Fetal biometry is an important component in understanding overall fetal well-being. Deviations from typical growth may prompt further evaluation, possibly affecting delivery planning, and ultimately, neonatal outcomes (1). Growth curves are used to determine if biometric measurements are considered typical for a given gestational age; the data used to create growth curves and the population in which these curves are used can impact how these measurements are interpreted. Starting at birth, infant size measurements are evaluated using sex-specific growth curves, reflecting small but consistent growth differences between sexes (2). However, fetal growth curves are not typically interpreted in a sex-specific manner.

Our study focuses on one component of biometry, head circumference (HC), and the potential implications of using sex-neutral growth curves to interpret fetal HC. There are no universally accepted definitions for microcephaly and macrocephaly; however, < 3rd and > 97th percentiles are commonly accepted as the cutoffs for microcephaly and macrocephaly, respectively. The mean HC for male and female fetuses differs by 0.3–0.5 standard deviations (SDs), with males having larger heads, on average, than females. However, most fetal growth curves are not sex-specific. These differences in growth can impact how many fetuses of each sex are classified as having microcephaly or macrocephaly, possibly resulting in over- or under-diagnosis depending on sex.

If fetal HC follows a normal distribution, the difference in means between male and female HC allows us to predict the difference in the proportions of male and female fetuses with microcephaly and macrocephaly. However, not all fetal and infant growth curves follow a normal distribution for HC (3, 4). For example, two populations with the same mean and standard deviation, one distributed normally and one left-skewed, would have different values for <3rd and >97th percentiles. Therefore, knowing the difference in mean HC between male and female fetuses does not necessarily allow us to predict precisely how the proportions of male and female fetuses at the extremes of a distribution differ.

As illustrated during the Zika virus (ZIKV) epidemic, whether or not a fetus is classified as microcephalic can significantly impact evaluation and management (5). ZIKV is a single-stranded RNA virus that is often asymptomatic (6); when symptomatic, it generally causes a mild illness (7). ZIKV infection during pregnancy has been shown to cause nervous system abnormalities, including microcephaly, in some fetuses (8). The ZIKV epidemic highlights the importance of accurate evaluation of fetal HC. As described above, prenatal growth curves are not sex-specific, despite the fact that size differences between male and female fetuses have been shown to be clinically significant (9–12). Current diagnostic criteria for fetal HC are limited by the widespread use of sex-neutral growth curves. In this paper, we applied three commonly used sex-neutral growth curves to our population: Hadlock, Intergrowth-21st, and NICHD.

The Hadlock curves were published in 1984 as a reference based on a study of 361 pregnant women in Houston and continue to be widely used (13). They allow determination of HC z-scores using a cubic equation for mean HC by age and a non-varying standard deviation (SD). The 2014 Intergrowth-21st standards are based on data from eight different countries and allow determination of HC z-scores based on fractional polynomials for median HC and SD as a function of GA (14). Also released recently, the 2015 NICHD growth standards were based on data prospectively collected from 12 sites across the United States. NICHD published percentiles using quantile regression for each week of GA and separate growth curves for four race/ethnicity groups (3).

Evaluating real-world differences in the proportions of male and female fetuses classified as having atypical head size is an important step in understanding the clinical implications of using sex-neutral vs. sex-specific fetal growth curves and whether it is important to incorporate sex-specific curves when interpreting fetal growth. If proportions of males and females classified as microcephalic and macrocephalic differ from what we expect based on the means, this may also provide some information about the epidemiology of causes of microcephaly and macrocephaly in a population.

We hypothesize that more female than male fetuses meet criteria for microcephaly and more male than female fetuses meet criteria for macrocephaly when classifying second trimester HC measurements using three sex-neutral growth curves.

We performed a retrospective cohort study of pregnant women seen at Penn State Milton S. Hershey Medical Center (HMC) who delivered between 7/1/2012 and 6/30/2017. The Penn State Institutional Review Board approved the study and provided a waiver of informed consent (study number 00007195).

The final dataset for analysis contained 6,192 unique ultrasound visits for 5,454 women (Figure 2). Four percent of ultrasound visits occurred between 16 and 0/7 and 17–6/7 weeks gestation, 63.8% between 18 and 0/7 and 19–6/7 weeks gestation, 26.6% between 20 and 0/7 and 20–6/7 weeks gestation, and 5.6% between 21 and 0/7 and 21–6/7 weeks gestation. The sex recorded in the birth log indicated 3,186 male fetuses (51.5%) and 3,006 female fetuses (48.5%). For the 5,013 records that were linked to infant EHR data via newborn MRN, the EHR and birth log listed the same sex in 99.3% of the records. For the 6,156 records (99.4%) that were linked to the maternal EHR data, Table 1 shows the final maternal race/ethnicity groups used for the NICHD growth curve. Very few recorded HC values appeared implausible for GA; all HC values were retained in the analysis.

Female fetuses were more likely than male fetuses to be classified as microcephalic using the Hadlock (0.4% male, 1.4% female; OR F vs. M 3.7, 95% CI [1.9, 7.0], p < 0.001), Intergrowth-21st (0.5% male, 1.6% female; OR F vs. M 3.4, 95% CI [1.9, 6.1], p < 0.001), and NICHD (0.3% male, 1.6% female; OR F vs. M 5.6, 95% CI [2.7, 11.5], p < 0.001) curves (Table 2). Male fetuses were more likely than female fetuses to be classified as macrocephalic using the Intergrowth-21st (6.0% male, 1.5% female; OR M vs. F 4.3, 95% CI [3.1, 6.0], p < 0.001) and NICHD (4.7% male, 1.0% female; OR M vs. F 5.1, 95% CI [3.4, 7.6], p < 0.001) curves (Table 2). The overall prevalence of macrocephaly using the Hadlock curve was low, and the difference in prevalence by sex was not statistically significant.

There were differences in the overall proportions of fetuses classified as having microcephaly or macrocephaly depending on the growth curve used (Figure 3, Figure 4 and Table 3). Figure 3 shows scatter plots, separated by males and females, of fetal HC by gestational age with percentiles (3rd and 97th) overlaid for Hadlock and Intergrowth-21st growth charts. Figure 4 shows scatter plots, separated by male, female, and race/ethnicity, of fetal HC by gestational age percentiles (3rd and 97th) for each race/ethnicity category used in NICHD growth charts. These visuals help highlight the difference in the number of females vs. males that are classified as microcephalic or macrocephalic by the Hadlock, Intergrowth-21st, and NICHD growth curves.

Consistent with our hypotheses, we found that second-trimester female fetuses were approximately three times more likely than male fetuses to be classified as having microcephaly, and second-trimester male fetuses were approximately three times more likely than female fetuses to be classified as having macrocephaly when applying sex-neutral growth curves to our population. We also found wide variation in the proportion of fetuses meeting criteria for microcephaly or macrocephaly based on second-trimester measurements depending on which sex-neutral growth curve was applied to our population: Hadlock, Intergrowth-21st, or NICHD.

Prior studies have demonstrated differences in growth curves created separately for male and female fetuses (20, 21), and studies of various fetal growth parameters have demonstrated the clinical significance of sex differences (9–12). Sex-specific estimated fetal weight (EFW) curves predict stillbirth better than sex-neutral curves, and the between-sex differences for EFW are smaller than those for HC (approximately 0.2 SDs at 20 weeks) (9).

Rather than predicting expected proportions based on the mean and SD of a normal distribution, our study evaluated how many fetuses actually met criteria for microcephaly and macrocephaly when three different sex-neutral growth curves were applied to our population. This is important because HC does not necessarily follow a normal distribution and does not necessarily meet other criteria (identical SDs) required for predicting proportions from differences in means.

Our analysis, driven partially by the need to better understand prenatal diagnosis of microcephaly in the context of the recent Zika virus epidemic, demonstrates that ignoring known differences in fetal growth by sex can result in large differences in the proportion of male and female fetuses classified as having microcephaly or macrocephaly. We are unaware of any evidence of a true difference by sex in the prevalence of pathology that causes atypically small or large heads. Therefore, any discrepancy that we see in the proportions of male and female fetuses classified with microcephaly and macrocephaly is likely due to the methods used to interpret fetal growth. This artifact could lead to significant potential consequences.

There were significant differences by sex in the proportions of fetuses in our population classified as macrocephalic and microcephalic. When a fetus with a pathological condition associated with abnormal head size has a falsely normal HC measurement, the opportunity for early detection that could support further prenatal and/or neonatal planning may be missed. Alternatively, a healthy fetus with a somewhat large or small head may be subjected to unnecessary testing, which is not only burdensome to the patient, but also associated with significant maternal anxiety (22–24).

In addition to differences by sex, there were differences in the proportions of fetuses classified as microcephalic or macrocephalic depending on which growth curve was used. When all fetuses are evaluated together, the proportion of fetuses meeting criteria for macrocephaly using the NICHD and Intergrowth-21st curves was close to the expected proportion of 3%. Using the Hadlock curve, however, very few fetuses met this criterion. For all three curves, the proportion of fetuses meeting criteria for microcephaly was approximately 1/3 the expected proportion of 3%. This difference in proportions could reflect a real difference in the distribution of fetal HC at our institution compared to the populations underlying the evaluated growth curves or could represent measurement differences at our center for fetuses with smaller heads. If this under-identification of microcephalic fetuses is present at other centers, it is possible that a significant number of fetuses with prenatally identifiable pathology are currently not being identified.

Prior to recent advancements in prenatal sex determination, the challenge of accurately determining fetal sex made the utility of sex-specific fetal growth curves irrelevant for practical purposes. However, it is now technically feasible to use cell-free DNA screening to accurately and safely determine fetal sex (25, 26). Sex-specific growth curves could be useful in the subset of fetuses with a HC that is atypical or borderline atypical on a sex-neutral curve, such as <10th or >90th percentile. In those cases, determining fetal sex could allow evaluation using a sex-specific growth curve, thereby directing further evaluation, counseling, and possible intervention in a more informed context. Further research would be needed to evaluate the risks and benefits of this approach.

The poor fit of the Hadlock curves is important to note as it is one of the most widely used growth curves in clinical practice. The Hadlock curves were developed in 1984 based on data from a relatively small number (361) of women, all Caucasian and from the same area. Our findings support use of a more recent growth curve.

The primary strength of this study is the use of accurate sex data from birth to evaluate differences by sex in a relatively large population. The primary limitations are that it was done retrospectively in a single center and that we lacked data on long-term clinical outcomes.

In conclusion, applying sex-neutral fetal growth curves to second-trimester HC measurements classifies more female fetuses as microcephalic and more male fetuses as macrocephalic. Our findings support use of a growth curve more current than Hadlock for evaluation of HC in fetuses in our population. In the future, the creation of widely applicable sex-specific HC references to be used for fetuses with a HC at the edge of the expected range could improve prenatal identification of, and subsequent intervention or counseling for, fetuses with an atypical head size.