Symphysis-fundal height curve in the diagnosis of fetal growth deviations Curva de altura uterina por idade gestacional e diagnóstico de desvios do crescimento fetal

OBJECTIVE: To validate a new symphysis-fundal curve for screening fetal growth deviations and to compare its performance with the standard curve adopted by the Brazilian Ministry of Health. METHODS: Observational study including a total of 753 low-risk pregnant women with gestational age above 27 weeks between March to October 2006 in the city of João Pessoa, Northeastern Brazil. Symphisys-fundal was measured using a standard technique recommended by the Brazilian Ministry of Health. Estimated fetal weight assessed through ultrasound using the Brazilian fetal weight chart for gestational age was the gold standard. A subsample of 122 women with neonatal weight measurements was taken up to seven days after estimated fetal weight measurements and symphisys-fundal classifi cation was compared with Lubchenco growth reference curve as gold standard. Sensitivity, specifi city, positive and negative predictive values were calculated. The McNemar χ2 test was used for comparing sensitivity of both symphisys-fundal curves studied. RESULTS: The sensitivity of the new curve for detecting small for gestational age fetuses was 51.6% while that of the Brazilian Ministry of Health reference curve was signifi cantly lower (12.5%). In the subsample using neonatal weight as gold standard, the sensitivity of the new reference curve was 85.7% while that of the Brazilian Ministry of Health was 42.9% for detecting small for gestational age. CONCLUSIONS: The diagnostic performance of the new curve for detecting small for gestational age fetuses was signifi cantly higher than that of the Brazilian Ministry of Health reference curve. DESCRIPTORS: Embryonic and Fetal Development. Uterus, growth & development. Gestational Age. Ultrasonography, Prenatal. Prenatal Diagnosis. Validation Studie. Uterine height curve and fetal growth Freire DMC et al Serial measurement of symphysis-fundal height (SFH) during pregnancy is a major clinical tool to evaluate fetal growth as well as to identify other pregnancy complications.a It is an easy-to-use low-cost tool, although it has not been widely confi rmed as a reliable measurement in the prenatal care routine. Belizan et al4 showed these serial measurements greatly improve the predictive capacity of detecting fetal growth abnormalities. The Brazilian Ministry of Health (BMoH) recommends the use of a standard SFH curve developed by Fescina et al.9 Normal uterine growth is within the 10th and the 90th percentiles.b However, studies have reported different SFH patterns for Brazilian populations,10,18,21 and these differences could be explained not only by specifi c characteristics of ethnic groups but also by other factors such as different measurement techniques, risk level, socioeconomic and nutritional conditions, gestational age estimates and number of examiners. RESUMO OBJETIVO: Validar curva de referência de altura uterina por idade gestacional para o rastreamento de desvios do crescimento fetal e comparar sua performance com a curva-padrão adotada pelo Ministério da Saúde do Brasil. MÉTODOS: Estudo observacional que envolveu 753 gestantes de baixo risco de João Pessoa, PB, entre março e outubro de 2006, com idade gestacional acima de 27 semanas. A altura uterina foi medida de acordo com técnica preconizada pelo Ministério da Saúde. O padrão-ouro foi o peso fetal, estimado pelo ultrassom com base na curva de referência brasileira por idade gestacional. Uma subamostra de 122 casos com pesos neonatais obtidos até sete dias depois da estimativa do peso fetal, a classifi cação da altura uterina foi comparada com a curva de Lubchenco como padrão-ouro. A sensibilidade, a especifi cidade e os valores preditivos positivo e negativo foram calculados. Para comparar o desempenho da sensibilidade entre ambas as curvas de altura uterina, utilizou-se o teste χ2 de McNemar. RESULTADOS: A sensibilidade da nova curva para a detecção de fetos pequenos para a idade gestacional foi de 51,6%, enquanto a da curva do padrãoouro foi signifi cativamente menor (12,5%). Na subamostra que teve o peso neonatal como padrão-ouro, a sensibilidade da nova curva de referência foi de 85,7%, enquanto a do Ministério da Saúde exibiu 42,9% de sensibilidade para a detecção de fetos pequenos para a idade gestacional. CONCLUSÕES: A capacidade diagnóstica da nova curva de referência para detectar fetos pequenos para a idade gestacional foi signifi cativamente melhor do que a da curva recomendada pelo Ministério da Saúde. DESCRITORES: Desenvolvimento Embrionário e Fetal. Útero, crescimento & desenvolvimento. Idade Gestacional. Ultrassonografi a Pré-Natal. Diagnóstico Pré-Natal. Estudos de Validação.

Serial measurement of symphysis-fundal height (SFH) during pregnancy is a major clinical tool to evaluate fetal growth as well as to identify other pregnancy complications.a It is an easy-to-use low-cost tool, although it has not been widely confi rmed as a reliable measurement in the prenatal care routine.Belizan et al 4 showed these serial measurements greatly improve the predictive capacity of detecting fetal growth abnormalities.The Brazilian Ministry of Health (BMoH) recommends the use of a standard SFH curve developed by Fescina et al. 9 Normal uterine growth is within the 10 th and the 90 th percentiles.b However, studies have reported different SFH patterns for Brazilian populations, 10,18,21 and these differences could be explained not only by specifi c characteristics of ethnic groups but also by other factors such as different measurement techniques, risk level, socioeconomic and nutritional conditions, gestational age estimates and number of examiners.

CONCLUSÕES:
A capacidade diagnóstica da nova curva de referência para detectar fetos pequenos para a idade gestacional foi signifi cativamente melhor do que a da curva recomendada pelo Ministério da Saúde.The relationship between SFH and gestational age can provide clinical indications of fetal growth deviations.However, studies have found inconsistent performance results. 4,9,11,19,21Despite the lack of a single reliable method to identify fetal growth deviations, combined clinical and ultrasound methods improve diagnosis. 5,6,15,18As for the diagnosis of fetal growth restriction, newborn's weight is only a fi nal measurement which does not allow any kind of intervention to manage fetal weight gain. 5,6On the other hand, fetal weight estimates using ultrasound provides data that enable to diagnose conditions that may be impairing fetal growth. 5,6However, the SFH curve is a non-invasive, easy-to-use, low-cost tool that may be used to support important clinical decisions for either diagnostic or therapeutic interventions in developing countries where access to sophisticated technological resources for the assessment of fetal growth is not always available.Although SFH curves are usually used to screen fetal growth restriction, there are concerns about their performance.

DESCRITORES:
Freire et al 10 have demonstrated that the UH growth curve developed with low risk pregnant women from Joao Pessoa (Northeastern Brazil), differs signifi cantly from that adopted by the BMoH 9 as the standard for the Brazilian population.
Differences between both curves, the one from Freire et al 10 and that from Fescina et al 9 used by the BMoH, raise the hypothesis that they perform differently when identifying normal fetal growth and any deviations.Therefore, since one of the main applications of SFH measurements in prenatal care is for screening fetal growth deviations, a the standard curve may not be effective enough for this purpose, over or under diagnosing those conditions because it is based in a sample of women with a pattern of fetal growth different from that of the Brazilian population.The objective of the present study was to validate the SFH growth curve as a standard method for the screening of fetal growth deviations and to compare its performance with the pattern curve adopted by the Brazilian Ministry of Health.

METHODS
The validation of a SFH reference curve for gestational age was based on a curve developed by Freire et al. 10 The present study was carried out in a university hospital and a public maternity hospital in the city of João Pessoa, Northeastern Brazil, from March to October 2006.These hospitals were selected as they are reference centers of prenatal care in the public health service and they have higher demand of pregnant women.Eligibility of pregnant women was assessed in the ultrasound department of these hospitals.The sample size was calculated based on an estimated prevalence of 12% intra-uterine growth restriction in this population.a An 80% sensitivity for SFH measurements was assumed for detection of actual fetal growth restriction in the sample, with a sampling error of 9.2% and type I error of 0.05.The estimated sample size was 750 pregnant women.A sample of 753 pregnant women was selected according to free demand for each prenatal service.
The inclusion criteria included gestational age above 26 weeks, with a known last menstrual period or confi rmed by ultrasound during the fi rst half of pregnancy.Women who were pregnant with twins, fetal death or any fetal malformations that prevented the assessment of fetal weight and the amniotic fl uid index by ultrasound were excluded.The following variables were studied: SFH, gestational age, ultrasound-derived estimated fetal weight (EFW), and fetal growth deviation (categorized by gestational age in small -lower than the 10 th percentile; large -higher than the 90 th percentile; and adequate -between the 10 th and the 90 th percentiles).Data was also collected for the following control variables: age; body mass index (BMI, classifi ed by reference values for each gestational age according to Atalah et al 1 as recommended by the BMoH); ethnic group/skin color; number of pregnancies; parity; previous caesarean sections; previous abortions and smoking habits.a Fetal weight was estimated by ultrasound using Hadlock et al formula 13 as described elsewhere. 5he gold standard for classifying EFW was adapted from the curve developed by Cecatti et al 5 with normal EFW percentiles established by ultrasound according to gestational age in a Brazilian population using third degree polynomial adjustments.We opted for this method due to high linear correlation previously found between EFW and neonatal weight. 6H measurements were taken in pregnant women attending prenatal care visits for routine ultrasound examination.Data was recorded in standard forms by the examiner.Ultrasound scans were performed with GE ® Voluson 730 PRO and GE ® Logic 500 equipment, both with multifrequency (2.5-5.0MHz) convex probes.
For statistical analysis, SFH and EFW values for each pregnant woman in the study were plotted on their related charts as a function of gestational age and were compared to related standards.To validate Freire et al SFH growth curve, 10 sensitivity, specifi city, positive and negative predictive values were calculated for screening fetal growth deviation.The gold standard for diagnosing SGA and large for gestational age (LGA) fetuses was the classifi cation of fetal weight using ultrasoundderived EFW reference curve developed by Cecatti et al. 5 The same procedure was then performed with the BMoH SFH curve.For the comparison of sensitivity, the McNemar's χ 2 test was used for non-independent samples with Yates' correction.
In a subsample of 122 women SFH measurements were taken up to seven days before delivery.Additionally, SFH classifi cation was compared with neonatal weight classifi cation based on Lubchenco et al's curve 17 as the gold standard.SFH sensitivity for the diagnosis of fetal growth deviations was also compared with that of the BMoH standard SFH curve.A level of signifi cance of 5% was set.For statistic analysis EpiInfo 2000 and Excel™ software were used.

RESULTS
The age of the 753 women studied ranged between 13 and 45 years old, with mean age of 26.5.The predominant ethnic group/skin color was white (46.6%);BMI ranged between 18.9 kg/m 2 and 37.8 kg/m 2 with mean BMI of 27 kg/m 2 .Most of them did not smoke (97.6%).Table 1 shows that 47.5% were primigravida, 50% had never had a delivery and 17.2% had previous cesarean sections.Gestational age ranged between 29 and 41 weeks, and most were over 32 weeks of pregnancy.
Figure 1 shows the distribution of SFH measurements against the 10 th percentile and the 90 th percentile in Freire et al SFH curve. 10The majority of cases were considered adequate (64.3%) and 10.5% were SGA.Compared to Cecatti et al curve, 5 the proportion of adequate growth increased (70.1%) while the proportion of SGA (8.5%) and LGA (21.4%) decreased.
The sensitivity of SFH curve for the diagnosis of SGA and LGA, using the ultrasound-derived EFW curve as gold standard, is outlined in Table 2. Table 3 shows the sensitivity of SFH curve for the diagnosis of SGA and LGA using the neonatal weight curve as gold standard.The sensitivity of Freire et al 11 curve for detecting SGA was 51.6% while that of the BMoH curve was only 12.5%, a statistically signifi cant difference.When using actual neonatal weight, the sensitivity of the SFH curve was also higher for detecting SGA (85.7%) than that of the BMoH curve (42.9%).

DISCUSSION
This study validated the SFH curve developed by Freire et al 10 and compared its performance with that of the BMoH curve a in a population of pregnant women under conditions similar to those seen in public health facilities in Brazil.This sample characteristic may be attributed to BMI distribution in the population as a signifi cant association between BMI and SFH measurement was seen (data not shown).Almost half of pregnant women with fetal growth lying above the 90 th percentile was classifi ed as overweight or obese, a higher proportion than those below the 90 th percentile in the SFH curve (23.1%) (p<0.0001;data not shown).The BMI in the sample seems to refl ect the behavior of ultrasound-derived EFW, which showed 8.5% of SGA and 21.4% of LGA.BMI was also signifi cantly associated with EFW (p<0.001;data not shown) as the proportion of overweight or obese pregnant women was signifi cantly higher in the LGA (41.0%) than in the non-LGA group (25.7%).This may explain the high prevalence of LGA in the sample, as well as refl ect the so-called "obesity epidemics" as currently described in Western societies.
In general, studies on fetal growth curves among Brazilian women 19 use Hadlock et al curve 13 as a reference.However, a fetal growth curve created from a specifi c Brazilian population would be more appropriate, so the present study used the EFW curve, which was developed based on a sample of Brazilian women 5,6 to assess SFH for diagnosing SGA and LGA.There is a wide variation in the performance of SFH measurements as a method for screening fetal growth  resource in prenatal care.Ideally, all measurements of these different studies could be pooled in a common Brazilian reference SFH curve.
According to Freire et al, 10 the SFH growth curve of pregnant women from public hospitals in the city of João Pessoa signifi cantly differs from the standard curve recommended by the BMoH. 9Some other studies have investigated curves developed from a Brazilian population of pregnant women, however these curves differ methodologically from the present curve, especially in the SFH measurement technique and variation between examiners. 18though Neilson 20 claimed there is a lack of evidence indicating any real benefi t of using SFH measurement in prenatal care, many countries have prenatal care protocols recommending its use along with reference curves designed for the population of interest.b Despite technological improvements, the real benefi t of obstetric ultrasound, when used in the clinical routine as a screening method for fetal growth restriction in low-risk populations, is not yet supported by strong scientifi c evidence. 2,14Considering that even in developed countries it is estimated that approximately 50% of SGA fetuses are only identifi ed at birth or death, and that this number is likely to be higher in developing countries, it is reasonable to accept the use of a low-cost method like the SFH measurement, even without excellent sensitivity.In a recent study Lindqvist & Molin 16 showed that SGA fetuses that were not diagnosed during prenatal care have a four times greater risk of developing unfavorable outcomes than those diagnosed during prenatal care.This data emphasizes the need for assessing feasible screening methods that are appropriate to the reality of motherinfant health care.
One of the limitations of the present study is that we could not correlate neonatal outcome in the entire sample as only 122 cases (16.2%) had neonatal weight determined up to seven days after SFH and EFW measurements.However, a high linear correlation between EFW and neonatal weight (r = 0.94) was found in this subsample, which was considered to be more than enough to ascertain a strong association. 11herefore, this fi nding supported the classifi cation of EFW as the gold standard to diagnose fetal growth deviation in this study.Another limitation is regarding its cross-sectional design.As each pregnant woman had her SFH measurement assessed only once, it was not possible to evaluate the individual standard of SFH growth.Hypothetically, some undetected cases of SGA (false negatives) could have a reduction of SFH between two consecutive measurements.This hypothesis is consistent with new conceptual aspects of fetal growth restriction, which consider detecting a reduction in fetal weight a more important diagnostic criterion than the use of predetermined percentiles in fetal weight curves. 3,7,22sed on these results, important questions are posed in the area of public health.For instance, what would be the economic and perinatal impact of using a new and better SFH curve in the Brazilian population?A nationwide cohort study on maternal weight gain and fetal growth patterns would be the desirable for better understanding this relationship.
The study protocol was approved by the Institutional Review Board of the Centro de Ciências da Saúde da Universidade Federal da Paraíba (3/23/2005) and the Instituto e Maternidade Cândida Vargas (3/11/2005).

Table 1 .
Percentage distribution of pregnant women according to demographic and obstetric characteristics.João Pessoa, Northeastern Brazil, 2006.

Table 2 .
Performance of the symphysis-fundal height curve for the diagnosis of small for gestational age and large for gestational age (n=753) with the ultrasound-derived estimated fetal weight curve as gold standard and comparison with the symphysisfundal height curve of the Brazilian Ministry of Health.João Pessoa, Northeastern Brazil, 2006.

Table 3 .
Performance of the symphysis-fundal height curve for the diagnosis of small for gestational age and large for gestational age (n=122) with the neonatal weight curve as the gold standard and comparison with the symphysis-fundal height curve of the Brazilian Ministry of Health.João Pessoa, Northeastern Brazil, 2006.Distribution of 753 symphysis-fundal height measurements for the 10th percentile and the 90th percentile of Freire's curve.João Pessoa, northeastern Brazil, 2006.