Materials and Methods

This study was based on 4 cohorts (Asturias province, Gipúzcoa province, Sabadell city, and Valencia city) of the larger Infancia y Medio Ambiente (Environment and Childhood) (INMA) Project. Patient recruitment and follow-up procedures have been reported in detail elsewhere. A total of 2644 eligible pregnant women agreed to participate and met the inclusion (≥16 years of age, singleton pregnancy, intention to deliver at the reference hospital) and exclusion (no communication handicap, no fetuses with malformations, no assisted conception) criteria. Women were followed-up during pregnancy and their children enrolled at birth and followed up until age 14 months. After excluding women who withdrew, were lost to follow-up, or underwent abortions or fetal deaths, a total of 2506 pregnant women were monitored through delivery. The final study sample included 2104 children with complete data on HC measurements and Bayley mental and motor scores. Both preterm (<37 weeks of gestation) and term children were included. All women provided written informed consent prior to participating in the study and the research protocol was approved by the clinical research ethics committee of the Municipal Institute of Health Care, Barcelona, Spain. The revised version of Helsinki declaration was followed. The ethics committees of the individual study centers also provided approval. All data were entered by the data manager of INMA Project ( http://www.proyectoinma.org/en_index.html ).

Ultrasound examinations for all women were scheduled in weeks 12, 20, and 34 of gestation. Measurements of HC were performed by trained obstetricians. We also had access to the records of any other ultrasound performed in the same hospital unit during pregnancy, thus allowing us to obtain from 2-7 valid ultrasonograms per woman between 7-42 weeks of gestation. Gestational age was established using early crown-rump length when the difference with gestational age based on self-reported last menstrual period was ≥7 days. Data in which this difference was >3 weeks or ±4 SD from the study mean were eliminated (n = 16) to avoid possible bias due to typographical errors in scan records or last menstrual period data. All measurements were performed in millimeters, obtained by transabdominal ultrasound examination (Voluson 730 Pro and 730 Expert; Siemens, Berlin, Germany) and followed standardized procedures. Ultrasonographists conducted a validation study in the Sabadell cohort to determine interobserver reliability. Intraclass correlation coefficients were in the range of 0.80–0.91 and coefficients of variation were <5%. Infant HC at birth was assessed by a nurse when the newborn arrived to the hospital ward within the first 12 hours of life.

Linear mixed models using the Hadlock algorithm were used to obtain longitudinal growth curves for HC. Box-Cox transformations were used to normalize the distribution and were modeled as a polynomial of gestational age in days until degree 3. Physiological determinants of growth and their interactions with gestational age were evaluated using the likelihood ratio test ( P < .05) through a forward selection procedure. Models were adjusted for physiological factors known to affect fetal growth to get an individualized rather than a population-based growth standard. Maternal and paternal height, age, parity, ethnicity, prepregnancy weight, and fetal sex were considered here. The use of individualized standards is expected to reduce misclassification of small for gestational age by excluding constitutionally small babies and including those within normal population limits who should have reached a greater size.

Time between ultrasonograms was used to model the correlation structure for within-patients errors. Gestational age, fetal sex, parity, ethnicity, and indicator variables for mothers with ultrasonograms close in time were considered to estimate their variance (heteroscedasticity). Random effects on the intercept, slope (days of gestation), or both were allowed and tested using the likelihood ratio test ( P < .05). Finally, goodness of fit was assessed by consideration of normality and independence of residuals.

Mean, SD, and predictions for weeks 12, 20, and 32 conditioned on the nearest HC measurement were obtained from fetal curves and were used for calculating unconditional z-scores and conditional SD scores. The unconditional z-score at a certain time point describes the size at that time. The SD score at the end of a time interval conditioned on the value at the starting point describes the growth experienced in this interval. Unconditional z-scores (in mm) were obtained at 12, 20, and 32 weeks of gestation and conditional SD scores were obtained for the intervals: 12-20 and 20-32 weeks. More detailed information is published elsewhere.

Infants were also classified as microcephalic, normocephalic, or macrocephalic according to HC measurements at birth adjusted for gestational age and sex. We used the whole study sample distribution of this measurement as a population reference to categorize the children among these 3 categories. We used the cutoff point of the 10th and 90th percentiles. Children were considered microcephalic if their values were <10th percentile, macrocephalic if their values were >90th percentile, and normocephalic if their values were within these 2 cutoff points. This percentile cutoff criterion is commonly used in population-based studies.

Neuropsychological development was assessed at mean age of 14.7 months (range, 11–23 months) using the mental and psychomotor scales of the Bayley Scales of Infant Development, version I (BSID). Twelve specially trained neuropsychologists conducted all of the testing at primary care centers in the presence of the mothers. Testing was conducted according to a strict protocol, included neuropsychologist training, and for a small number of children, multiple neuropsychologist evaluations were performed and where there was a discrepancy, results were reached by consensus. Children whose Bayley tests were of poor quality due to underlying pathology (Down syndrome, autism, n = 16) or did not cooperate in the testing (due to, eg, fatigue, bad mood, illness, n = 136) were categorized by the neuropsychologist as having a poor-quality test (n = 152). All children were included in the main analysis to avoid differences with the general population. Further sensitivity analyses excluding such children were performed to explore potential inconsistencies with the main results ( Appendix , Supplementary Tables 1-3 ).

BSID showed good psychometric characteristics based on our sample, such as internal consistency and reliability. BSID raw scores were standardized according to child’s age in days at the time of testing using a parametric method for the estimation of age-specific reference intervals. Scores were then normalized to a mean of 100 points with a SD of 15 points so as to homogenize the scales. These new indices were created for both the mental (Mental Development Index) and motor (Psychomotor Development Index) scales to obtain indices in accordance with a local normative sample and to avoid the use of US norms provided in the manual.

Information on parental education, social class, country of birth (Spain, foreign), age at birth, parity, maternal smoking, maternal pregnancy complications, maternal alcohol intake, and marital status was obtained through questionnaires administered during the first and third trimesters of gestation. Parity was defined as the number of previous pregnancies that lasted at least 22 weeks. Women were considered smokers if they reported smoking at the baseline interview (week 12) or week 34 of pregnancy. Information related to the child’s gestational age, sex, and type of delivery was obtained by our fieldwork nurses from clinical records at birth. Child postnatal health information, such as on infection and allergy episodes, was obtained through questionnaires (6th and 14th months). All questionnaires were administered face-to-face by trained interviewers. All such factors were taken into consideration as potential confounders.

According to power calculations, the minimum sample size required for present study was 883 patients (β = 2, α = 0.05, power = 0.80). The calculations were based on an expected difference of means of 2 points (outcome) between 2 perinatal HC groups in the independent variable. Bivariate analysis of predictive variables (HC measurements), neuropsychological outcomes (mental and psychomotor scales of the BSID), and covariates were carried out first to detect potential confounders ( P < .25).

Generalized Additive Models and multivariable linear regression analyses were performed to assess the association between HC measurements and BSID outcomes. All basic models were adjusted for sex and gestational age at delivery. Linear regression models for mental and psychomotor scores considered all potential confounding variables using enter regression methods. The final multivariate models were built based on this and previous studies. The final models were adjusted for gestational age, maternal age at birth, maternal and paternal body mass index, maternal smoking, parity, fetal sex, country of birth of the father and mother, preterm birth (<37 weeks), maternal social class (included as mandatory variable, 3 categories), maternal education (included as mandatory, 3 categories), cohort location (included as mandatory), and psychologist (included as mandatory). P values < .05 were considered statistically significant. Statistical analyses were carried out using SPSS version 12.0 (IBM, Armonk, NY) and Stata version 12 (StataCorp, College Station, TX).

Results

Characteristics of the 2104 infants and their parents are shown in Table 1 . The average HC at birth was 343.1 mm (SD 13.8). Children were classified as microcephalic with a HC at birth of <325.0 mm, and macrocephalic with a HC at birth of >360.0 mm. Mean mental and psychomotor scores were 98.3 (SD 16.3) and 99.1 (SD 15.6), respectively.

Women who were excluded from the analysis had a slightly lower maternal age at birth (30.4 vs 31.6 years old), gestational age (39.1 vs 39.7 weeks), and social class, and a higher parity and foreign origin prevalence (10.9 vs 7.8%). Further, their children presented with a slightly smaller HC at birth (341.2 mm vs 343.1 mm). However, no significant differences were observed between participants and nonparticipants in relation to the remaining variables presented in Table 1 , including HC and BSID scores ( P > .30).

The Figure depicts the relationship between HC at 20 weeks and BSID mental and psychomotor scores using Generalized Additive Models. Throughout the entire range of HC, the confidence limits of the association crossed the null value indicating no association with either BSID outcome. Similar patterns were observed with further prenatal and perinatal HC measurements described in Table 1 (data not shown). In results from linear regression models, there were also no clear associations observed between HC and BSID mental or psychomotor scores ( Table 2 ).

Adjusted linear relation between head circumference and Bayley scores (GAM)

Adjusted (for infant sex and gestational age at delivery) linear relation between head circumference measurement at 20 weeks of gestation and Bayley A , Mental Development Index (MDI) and B , Psychomotor Development Index (PDI) scores (Generalized Additive Models [GAM]). Weak linear pattern in limits of significance ( P values for linearity test: .08 and .10, respectively). Linearity pattern in relationship between 2 variables does not necessary indicate significant association in regression models shown in Table 2 .

Álamo-Junquera. Head growth and neurodevelopment. Am J Obstet Gynecol 2015 .

Stay updated, free articles. Join our Telegram channel

Join