated, and the majority of studies have been negative for any association. Moore et al¹⁷⁶ examined a large number of infants (over 2000, half of them exposed to ultrasound) and found a small but statistically significant lower mean birth weight of exposed versus non-exposed infants. However, information was collected several years after exposure, no indications for the examination are known, and no exposure information is available. This is very often the major problem in analyzing these reports. In a later study, the authors concluded that the relationship of ultrasound exposure and reduced birth weight may be due to shared common risk factors, which lead to both exposure and a reduction in birth weight.¹⁷⁷ Another retrospective study, with Moore as a coauthor, reported a 2.0 greater risk of low birth weight after 4 or more exposures to diagnostic ultrasound.¹³³ These results were not reproduced in other retrospective studies.¹⁷⁶ In a large study (originally 10,000 pregnancies exposed to ultrasound matched with 500 controls) with a 6-year follow-up, Lyons et al¹⁷⁸ did not find differences in birth weight (nor increased congenital malformations, chromosomal abnormalities, infant neoplasms, speech or hearing impairment, or developmental problems).

Newnham et al¹⁷¹ performed a randomized control trial including more than 2800 parturients. Of these, about half received 5 ultrasound imaging and Doppler flow studies at 18, 24, 28, 34, and 38 weeks gestation, and half received a single ultrasound imaging at 18 weeks. They found an increased risk of IUGR when exposed to frequent Doppler examinations, possibly via some effects on bone growth. However, when children from the last mentioned study were examined at 1 year of age, there were no differences between the study and control groups. In addition, after examining their original subjects after 8 years, no evidence of long-term adverse impact in neurological outcome was noted by the same group.¹⁷⁹ Similarly, no harmful effect of a single or 2 prenatal scans on growth were found in several randomized studies.^180,181 In fact, in some studies, birth weight was slightly higher in the scanned group, but not significantly so, except in one.¹⁸² In conclusion, decreased birth weight has been extensively analyzed after DUS exposure in utero, and it does not appear that such exposure is associated with reduced birth weight, although Doppler exposure may have some risks.¹³⁶ In a few studies that appear to favor such an effect, a major problem is that there is an important confounding factor: many studies include pregnancies at risk for IUGR due to existing maternal or fetal conditions.

A second major potential effect extensively evaluated is delayed speech. In an attempt to determine if there is an association between prenatal ultrasound exposure and delayed speech in children, Campbell et al¹⁷² studied 72 children with delayed speech and found a higher rate of ultrasound exposure in utero than the 144 control subjects. Some issues render these results less valid: there was neither a dose-response effect nor any relationship to time of exposure, and many of the records were more than 5 years old. Another study of over 1100 children exposed to ultrasound in utero and over 1000 controls found no significant differences in delayed speech, limited vocabulary, or stuttering.¹⁸³

Dyslexia is another widely studied subject. In one study over 4000 children, aged 7 to 12, exposed to ultrasound in utero were used as a study group and compared to matched controls to evaluate the appearance of adverse effects.¹⁷³ Seventeen outcomes measures were examined, at birth (APGAR scores, gestational age, head circumference, birth weight, length, congenital abnormalities, neonatal infection, and congenital infection) or in early infancy (hearing, visual acuity and color vision, cognitive function, and behavior). No significant differences were found, except for a significantly greater proportion of dyslexia in those children exposed to ultrasound. However, the authors indicated that this could be an incidental finding, given the design of the study and the presence of several confounding factors that could have contributed to the possible dyslexia finding. On the other hand, it should be noted that exposure conditions were probably much lower than modern ultrasound systems, given that the fetal examinations were performed from 1968 to 1972. Subsequently, a long-term follow-up study was performed on over 2100 children.^180,184 End points included evaluation for dyslexia along with additional hypotheses including an examination of non–right-handedness (see below) said to be associated with dyslexia. These studies^{185,186 ans 187} included the specific examination of more than 600 children with various tests for dyslexia such as spelling and reading. No statistically significant differences were found between ultrasound-exposed children and controls for reading, spelling, arithmetic, or overall performance as reported by teachers. Specific dyslexia tests showed similar rates of occurrence among scanned children and controls in reading, spelling, and intelligence scores, and no discrepancy between intelligence and reading or spelling. Since the original finding of dyslexia was not confirmed in subsequent randomized controlled trials, it is considered unlikely that routine ultrasound screening exams can cause dyslexia. However, these studies did raise the issue of laterality (in terms of handedness), which is discussed below.

The issue of non–right-handedness as a result of prenatal exposure has caused much ink to be used in extensive discussions and reports. The first report of a possible link between prenatal exposure to ultrasound and subsequent non–right-handedness in insonated children was published in 1993 by Salvesen et al,¹⁸⁵ but according to the authors, "only barely significant at the 5% level." In a later analysis of the data, they described that the association was restricted to males.¹⁸⁶ A second group of researchers (with Salvesen, the main author of the first study, included but with a new population, in Sweden as opposed to Norway) published similar findings of a statistically significant association between ultrasound exposure in utero and non–right-handedness in males.¹⁷⁴ Salvesen then published a meta-analysis of these 2 studies and of previously unreported results.¹⁷³ No difference was found in general, but a small increase in non–right-handedness was present when analyzing boys separately. No valid mechanistic explanation is given in the studies to explain the findings. In conclusion, although there may be a small increase in the incidence of non–right-handedness in male infants, there is not enough evidence to infer a direct effect on brain structure or function or even that non–right-handedness is an adverse effect.

Other end points that have been considered but not found to be associated with ultrasound exposure include congenital malformations and malignancies.¹⁸⁹

There has been no epidemiological study published on populations scanned after 1992, when regulations were altered and acoustic output of diagnostic instruments were permitted to reach levels many times higher than previously allowed (from 94 to 720 mW/cm² I_SPTA for fetal applications). There are no epidemiological studies related to the output display standard (thermal and mechanical indices) and clinical outcomes. The safety of new technologies such as harmonic imaging and three-dimensional (3D) ultrasound, as well as that of probe self-heating, needs to be investigated.

Clinical Exposimetry

Unfortunately there is no way to perform actual sonographic exposure measurements in the human fetus. Pressure, intensity, and power are not measured in situ, but are estimated from laboratory obtained measurements. Several tissue models have been developed to help with this estimation, depending mostly on approximate attenuation coefficients for various tissues or beam paths.^29,47 A large range of variation is expected secondary to individual patient characteristics, such as weight and thickness of tissues.¹⁹⁰ Because of these possible variations, the reasonable worst-case scenario is usually considered. There are scarce data on instruments' acoustic output (nor patient acoustic exposure) for routine clinical ultrasound examinations. Acoustic output was recorded in several prospective observational studies investigating first-trimester ultrasound,¹⁹¹ Doppler studies,¹⁹² and 3D/four-dimensional (4D) studies.¹⁹³ Basically, first-trimester ultrasound was associated with very low TI values (with a mean of 0.2 ± 0.1).¹⁹⁰ The TI was significantly higher in the pulsed wave Doppler (mean 1.5 ± 0.5, range 0.9-2.8) and color flow imaging studies (mean 0.8 ± 0.1, range 0.6-1.2) as compared to B-mode ultrasound (mean 0.3 ± 0.1, range 0.1-0.7; P < .01).¹⁹⁰ In the same study, TI was above 1.5 in 43% of the Doppler studies.¹⁹² Mean TI during the 3D (0.27 ± 0.1) and 4D examinations (0.24 ± 0.1) was comparable to the TI during the B-mode scanning (0.28 ± 0.1; P = .343).¹⁹³ Figures 1-3,1-4,1-5,1-6, and 1-7 are examples of actual screen shots during clinical exams, for B-mode, color Doppler, M-mode, spectral Doppler, and 3D acquisition, respectively. Figure 1-8 demonstrates that extremely elevated TIs are easily reachable with spectral Doppler, although in manufacturer's fetal setting.

The other side of the equation is, "What are we looking for?" Ultrasound is neither radiation nor thalidomide, and it is certain that ultrasound does not kill fetuses, does not cause limb amputations, and does not cause gross structural anomalies.¹⁹⁴ But are we looking where we should, and have we studied enough cases in a scientific fashion, looking at subtle changes? The answer is clearly, "No." We have been looking for macroscopic, gross findings and have not found any, but is it possible that harmful effects of ultrasound have been missed because the wrong time frame reference was used? Two possible factors are described for such errors.¹⁹⁵ If one uses a term human pregnancy (280 days [40 weeks]) to life expectancy of 70 years (25,550 days) ratio, then 7 in-utero days are comparable to about 631 ex-utero days. Therefore, it is conceivable that a much shorter time interval (1 day) should be used to group fetuses to evaluate effects, not intervals of 1 or more weeks as is usually done. Furthermore, there is also a potential "dilution error." Assuming an event has a background rate of 10% in the general population but occurs in 100% of fetuses exposed on day 35, if a large number (for instance, 1000) of fetuses exposed on that particular day are examined, the incidence will be 100%, ie, 90% increase over the control population (background rate of 10%). But if we assume 1000 fetuses are exposed per day for 12 weeks, this represents 84,000 scans, and only 11.1% will be affected (all 1000 scanned on day 30 and 10% [the background rate] of all 83,000 others [8300], or 9300 total), an increase of only 1.1% (1.07 to be precise) over the background rate of 8400 (10% of 84,000), which is very difficult to extract and observe, but still present in 100% of the fetuses exposed at the critical time (day 35 in the above example). The actual numbers are probably even more complicated since more than 1000 fetuses are scanned every day, the background rate of major anomalies is 3% to 4% in the general population and much lower for non-gross macroscopic findings, and furthermore, the hit rate of any teratological agent is rarely 100%. This points to the need for extensive, well-planned research—a goal very difficult to accomplish, given that the majority of pregnant women who receive prenatal care will have 1 or several DUS scans during their pregnancy.

Nonmedical Ultrasound

Nonmedical ultrasound refers to the performance of obstetrical ultrasound with no medical indication but to provide the mother/parents-to-be with images or video clips of the fetus (on hard copy, tape, CD, or DVD), also called "scanning for pleasure."¹⁹⁶ There are several reasons why most official organizations are opposed to this practice, such as issues of training of the providers, quality and nature of the scans, feedback to the "customers," and risks that these customers will not have a regular, clinical exam. But perhaps the most obvious reason for the resistance to these scans is the safety issue. For instance, the FDA is strongly opposed, stating, "… ultrasound energy delivered to the fetus cannot be regarded as completely innocuous. Laboratory studies have shown that diagnostic levels of ultrasound can produce physical effects in tissue, such as mechanical vibrations and rise in temperature. Although there is no evidence that these physical effects can harm the fetus, public health experts, clinicians and industry agree that casual exposure to ultrasound, especially during pregnancy, should be avoided."¹⁹⁷ The FDA goes further and indicates, "Persons who promote, sell or lease ultrasound equipment for making 'keepsake' fetal videos should know that the FDA views this as an unapproved use of a medical device. In addition, those who subject individuals to ultrasound exposure using a diagnostic ultrasound device (a prescription device) without a physician's order may be in violation of state or local laws or regulations regarding use of a prescription medical device."^197,198 Equally opposed to the nonclinical use of DUS are the American Institute of Ultrasound in Medicine (AIUM), the American College of Obstetrics and Gynecology (ACOG), and the European Committee for Medical Ultrasound Safety (ECMUS). The AIUM's most recent statement is, "The AIUM advocates the responsible use of diagnostic ultrasound … [and] strongly discourages the non-medical use of ultrasound. … The use of either two-dimensional (2D) or three-dimensional (3D) ultrasound to only view the fetus, obtain a picture of the fetus or determine the fetal gender without a medical indication is inappropriate and contrary to responsible medical practice. Although there are no confirmed biological effects on patients caused by exposures from present diagnostic ultrasound instruments, the possibility exists that such biological effects may be identified in the future. Thus ultrasound should be used in a prudent manner to provide medical benefit to the patient."¹⁹⁹ Similarly, the ECMUS's statement includes the following: "The embryonic period is known to be particularly sensitive to any external influences. Until further scientific information is available, investigations should be carried out with careful control of output levels and exposure times. With increasing mineralization of the fetal bone as the fetus develops the possibility of heating fetal bone increases."²⁰⁰ More recently the World Federation of Ultrasound in Medicine and Biology (WFUMB) and the International Society of Ultrasound in Obstetrics and Gynecology (ISUOG) issued a joint statement with identical conclusions: "The WFUMB and ISUOG disapprove of the use of ultrasound for the sole purpose of providing souvenir images of the fetus. … Furthermore, ultrasound should be employed only by health professionals who are well trained and updated in ultrasound clinical usage and bioeffects."²⁰¹

Official Positions

Many national and international organizations or societies have issued official statements regarding the epidemiology, bioeffects, and safety of ultrasound, as well as the nonmedical usage of ultrasound such as the AIUM, WFUMB, British Medical Ultrasound Society (BMUS), and European Committee of Medical Ultrasound Safety (ECMUS). They all state, in one way or another, that ultrasound appears safe if performed for clinical indications by appropriately trained personal, but that prudence is recommended because of the possibility of yet unknown deleterious effects. For instance, the AIUM has several statements available on its Web site for epidemiology,¹⁶⁷ prudent use,¹⁹⁹ and keepsake fetal imaging.²⁰¹ The Keepsake Fetal Imaging statement contains a clear "safety clause" particularly addressing pulsed Doppler: "Although the general use of ultrasound for medical diagnosis is considered safe, ultrasound energy has the potential to produce biological effects. Ultrasound bioeffects may result from scanning for a prolonged period, inappropriate use of color or pulsed Doppler ultrasound without a medical indication, or excessive thermal or mechanical index settings."²⁰¹

The sonographer and sonologist are interested in knowing how to keep the examination safe. One needs to provide recommendations based on scientific evidence. This is a difficult task. In terms of clinical exposure, what should be recommended? A general recommendation is that DUS should be used only when indicated, and minimal exposure should be used to obtain the diagnostic images. Furthermore, exposure time should be kept as short as possible.¹¹²

Several organizations have actually published recommendations, based more or less on scientific data. The most rigorous is the BMUS. Their 1999 statement declares, "For equipment for which the safety indices are displayed over their full range of values, the TI should always be less than 0.5 and the MI should always be less than 0.3^*."

When the safety indices are not displayed, T_max should be less than 1°C and MI_max should be less than 0.3. Frequent exposure of the same subject is to be avoided."²⁰⁰ They have very strict recommendations for maximum allowed exposure time, depending on the TI (Table 1-3). In 2009 they updated their recommendations and these are, at the moment, the most detailed guidelines for safe use of DUS in medicine in general and obstetrics in particular (http://www.bmus.org/policies-guides/pg-safety03.asp).

The WFUMB offers some scientific rationalization, stating that diagnostic exposure resulting in a temperature rise of no more than 1.5°C above normal physiological levels (37°C) may be used clinically without reservation on thermal grounds. Furthermore, diagnostic exposure that elevates embryonic and fetal in situ temperature above 41°C (4°C above normal temperature) for 5 minutes or more should be considered potentially hazardous.⁴⁵ In addition, in febrile patients, extra precaution may be needed to avoid unnecessary additional embryonic and fetal risk from ultrasound examinations. Precautions are much softer regarding mechanical phenomena, which, in the absence of gas nuclei (as is the case in fetal lungs and bowels and assuming no use of contrast agents) are probably negligible.

Scientists continue to be interested in biophysics of ultrasound and remain worried about potential harmful effects. Hence, research in this area is continuing. Ideally, epidemiological studies should be performed on large populations, blindly randomizing 50% to ultrasound testing and 50% to no testing. Given the extensive indications for DUS in pregnancy and the fact that most (and in certain countries, all) pregnant patients are referred for an ultrasound examination, this would be extremely difficult to realize in a human population. More accurate techniques to measure in-vivo real exposure may appear, allowing more precise assessment of safety, possibly by generating actual safety indices. In the meantime, areas of uncertainty persist and caution is justified, particularly in Doppler mode, early in pregnancy, but also when insonating the fetal skull for relatively long periods. Education of the end users will continue to be vital to maintaining the good safety record of ultrasound and preventing possible harmful bioeffects.

Highlighted References