Ultrasound safety in early pregnancy: reduced energy setting does not compromise obstetric Doppler measurements

Authors

  • R. K. Sande,

    Corresponding author
    1. Clinical Fetal Physiology Research Group, Department of Clinical Medicine, University of Bergen, Bergen, Norway
    2. Fetal Medicine Unit, Department of Obstetrics and Gynecology, Haukeland University Hospital, Bergen, Norway
    • Clinical Fetal Physiology Research Group, Department of Clinical Medicine, University of Bergen, Bergen, Norway

    Search for more papers by this author
  • K. Matre,

    1. Institute of Medicine, University of Bergen, Bergen, Norway
    Search for more papers by this author
  • G. E. Eide,

    1. Centre for Clinical Research, Haukeland University Hospital, Bergen, Norway
    2. Department of Public Health and Primary Health Care, University of Bergen, Bergen, Norway
    Search for more papers by this author
  • T. Kiserud

    1. Clinical Fetal Physiology Research Group, Department of Clinical Medicine, University of Bergen, Bergen, Norway
    2. Fetal Medicine Unit, Department of Obstetrics and Gynecology, Haukeland University Hospital, Bergen, Norway
    Search for more papers by this author

Abstract

Objectives

We hypothesized that first-trimester Doppler ultrasonography can be carried out at lower output energies than the currently advocated limits without compromising clinically important information.

Methods

We recruited 42 pregnant women for an ultrasound examination at 12 weeks' gestation. Twenty-one women were examined with a transvaginal transducer, the rest with a transabdominal transducer. We used pulsed Doppler to measure pulsatility index (PI) and peak systolic velocity (PSV) in five clinically relevant fetal and maternal blood vessels. The energy indicator thermal index for bone (TIb) was set at 1.0, 0.5 and 0.1. Each measurement was repeated three times. A mixed linear regression model accounting for correlation between measurements was used to assess the effect of different TIb levels and transducers.

Results

We were able to visualize the vessels by color Doppler and measure PI and PSV in all vessels at all energy levels in all the participants with the exception of the ductus venosus in two participants, yielding 1872 recordings for statistical analysis. A reduction in TIb from 1.0 to 0.5 and 0.1 had no effect on the PI or PSV values, nor was there any trend towards higher parameter variance with decreasing TIb. There was no difference between measured values of PI and PSV between the transducers, but the transabdominal technique was associated with a greater parameter variance.

Conclusion

Reliable first-trimester Doppler data can be obtained with output energy reduced to a TIb of 0.5 or 0.1. Copyright © 2012 ISUOG. Published by John Wiley & Sons, Ltd.

Introduction

First-trimester Doppler ultrasound measurements have found increasing use in recent years1, 2. Evidence for the safety of this procedure is insufficient, and caution has been recommended3, 4. A Cochrane review concluded that no consistent serious adverse effects of obstetric ultrasound have yet been demonstrated5. However, studies suggest that there is an association between ultrasound screening at 18 weeks' gestation and non-right handedness in boys6, 7. Furthermore, an increased number of growth-restricted fetuses has been observed after exposing a study population to continuous-wave Doppler examinations five times during pregnancy8.

In 1997 the maximum output energy permitted by the American Food and Drug Administration (FDA) was increased from 100 to 720 mW/cm2. It was also decided that ultrasound machines capable of exceeding the previous limit had to display the thermal index (TI) and mechanical index (MI) on the screen during examination. The current recommendation of the Bioeffects and Safety Committee of the International Society of Ultrasound in Obstetrics and Gynecology is that the TI should not exceed a value of 1.0 for first-trimester Doppler examinations9. However, it has been found that ultrasound operators frequently exceed these limits during Doppler recording10. This is in line with the finding that ultrasound specialists commonly have insufficient knowledge of TI, MI and how to adjust the output power level of their machines11.

Two studies have shown effects on neuronal migration and liver apoptosis in rodents at output energies that are used in obstetric ultrasound, with a dose–response effect in both studies12, 13. These studies are given added relevance when taken in conjunction with the effects on growth and handedness described in humans6, 8.

In this study we test the hypothesis that the output energy in first-trimester Doppler examination can be reduced below the currently recommended upper limits and that reliable results can still be obtained.

Methods

We invited 136 pregnant women from among those referred for routine ultrasound scans to participate in the study, from whom 56 were accepted after giving written informed consent. We registered maternal age, maternal height and weight, fetal age at examination, estimated date of delivery and duration of the scan. Fetal age was assigned based on head circumference14. At birth we registered the weight, length, sex and gestational age (GA) of the neonate. Apgar score was also noted, as was information on any transfer to the neonatal intensive care ward.

We used a GE Vivid 7 Dimension ultrasound machine (GE Vingmed Ultrasound, Horten, Norway) with an M4S 2.5-MHz transabdominal sector transducer and an E8C 4.5-MHz transvaginal transducer. The transducers were provided for the study by GE Vingmed. They were unused prior to the study and their output power was measured by technicians at the company before shipment. After completion of the study the transducers were re-tested in the same manner.

We visualized the fetal middle cerebral artery, ductus venosus, one umbilical artery and both maternal uterine arteries. Using pulsed Doppler we recorded blood velocities and applied the auto-trace function of the scanner to determine the peak systolic velocity (PSV), end-diastolic velocity and the time-averaged maximum velocity in each of the vessels. The pulsatility index (PI) was calculated from these values15, 16.

The TI for bone (TIb) was used as an indicator of output energy17. Each procedure was carried out three times at a TIb of ≤ 1.0. The power was then reduced to fit TIb ≤ 0.5. The settings were optimized, using gain, scale and other post-processing options, before the recording was repeated three times. Similarly, the procedure was repeated at TIb ≤ 0.1. We recorded the scanner settings for each vessel and energy level (power level, frequency, pulse rate, sample volume depth and size and whether or not color Doppler was frozen while the pulsed Doppler was recorded) for later reproduction in vitro. After completion of the study, the settings for each of our measurements were reproduced by GE in Horten, and the corresponding energy levels were assessed using a hydrophone in water.

The examiner was blinded to the value of the blood-flow parameters by use of a ‘post-it’ sticker applied to the relevant part of the screen. A single operator (R.K.S.) carried out all scans, recordings and measurements. We used video recording in two sessions in order to assess the time taken by the various parts of the procedure.

The software Statistical Package for the Social Sciences (SPSS 18.0; SPSS Inc., Chicago, IL, USA) was used for our analysis, which was restricted to the velocity parameters PI and PSV, which we considered to be the most relevant.

A mixed linear regression model18, accounting for correlation between measurements within individuals, was used to relate PI and PSV to the potential predictor variables TIb and transducer, adjusting for GA, body mass index (BMI) and the various vessels. A compound symmetry correlation structure was assumed. TIb, GA and BMI were coded as interval variables, while transducer and vessels were coded as categorical variables. The SD was calculated within each group of three measurements of each vessel at the same energy level for each participant; P≤0.05 was considered statistically significant.

The present cross-sectional observational study was part of the larger project ‘Safe and Sound’ approved by the Regional Committee for Medical Research Ethics (REK Vest 2009/251). The first 20 women included in the project were examined at 18 weeks' gestation. The data from this part of the study are not included in the current paper, but were used for an interim analysis to calculate the appropriate size of the study population. We used SPSS Sample Power 2.0 for the power calculation, selecting the power calculation for equivalence studies, setting the acceptable difference at 10%, alpha at 0.05, and the power to a minimum of 80%. This analysis indicated that an appropriate sample size for the study would be 42 women, but we expanded the number to 56 to compensate for withdrawals and exclusions.

Results

Of the original 56 women who agreed to join the study, four were excluded; three had a miscarriage and one presented with twins. Another 10 were excluded because of technical problems with data storage in the scanner, which led to incomplete sets of data. Thus the final study group comprised 42 women, of whom we examined 21 with the transabdominal transducer and 21 with the transvaginal one.

Characteristics of the participants are presented in Table 1. Three neonates were transferred to the neonatal intensive care unit. One suffered from mild respiratory problems and was observed for one night. Another was delivered at a GA of 27 + 2 weeks because of pre-eclampsia, placental insufficiency and fetal growth restriction. The third had complications during delivery and was observed for 36 h. One woman presented with intrauterine fetal death at term. None of the remaining children had a 5-min Apgar score below 7. The study population gave birth to 23 girls and 17 boys, and two of the participants had not yet given birth at the time this paper was submitted.

Table 1. Characteristics of the study population (n = 42)
ParameterMeanMedianRange
  • *

    n = 40. BMI, body mass index.

Maternal age at examination (years)29.02922 to 35
Maternal weight at examination (kg)67.36645 to 102
Maternal height at examination (cm)168168155 to 180
Maternal BMI (kg/m2)23.62319 to 31
Gestational age at examination (weeks)12 + 212 + 111 + 1 to 13 + 3
Active ultrasound time (min)29.12814 to 48
Gestational age at birth (weeks)*40 + 040 + 327 + 2 to 42 + 1
Birth weight (g)*36203610670 to 5350
Length at birth (cm)*50.75131 to 59

Testing the energy output from the probes revealed a considerable range in spatial peak temporal average intensity (SPTA) for a given on-screen value of TIb (Table 2). Only the measurements done at precisely TIb 1.0, 0.5 and 0.1 and the corresponding values for SPTA have been included in the table. All values were below the current maximum of 720 mW/cm2 recommended by the FDA.

Table 2. Values of thermal index for bone (TIb) used in the study and corresponding output energy (given as spatial peak temporal average intensity) in an experimental setting
 Spatial peak temporal average intensity (mW/cm2)
 Transabdominal transducerTransvaginal transducer
TIbnMedianRangeQuartilesnMedianRangeQuartiles
1.041175.8149.0–222.0171.0, 205.04792.540.7–378.080.2, 327.0
0.55484.572.7–88.777.0, 86.18545.314.3–182.037.8, 150.0
0.19921.211.4–32.718.5, 21.7988.84.04–45.07.48, 27.7

Using color Doppler, we were able to visualize all five vessels at all three energy levels (Figures 1 and 2) in all but two of the participants. One was a case of isolated ductus venosus agenesis and in the other the fetal position did not permit visualization of the ductus venosus by color Doppler. The recordings yielded 1872 observations for statistical analysis.

Figure 1.

Color and pulsed-Doppler transvaginal ultrasound images of the ductus venosus at decreasing power levels: (a) thermal index for bone (TIb) ≤ 1.0; (b) TIb ≤ 0.5; (c) TIb ≤ 0.1.

Figure 2.

Color and pulsed-Doppler transabdominal ultrasound images of the right uterine artery at decreasing power levels: (a) thermal index for bone (TIb) ≤ 1.0; (b) TIb ≤ 0.5; (c) TIb ≤ 0.1.

There was no effect on PSV or PI when we reduced the output energy from TIb ≤ 1.0 to ≤ 0.5 and ≤ 0.1 (Table 3). We found no difference between the measurements done with the transabdominal transducer and those done with the transvaginal transducer. The five vessels we examined were different from each other with respect to both PSV and PI. There was no effect of BMI or GA on our data.

Table 3. Effect of reduced thermal index for bone (TIb), transducer, body mass index (BMI) and gestational age (GA) on peak systolic velocity recording and pulsatility index from middle cerebral artery (MCA), umbilical artery (UA), ductus venosus (DV) and uterine arteries (UtAs). A mixed linear regression model accounting for correlation between measurements within individuals was used and a compound symmetry correlation structure was assumed (n = 42)
 Effect on peak systolic velocity (cm/s)Effect on pulsatility index
ParameterEffect (95% CI)PEffect (95% CI)P
  • *

    TIb from ≤ 1.0 to ≤ 0.5 and ≤ 0.1.

  • P for trend.

  • Transabdominal and transvaginal transducers, latter used as reference.

  • §

    Left UtA used as reference vessel.

TIb*0.373 (−1.18 to 1.93)0.638− 0.0430 (−0.0904 to 0.00440)0.075
Transducer3.38 (−0.987 to 7.74)0.1260.108 (−0.0495 to 0.266)0.173
Vessels 0.001 0.001
 Left UtA§0 0 
 MCA− 43.5 (−45.3 to − 41.7) 0.447 (0.392 to 0.501) 
 UA− 43.4 (−45.1 to − 41.5) 0.465 (0.410 to 0.520) 
 DV− 25.6 (−27.4 to − 23.8) − 0.386 (−0.442 to − 0.330) 
 Right UtA− 8.48 (−10.3 to − 6.67) − 0.175 (−0.229 to − 0.120) 
GA (per 10 days)1.13 (−4.18 to 6.43)0.669− 0.0907 (−0.282 to 0.101)0.344
BMI (per 10 kg/m2)0.838 (−6.35 to 8.03)0.8150.0538 (−0.206 to 0.313)0.677

There was no trend towards increasing parameter variance with decreasing output energy (Table 4). There was, however, an increase in SD of the measurements when using the abdominal transducer; P = 0.020 and 0.004 for the SD of PSV and PI, respectively. The five vessels were different from one another with respect to the SD of the measurements, with the exception of the two uterine arteries, for which there was no significant difference. Again, there was no effect of BMI or GA on our data.

Table 4. Effect of reduced thermal index for bone (TIb), transducer, body mass index (BMI) and gestational age (GA) on SD of measured values of peak systolic velocity and pulsatility index from the middle cerebral artery (MCA), umbilical artery (UA), ductus venosus (DV) and uterine arteries (UtAs). A mixed linear regression model accounting for correlation between measurements within individuals was used and a compound symmetry correlation structure was assumed (n = 42)
 Effect on SD of peak systolic velocity (cm/s)Effect on SD of pulsatility index
ParameterEffect (95% CI)PEffect (95% CI)P
  • *

    TIb from ≤ 1.0 to ≤ 0.5 and ≤ 0.1.

  • P for trend.

  • Transabdominal and transvaginal transducers, latter used as reference.

  • §

    Left UtA used as reference vessel.

TIb*− 0.0541 (−0.146 to 0.0380)0.250− 0.000842 (−0.00557 to 0.00388)0.727
Transducer0.248 (0.0410 to 0.456)0.0200.0134 (0.00453 to 0.0222)0.004
Vessels 0.001 0.001
 Left UtA§0 0 
 MCA− 1.25 (−1.36 to − 1.14) 0.0160 (0.0105 to 0.0214) 
 UA− 1.40 (−1.51 to − 1.29) 0.00461 (−0.000863 to 0.0101) 
 DV− 0.638 (−0.747 to − 0.530) − 0.00355 (−0.00910 to 0.00201) 
 Right UtA− 0.302 (−0.409 to − 0.196) 0.00268 (−0.00280 to 0.00815) 
GA (per 10 days)− 0.164 (−0.416 to 0.0887)0.197− 0.00494 (−0.0157 to 0.00581)0.358
BMI (per 10 kg/m2)− 0.0127 (−0.355 to 0.329)0.9410.0114 (−0.00319 to 0.0259)0.122

The two video-recorded examinations showed that a total of approximately 6.5 min had been spent on color Doppler and 3.5 min on pulsed Doppler within a typical session of 34 min (Table 5). Each pulsed Doppler recording lasted on average 3.3 s; median, 2.3 s, maximum, 12 s.

Table 5. Time spent in different modalities during a typical examination*
    Doppler time
TransducerTotal examination timeFreeze timeActive ultrasound timeTotalColor DopplerPulsed Doppler
  • *

    Based on two videotaped examinations, one with each transducer.

Transvaginal33 min 0s16 min 33s16 min 27s9 min 4s6 min 2s3 min 2s
Transabdominal35 min 0s14 min 55s20 min 5s10 min 42s6 min 51s3 min 51s

Discussion

We have shown that the output energy for first-trimester pulsed-Doppler transvaginal and transabdominal ultrasound can be reduced from the recommended limit of TIb of 1.0 to ≤ 0.5 and ≤ 0.1 without affecting commonly used blood-velocity parameters. Our results raise the question as to whether the recommendation regarding power should be revised. Particularly at this early stage of pregnancy it should be of general interest to start any Doppler examination with a TIb ≤ 0.1 (or ≤ 0.5) from where, under exceptional insonation conditions, the operator may increase the power setting.

Many operators neither observe the TI setting nor adjust it11. We suspect that it is even rarer for the TI to be set according to the exposed tissue, i.e. TIs for soft tissue, TIc for cranial insonation and TIb for bone. In the present study we chose TIb, which is appropriate for bone exposure but overestimates tissue heating when used in soft tissues.

It is also interesting to note that the adjustments to lower energy levels never reached the level of effect that occurred with the change of transducer, the transabdominal recording showing higher variation than the transvaginal. It is obvious that energy is important for the quality of the recording, higher energy giving a more favorable signal-to-noise ratio. However, since reduced power within the range used in the present study did not affect the velocity parameters in a setting where repeat measurements were carried out in the same subject by the same observer with the same equipment, it is likely that the effects of this important reduction in power are insignificant when compared with intraobserver variation and other challenges in clinical examination19. It is also possible that a further reduction in power could be achieved in many cases before a significant impact on the velocity parameters occurs. A more developed algorithm and technology for low levels of power setting would be helpful. At levels of TI ≤ 0.1 the adjustment steps are commonly too few or too crude to permit a finer tuning of power, which would be of interest if exposure should be restricted to a minimum.

Diagnostic ultrasound is an important part of modern obstetrics20. Earlier studies on ultrasound safety have broadly fallen into two categories: epidemiological studies on humans, which have shown little or no harmful effect of diagnostic ultrasound5, and animal studies, where ultrasound in high relative doses or with long exposure times have shown harmful effects12, 13. Additionally, there is evidence to suggest that the safety limits are often exceeded10. One orthopedic study demonstrated an increase in bone volume when fractures of the fibula were treated with low-intensity pulsed ultrasound21. The reported peak energy of 30 mW/cm2 is well within the range used in obstetric ultrasound.

The current study was carried out with one set of equipment, and it is not certain that the results can be generalized to all scanners. A replication of the study in a different setting would therefore be welcome. However, we believe that the principle is applicable to most commercially available machines used in obstetric ultrasound.

It may also be argued that the study was insufficiently powered to show an effect. Increased numbers do indeed increase the probability of showing smaller effects. However, we would maintain that the effect of an increased number of participants would be so small as to be negligible in the clinical context where transducers, machines and operators show variation greater than that caused by changes in the power setting, at least within the range explored in the present study19.

When comparing the SPTA tested in vitro with the on-screen values for TIb during the measurements taken in pregnancy, we found a considerable variation in correlation, which was widest for the transvaginal transducer. Only when the TIb was ≤ 0.1 did we consistently get values for SPTA for the transvaginal transducer below the pre-1997 upper limit of 100 mW/cm2. Somewhat better consistency was found when testing the transabdominal transducer. The question then arises whether the on-screen display is sufficiently reliable for assessing the risk associated with a given first-trimester Doppler recording. This is further complicated by the fact that the energy measurements were done in water, a medium with properties very different from those of living tissue, e.g. when considering linear versus non-linear acoustics22. It is important that ultrasound operators keep the uncertainty of the on-screen display values in mind, in particular when performing first-trimester Doppler studies, the safety of which remains unproven.

The present findings that first-trimester Doppler ultrasound can be carried out at a TIb of 0.1 with no loss of clinical information raises the question whether such a level should be the recommended preset for Doppler examinations in early pregnancy.

Acknowledgements

GE Vingmed Ultrasound, Horten, Norway, tested the SPTA intensities of the transducers at no cost to us. The project was funded by grant No. 440020 from the Western Norway Regional Health Authority.

Ancillary