To determine whether classification differences were consistent across growth curves, we classified HC using three different growth curves. We used growth curves that are either commonly used (Hadlock and Intergrowth-21st) and/or were designed to be representative of the U.S. population (Intergrowth-21st and National Institute of Child Health and Human Development [NICHD]) to determine the proportion of female and male fetuses classified as having microcephaly (<3rd percentile, z-score < −1.88) or macrocephaly (>97th percentile, z-score >1.88) in the data subset of our sample described above. While the Society for Maternal-Fetal Medicine (SMFM) provides recommendations for standardizing the evaluation of fetal HC in the context of Zika virus exposure (5 ), there are no universal definitions for microcephaly and macrocephaly; the 3rd and 97th percentiles were chosen because they are commonly used and because the information provided in the NICHD curves does not allow direct calculation of other potential cutpoints for microcephaly and macrocephaly. All three evaluated growth curves are sex-neutral, using a single set of curves for both sexes.
We used cubic interpolation to calculate values of the 3rd and 97th percentiles for integer values of GA in days (17 (link)). NICHD percentiles were published separately for four specific race/ethnicity groups: Asian/Pacific Islander, Hispanic, Black non-Hispanic, and White non-Hispanic. There were no published NICHD percentiles that were nonspecific for race/ethnicity. We used the mean of the four race/ethnicity-specific values at each GA to create percentiles for a fifth group, deemed “Uncategorized.” For the NICHD analyses only, we excluded the small number of ultrasounds that could not be linked with maternal data. Women with a recorded race/ethnicity that did not fit in these categories, or whose maternal data were available but missing race/ethnicity data, were evaluated using the “Uncategorized” percentiles. Although there are limits of the reliability and precision of race/ethnicity data, the EHR was the only potential source of race/ethnicity data for this population and therefore the only way to evaluate our data compared to the US-based NICHD percentiles.
We used cubic interpolation to calculate values of the 3rd and 97th percentiles for integer values of GA in days (17 (link)). NICHD percentiles were published separately for four specific race/ethnicity groups: Asian/Pacific Islander, Hispanic, Black non-Hispanic, and White non-Hispanic. There were no published NICHD percentiles that were nonspecific for race/ethnicity. We used the mean of the four race/ethnicity-specific values at each GA to create percentiles for a fifth group, deemed “Uncategorized.” For the NICHD analyses only, we excluded the small number of ultrasounds that could not be linked with maternal data. Women with a recorded race/ethnicity that did not fit in these categories, or whose maternal data were available but missing race/ethnicity data, were evaluated using the “Uncategorized” percentiles. Although there are limits of the reliability and precision of race/ethnicity data, the EHR was the only potential source of race/ethnicity data for this population and therefore the only way to evaluate our data compared to the US-based NICHD percentiles.
Free full text: Click here
