Objective
We sought to apply customized standards to ultrasound-derived estimated fetal weight (EFW), and assess the frequency of abnormal growth when compared to population-based standards. We also evaluated association with adverse perinatal outcomes.
Study Design
This was a historical cohort using prenatal ultrasound examination data at ≥24 weeks over a 1-year period. Ultrasound-derived EFW and growth percentile (population-based EFW [popEFW]) were reported and compared to a customized EFW (custEFW).
Results
In all, 782 women met inclusion criteria. More fetuses were identified as small for gestational age (SGA) (15.1% vs 3.8%; P < .0001) and large for gestational age (LGA) (6.8% vs 1.7%; P < .0001) using custEFW, and adverse perinatal outcomes were more frequent among those identified by the custEFW compared to the popEFW. Both SGA and LGA diagnosed by custEFW were predictive of a neonatal SGA (positive likelihood ratio, 8.64) and LGA (positive likelihood ratio, 15.4).
Conclusion
CustEFW was a better predictor of abnormal birthweight and adverse outcomes compared to traditional popEFW standards.
Fetal growth abnormalities are associated with increased perinatal morbidity and mortality, both acute and in the long term. Consequently, the presence of abnormal fetal growth detected during prenatal ultrasound evaluation is a frequent cause of antenatal testing and a major reason for medically indicated delivery. Estimating gestational age solely by patient-reported last menstrual period is plagued with errors that contribute to inaccurate determinations of fetal growth abnormalities leading to unnecessary interventions in the late preterm and early term period. Inaccuracies in menstrual dating have led to the widespread use of ultrasound to improve the accuracy of gestational age estimation and the creation of new fetal growth curves, from which deviations are considered pathologic.
Customized growth curves that take into consideration maternal height, weight, parity, ethnicity, and fetal sex have been used to determine the expected birthweight for a pregnancy in optimal conditions. Numerous studies have demonstrated that the use of customized birthweight improves the identification of actual pathologic growth and is better correlated with adverse neonatal outcomes when compared to population-derived growth curves. In 2009, Gardosi and Francis created customized growth standards for the US population based on >30,000 pregnancies. Our hypothesis was that customized estimated fetal weight (custEFW) percentiles based on ultrasound-derived estimated fetal weight (EFW) would considerably improve the prediction of adverse perinatal outcomes.
Materials and Methods
This was a historic cohort study of women who delivered at a single tertiary hospital over a 1-year period (July 1, 2010, through June 30, 2011). A computerized perinatal database was used to identify women delivered at ≥24 weeks 0 days at Memorial Hermann Hospital–Texas Medical Center in Houston, TX. Women with multiple gestation, antepartum stillbirth, or fetal structural/chromosomal abnormalities were excluded. Study personnel reviewed maternal and neonatal hospital charts of all eligible women to obtain pertinent clinical and outcome data. In cases where women had >1 ultrasound during the pregnancy, the ultrasound latest in gestational age was included for study. Ultrasound-derived EFW was calculated utilizing the formula of Hadlock et al and the EFW percentiles (population-based EFW [popEFW]) were reported using the table devised by Williams et al. The table of Williams et al was developed from a population of >2 million births in the state of California, is used in our routine clinical practice, and only adjusts for gestational age. A customized birthweight percentile based on the US population has been described that adjusted for maternal height, weight, ethnicity, parity, and fetal sex (Bulk Centile Calculator, version 6.4 US Edition; Perinatal Institute, Birmingham, UK). In this analysis, we utilized the EFW by ultrasound parameters (formula of Hadlock et al ) and determined a custEFW using this same formula. The frequency of infants who were small for gestational age (SGA), defined as EFW <10th%ile, vs large for gestational age (LGA), defined as EFW >90th%ile, were compared between methods.
Secondary outcomes studied were SGA by customized newborn birthweight, LGA by customized newborn birthweight, preterm birth (PTB) <37 weeks, PTB <34 weeks, cesarean delivery (CD) for fetal indications, neonatal intensive care unit (NICU) admission, CD for arrested labor, and birthweight >4000 g. We then evaluated the association of ultrasound performed at 24-28 and 28-32 weeks to actual birthweight >90th%ile or <10th%ile. SPSS statistical software (PASW Stats version 18.0; IBM Corp, Armonk, NY) was used for data analysis. We calculated and reported χ 2 , sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and likelihood ratios where appropriate. This study was approved by both university and hospital institutional review boards.
Results
During the 1-year study period, there are approximately 4800 births at our institution; 782 pregnancies met the study inclusion criteria and had complete data available for analysis. There were 93 ultrasounds performed between 24-28 weeks, 168 between 28-32 weeks, 325 between 32-36 weeks, and 196 performed at >36 weeks for which outcome data were available.
The patient demographic and clinical characteristics are summarized in Table 1 . Eighty-nine fetuses were classified as SGA and 42 as LGA by only custEFW in contrast to only popEFW, which identified 1 fetus as SGA and 2 fetuses as LGA that were not identified by the custEFW. On prenatal ultrasound, more fetuses were identified as SGA using custEFW percentile compared to popEFW percentile (15.1% vs 3.8%, P < .0001; relative risk [RR], 163.2; 95% confidence interval [CI], 22.4–1186.5). Of all fetuses identified as SGA by prenatal ultrasound, 48.4% diagnosed by the custEFW percentile and 14.8% diagnosed by the popEFW percentile were ultimately diagnosed as SGA at birth (RR, 7.1; 95% CI, 5.1–9.8; and RR, 13.3; 95% CI, 5.8–30.4, respectively). All secondary outcome measures (PTB <34 weeks, PTB <37 weeks, NICU admission, and CD for fetal indications) were also more frequent among those diagnosed as SGA by custEFW percentile ( Table 2 ). Additionally, more fetuses were identified as LGA using custEFW percentile compared to popEFW percentile (6.8% vs 1.7%; P < .0001; RR, 75.7; 95% CI, 17.2–332.5). Of those fetuses identified as LGA by prenatal ultrasound, 44.6% diagnosed by the custEFW percentile and 12.5% diagnosed by the popEFW percentile were ultimately diagnosed as LGA at birth (RR, 11.6; 95% CI, 7.3–18.4; and RR, 15.1; 95% CI, 5.3–43.5, respectively). The remaining outcomes are further characterized in Table 3 . We calculated sensitivity, specificity, PPV, NPV, and likelihood ratios for outcomes, which are further described in Tables 4 and 5 .
| Characteristics | n = 782 |
|---|---|
| Maternal age, y | 26.6 ± 6.1 |
| Race/ethnicity | |
| White | 155 (19.8) |
| Black | 361 (46.2) |
| Hispanic | 209 (26.7) |
| Other | 57 (7.3) |
| BMI, kg/m 2 | 32.6 ± 8.1 |
| Government insurance | 400 (51.2) |
| Birthweight, g | 2961 ± 705 |
| Nulliparity | 325 (41.6) |
| CD | 313 (40) |
| CD for fetal indications | 87 (11.1) |
| Preterm birth <37 wk | 192 (24.6) |
| Preterm birth <34 wk | 65 (8.3) |
| NICU admission | 123 (15.7) |
| Outcome | SGA by popEFW centile | SGA by custEFW centile | P value |
|---|---|---|---|
| PTB <37 wk, n = 192 | 10 (5.2%) | 56 (29.2%) | < .0001 |
| PTB <34 wk, n = 65 | 0 (0%) | 19 (29.2%) | < .0001 |
| CD for fetal indications, n = 87 | 8 (9.2%) | 24 (27.6%) | < .0001 |
| NICU admission, n = 123 | 13 (10.6%) | 46 (37.4%) | < .0001 |
| Outcome | LGA by popEFW centile | LGA by custEFW centile | P value |
|---|---|---|---|
| Birthweight >4000 g, n = 28 | 6 (21.4%) | 9 (32.1%) | .007 |
| Customized birthweight >90th%ile, n = 56 | 7 (12.5%) | 25 (44.6%) | .037 |
| CD for arrested labor, n = 68 | 2 (2.9%) | 7 (10.3%) | .009 |
| Variable | PopEFW <10th%ile | CustEFW <10th%ile |
|---|---|---|
| NICU admission | Sensitivity 10.6% | Sensitivity 37.4% |
| Specificity 97.4% | Specificity 89.1% | |
| PPV 43.3% | PPV 39% | |
| NPV 85.4% | NPV 88.4% | |
| LR+ 4.1 | LR+ 3.4 | |
| LR– 0.92 | LR– 0.70 | |
| CD for fetal indications | Sensitivity 9.2% | Sensitivity 27.6% |
| Specificity 96.8% | Specificity 86.5% | |
| PPV 26.7% | PPV 20.3% | |
| NPV 89.5% | NPV 90.5% | |
| LR+ 2.9 | LR+ 2.04 | |
| LR– 0.94 | LR– 0.84 | |
| PTB <37 wk | Sensitivity 5.2% | Sensitivity 29.2% |
| Specificity 96.6% | Specificity 89.5% | |
| PPV 33.3% | PPV 47.5% | |
| NPV 75.8% | NPV 79.5% | |
| LR+ 1.54 | LR+ 2.78 | |
| LR– 0.98 | LR– 0.79 | |
| Sensitivity 0.8% | Sensitivity 29.2% | |
| Specificity 95.8% | Specificity 86.2% | |
| PPV 1.6% | PPV 16.1% | |
| NPV 91.3% | NPV 93.1% | |
| LR+ 0.18 | LR+ 2.1 | |
| LR– 1.0 | LR– 0.82 |
| Variable | PopEFW>90th%ile | CustEFW>90th%ile |
|---|---|---|
| Customized birthweight >90th%ile | Sensitivity 12.5% | Sensitivity 44.6% |
| Specificity 99.2% | Specificity 96.1% | |
| PPV 53.8% | PPV 47.2% | |
| NPV 93.6% | NPV 95.7% | |
| LR+ 15.1 | LR+ 11.6 | |
| LR– 0.88 | LR– 0.58 | |
| Birthweight >4000 g | Sensitivity 21.4% | Sensitivity 33.3% |
| Specificity 99.1% | Specificity 94.2% | |
| PPV 46.2% | PPV 17% | |
| NPV 97.1% | NPV 97.5% | |
| LR+ 23.1 | LR+ 5.7 | |
| LR– 0.79 | LR– 0.71 | |
| CD for arrested labor | Sensitivity 2.9% | Sensitivity 10.3% |
| Specificity 97.9% | Specificity 93.5% | |
| PPV 28.5% | PPV 30.4% | |
| NPV 78.4% | NPV 80% | |
| LR+ 1.44 | LR+ 1.58 | |
| LR– 0.99 | LR– 0.96 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
