Introduction

Relationships with caregivers are foundational to survival and play a role in risk and resilience1. Research based on attachment theory has shown that how caregivers think and feel about their child (their mental representations) affects their caregiving relationships2. These mental representations begin forming during pregnancy3.

Caregivers’ thoughts and feelings about their baby prior to birth (while the technical term is fetus, we use the term “baby” here, given this is the term most participants choose to use) are predictive of later caregiver behavior and caregiver–child attachment2,4,5. These thoughts and feelings include beliefs about who their child is as a person and are termed mental representations or internal working models of the child6,7,8. In a recent meta-analysis of 14 studies consisting of 1862 parents, the weighted mean effect size suggested a modest, though robust, association between prenatal representations and postnatal caregiving behaviors6. Expecting parents’ representations of their child characterized as detached, inconsistent, or preoccupied are associated with a greater likelihood of insecure caregiver–child attachment relationships and more negative caregiving behaviors, such as lower sensitivity and greater intrusiveness4,5,9,10,11.

In addition to the assessment of representations of the baby prenatally, recent work focusing specifically on the descriptions expecting parents provided of the baby and the tone of the words/phrases used to describe them demonstrated a link with later caregiving toward the child10. Specifically, in a sample of 120 pregnant people interviewed in their third trimester, participants were asked to provide up to five words or phrases to describe the personality of the baby they were pregnant with. Researchers assigned each description as positive or negative in tone (that is, semantic or emotional tone, capturing the connotation of the descriptor). At a follow-up caregiver–child interaction session when children were 12 months old, babies described with more positive words, compared to negative words, experienced higher levels of caregiver sensitivity, warmth, and engagement. Those described with more negative words, compared to positive words, received less sensitive and warm care from their mothers12.

Given the potential importance of caregivers’ representations about their baby, a number of studies have examined factors that may contribute to variation in them. Experiences from childhood13, exposure to stressful events14, social support in pregnancy15, number of other children, and whether the current pregnancy was planned16 have each been associated with prenatal representations. These studies point to contextual influences on prenatal representations and suggest the possibility that other contexts experienced by pregnant people (e.g., healthcare visits during pregnancy) may be important in establishing a parent’s emerging representation of their baby.

Ultrasound examinations are considered standard prenatal care in the United States17 and are growing in use across the world following recommendations made by the World Health Organization18. In order to inform guidelines for how, when, and if to conduct ultrasound examinations in pregnancy, Moncrieff and colleagues19 completed a systematic review of 80 publications. The authors identified four themes indicating that, first, many pregnant people view ultrasound examinations as an integral part of the pregnancy experience. Second, the majority of pregnant people reported trust in the visual technology of ultrasounds to provide them reassurance of fetal existence and well-being. Third, pregnant people and providers alike describe difficulty in balancing the possible unknown outcome and anticipated versus resulting emotional response of ultrasound results. For example, adverse clinical findings from the exam (e.g., fetal abnormality) can lead a pregnant person to experience anxiety, grief, or emotional distancing from the baby, which is often in stark contrast to the expected joy pregnant people report in seeing their baby for the first time. And lastly, pregnant people and their partners generally expressed a preference for ultrasound technicians to be warm, engaging, and inclusive, recognizing the ultrasound as a special part of the pregnancy experience. However, many reported that this was not reflective of their actual experiences. The narrative synthesis concluded that ultrasound provider behaviors and demeanor were influential to pregnant people’s experiences.

A recent systematic review considered the impact of prenatal imaging on parental prenatal attachment20. The authors identified six themes, several of which align with the results of the previous review—that ultrasound scans are an integral part of pregnancy, a desire for knowledge and understanding of the scan, and the importance of the parent–technician partnership during the scan. In addition, Skelton and colleagues20 highlighted that the scan experience extends beyond the scan appointment, as expecting parents prepare for the exam and share the results with others, and ultrasound scans help to create a social identity for the baby.

The Moncrieff19 and Skelton20 reviews point to the saliency of ultrasounds during prenatal care visits. These experiences are often emotionally activating, which intensifies their saliency21. According to the prior entry hypothesis, salient information is given priority access to perceptual processing, aiding in attention selection22. There appears to be a direct link between attention selection and later memory processes23,24,25, indicating that information prioritized due to perceived importance or the emotions experienced during the event is more likely to be remembered. As such, the uncertainty about the information obtained from ultrasound exams, combined with the potential for emotional reactions to seeing the baby, likely contribute to increased memory of what was communicated, including by prenatal care providers, during those experiences.

For many pregnant people, prenatal care visits serve to both legitimize the pregnancy and provide the opportunity to view the fetus as a person19,20. These visits allow expectant parents to imagine a future for their baby and envision potential characteristics and personality traits. The statements ultrasound providers make during these exams have been postulated to influence pregnant people’s representations of their baby. In a small observational study of 22 pregnant women’s ultrasounds, Walsh26 found that providers may inhibit, amplify, or otherwise influence expectant parents’ mental representations of their baby. In fact, at each of the observed ultrasound exams, providers made attributions to fetal behavior and many assigned personality characteristics26. The reported fetal descriptors from providers ranged from positive (e.g., “(the baby) is comfy right where she is”) to negative (e.g., describing the baby as “uncooperative” and “stubborn”). In the first instance, the prospective mother reportedly shared, “I just love that she’s happy as a clam inside of me,” while in the second instance, the prospective parents reportedly stated, “I was afraid this baby would be stubborn because everybody in his family is like that” and “It’s true. I guess I’m not surprised the baby is stubborn”26. Although preliminary, this observational study provides evidence that the manner in which providers interact with expecting parents during ultrasound exams may influence prenatal representations of their baby. Further, the representational influence likely varies depending on the provider’s ascribed narrative and interpersonal behaviors.

In the present article, we present two complementary studies to investigate the role and impact of statements made about the baby during prenatal care visits. In Study 1, a longitudinal observational study, we examined the association between prenatal descriptions of the baby and later psychopathology problems at child age of 18 months. This study is the first to investigate the prospective association between prenatal descriptions of the baby and later child outcomes. We expected that prenatal descriptions with greater negative tone would be associated with more problems in toddlerhood. Next, we tested if prenatal descriptions attributed to experiences from prenatal care visits differed in tone from prenatal descriptions without a stated connection to these visits. Studies highlight the importance of prenatal care providers’ narrative and interpersonal behaviors in shaping parent perceptions19,26,27,28. While prenatal care visits provide an opportunity to support parent–baby bonding19,26,27,28,29, high levels of provider burnout30 and contextual factors, including inability to obtain required images, resulting in repeated scans31, provide a foundation for hypothesizing that experiences during prenatal care visits have the potential for more negative interactions. Given these challenges, we hypothesized that descriptions attributed to experiences from prenatal care visits would be more negative. In Study 2, we conducted an experiment in which participants watched a video online and were asked to imagine receiving an ultrasound exam during pregnancy and then report descriptions of the baby. Participants were randomized to one of three video conditions, differing only in statements made about the fetus by the healthcare provider. Our pre-registered hypotheses were (1) that participants in the “blaming of baby” condition would report more negative tone in describing the baby than participants in either the “technical issues” or “parent–child relationship” conditions, and (2) that participants in the “parent–child relationship” condition would report more positive tone in describing the baby than participants in the other conditions.

Methods

Study 1

Study 1 Preregistration

Study 1 was not pre-registered.

Study 1 Participants

Participants (N = 325) were pregnant people 19–45 years of age (M = 31.26, SD = 4.97) recruited from a large metropolitan city in the central southeastern part of the United States. Participants were recruited from local obstetric clinics, printed advertisements, and social media advertising. Participants were required to be (a) US citizen or permanent resident, (b) at least 18 years of age, (c) fluent in English, and (d) between 11 and 38 weeks gestation in pregnancy. Of the 325 participants enrolled, 5 participants were missing the data necessary for the analyses (i.e., 1 participant withdrew from the pregnancy session prior to completion, 2 elected not to participate in the Working Model of the Child Interview portion of the study, and the data from 2 participants were lost due to data input errors). All participants identified as women. Participant pregnancy ranged from 11 to 38 weeks in gestation (M = 3.06, SD = 6.00). See Supplementary Table S1 for full demographic information.

Although recruitment has been completed, longitudinal follow-up visits are ongoing. A portion of participants (n = 175) had data available from the 18-month follow-up session. See the Supplementary Table S2 for availability and attrition details. The demographic characteristics of the 18-month sample largely mirrored those of the full sample, with the exception that participant age ranged from 21 to 46 years (M = 33.33, SD = 5.10) given this assessment was at infant age 18-months (range from 17.5 to 20 months, M = 18.50, SD = 1.23; 57.65% male). Of the 175 participants enrolled at 18-months, one had not provided prenatal descriptions and one did not complete the assessment of child emotional and behavioral problems. As such, 173 participants were used for analyses.

Study 1 data collection procedures

All recruitment and study procedures performed were in accordance with the prevailing ethical standards. The Vanderbilt University Institutional Review Board approved the study. Participants completed their first session in pregnancy either in the lab or remotely due to closures resulting from the COVID-19 pandemic. Relevant to the current study, the first session consisted of informed consent procedures and a semi-structured clinical interview. All interviews were conducted by trained study personnel. Eligible participants were invited to return for a follow-up visit when the child was approximately 18 months of age. Although not relevant to the current study, only families in which the infant was eligible for participation in an MRI visit post-birth (i.e., infant was not born before 36 weeks and had no medical history of neurological injury or impairment) were invited to participate in the follow-up visit. At child age 18 months, participants completed questionnaires related to child functioning and clinical outcomes. All participants provided informed consent prior to participating in each session. Participants were compensated $15 USD/hour.

Study 1 Measures

Prenatal description of the child

As part of the Prenatal Working Model of the Child Interview32, participants were asked to provide 5 descriptors to describe their unborn child’s personality. Specifically, participants were instructed, “I want you to think about your child’s personality. I am going to ask you to pick five words (adjectives) to describe your child’s personality now.” Descriptors were recorded in the order they were provided. Approximately halfway through data collection, interviewers were instructed to provide an additional prompt at the end of the interview to consider whether there were any additional descriptors related to prenatal care visits when prompted: “Sometimes parents report that the descriptors they use to describe their child’s personality come from their experiences during prenatal exams, like ultrasound exams or prenatal care visits. Do you recall any comments made about your baby in these appointments that really stuck with you?” There were 126 interviews prior to the addition of this question and 194 after the addition. For the first 126 interviews, audio recordings were reviewed for whether a descriptor was described as related to a prenatal care visit. All of these descriptions were coded as spontaneous (i.e., in response to the personality prompt, without specific prompting regarding prenatal care visits from the interviewer). For the last 194 interviews, interviewers marked whether a participant spontaneously provided a descriptor that was related to a prenatal care visit or were prompted (i.e., provided in response to the additional prompt).

We used two approaches to characterize the tone of the descriptors provided. First, we obtained expert ratings33 for each description, which was available in continuous score (from −1 to 1) and categorical (Negative = −1, Neutral = 0, Positive = 1) formats. Out of 1596 descriptions, 187 (11.7%) did not match descriptions within the expert rating dataset. In the event that a provided descriptor in the current sample did not match a descriptor rated by experts, the authorship team independently rated and reached consensus to interpolate the related category score. Expert tone across all prenatal descriptions ranged -1 to 1 (Mean = 0.61, SD = 0.54). Second, we used Linguistic Inquiry and Word Count (LIWC-22)34 given this is an alternative method for sentiment analysis with wide-scale adoption. For LIWC methods and analyses, please see the Supplementary Method.

Child Behavior Checklist

Participants completed the Child Behavior Checklist 1.5–5 (CBCL)35 to assess behavior and emotional problems in their young children at the 18-month visit. The CBCL is a 99-item parent-report measure that is valid for children aged 1.5–5 years wherein parents report problem behaviors within the past two months. Items cover a range of concerns, including Emotionally Reactive, Anxious/Depressed, Somatic Complaints, Withdrawn, Attention Problems, Aggressive Behavior, and Sleep Problems. For each item, one of the following responses is selected: 0 (not true), 1 (somewhat or sometimes true), or 2 (very true or often true). The Total Problems Score is calculated by taking the sum of all 99 item responses, with a possible range of 0–198. In the present sample, the Total Problems raw score was used in analyses and ranged from 0 to 57 (M = 14.88, SD = 10.19).

Study 1 data analysis plan

All variables were examined for normal distribution according to skewness and kurtosis values using the moments package36 and manipulated as needed for analysis with the dplyr37 and tidyverse38 packages. To test hypothesis 1, we implemented correlations to examine the associations between the tone of the prenatal descriptors participants used to describe their baby and 18-month child outcomes. We used Pearson correlations to test prenatal descriptor tone using the stats package included in the R base package39 (version 2023.09.0 + 463). Descriptor tone was operationalized across two measurements, including (a) the average tone of all descriptors participants provided and (b) the most negative tone provided across all descriptors provided. To test hypothesis 2, we used the lmtest40, lm.beta41, and betaDelta42 packages to conduct two regression models with mean tone from prenatal descriptors regressed onto whether that descriptor was reported to come from a prenatal care visit (yes or no). We used the sandwich package43,44 to conduct regression with cluster robust standard errors (CRSE) to account for the nested nature of the descriptor data within persons. We covariated whether the descriptor from a prenatal care visit was spontaneous or prompted, descriptor order, and gestational age45. All figures were created with the ggplot2 package46. All statistical analyses were conducted using two-sided tests.

Study 2

Study 2 Preregistration

Study aims, approach, and hypotheses were pre-registered prior to data collection (https://osf.io/pdnrq).

Study 2 Sample Size

Although our final sample size was 161, we had planned to recruit a sample size of 159 based on the results of a power analysis (conducted in G*Power)47 for an ANOVA estimating a medium effect size of Cohen’s f = 0.25 and using an α = 0.05 with 80% power. This sample size also supported sufficient power for the planned chi-squared analyses according to an a priori sensitivity analysis, which estimated effects w > 0.22.

Study 2 participants and data collection procedures

We recruited participants through Prolific, an online participant recruitment platform. We collected data in batches to confirm eligibility criteria via participant report within the study and assess if participants had passed the attention checks. We initially over-recruited for participation in the study to ensure the planned minimal sample size was met after removing participants who did not pass attention checks (n = 30). As such, we completed recruitment with 161 participants. Participant (N = 161) age ranged from 19 to 40 (M = 30.25, SD = 5.82). All demographic information, including sex and gender identity, was provided by participants. See Supplementary Table S4. Eligibility criteria included: English-speaking biological females, aged 18–40 years old, citizens or permanent residents of the United States, and current residents of the United States. Participants were not required to be pregnant at the time of participation. The online study description remained consistent for all participants, regardless of participant parity or current pregnancy: “This study includes asking you to consider your thoughts and feelings given a specific scenario. We ask that you do your best to place yourself in the scene and imagine how you would react given the situation, regardless of whether you have experienced a similar event. We will then ask you to report on your thoughts and feelings, in addition to some demographic questions. You will watch a series of short videos.” The compensation rate was $12 USD/hour.

Study 2 study procedures

Participants first completed consent, then were prompted to check their video and audio settings. Once confirmed, they were presented with videos instructing them to imagine an experience, and finally completed a series of questionnaires which included attention checks. For the imagined experience, participants were randomly assigned to one of three conditions related to a prenatal care visit: (1) blaming of baby [BOB], (2) technical issues [TI], and (3) parent–child relationship [PCR] (described below).

In each condition, the prenatal care visit began with an identical introduction wherein the narrator instructed the participant to imagine that they were 20-weeks pregnant with a child they intended to keep and introduced the purpose of the anatomy ultrasound exam. Next, all conditions viewed the same ultrasound video; however, the audio commentary varied by condition. Scripts were matched in approximate word count (BOB: 81 words, TI: 83 words, PCR: 81 words), and in each, the ultrasound technician explained that she was unable to capture the necessary images which necessitated a follow-up appointment. In order to hold the vocal tone constant, recordings were produced using an online voice tool (WellSaid Labs). In the BOB condition, the ultrasound technician attributed the difficulty capturing images to the baby’s lack of cooperativeness. In the TI condition, the ultrasound technician attributed the difficulty capturing the images to the combination of fetal position and the equipment. In the PCR condition, the ultrasound technician did not make an attribution for why the images could not be captured and instead focused on the participant being able to spend time seeing the baby again at a subsequent ultrasound exam. Conditions were randomly assigned via Qualtrics, the online platform used to administer the experiment and the following questionnaires. Condition scripts and videos are available here: https://osf.io/g2sb7/?view_only=68f55e268a1a45ccb9fdd03c8c5a0c37.

Study 2 Measures

Prenatal description of the child

Participants were instructed to provide descriptions of the baby (i.e., “Now, please pick five words or phrases to describe the personality of the baby you were asked to imagine in the video. Feel free to take your time.”). Five discrete text boxes were provided for participant responses. We rated the tone of descriptors provided in the same way as Study 1. The authorship team independently rated and reached consensus to interpolate the related category score. Out of 788 descriptors, 244 (31%) were rated by the authorship team.

Difficulty imagining the ultrasound scene

Given the potential difficulty in imagining the scene, participants were asked to rate, “How difficult was it for you to imagine the ultrasound scene?” Answers range from 1 to 10, with 1 = Not difficult at all to 10 = All the effort possible.

Study 2 Hypotheses

Preregistered hypotheses were:

  1. 1.

    Participants in the BOB condition would describe the baby using more negative tone than participants in the TI and PCR conditions.

  2. 2.

    Participants in the PCR condition would describe the baby using more positive tone than participants in the TI and BOB conditions.

Hypotheses were examined according to overall tone, such that (a) the mean score across the 5 descriptors provided operationalized the participant’s overall impression of the baby. Additionally, we examined our hypotheses according to (b) if any decidedly negative descriptors were used to describe the baby. The full hypotheses are available at: https://osf.io/pdnrq.

In addition to our pre-registered hypotheses, we also examined if condition was associated with the frequency of using at least one positive descriptor to describe the baby. This addition was added to test the possible benefits of the PCR condition above and beyond decreasing the likelihood of negative descriptors. In particular, we were interested in offering plausible suggestions to ultrasound technicians for what may support parent–baby bonding, though we formalized this analysis plan after we completed the pre-registration.

Study 2 data analysis plan

We used ANOVA to examine the main effect of condition (BOB, TI, PCR) on the tone of participants’ overall impression of the baby, and estimated the effect size with the effectsize package48. We used pairwise contrasts using the Least Significant Differences formula via the agricolae package49 to test each condition compared to the others. We used the effsize package in R to calculate Cohen’s d50. Given the potential difficulty in imagining the scene, the same analyses were conducted as an ANCOVA with participants’ scores on difficulty imaging the ultrasound scene as a covariate with the rstatix package51. The number of descriptors provided was also added as a covariate in this analysis. In addition to these pre-registered covariates, we completed an additional ANCOVA which included participant parity and current pregnancy. We then used rcompanion52 to compute chi-squared tests to examine the effect of condition on the use of at least one negative descriptor to describe the baby, and odds ratios were calculated. All figures were created with ggplot246, ggdist53,54, and gghalves55.

In addition to our pre-registered hypotheses, we also examined if condition was associated with the frequency of using at least one positive descriptor to describe the baby. We used chi-square tests to examine the effect of condition on the use of at least one positive descriptor to describe the baby, and odds ratios were calculated. All statistical analyses were conducted using two-sided tests.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Results

Study 1

Study 1 Descriptive Statistics

As noted above, the first portion of data collection did not include an additional interview prompt to ask about experiences during prenatal care visits. Among these participants, 28% spontaneously provided at least one description of their baby that came from a prenatal care visit (n = 35/126). In the second portion of data collection, following the introduction of the additional prompt, approximately 70% of participants provided at least one description that came from a prenatal care visit (n = 135/194). Of all prenatal descriptions provided, the majority of descriptions of the baby were positive (see Table 1).

Table 1 Prenatal description counts by tone category
Full size table

Bivariate associations were examined amongst study variables. Order of descriptors shared a small positive association with expert-rated tone (r(1584) = 0.07 [0.02, 0.12], p = 0.005), such that descriptions provided later tended to be more positive. Gestational weeks of the current pregnancy shared a small association with expert-rated overall tone (r(318) = −0.18 [−0.29, −0.08], p < 0.001) such that descriptions provided later in pregnancy tended to be more negative. Participant parity did not share a statistically significant association with expert rated overall tone (r(317) = −0.09 [−0.20, 0.02], p = 0.118).

Study 1 prenatal description and 18-month child behavioral and emotional difficulties

Average tone of descriptors of the baby provided during pregnancy was prospectively negatively associated with 18-month child CBCL total score, such that more negative tone was moderately associated with higher levels of child total problems (r(171) = −0.32 [−0.45, −0.18], p < 0.001). When considering the most negative tone across all prenatal descriptors provided, the negative association between tone and CBCL total score was observed (r(171) = −0.28 [−0.41, −0.14], p < 0.001).

Study 1 comparing descriptions from prenatal care visits vs. other sources

Prenatal descriptor tone was more negative, with a medium effect size, if the descriptor came from a prenatal care visit, relative to descriptions not attributed to this source (F(3, 1581) = 67.49, p < 0. 001, R2 = 0.11; t = −7.52, p < 0.001, b = −0.44, β =−0.30 [−0.36, −0.24]). Whether the descriptor was provided spontaneously or after prompting was not associated with tone (t = −1.29, p = 0.198, b = −0.12, β =−0.06 [−0.13, 0.01]). The association between prenatal descriptor tone and prenatal care visit attribution remained statistically significant when gestational weeks at the time of the interview was included as a covariate in the model (F(4,1580) = 54.29, p < 0.001, R2 = 0.12), and gestational weeks was negatively associated with prenatal descriptor tone with a small effect size (t = −3.40, p = 0.001, b = −0.01, β = −0.09 [−0.13, −0.04]).

In order to examine if the association between prenatal descriptor tone and child outcomes was moderated by experience attribution (e.g., if descriptors were associated with prenatal care visits or not), we conducted additional analyses of the interaction between tone and attribution source using hierarchical linear regression. First, we examined average prenatal tone across all descriptors provided in relation to child total problems. The overall model examining child outcomes at 18-months using the CBCL total problems scale was statistically significant (F(3,169) = 7.41, p < 0.001, R2 = 0.12). We found that the prenatal descriptor tone negatively related to 18-months CBCL total problems with a medium effect size (t = −3.66, p < 0.001, b = −20.02, β = −0.42 [−0.64, −0.20]) and whether any of the descriptors were related to prenatal care visits did not moderate this association (main effect of prenatal care visit: t = −1.28, p = 0.202, b = −2.56, β = −0.10 [−0.24, 0.05]; interaction of tone by prenatal care visit (t = 0.82, p = 0.413, b = 5.94, β = 0.09 [−0.13, 0.31]). See Fig. 1. Results remained consistent when including gestational weeks in the model (F(4,168) = 5.53, p < 0.001, R2 = 0.12). Gestational weeks was not statistically significantly associated with CBCL total problems (t = 0.17, p = 0.863, b = 0.03, β = 0.01 [−0.13, 0.16]).

Fig. 1: Child behavioral and emotional problems in association with prenatal description tone.
figure 1

Figure (n = 173) depicts that child behavioral and emotional problems assessed at age 18-months are associated with prenatal description tone, represented according to if the description was attributed to a prenatal care visit (green circles) or not (purple circles). When considering prenatal descriptions related to a prenatal care visit, average tone of descriptors of the baby was prospectively negatively associated with 18-month child CBCL total score, such that more negative tone was moderately associated with higher levels of child total problems (r(90) = −0.29 [−0.46, −0.09], p = 0.006) When considering prenatal descriptions not related to prenatal care visits, average tone of descriptors of the baby was prospectively negatively associated with 18-month child CBCL total score, such that more negative tone was moderately associated with higher levels of child total problems (r(79) = −0.40 [−0.57, −0.20], p < 0.001).

Full size image

Second, we repeated this approach using each participant’s most negative descriptor’s tone. We examined if the association between the most negative prenatal descriptor tone and child outcomes was moderated by experience attribution (e.g., if any descriptors were associated with prenatal care visits or not). The overall model examining child outcomes at 18-months using the CBCL total problems scale was statistically significant (F(3,169) = 5.82, p < 0.001, R2 = 0.09), and the most negative tone descriptor negatively related to 18-months CBCL total problems with a medium effect size (t = −3.63, p < 0.001, b = −8.16, β = −0.39 [−0.59, −0.19]). Whether the descriptors were related to prenatal care visits did not moderate this association (main effect of prenatal care visit: t = −1.33, p = 0.186, b = −2.73, β=−0.10 [−0.26, 0.05]; interaction of tone by prenatal care visit: t = 0.97, p = 0.335, b = 3.19, β = 0.10 [−0.10, 0.30]).

Study 2

Study 2 descriptive statistics

There were no statistically significant differences between groups based on demographic information (Supplementary Table S4). Of all participant prenatal descriptions provided, the majority were positive (n = 426; 54%) compared to negative (n = 216; 27%) and neutral (n = 146; 19.5%). The three most commonly used descriptions were handful, stubborn, and difficult for the BOB condition; active, small, and stubborn for the TI condition; and happy, beautiful, and active for the PCR condition.

Study 2 condition differences in the overall tone of descriptions of the baby

Tone statistically significantly differed as a function of condition (F(2,158) = 51.24, p < 0.001, η2 = 0.39 [0.30, 1.00]) as expected, such that the tone of descriptions was more negative for participants in the BOB condition relative to the TI condition with medium effect size (Mean difference = −0.36, p < 0.001, d = −0.77 [−1.17, −0.38]) and PCR condition with large effect size (Mean difference = −0.80, p < 0.001, d = −1.89 [−2.35, −1.43]). See Fig. 2. LSD post hoc analyses also demonstrated that participants in the TI condition provided more negative tone relative to the PCR condition with a large effect size (Mean difference = −0.44, p < 0.001, d = 1.32 [0.89, 1.74]). Condition differences remained when accounting for difficulty imaging the scene and number of descriptors provided, F(2,156) = 51.03, p < 0.001. Similarly, condition differences remained when accounting for participant parity, current pregnancy, difficulty imagining the scene, and number of descriptors provided, F(2,154) = 48.40, p < 0.001. All covariates in this model were statistically non-significant, all ps > 0.154. Together, these results demonstrate support for Hypotheses 1A and 2A.

Fig. 2: Condition differences in the overall tone of descriptions of the baby.
figure 2

Figure (N = 161) depicts the distributions of overall tone of descriptions of the baby provided by participants randomized to blaming of baby (represented in purple), parent–child relationship (represented in green), and technical issues (represented in orange) conditions. Distributions are represented as scatter, box, and kernel density plots.

Full size image

Study 2 condition differences in using negative words to describe the baby

Chi-square analyses demonstrated that condition was statistically significantly associated with differences in tone with a large effect size (χ2(2) = 36.61, p < 0.001, V = 0.48, 95%CI [0.36, 0.60]) such that participants in the BOB condition were much more likely to describe the baby with at least one negatively rated descriptor than those in the PCR or TI conditions (see Table 2 and Fig. 3). Together, these results demonstrate support for Hypotheses 1B and 2B.

Table 2 Odds of using at least one descriptor to describe the child prenatally
Full size table
Fig. 3: Condition differences in the use of at least one negative word to describe the baby.
figure 3

Figure (N = 161) depicts the count of participants who used at least one negative word to describe the baby according to condition. Blaming of baby condition counts are represented in purple, parent–child relationship condition counts are represented in green, and technical issues condition counts are represented in orange.

Full size image

In addition to our pre-registered hypotheses, we also examined if condition was associated with the frequency of using at least one positive descriptor to describe the baby. Chi-square analyses indicated that condition was statistically significantly associated with differences in positive descriptor use also with a large effect size (χ2(2) = 28.23, p < 0.001, V = 0.42, 95%CI [0.32, 0.53]) such that participants in the PCR condition were much more likely to describe the baby with a positively rated descriptor than those in the BOB or TI conditions (see Table 2).

Discussion

Study 1 Discussion

Study 1 used a longitudinal design to examine prospective associations between expecting parents’ prenatal descriptions of their child and the child’s behavioral and emotional difficulties in toddlerhood. We found that a more negative tone in describing the child prenatally was associated with more reported child difficulties at 18 months old. Additionally, we demonstrated that descriptors attributed to prenatal care visits were more negative in tone than other descriptors. Lastly, we found that the association between tone of the descriptions and the child’s behavioral and emotional difficulties was not moderated by whether a description was attributed to an experience during a prenatal care visit. This suggests that the association between an expecting parent’s thoughts and feelings about their child prenatally are prospectively associated with child difficulties, regardless of the origins of those descriptions.

Numerous studies have established the link between the prenatal mental representations held by expecting parents hold about their baby and later attachment and caregiving behavior2,4,5,6,10,11,12. The present study builds upon this work by demonstrating that prenatal representations are prospectively associated with child behavioral outcomes. We propose that caregiving is likely the key mechanism underlying this prospective association. Caregiving relationships are the foundation of early development context1, and previous research has shown that parent behaviors directly influence child behaviors56. Moreover, studies have found that parental attachment is associated with developmental outcomes in children57. Taken together, this literature supports the idea that prenatal representations may shape child behavior by influencing the quality of postnatal caregiving.

Identifying the factors that may contribute to variation in prenatal representations is a critical next step in further understanding the development of representations and how best to support caregivers and their children. In addition to other factors13,14,15,16, we found that the experiences expecting parents have during prenatal care visits may be associated with prenatal representations. Further, these prenatal care visit experiences may increase the likelihood that the parent forms a more negative view of their child. Prenatal care visits—especially those which include ultrasound exams of the fetus—are considered to be an integral part of the pregnancy experience and quite salient emotionally, wherein providers use trusted technology to examine the baby19. As such, these providers may have an outsized influence on an expecting parent’s representation of their child. This idea has been investigated via observation previously26, and the results of this study extend this work to demonstrate the associations between the tone of the descriptions provided by expecting parents and child outcomes almost 2 years later.

Study 1 Limitations

Strengths of Study 1 include the longitudinal approach with more than 1,500 descriptors collected via structured interview. Limitations of Study 1 include the limited representativeness of the sample and the observational nature of data collection that precludes causal inferences, making it difficult to establish a direct cause-and-effect relationship between the variables studied. While the prospective associations observed are compelling and support our hypotheses that negative comments made about a fetus during prenatal care visits may initiate a cascade of negative thoughts about the child, affecting child outcomes, there may be extraneous factors moderating the association. For example, it is reasonable to wonder if pregnant people more prone to negative biases (e.g., perhaps those with a life history of depression58) would both remember more negative comments made about their baby during a prenatal care visit and be prone to incorporate these comments into their representation of the baby. This alternative hypothesis has been supported post-birth, wherein maternal mood was the most robust predictor of maternal perceptions of their child’s behavior59. Additionally, fetal behavior itself could be responsible for variation in the descriptions provided (e.g., less fetal movement is associated with higher behavioral inhibition assessed at 7–14 years60). To rule out this rival hypothesis that individual differences fully explain the tone in which a baby is described, we designed an experimental study. In Study 2, participants were randomly assigned to one of three experiences by viewing a video of a pregnancy ultrasound exam, in which the only difference was statements made about the fetus by the healthcare provider.

Study 2 Discussion

Study 2 was a pre-registered experiment designed to examine whether statements made about a fetus during a prenatal care visit influenced descriptions of that baby. Study participants were randomly assigned to one of three conditions during an imagined prenatal care visit, in which an ultrasound technician provided different explanations regarding the need to return for another ultrasound given that not all required images were obtained. Conditions varied in the content of these explanations, such that technicians either blamed the baby, referred to limitations in the equipment, or used relationally-focused language. The experimental approach in Study 2 allowed us to examine causal claims about the role of prenatal care experiences and descriptions of the child, given that random assignment ensures that any differences observed can be attributed to the assigned conditions rather than pre-existing differences among participants.

Results across coding schemes supported our pre-registered hypotheses such that participants in the condition where the ultrasound technician blamed the baby provided child descriptions characterized as much more negative in tone than participants in the other two conditions. Conversely, participants in the relationally-focused condition, in which the opportunity to see the baby again was highlighted, provided child descriptions characterized as much more positive in tone than participants in the other two conditions. The condition in which ultrasound providers explained that fetal position and equipment misalignment were responsible for the need to repeat the visit at a later date resulted in intermediate outcomes in terms of the tone of the descriptions (in between the two other conditions in the proportion of use of negative and positive descriptions). While this outcome is in line with our hypotheses, we included a condition focused on the technical constraints of ultrasounds based on pregnant people’s reported preference for explaining procedures19 without euphemisms61. A “sticking to the facts” style may be the most easily adopted, authentic interpersonal style; although if a desired outcome is more positive views of the baby during pregnancy, the selection of this approach is likely to underperform relative to the relationally-focused approach.

The notable strength of Study 2 is that these findings refute otherwise plausible hypotheses that links between descriptions and child functioning are fully explained by individual differences in the person providing the descriptions. Specifically, individual differences in respondents are controlled through the use of random assignment into conditions. Further, participants viewed the same video of a fetus moving during an ultrasound procedure in this experiment.

Study 2 Limitations

This study is not without its limitations. Study 2 limitations include that participants were not selected based on their pregnancy status and instead were provided instructions to watch a video and imagine themselves in the scene. We adjusted for self-reported imaginal ability, which did not alter the results, but our participants’ experience still greatly differs from experiencing an actual ultrasound during one’s own pregnancy. Here, participants were not reporting on their own fetus in order to avoid random assignment to personal prenatal care experiences and reduce the risk of harm caused by providing negative statements about actual individuals (e.g., random assignment to the BOB condition). Together, Studies 1 and 2 provide complementary information in demonstrating the effect of prenatal care experiences on mental representations of a baby.

General Discussion

Summary of Findings

We investigated (1) the prospective association between prenatal descriptions of one’s baby and that child’s later behavioral and emotional difficulties and (2) the role of prenatal care visit experiences in descriptions of a baby. In Study 1, we found a small prospective association between the tone of expecting parents’ prenatal description of their child and child emotional and behavioral difficulties in toddlerhood. Prenatal descriptions were more negative, with a medium effect, if the descriptor was related to a prenatal care visit. In Study 2, we experimentally controlled what participants, who were instructed to imagine an ultrasound exam experience, were told by a healthcare provider when the exam did not result in the acquisition of all necessary images. While the majority of the descriptions used were positive across both studies, tone varied both within and across individual descriptions. Collectively, our findings provide evidence that prenatal care experiences influence representations, and that the tone of those representations (both average tone and the most negative descriptor used) are associated with later child functioning approximately 2 years later.

Broader implications

Prenatal care visits, particularly those including ultrasound imaging, influence the pregnancy experience and are associated with feelings of connection to the developing baby19,20,26,27,29,45,62. During these visits, interactions with prenatal care providers can shape parents’ perceptions of their unborn child—whether they develop positive, negative, or neutral perceptions26. However, these visits can sometimes become sources of tension, as more than 10% of ultrasound scans in pregnancy require repeat examination, mainly due to the inability to acquire all the necessary diagnostic information during the initial scan31. This may lead to increased anxiety, confusion, and frustration for both the prospective parents and prenatal care providers. In fact, recent estimates indicate that more than 90% of obstetric sonographers/ultrasound technicians met burnout thresholds for exhaustion and disengagement30. In these challenging moments, providers may attempt to explain why they are unable to complete the exam. Interviews from the pregnant participants in Study 1 provide first-person accounts that providers called their baby ‘stubborn’ or ‘uncooperative’ in such situations. While these descriptions may aim to reassure parents that the need for repeat scans is not due to provider error nor fetal abnormalities, it is likely that those who use such characterizations may not realize how their words can shape parents’ perceptions of their child. The broader implications of these studies suggest the need to consider the importance of how children, even in gestation, are described. Guidance regarding how to communicate with patients in these circumstances may be warranted, similar to recommendations to avoid euphemisms in medical care61,63.

Future work would benefit from testing alternative language, or “scripts” that healthcare providers may use to avoid negatively influencing expecting parents during prenatal care visits. More broadly, this may result in guidance to avoid all attributions about the baby’s intent or personality in prenatal care exams to instead focus on the procedures of the exam and the developing relationship between expecting parents and their child. While these providers are indeed experts in their field, including in fetal physical development, assessing the baby’s physical development is not equivalent to being all-knowing of the baby’s personality. For example, there has been work that fetal movement may be prospectively associated with child temperament (e.g., less fetal movement is associated with higher behavioral inhibition assessed at 7–14 years60); however, it is unknown if the providers in these studies made comments about those observations, which in turn might have influenced later parent report of temperament. Further, Study 2 demonstrates that, while there may be an association between fetal movement and temperament, randomly assigned provider narratives also impact participant reports of child temperament even when actual fetal movement is held constant. Moncrieff and colleagues19 point to a need for additional training in navigating the interpersonal aspects of prenatal care exams, and our findings suggest that this training should consider both the content of provider comments to patients during prenatal care visits and the tone of that content.

Conclusion

These studies are the first to investigate prospective associations between prenatal descriptions and child-specific outcomes, to do so in relation to whether descriptions came from prenatal care experiences, and to illustrate experimentally that descriptions about a fetus influence beliefs about that child’s personality. The current studies together provide evidence that prenatal care visit experiences are associated with, and may play a causal role on, expecting parents’ representations of their baby. These studies suggest that there are likely unintended consequences to statements made by healthcare providers to expecting parents during prenatal care exams. Parents’ prenatal representations are associated with a host of relational and child outcomes. Given the heightened saliency of ultrasound exams, trust in medical professionals, and the importance of the caregiver–child relationship, statements made about a fetus during prenatal care exams may play a role in influencing a child’s future.