Handedness in the Helsinki Ultrasound Trial

Authors


Abstract

Objectives

To determine whether exposure to prenatal ultrasound increases non-right-handedness in boys.

Methods

The association between exposure to prenatal ultrasound and handedness was tested, using logistic regression analysis, in the Helsinki Ultrasound Trial data. We applied an intention-to-treat approach in this analysis of a subset of 4150 subjects whose parents answered a follow-up questionnaire on handedness when the children were aged 13–15 years.

Results

The odds ratio for non-right-handedness of children who had been exposed to prenatal ultrasound was 1.16 (0.98–1.37) for all subjects, 1.12 (0.89–1.41) for boys and 1.24 (0.97–1.58) for girls.

Conclusions

We could not confirm the hypothesis that prenatal ultrasound exposure and handedness are associated. Our findings were independent of the particular definition of handedness used, whether it was considered according to the writing hand alone or defined using a laterality quotient. Copyright © 2011 ISUOG. Published by John Wiley & Sons, Ltd.

Introduction

There is some evidence that the determination of human handedness takes place during prenatal development1 and genetic effects have been reported to explain about one quarter of the variance in handedness2, 3. It is also well known that left-handedness is more prevalent among males4. Furthermore, although non-right-handedness (NRH) is a natural phenomenon in general5, it may result from a brain injury6.

Two previous randomized controlled trials (RCTs) have raised concern that NRH might be increased in males who have been exposed to prenatal ultrasound7, 8. A recent meta-analysis of four studies9, including two RCTs (OR, 1.26) and two cohort studies (OR, 1.17), indicated that males who were exposed to prenatal ultrasound were more likely to be NRH than were males who were not exposed to prenatal ultrasound. Thus, the effect of prenatal ultrasound exposure on handedness, especially among boys, needs further investigation. The aim of this study was to investigate whether prenatal exposure to ultrasound is associated with an increase of NRH in boys.

Methods

Our study was a RCT based on the randomized Helsinki Ultrasound Trial and the detailed study design has been described in a previous paper10. This trial investigated the effect of ultrasound screening on perinatal mortality, randomizing into cases or controls on an intention-to-treat basis 9310 pregnant women at the time of their first visit to the prenatal clinic in 1985–1987. Of these 9310 pregnancies, 8662 delivered successfully. When preparing the questionnaires in 1999, we were able to trace addresses for 7773 mothers who had been randomized into prenatal ultrasound (n = 3915) and control (n = 3858) groups. If a mother had more than one pregnancy during the trial period, these were recorded separately, while a multiple pregnancy had a single record, with the twin status noted. During 2001–2002, when the children were aged 13–15 years, questionnaires were posted out with an accompanying cover letter that reminded the parents of the study during the pregnancy and informed them about the purpose and importance of the ongoing study. Answering and returning the questionnaire was regarded as indicating consent.

The handedness of offspring was measured by means of a parental report as part of a longer questionnaire that included the Child Behavior Checklist designed by Achenbach and Edelbrock11. There was a yes/no question about familial left handedness (existence of another left-handed member of the family). Parents indicated whether their child used their left or right hands or both hands equally when performing five activities: writing, throwing a ball, holding scissors, holding a knife and holding a spoon. We calculated the laterality quotient (LQ) based on the Edinburgh Handedness Inventory method12. The difference between the right- and left-hand scores was divided by their sum and the result was multiplied by 100. This formula yielded a LQ that ranged from − 100 (totally left-handed) to + 100 (totally right-handed). We categorized subjects as right-handed when the LQ was + 100, the rest being NRH. We used a LQ of + 70 as an alternative cut-off as well as NRH based on writing-hand alone to test whether the particular classification of NRH used would affect our analysis. When a woman had responded regarding more than one child, only one, selected arbitrarily, was included in the study.

The study was approved by the ethics committee of Helsinki University Central Hospital.

Statistical methods

Using logistic regression analysis, we first tested all children together and then boys and girls separately because of the known higher prevalence of NRH in boys4. Testing boys alone also directly addressed the initial hypothesis presented by Salvesen and Eik-Nes13. According to their meta-analysis based on Norwegian and Swedish follow-up studies, if boys alone were analyzed according to intention-to-treat, the OR for NRH was 1.26 (95% CI, 1.03–1.54).

We also carried out per-protocol analysis (analysis according to actual exposure status), which would be expected to show the size of any effect but at the cost that it may be confounded. We then carried out subgroup analysis according to trimester(s) during which exposure occurred (Model 1), using 85 days and 170 days as cut-offs for the start of the second and third trimesters, respectively, and then introduced as confounders smoking status and handedness of any first-degree relative (Model 2). Smoking status was pooled into three classes: non-smokers (reference group), mothers who had stopped because of pregnancy and current smokers. We also investigated dose–response effects, which would be particularly important if the intention-to-treat analysis showed an association between prenatal ultrasound exposure and handedness.

Multiple selection correction was used because of hypothesis generation. Among individuals there were subgroups such as controls that had ultrasound during pregnancy and cases that had extra exposure when they were exposed to ultrasound mostly outside the planned time window. This is why some combined subgroup analyses with respect to exposure were carried out.

There was no significant difference in randomization status between subjects with or without handedness data. Because of the known genetic effects on handedness2, 3, we included familial NRH (at least one NRH first-degree relative) as a covariate in the logistic regression analyses. Maternal smoking during pregnancy was also included as a covariate to control for possible pathological effects on handedness.

Results

Of the 7773 initially randomized eligible subjects for whom contact details could be traced, 4174 passed date-of-birth and gender checks against information available from birth records and CPR (Central Population Registry) records. Of these, 24 had missing LQ data. Our analysis thus included 4150 subjects (2013 boys and 2137 girls), i.e. 53% of those contacted. Of these, 2112 (981 boys and 1131 girls) were cases and 2038 (1032 boys and 1006 girls) were controls.

There was a significant sex difference in the prevalence of NRH between boys and girls (Table S1 online). Using a LQ of + 100 to define right-handedness, the prevalence of NRH was 17.1% in boys and 14.4% in girls (chi-square = 5.81, P = 0.02) and using a LQ of + 70, the prevalence of NRH was 13.2% in boys and 10.0% in girls (chi-square = 10.1, P = 0.002). Based on writing hand alone, the prevalence of NRH was significantly higher in boys (10.4%) than in girls (7.6%) (chi-square = 10.0, P = 0.002).

Prenatal ultrasound exposure did not increase the probability of NRH (defined as LQ ≠ + 100): OR, 1.16 (95% CI, 0.98–1.37) (Table 1). The result was similar using the alternative + 70 LQ cut-off. Similarly, there was no effect of ultrasound exposure on handedness when boys and girls were analyzed separately. The results were similar when analyzing writing hand alone (Table S2).

Table 1. Association between prenatal exposure to ultrasound and handedness, analyzed by intention-to-treat using logistic regression (n = 4150)
 ControlsCases 
LQ cut-off/subgroupnNRH (n)nNRH (n)OR (95% CI)
  1. Laterality quotient (LQ) was based on Edinburgh Handedness Inventory method12, those at or above the cut-off being defined as right-handed and those below it as non-right-handed (NRH). OR, odds ratio.

LQ + 100     
 All203830021123531.16 (0.98–1.37)
 Boys10321699811761.12 (0.89–1.41)
 Girls100613111311771.24 (0.97–1.58)
LQ + 70     
 All203822121122581.14 (0.95–1.38)
 Boys10321319811341.09 (0.84–1.41)
 Girls10069011311241.25 (0.94–1.67)

Table 2 shows the per-protocol subgroup analyses for girls and boys separately. Logistic regression analyses indicated there to be no trimester-specific ultrasound exposure effect on handedness, either in boys (Table 3) or in girls (Table 4). Instead, having at least one NRH first-degree relative significantly increased the probability of NRH in Model 2 of both sexes (P < 0.05). Moreover, maternal smoking cessation because of pregnancy increased the probability of NRH in boys (P < 0.05), but this effect was not evident when familial NRH was included in the analysis. Similar results to those obtained for analysis according to LQ of + 100 were obtained when the LQ cut-off was + 70 or when writing handedness alone was considered (Tables S3–8 online). However, the effect of maternal smoking cessation because of pregnancy in boys was not significant when we used the LQ cut-off of + 70 or writing handedness alone.

Table 2. Association between prenatal exposure to ultrasound and handedness, analyzed according to exposure and sex using logistic regression
 No ultrasound exposureUltrasound exposure 
GroupnNRH (n)nNRH (n)OR (95% CI)*
  • Handedness was defined using a laterality quotient (based on Edinburgh Handedness Inventory method12) cut-off of + 100, i.e. those with LQ + 100 were defined as right-handed and those with LQ below this as non-right-handed (NRH).

  • *

    Confidence limits corrected for 24 separate tests.

  • True controls were those randomized to the control group that did not receive a scan.

  • ‘Drop-ins’ were controls that were scanned; ‘cases with extra scans’ underwent extra scans, including outside weeks 16–20.

  • §

    Per-protocol cases were cases that were scanned only as planned, during weeks 16–20. OR, odds ratio.

Boys     
 True controls and cases with no scan263441.0
 Drop-ins and cases with extra scans13132110.95 (0.55–1.67)
 Cases per-protocol§437901.29 (0.69–2.41)
Girls     
 True controls and cases with no scan270491.0
 Drop-ins and cases with extra scans14121970.73 (0.43–1.26)
 Per-protocol cases§455620.71 (0.37–1.35)
Table 3. Results of univariable and multivariable logistic regression models for the prediction of non-right-handedness (NRH) in boys
 NRH (n)/total (n) Adjusted OR (95% CI)
Indicator variableIndicator variable +Indicator variable −Unadjusted OR (95% CI)Model 1Model 2
  • 95% CIs Bonferroni corrected for 24 separate tests.

  • *

    For each of the trimesters the indicator variable was scored + if any ultrasound exposure occurred during that trimester and—if there was no ultrasound exposure during that trimester. Cut-offs for the start of second and third trimesters were 85 days and 170 days.

  • Smoking during pregnancy/cessation of smoking because of pregnancy.

  • P < 0.05. +, present; −, absent; OR, odds ratio.

Trimester 1*45/267300/17460.98 (0.57–1.68)1.02 (0.59–1.77)0.94 (0.52–1.68)
Trimester 2*274/152571/4881.29 (0.82–2.00)1.31 (0.84–2.05)1.26 (0.80–2.00)
Trimester 3*139/928206/10850.75 (0.52–1.09)0.74 (0.51–1.07)0.77 (0.52–1.14)
Smoking cessation55/212237/14331.77 (1.04–3.00) 1.71 (0.98–2.96)
Smoking41/302237/14330.79 (0.45–1.39) 0.79 (0.44–1.40)
NRH first-degree relative114/475215/14881.87 (1.25–2.79) 1.83 (1.21–2.77)
Table 4. Results of univariable and multivariable logistic regression models for the prediction of non-right-handedness (NRH) in girls
 NRH (n)/total (n) Adjusted OR (95% CI)
Indicator variableIndicator variable +Indicator variable −Unadjusted OR (95% CI)Model 1Model 2
  • 95% CIs Bonferroni corrected for 24 separate tests.

  • *

    For each of the trimesters the indicator variable was scored + if any ultrasound exposure occurred during that trimester and—if there was no ultrasound exposure during that trimester. Cut-offs for the start of second and third trimesters were 85 days and 170 days.

  • Smoking during pregnancy/cessation of smoking because of pregnancy.

  • P < 0.05. +, present; −, absent; OR, odds ratio.

Trimester 1*30/273278/18640.70 (0.38–1.32)0.70 (0.37–1.32)0.59 (0.29–1.21)
Trimester 2*240/164068/4971.08 (0.69–1.71)1.09 (0.69–1.72)1.02 (0.63–1.65)
Trimester 3*144/1013164/11240.97 (0.66–1.42)0.99 (0.67–1.44)1.03 (0.69–1.54)
Smoking cessation36/237201/14801.14 (0.62–2.08) 1.14 (0.61–2.13)
Smoking59/347201/14801.30 (0.79–2.14) 1.24 (0.74–2.09)
NRH first-degree relative88/489198/15831.54 (1.00–2.36) 1.57 (1.01–2.43)

Discussion

Although NRH was significantly more common in boys than in girls, we could not confirm an association between prenatal ultrasound and NRH in either the whole sample or in the subgroup of boys. The determination of handedness is apparently associated with other factors such as the handedness of family members. Our findings were independent of the particular definition of handedness used, whether it was considered according to the writing hand alone or defined using a LQ.

The meta-analysis of Torloni et al.9 was based on two RCTs in Norway7, 14, 15 and a study in three Swedish hospital districts8, 16. The Norwegian and Swedish results were for boys exposed to ultrasound before 133 or 154 gestational days, while in our study ultrasound exposure was between 85 and 170 days of gestation. The non-significant result for boys in our study differed from the results of this meta-analysis9, 13.

A strength of our study, which compared well in this respect with the Swedish16 and Norwegian14, 15 randomized trials, was the large number of subjects in the final analysis. Our study had 4150 subjects with information on handedness, constituting 53% of the initially randomized cohort of 7773 eligible subjects for whom contact details could be traced. The Swedish study had 3052 (65%) and the Norwegian study had 1663 (69%) individuals with handedness data.

Handedness is known to have a familial component2, 3. In our findings the familial association (Model 2) was strongest. It might be argued that the parents being the source of information could have caused bias in our study. However, it is unlikely that a child's handedness and a parent's answer to the familial handedness question would have any confounding causal association.

The intention-to-treat method is considered to be confounder-free when all subjects are analyzed, but is not commonly regarded as such when subjects are analyzed in subgroups. In the same sense, the lower the response rate to a survey following a trial, the larger the possibility of confounders. It would be hard to determine the exact nature of a mechanism by which handedness, randomization status and answer status could be associated. However, as Salvesen et al. state in their paper7, the Swedish and Norwegian studies had no difference in response rate between cases and controls, and our data were similar17.

Thermal and mechanical indices have been introduced to indicate regulatory limits for ultrasound, and their correct use is generally believed to be sufficient to ensure safety in human applications. Many of the concerns regarding ultrasound safety are historically based, with a lot of the available epidemiological data having been gathered more than 20 years ago9. Speculations regarding ultrasound safety in humans also refer to the difficulty in precisely calculating wave characteristics such as wave amplitude fluctuations and excessive local temperature increases, as well as potential unknown vulnerabilities of the growing tissue, possible bone–tissue interactions and maternal fever, which could lead to a vulnerable rise of temperature18–23. Experimental animal research has indicated that ultrasound may affect neuronal migration in the growing fetus24. However, the finding is not generalizable to prenatal ultrasound scanning in humans, because, in animal trials, the proportion of ultrasound exposure to body size differs from that in human scanning. Moreover, the time and amount of exposure tend to be more regulated in medical use.

The efficiency of devices in clinical practice has improved considerably in recent years. Yet such speculations regarding risk are needed, if only to remind us that there are still things we do not know for sure. Sometimes, risk is a matter of thresholds, and this is addressed by the ALARA (As Low As Reasonably Achievable) principle.

In conclusion, we could not confirm the association between handedness and prenatal ultrasound exposure. Our results suggest that other factors, such as familial left-handedness, may contribute to the determination of handedness.

SUPPORTING INFORMATION ON THE INTERNET

The following supporting information may be found in the online version of this article:

Table S1 Prevalence of non-right-handedness by sex and case-control status

Table S2 Association between prenatal exposure to ultrasound and handedness (writing hand), analyzed by intention-to-treat using logistic regression (n = 4156)

Table S3 Association between prenatal exposure to ultrasound and handedness (laterality quotient, cut-off + 70), analyzed according to exposure and sex using logistic regression

Table S4 Association between prenatal exposure to ultrasound and handedness (writing hand), analyzed according to exposure and sex using logistic regression

Table S5 Results of univariable and multivariable logistic regression models for the prediction of non-right-handedness (laterality quotient, cut-off + 70) in boys

Table S6 Results of univariable and multivariable logistic regression models for the prediction of non-right-handedness (writing hand) in boys

Table S7 Results of univariable and multivariable logistic regression models for the prediction of non-right-handedness (laterality quotient, cut-off + 70) in girls

Table S8 Results of univariable and multivariable logistic regression models for the prediction of non-right-handedness (writing hand) in girls

Acknowledgements

We owe thanks to Dr O. Karjalainen and Prof. P. Ylöstalo for designing the Helsinki Ultrasound Trial. This study was funded by the Helsinki University Central Hospital and Rinnekoti Research Foundation.

Ancillary