Original Paper Published online: February 3, 2015
Fetal Diagn Ther DOI: 10.1159/000368829
Neonatal Acid-Base Status in Term Fetuses: Mathematical Models Investigating Cerebroplacental Ratio and Birth Weight José Morales-Roselló b Asma Khalil a José Alberola-Rubio b, c David Hervas-Marín d Maddalena Morlando a Amar Bhide a Aris Papageorghiou a Alfredo Perales-Marín b Baskaran Thilaganathan a
a
Fetal Medicine Unit, St George Hospital, London, UK; b Servicio de Obstetricia, Hospital Universitario y Politécnico La Fe, Valencia, c Grupo Bioelectrónica (I3BH), Universidad Politécnica, Valencia, and d Unidad de Bioestadística, Instituto de Investigación Sanitaria La Fe, Valencia, Spain
Key Words Fetal Doppler · Cerebroplacental ratio · Late-onset fetal growth restriction · Neonatal acid-base status · Neonatal pH
threaten fetal well-being at term: a very low birth weight and a very low fetal CPR. Our findings suggest that the importance of fetal hemodynamics in determining the acid-base status at birth surpasses that of fetal weight. © 2015 S. Karger AG, Basel
© 2015 S. Karger AG, Basel 1015–3837/15/0000–0000$39.50/0 E-Mail [email protected] www.karger.com/fdt
Introduction
The current approach to late-onset fetal growth restriction (FGR) focuses on small for gestational age fetuses (SGA). The majority of SGA fetuses result in healthy constitutional infants [1]. Poor outcome is more likely in very low-birth-weight (BW) fetuses (less than 3rd percentile), or in SGA fetuses presenting with abnormal fetal Doppler. This distinction between true FGR and constitutional SGA is relevant, as according to the most recent FGR management guidance, constitutional SGA fetuses do not need labor induction before 40 weeks [1]. The fact that fetal Doppler influences the prognosis of SGA fetuses prompted us to evaluate the existence of similar hemodynamic changes in appropriate (AGA) and large for gestational age fetuses (LGA) and reported the existence of Doppler changes in AGA and LGA fetuses, Prof. Baskaran Thilaganathan Fetal Medicine Unit, St George’s University of London Cranmer Terrace London SW17 0RE (UK) E-Mail basky @ pobox.com
Downloaded by: Lund University Libraries 130.235.136.21 - 5/12/2015 4:08:13 PM
Abstract Objective: Cerebroplacental ratio (CPR) is emerging as a marker of fetal hypoxia at term. The aim of this study was to demonstrate graphically the interrelationships among CPR, birthweight (BW), and neonatal pH, and construct 2D and 3D representations of the areas with potential low pH. Methods: This was a retrospective study of 2,927 term fetuses evaluated according to BW and CPR. The outcome was the acidbase status at birth. Multivariate relationships among CPR, BW, and arterial and venous pH were depicted in 3D scattergrams. Subsequently, trend surfaces were calculated and represented in 2D contour graphs. Finally, 3D representations were constructed by smothering pH data using moving average filters. Results: The trend surfaces and the 2D and 3D contour graphs showed the complex association among the three variables. Although pH changed with CPR and BW, the influence of the BW was smaller than the influence of the CPR, with this effect being more evident in the venous than in the arterial pH. Conclusions: Two scenarios
Materials and Methods This was a retrospective cohort study performed in two tertiary centers, including a database of 2,927 singleton term fetuses used in an earlier study that had an ultrasound scan within 14 days before delivery. The umbilical (UA) and middle cerebral (MCA) arteries were examined in all fetuses using color Doppler according to a standard protocol [7, 8], and CPR was calculated as the simple ratio between the MCA PI and the UA PI [9]. Only one (the last) examination per fetus was included. One third of the examinations belonged to unselected fetuses referred to the day assessment unit at La Fe Hospital and two thirds to fetuses referred to the fetal medicine unit at St. Georges’s Hospital. Regarding the latter, we appreciated that the data of this retrospective study were influenced by the biases of selective assessment of a population referred for scan assessments at term, a non-routine procedure in the United Kingdom. However, this was mitigated by the relatively large dataset of prospectively collected data and because the majority of women were, in fact, assessed routinely in an unselected fashion for a post-date scan as per local protocol. Ultrasound examinations were performed with Voluson E8/ E6/730 ultrasound machines using 2–8 MHz convex probes, during fetal quiescence, in the absence of fetal tachycardia, and keeping the insonation angle with the examined vessels as small as possible. Gestational age (GA) was determined according to the crown-rump length in the first trimester. Pregnancies complicated by congenital fetal abnormalities or stillbirths were excluded.
2
Fetal Diagn Ther DOI: 10.1159/000368829
After labor, BW was obtained and umbilical cord pH was immediately measured in the umbilical arterial and venous samples using Radiometer ABL80-FLEX and ABL800-FLEX pH meters. Although we did not stratify data according to the different labor outcome, we considered that circumstances influencing pH were equally distributed along the whole range of CPR MoM and BW centiles values, and therefore did not bias mathematical calculations. In order to correct for GA in weeks, BW values were converted into centiles using the method described earlier by Yudkin [10]. To simplify calculations, we used the same percentiles for male and female fetuses. However, this did not influence the results, as the error was small and equally distributed along the whole range of CPR MoM and BW percentiles. CPR values were converted into multiples of median (MoM) dividing values by the 50th percentile at each gestational age. CPR medians (50th percentile) were those used in a recent study [2] and were represented by the following equation: CPR 50th percentile = –1.3841 + (0.22659 × GA in weeks) – (0.003743 × GA in weeks [2]). Multivariate relationships among CPR, BW Centile, and arterial and venous pH values were depicted in three-dimensional (3D) scattergrams. Afterward, a trend surface was adjusted to the scattergrams by generalized additive models using thin plate smoothers [11]. Use of generalized additive models was motivated by the evident non-linear relationship among the three variables. The degrees of freedom of the smoothers, and consequently the flexibility of the fitted trend surfaces, were estimated by generalized cross-validation. To provide a better numerical interpretation of the models, the adjusted surfaces were also represented in 2D contour graphs with predictive variables in the axes and the corresponding pH values were represented as colored contour lines. In addition, in order to obtain a virtual 3D representation of the model, 3D surface contour graphs were constructed by smothering pH data using a moving average filter. However, the 3D generalized additive and 2D contour models proved the mathematical trend. The moving average filters 3D representation was applied to increase the comprehension of the scenario described earlier. Our aim was to represent fetal CPR, BW, and pH, using this final mathematical analysis in a relief map to reveal the interrelationships among the indicated parameters. Statistical analysis and graphs were performed with Matlab® R2012a (The Mathworks Inc®, Natick, Mass., USA) and R-software® (version 3.1.0), including package mgcv® (version 1.7–27).
Results
The study population included 2,927 pregnancies, of which 1,580 (54%) were male and 1,347 (46%) were female. The mean maternal age was 31.1 years (SD 5.5, range 14–52), and the mean BW was 3,471 g (SD 533, range 1,645–5,730). The mean gestational age at ultrasound examination was 40.2 weeks (SD 1.3, range 37– 41.9), and at delivery it was 40.9 weeks (SD 1.2, range Morales-Roselló et al.
Downloaded by: Lund University Libraries 130.235.136.21 - 5/12/2015 4:08:13 PM
indicating placental insufficiency. We described a novel hemodynamic model for the diagnosis of FGR based on Doppler CPR multiples of the median (MoM), instead of BW centiles [2]. This approach characterized for the first time two areas with potential risk: one based on the classical BW centile (size model) and another based on the existence of Doppler anomalies, independently of BW centile (hemodynamic model), which demonstrated that a proportion of AGA and LGA fetuses were, in fact, at risk of poor outcome. The finding that apparently ‘normal’ fetuses with low CPR MoM were failing to reach their growth potential was recently supported by the fact that AGA fetuses with low CPR presented more abnormal acid-base status than their counterparts with normal CPR [3]. Descriptions of adverse outcomes at birth are frequently dependent on subjective pediatrician diagnosis. Among quantifiable methods, acid-base status is probably the most extended way to evaluate neonatal outcome [4]. Both Doppler [5] and BW [6] are associated with birth pH, but the characteristics of these associations (magnitude and distribution) in SGA, AGA, and LGA remain unclear. The main aim of this study was to demonstrate graphically the interrelationships among fetal Doppler, BW, and neonatal pH, and construct 2D and 3D representations of those areas with potential low pH.
Color version available online
7.5 7.34 7.33
7.3 7.2
Venous pH
Venous pH
7.4
7.1 7.0
7.32 7.31 7.30 7.29
6.9 2.5 2.0
a
40 60 BW ce ntile
80
1.0 100
0.5
M
1.5
Mo
20
0
CP R
0
7.28
20
b
40 60 BW centile
80
100
2.5 2.0 1.5 oM 1.0 M 0.5 CPR
7.5 7.27
7.3
7.26
7.2
Arterial pH
7.1 7.0
7.24 7.23
6.9 2.5 2.0 60 BW centile
80
100
0.5
oM
1.5 1.0
40
7.22
0
RM
20
CP
0
c
7.25
d
20
40 60 BW centile
80
2.5 2.0 1.5 1.0 oM RM 100 0.5 P C
Fig. 1. a 3D scattergram of the venous pH with its trend surface calculated by mean of generalized additive modeling (R2 = 0.014, p < 0.001). b Venous pH trend surface calculated by mean of generalized additive modeling. The pH scale has been reduced to highlight the trend. A clear non-linear trend is shown, with the lowest pH values at the lowest BW Centile and CPR MoM values. c 3D
scattergram of the arterial pH with its trend surface calculated by mean of generalized additive modeling (R2 = 0.007, p < 0.001). d Arterial pH trend surface calculated by the mean of generalized additive modeling. The pH scale has been reduced to highlight the trend. A mild non-linear trend is shown, with the lowest pH values at the lowest BW Centile and CPR MoM values.
37–43.3). The mean interval between ultrasound and delivery was 5.1 days (SD 3.5, range 0–14). Results are shown in figures 1–3. The generalized additive model for venous pH included a smoother with 3.32 degrees of freedom (adjusted R2 = 0.014, p < 0.001). The trend surface with the scattered values is shown in
figure 1a. In figure 1b, the pH scale has been reduced to highlight the trend. The complex association among the three variables is shown in figure 2a, which represents a 2D contour graph of the additive model results for venous pH resembling the isobaric lines of a pressure map. Both ex-
Neonatal pH in Fetuses with Low CPR
Fetal Diagn Ther DOI: 10.1159/000368829
3
Downloaded by: Lund University Libraries 130.235.136.21 - 5/12/2015 4:08:13 PM
Arterial pH
7.4
2.0
1.0
1.0
0.5
0.5 20
40
60
80
0
100
BW centile
Fig. 2. 2D Contour graph representing the additive model results for venous (a) and arterial (b) pH depicted mimicking the isobaric lines of a pressure map. Both explicative variables are at the axes and the different pH values are represented as contour lines. The highest pH values are at the top-right side of the graph while the
7.31 7.29 7.27
7.25 1.10 1.05 1.00 90 100 CPR 0.950.90 60 70 80 50 Mo 30 40 0.85 M ntile 0.80 0 10 20 a BW ce
40
60
80
100
BW centile
lowest pH values are at the bottom-left side. Venous pH clearly changes with CPR MoM and BW centile values but their perceived effects are not similar, being the effect of the BW centile smaller than the effect of the CPR MoM. Arterial pH changes with CPR MoM but, in this case, BW centile effect is almost negligible.
7.35 7.33
20
b
b
7.28 7.26 7.24 7.22 7.20 7.18 1.10
1.05 1.00 CPR 0.950.90 Mo 0.85 M
70 50 60 30 40 tile n 20 e c BW 0.80 0 10
100 80 90
constructed using a 46 points moving average filter. Colors resemble the levels of a geographic relief map. Low umbilical cord pH values coincided with abnormal Doppler values (low CPR MoM)
and low BW centile. The lowest pH values occupied the confluence of these territories. Conversely, the highest pH values were situated in those territories where both CPR MoM and BW centile were normal. This phenomenon is more clearly seen in the analysis of the venous pH.
plicative variables are at the axes, and the different pH values are represented as contour lines. The highest pH values are at the top-right side of the graph, whereas the lowest pH values are at the bottom-left side. Venous pH clearly changes with CPR MoM and BW centile values, but their perceived effects are not similar, with the effect of the BW centile being smaller than that of the CPR MoM. This phenomenon explains why the lowest pH
values are expected at the bottom-left zone of the contour graph. The arterial pH model presented 2.25 degrees of freedom (adjusted R2 = 0.007, p < 0.001). The trend surface with the scattered values is shown in figure 1c. In figure 1d, the pH scale has been reduced to highlight the trend. Figure 2b represents the 2D contour graph of the additive model results for arterial pH. The highest pH values are
Fig. 3. 3D surface contour graphs representing the umbilical cord venous (a) and arterial (b) pH against BW centile and CPR MoM
4
Fetal Diagn Ther DOI: 10.1159/000368829
Morales-Roselló et al.
Downloaded by: Lund University Libraries 130.235.136.21 - 5/12/2015 4:08:13 PM
0
a
Umbilical cord venous pH
1.5
Color version available online
1.5
Color version available online
2.0 CPR MoM
2.5
Umbilical cord arterial pH
CPR MoM
2.5
Discussion
Graphically, all 2D and 3D representations depicted two areas of high risk for low arterial and venous pH: one including low BW centile, regardless of CPR MoM, and another including low CPR MoM values, regardless of BW centile. Regarding BW centile, recent research has shown a clear association between smallness for gestational age and poor outcome [12], especially in the presence of Doppler anomalies [13, 14]. However, studies evaluating pH or labor outcome in term AGA fetuses are scarce, probably because AGA fetuses have been considered free of potential risks associated with FGR. Regarding CPR, there are a few studies relating abnormal Doppler examination at term with poor outcome. In addition, most of them focus only on SGA fetuses [15, 16]. In a study published in the early 1990s, 215 fetuses were assessed days before labor using fetal Doppler. Of those, 16 and 17 were SGA and AGA fetuses with adverse perinatal outcomes, respectively. MCA Doppler was the best predictor of adverse outcome in AGA fetuses, whereas prediction was better achieved with UA Doppler and CPR in case of the SGA fetuses [17]. In a recent study, an association between low CPR and poor labor outcome was also established in AGA fetuses [18]. In this study, however, examinations were performed just before the active phase of labor and were not clearly independent of changes caused by the onset of labor and labor contractions [19–26]. Neonatal pH in Fetuses with Low CPR
Although one study described an increased frontal perfusion in AGA fetuses [27], studies have not investigated the interrelationship of fetal Doppler with BW and neonatal pH in term fetuses. Our study is novel in this regard, and in the sense that it applies a new approach to evaluate these associations constructing 2D and 3D representations. These representations present a scenario with two areas of potential risk: first, the presence of a low BW (which is already known), and second, the finding of an abnormal CPR (which is novel). Several shortcomings must be, however, acknowledged in this study. The first is that, as we mentioned earlier, we were not able to analyze the data according to the different modes of delivery due to the number of cases. Therefore, intrapartum factors influencing labor outcome such as the use of oxytocin or prostaglandins in case of induced labor were not evaluated. In addition, it could also be argued that the election of a 46-point moving average filter instead of a higher or lower one was an arbitrary decision. Although this is true, the 3D figure was created to visualize the result of our earlier mathematical model and in this regard, the 46-point filter provided the best visual approach. We would like to emphasize that the clinical application of the pH prediction model using these variables is likely to be poor, as the trends are not strong enough to perform an accurate pH prediction in all fetuses. BW cannot predict pH, but those fetuses with very low BW are prone to low pH. In the same way, CPR cannot predict pH but those fetuses with very low CPR are also prone to abnormal pH. Therefore, we can propose that those fetuses with very abnormal BW and/or very abnormal CPR are at a higher risk of adverse outcomes due to low neonatal pH. According to these findings, we suggest that, in addition to fetal biometry and estimated fetal weight, MCA PI and UA PI should be measured in all term fetuses to detect the presence of an abnormal CPR, a finding which, in case of very abnormal values, and independently of fetal biometry, could place the fetus at risk for abnormal neonatal pH. Finally, the fact that CPR is associated with arterial and venous pH more than BW highlights the importance of assessment of the fetal Doppler during ultrasound examination of the term fetus. In addition, the fact that this association is always higher for venous pH suggests that at least at term the venous pH is probably a better reflection of placental underperfusion than arterial pH [28–30]. In summary, our findings suggest that two scenarios may threaten fetal well-being at term: a very low fetal weight and a very low CPR, whose importance seems to surpass that of low fetal weight. Fetal Diagn Ther DOI: 10.1159/000368829
5
Downloaded by: Lund University Libraries 130.235.136.21 - 5/12/2015 4:08:13 PM
at the top-right side of the graph, whereas the lowest pH values are at the bottom-left side. Arterial pH also changes with CPR MoM but, in this case, BW centile effect is almost negligible. However, the lowest pH values are still found at the bottom-left zone of the graph. Figure 3a and b represent the 3D surface contour graphs of the venous and arterial pH against CPR and BW centile resembling the levels of a geographical relief map. Although several moving average filters were applied, the clearest 3D representations were obtained using a 46-point moving average filter. In general, low pH values coincided with two different areas: one including abnormal CPR measurements independently of fetal BW centile, and another including low BW centile values independently of fetal CPR. Interestingly, the lowest arterial and venous pH values were situated in the confluence of low CPR and BW, whereas the highest pH values were placed in the confluence of normal CPR and BW.
References
6
Fetal Diagn Ther DOI: 10.1159/000368829
13 Hershkovitz R, Kingdom JC, Geary M, Rodeck CH: Fetal cerebral blood flow redistribution in late gestation: identification of compromise in small fetuses with normal umbilical artery Doppler. Ultrasound Obstet Gynecol 2000;15:209–212. 14 Eixarch E, Meler E, Iraola A, Illa M, Crispi F, Hernandez-Andrade E, Gratacos E, Figueras F: Neurodevelopmental outcome in 2-yearold infants who were small-for-gestational age term fetuses with cerebral blood flow redistribution. Ultrasound Obstet Gynecol 2008;32:894–899. 15 Murata S, Nakata M, Sumie M, Sugino N: The Doppler cerebroplacental ratio predicts nonreassuring fetal status in intrauterine growth restricted fetuses at term. J Obstet Gynaecol Res 2011;37:1433–1437. 16 Mari G, Deter RL: Middle cerebral artery flow velocity waveforms in normal and small-forgestational-age fetuses. Am J Obstet Gynecol 1992;166:1262–1270. 17 Hecher K, Spernol R, Stettner H, Szalay S: Potential for diagnosing imminent risk to appropriate- and small-for-gestational-age fetuses by Doppler sonographic examination of umbilical and cerebral arterial blood flow. Ultrasound Obstet Gynecol 1992;2:266–271. 18 Prior T, Mullins E, Bennett P, Kumar S: Prediction of intrapartum fetal compromise using the cerebroumbilical ratio: a prospective observational study. Am J Obstet Gynecol 2013;208:124.e1–e6. 19 Morales-Roselló J, Hervás-Marín D, PeralesMarín A: Proximity of term labor deepens the fall of Doppler impedance in the fetal cerebral arteries. J Matern Fetal Neonatal Med 2014; 27:283–290. 20 Siristatidis C, Salamalekis E, Kassanos D, Loghis C, Creatsas G: Evaluation of fetal intrapartum hypoxia by middle cerebral and umbilical artery Doppler velocimetry with simultaneous cardiotocography and pulse oximetry. Arch Gynecol Obstet 2004;270:265–270. 21 Severi FM, Boni C, Bruni L, Bocchi C, Aguiar RA, Reis FM, Petraglia F: The increase of blood flow in the fetal middle cerebral artery correlates with the onset of labor at term. Reprod Sci 2008;15:584–590. 22 Siristatidis C, Salamalekis E, Kassanos D, Loghis C, Creatsas G: Evaluation of fetal intrapartum hypoxia by middle cerebral and umbilical artery Doppler velocimetry with simultaneous cardiotocography and pulse oximetry. Arch Gynecol Obstet 2004; 270: 265– 270.
23 Siristatidis C, Salamalekis E, Kassanos D, Creatsas G: Alterations in Doppler velocimetry indices of the umbilical artery during fetal hypoxia in labor, in relation to cardiotocography and fetal pulse oximetry findings. Arch Gynecol Obstet 2005;272:191–196. 24 Kassanos D, Siristatidis C, Vitoratos N, Salamalekis E, Creatsas G: The clinical significance of Doppler findings in fetal middle cerebral artery during labor. Eur J Obstet Gynecol Reprod Biol 2003;109:45–50. 25 Siristatidis C, Kassanos D, Salamalekis G, Creatsa M, Chrelias C, Creatsas G: Cardiotocography alone versus cardiotocography plus Doppler evaluation of the fetal middle cerebral and umbilical artery for intrapartum fetal monitoring: a Greek prospective controlled trial. J Matern Fetal Neonatal Med 2012; 25: 1183–1187. 26 Sütterlin MW, Seelbach-Göbel B, Oehler MK, Heupel M, Dietl J: Doppler ultrasonographic evidence of intrapartum brain-sparing effect in fetuses with low oxygen saturation according to pulse oximetry. Am J Obstet Gynecol 1999;181:216–220. 27 Mula R, Savchev S, Parra M, Arranz A, Botet F, Costas-Moragas C, Gratacos E, Figueras F: Increased fetal brain perfusion and neonatal neurobehavioral performance in normally grown fetuses. Fetal Diagn Ther 2013;33:182– 188. 28 Parra-Saavedra M, Crovetto F, Triunfo S, Savchev S, Peguero A, Nadal A, Parra G, Gratacos E, Figueras F: Placental findings in late-onset SGA births without Doppler signs of placental insufficiency. Placenta 2013; 34: 1136–1141. 29 Khalil AA, Morales-Rosello J, Elsadigg M, Khan N, Papageorghiou A, Bhide A, Thilaganathan B: The association between fetal Doppler and admission to neonatal unit at term. Am J Obstet Gynecol 2014;pii:S00029378(14)01055-2. DOI: 10.1016/j.ajog.2014. 10.013. Epub ahead of print. 30 Khalil AA, Morales-Rosello J, Morlando M, Hannan H, Bhide A, Papageorghiou A, Thilaganathan B: Is fetal cerebroplacental ratio an independent predictor of intrapartum fetal compromise and neonatal unit admission? Am J Obstet Gynecol 2014;pii:S0002-9378 (14)01079-5. DOI: 10.1016/j.ajog.2014.10.024. Epub ahead of print.
Morales-Roselló et al.
Downloaded by: Lund University Libraries 130.235.136.21 - 5/12/2015 4:08:13 PM
1 Figueras F, Gratacós E: Update on the diagnosis and classification of fetal growth restriction and proposal of a stage-based management protocol. Fetal Diagn Ther 2014;36:86–98. 2 Morales-Roselló J, Khalil A, Morlando M, Papageorghiou A, Bhide A, Thilaganathan B: Changes in fetal Doppler indices as a marker of failure to reach growth potential at term. Ultrasound Obstet Gynecol 2014;43:303–310. 3 Morales-Roselló J, Khalil A, Morlando M, Bhide A, Papageorghiou A, Thilaganathan B: Poor neonatal acid-base status in term fetuses with low cerebroplacental ratios. Ultrasound Obstet Gynecol 2014. DOI: 10.1002/uog.14647. Epub ahead of print. 4 Thorp JA, Rushing RS: Umbilical cord blood gas analysis. Obstet Gynecol Clin North Am 1999;26:695–709. 5 Arduini D, Rizzo G, Romanini C, Mancuso S: Are blood flow velocity waveforms related to umbilical cord acid-base status in the human fetus? Gynecol Obstet Invest 1989;27:183–187. 6 Lindqvist PG, Molin J: Does antenatal identification of small-for-gestational age fetuses significantly improve their outcome? Ultrasound Obstet Gynecol 2005;25:258–264. 7 Acharya G, Wilsgaard T, Berntsen GK, et al: Reference ranges for serial measurements of umbilical artery Doppler indices in the second half of pregnancy. Am J Obstet Gynecol 2005;192:937–944. 8 Bahlmann F, Reinhard I, Krummenauer F, et al: Blood flow velocity waveforms of the fetal middle cerebral artery in a normal population: reference values from 18 weeks to 42 weeks of gestation. J Perinat Med 2002;30:490–501. 9 Baschat AA, Gembruch U: The cerebroplacental Doppler ratio revisited. Ultrasound Obstet Gynecol 2003;21:124–127. 10 Yudkin PL, Aboualfa M, Eyre JA, Redman CW, Wilkinson AR: New birthweight and head circumference centiles for gestational ages 24 to 42 weeks. Early Hum Dev 1987;15:45–52. 11 Wood SN: Stable and efficient multiple smoothing parameter estimation for generalized additive models. J Amer Statist Ass 2004; 99:673–686. 12 Savchev S, Sanz-Cortes M, Cruz-Martinez R, Arranz A, Botet F, Gratacos E, Figueras F: Neurodevelopmental outcome of full-term small-for-gestational-age infants with normal placental function. Ultrasound Obstet Gynecol 2013;42:201–206.
Fetal Diagn Ther DOI: 10.1159/000368829
Neonatal Acid-Base Status in Term Fetuses: Mathematical Models Investigating Cerebroplacental Ratio and Birth Weight José Morales-Roselló b Asma Khalil a José Alberola-Rubio b, c David Hervas-Marín d Maddalena Morlando a Amar Bhide a Aris Papageorghiou a Alfredo Perales-Marín b Baskaran Thilaganathan a
a
Fetal Medicine Unit, St George Hospital, London, UK; b Servicio de Obstetricia, Hospital Universitario y Politécnico La Fe, Valencia, c Grupo Bioelectrónica (I3BH), Universidad Politécnica, Valencia, and d Unidad de Bioestadística, Instituto de Investigación Sanitaria La Fe, Valencia, Spain
Key Words Fetal Doppler · Cerebroplacental ratio · Late-onset fetal growth restriction · Neonatal acid-base status · Neonatal pH
threaten fetal well-being at term: a very low birth weight and a very low fetal CPR. Our findings suggest that the importance of fetal hemodynamics in determining the acid-base status at birth surpasses that of fetal weight. © 2015 S. Karger AG, Basel
© 2015 S. Karger AG, Basel 1015–3837/15/0000–0000$39.50/0 E-Mail [email protected] www.karger.com/fdt
Introduction
The current approach to late-onset fetal growth restriction (FGR) focuses on small for gestational age fetuses (SGA). The majority of SGA fetuses result in healthy constitutional infants [1]. Poor outcome is more likely in very low-birth-weight (BW) fetuses (less than 3rd percentile), or in SGA fetuses presenting with abnormal fetal Doppler. This distinction between true FGR and constitutional SGA is relevant, as according to the most recent FGR management guidance, constitutional SGA fetuses do not need labor induction before 40 weeks [1]. The fact that fetal Doppler influences the prognosis of SGA fetuses prompted us to evaluate the existence of similar hemodynamic changes in appropriate (AGA) and large for gestational age fetuses (LGA) and reported the existence of Doppler changes in AGA and LGA fetuses, Prof. Baskaran Thilaganathan Fetal Medicine Unit, St George’s University of London Cranmer Terrace London SW17 0RE (UK) E-Mail basky @ pobox.com
Downloaded by: Lund University Libraries 130.235.136.21 - 5/12/2015 4:08:13 PM
Abstract Objective: Cerebroplacental ratio (CPR) is emerging as a marker of fetal hypoxia at term. The aim of this study was to demonstrate graphically the interrelationships among CPR, birthweight (BW), and neonatal pH, and construct 2D and 3D representations of the areas with potential low pH. Methods: This was a retrospective study of 2,927 term fetuses evaluated according to BW and CPR. The outcome was the acidbase status at birth. Multivariate relationships among CPR, BW, and arterial and venous pH were depicted in 3D scattergrams. Subsequently, trend surfaces were calculated and represented in 2D contour graphs. Finally, 3D representations were constructed by smothering pH data using moving average filters. Results: The trend surfaces and the 2D and 3D contour graphs showed the complex association among the three variables. Although pH changed with CPR and BW, the influence of the BW was smaller than the influence of the CPR, with this effect being more evident in the venous than in the arterial pH. Conclusions: Two scenarios
Materials and Methods This was a retrospective cohort study performed in two tertiary centers, including a database of 2,927 singleton term fetuses used in an earlier study that had an ultrasound scan within 14 days before delivery. The umbilical (UA) and middle cerebral (MCA) arteries were examined in all fetuses using color Doppler according to a standard protocol [7, 8], and CPR was calculated as the simple ratio between the MCA PI and the UA PI [9]. Only one (the last) examination per fetus was included. One third of the examinations belonged to unselected fetuses referred to the day assessment unit at La Fe Hospital and two thirds to fetuses referred to the fetal medicine unit at St. Georges’s Hospital. Regarding the latter, we appreciated that the data of this retrospective study were influenced by the biases of selective assessment of a population referred for scan assessments at term, a non-routine procedure in the United Kingdom. However, this was mitigated by the relatively large dataset of prospectively collected data and because the majority of women were, in fact, assessed routinely in an unselected fashion for a post-date scan as per local protocol. Ultrasound examinations were performed with Voluson E8/ E6/730 ultrasound machines using 2–8 MHz convex probes, during fetal quiescence, in the absence of fetal tachycardia, and keeping the insonation angle with the examined vessels as small as possible. Gestational age (GA) was determined according to the crown-rump length in the first trimester. Pregnancies complicated by congenital fetal abnormalities or stillbirths were excluded.
2
Fetal Diagn Ther DOI: 10.1159/000368829
After labor, BW was obtained and umbilical cord pH was immediately measured in the umbilical arterial and venous samples using Radiometer ABL80-FLEX and ABL800-FLEX pH meters. Although we did not stratify data according to the different labor outcome, we considered that circumstances influencing pH were equally distributed along the whole range of CPR MoM and BW centiles values, and therefore did not bias mathematical calculations. In order to correct for GA in weeks, BW values were converted into centiles using the method described earlier by Yudkin [10]. To simplify calculations, we used the same percentiles for male and female fetuses. However, this did not influence the results, as the error was small and equally distributed along the whole range of CPR MoM and BW percentiles. CPR values were converted into multiples of median (MoM) dividing values by the 50th percentile at each gestational age. CPR medians (50th percentile) were those used in a recent study [2] and were represented by the following equation: CPR 50th percentile = –1.3841 + (0.22659 × GA in weeks) – (0.003743 × GA in weeks [2]). Multivariate relationships among CPR, BW Centile, and arterial and venous pH values were depicted in three-dimensional (3D) scattergrams. Afterward, a trend surface was adjusted to the scattergrams by generalized additive models using thin plate smoothers [11]. Use of generalized additive models was motivated by the evident non-linear relationship among the three variables. The degrees of freedom of the smoothers, and consequently the flexibility of the fitted trend surfaces, were estimated by generalized cross-validation. To provide a better numerical interpretation of the models, the adjusted surfaces were also represented in 2D contour graphs with predictive variables in the axes and the corresponding pH values were represented as colored contour lines. In addition, in order to obtain a virtual 3D representation of the model, 3D surface contour graphs were constructed by smothering pH data using a moving average filter. However, the 3D generalized additive and 2D contour models proved the mathematical trend. The moving average filters 3D representation was applied to increase the comprehension of the scenario described earlier. Our aim was to represent fetal CPR, BW, and pH, using this final mathematical analysis in a relief map to reveal the interrelationships among the indicated parameters. Statistical analysis and graphs were performed with Matlab® R2012a (The Mathworks Inc®, Natick, Mass., USA) and R-software® (version 3.1.0), including package mgcv® (version 1.7–27).
Results
The study population included 2,927 pregnancies, of which 1,580 (54%) were male and 1,347 (46%) were female. The mean maternal age was 31.1 years (SD 5.5, range 14–52), and the mean BW was 3,471 g (SD 533, range 1,645–5,730). The mean gestational age at ultrasound examination was 40.2 weeks (SD 1.3, range 37– 41.9), and at delivery it was 40.9 weeks (SD 1.2, range Morales-Roselló et al.
Downloaded by: Lund University Libraries 130.235.136.21 - 5/12/2015 4:08:13 PM
indicating placental insufficiency. We described a novel hemodynamic model for the diagnosis of FGR based on Doppler CPR multiples of the median (MoM), instead of BW centiles [2]. This approach characterized for the first time two areas with potential risk: one based on the classical BW centile (size model) and another based on the existence of Doppler anomalies, independently of BW centile (hemodynamic model), which demonstrated that a proportion of AGA and LGA fetuses were, in fact, at risk of poor outcome. The finding that apparently ‘normal’ fetuses with low CPR MoM were failing to reach their growth potential was recently supported by the fact that AGA fetuses with low CPR presented more abnormal acid-base status than their counterparts with normal CPR [3]. Descriptions of adverse outcomes at birth are frequently dependent on subjective pediatrician diagnosis. Among quantifiable methods, acid-base status is probably the most extended way to evaluate neonatal outcome [4]. Both Doppler [5] and BW [6] are associated with birth pH, but the characteristics of these associations (magnitude and distribution) in SGA, AGA, and LGA remain unclear. The main aim of this study was to demonstrate graphically the interrelationships among fetal Doppler, BW, and neonatal pH, and construct 2D and 3D representations of those areas with potential low pH.
Color version available online
7.5 7.34 7.33
7.3 7.2
Venous pH
Venous pH
7.4
7.1 7.0
7.32 7.31 7.30 7.29
6.9 2.5 2.0
a
40 60 BW ce ntile
80
1.0 100
0.5
M
1.5
Mo
20
0
CP R
0
7.28
20
b
40 60 BW centile
80
100
2.5 2.0 1.5 oM 1.0 M 0.5 CPR
7.5 7.27
7.3
7.26
7.2
Arterial pH
7.1 7.0
7.24 7.23
6.9 2.5 2.0 60 BW centile
80
100
0.5
oM
1.5 1.0
40
7.22
0
RM
20
CP
0
c
7.25
d
20
40 60 BW centile
80
2.5 2.0 1.5 1.0 oM RM 100 0.5 P C
Fig. 1. a 3D scattergram of the venous pH with its trend surface calculated by mean of generalized additive modeling (R2 = 0.014, p < 0.001). b Venous pH trend surface calculated by mean of generalized additive modeling. The pH scale has been reduced to highlight the trend. A clear non-linear trend is shown, with the lowest pH values at the lowest BW Centile and CPR MoM values. c 3D
scattergram of the arterial pH with its trend surface calculated by mean of generalized additive modeling (R2 = 0.007, p < 0.001). d Arterial pH trend surface calculated by the mean of generalized additive modeling. The pH scale has been reduced to highlight the trend. A mild non-linear trend is shown, with the lowest pH values at the lowest BW Centile and CPR MoM values.
37–43.3). The mean interval between ultrasound and delivery was 5.1 days (SD 3.5, range 0–14). Results are shown in figures 1–3. The generalized additive model for venous pH included a smoother with 3.32 degrees of freedom (adjusted R2 = 0.014, p < 0.001). The trend surface with the scattered values is shown in
figure 1a. In figure 1b, the pH scale has been reduced to highlight the trend. The complex association among the three variables is shown in figure 2a, which represents a 2D contour graph of the additive model results for venous pH resembling the isobaric lines of a pressure map. Both ex-
Neonatal pH in Fetuses with Low CPR
Fetal Diagn Ther DOI: 10.1159/000368829
3
Downloaded by: Lund University Libraries 130.235.136.21 - 5/12/2015 4:08:13 PM
Arterial pH
7.4
2.0
1.0
1.0
0.5
0.5 20
40
60
80
0
100
BW centile
Fig. 2. 2D Contour graph representing the additive model results for venous (a) and arterial (b) pH depicted mimicking the isobaric lines of a pressure map. Both explicative variables are at the axes and the different pH values are represented as contour lines. The highest pH values are at the top-right side of the graph while the
7.31 7.29 7.27
7.25 1.10 1.05 1.00 90 100 CPR 0.950.90 60 70 80 50 Mo 30 40 0.85 M ntile 0.80 0 10 20 a BW ce
40
60
80
100
BW centile
lowest pH values are at the bottom-left side. Venous pH clearly changes with CPR MoM and BW centile values but their perceived effects are not similar, being the effect of the BW centile smaller than the effect of the CPR MoM. Arterial pH changes with CPR MoM but, in this case, BW centile effect is almost negligible.
7.35 7.33
20
b
b
7.28 7.26 7.24 7.22 7.20 7.18 1.10
1.05 1.00 CPR 0.950.90 Mo 0.85 M
70 50 60 30 40 tile n 20 e c BW 0.80 0 10
100 80 90
constructed using a 46 points moving average filter. Colors resemble the levels of a geographic relief map. Low umbilical cord pH values coincided with abnormal Doppler values (low CPR MoM)
and low BW centile. The lowest pH values occupied the confluence of these territories. Conversely, the highest pH values were situated in those territories where both CPR MoM and BW centile were normal. This phenomenon is more clearly seen in the analysis of the venous pH.
plicative variables are at the axes, and the different pH values are represented as contour lines. The highest pH values are at the top-right side of the graph, whereas the lowest pH values are at the bottom-left side. Venous pH clearly changes with CPR MoM and BW centile values, but their perceived effects are not similar, with the effect of the BW centile being smaller than that of the CPR MoM. This phenomenon explains why the lowest pH
values are expected at the bottom-left zone of the contour graph. The arterial pH model presented 2.25 degrees of freedom (adjusted R2 = 0.007, p < 0.001). The trend surface with the scattered values is shown in figure 1c. In figure 1d, the pH scale has been reduced to highlight the trend. Figure 2b represents the 2D contour graph of the additive model results for arterial pH. The highest pH values are
Fig. 3. 3D surface contour graphs representing the umbilical cord venous (a) and arterial (b) pH against BW centile and CPR MoM
4
Fetal Diagn Ther DOI: 10.1159/000368829
Morales-Roselló et al.
Downloaded by: Lund University Libraries 130.235.136.21 - 5/12/2015 4:08:13 PM
0
a
Umbilical cord venous pH
1.5
Color version available online
1.5
Color version available online
2.0 CPR MoM
2.5
Umbilical cord arterial pH
CPR MoM
2.5
Discussion
Graphically, all 2D and 3D representations depicted two areas of high risk for low arterial and venous pH: one including low BW centile, regardless of CPR MoM, and another including low CPR MoM values, regardless of BW centile. Regarding BW centile, recent research has shown a clear association between smallness for gestational age and poor outcome [12], especially in the presence of Doppler anomalies [13, 14]. However, studies evaluating pH or labor outcome in term AGA fetuses are scarce, probably because AGA fetuses have been considered free of potential risks associated with FGR. Regarding CPR, there are a few studies relating abnormal Doppler examination at term with poor outcome. In addition, most of them focus only on SGA fetuses [15, 16]. In a study published in the early 1990s, 215 fetuses were assessed days before labor using fetal Doppler. Of those, 16 and 17 were SGA and AGA fetuses with adverse perinatal outcomes, respectively. MCA Doppler was the best predictor of adverse outcome in AGA fetuses, whereas prediction was better achieved with UA Doppler and CPR in case of the SGA fetuses [17]. In a recent study, an association between low CPR and poor labor outcome was also established in AGA fetuses [18]. In this study, however, examinations were performed just before the active phase of labor and were not clearly independent of changes caused by the onset of labor and labor contractions [19–26]. Neonatal pH in Fetuses with Low CPR
Although one study described an increased frontal perfusion in AGA fetuses [27], studies have not investigated the interrelationship of fetal Doppler with BW and neonatal pH in term fetuses. Our study is novel in this regard, and in the sense that it applies a new approach to evaluate these associations constructing 2D and 3D representations. These representations present a scenario with two areas of potential risk: first, the presence of a low BW (which is already known), and second, the finding of an abnormal CPR (which is novel). Several shortcomings must be, however, acknowledged in this study. The first is that, as we mentioned earlier, we were not able to analyze the data according to the different modes of delivery due to the number of cases. Therefore, intrapartum factors influencing labor outcome such as the use of oxytocin or prostaglandins in case of induced labor were not evaluated. In addition, it could also be argued that the election of a 46-point moving average filter instead of a higher or lower one was an arbitrary decision. Although this is true, the 3D figure was created to visualize the result of our earlier mathematical model and in this regard, the 46-point filter provided the best visual approach. We would like to emphasize that the clinical application of the pH prediction model using these variables is likely to be poor, as the trends are not strong enough to perform an accurate pH prediction in all fetuses. BW cannot predict pH, but those fetuses with very low BW are prone to low pH. In the same way, CPR cannot predict pH but those fetuses with very low CPR are also prone to abnormal pH. Therefore, we can propose that those fetuses with very abnormal BW and/or very abnormal CPR are at a higher risk of adverse outcomes due to low neonatal pH. According to these findings, we suggest that, in addition to fetal biometry and estimated fetal weight, MCA PI and UA PI should be measured in all term fetuses to detect the presence of an abnormal CPR, a finding which, in case of very abnormal values, and independently of fetal biometry, could place the fetus at risk for abnormal neonatal pH. Finally, the fact that CPR is associated with arterial and venous pH more than BW highlights the importance of assessment of the fetal Doppler during ultrasound examination of the term fetus. In addition, the fact that this association is always higher for venous pH suggests that at least at term the venous pH is probably a better reflection of placental underperfusion than arterial pH [28–30]. In summary, our findings suggest that two scenarios may threaten fetal well-being at term: a very low fetal weight and a very low CPR, whose importance seems to surpass that of low fetal weight. Fetal Diagn Ther DOI: 10.1159/000368829
5
Downloaded by: Lund University Libraries 130.235.136.21 - 5/12/2015 4:08:13 PM
at the top-right side of the graph, whereas the lowest pH values are at the bottom-left side. Arterial pH also changes with CPR MoM but, in this case, BW centile effect is almost negligible. However, the lowest pH values are still found at the bottom-left zone of the graph. Figure 3a and b represent the 3D surface contour graphs of the venous and arterial pH against CPR and BW centile resembling the levels of a geographical relief map. Although several moving average filters were applied, the clearest 3D representations were obtained using a 46-point moving average filter. In general, low pH values coincided with two different areas: one including abnormal CPR measurements independently of fetal BW centile, and another including low BW centile values independently of fetal CPR. Interestingly, the lowest arterial and venous pH values were situated in the confluence of low CPR and BW, whereas the highest pH values were placed in the confluence of normal CPR and BW.
References
6
Fetal Diagn Ther DOI: 10.1159/000368829
13 Hershkovitz R, Kingdom JC, Geary M, Rodeck CH: Fetal cerebral blood flow redistribution in late gestation: identification of compromise in small fetuses with normal umbilical artery Doppler. Ultrasound Obstet Gynecol 2000;15:209–212. 14 Eixarch E, Meler E, Iraola A, Illa M, Crispi F, Hernandez-Andrade E, Gratacos E, Figueras F: Neurodevelopmental outcome in 2-yearold infants who were small-for-gestational age term fetuses with cerebral blood flow redistribution. Ultrasound Obstet Gynecol 2008;32:894–899. 15 Murata S, Nakata M, Sumie M, Sugino N: The Doppler cerebroplacental ratio predicts nonreassuring fetal status in intrauterine growth restricted fetuses at term. J Obstet Gynaecol Res 2011;37:1433–1437. 16 Mari G, Deter RL: Middle cerebral artery flow velocity waveforms in normal and small-forgestational-age fetuses. Am J Obstet Gynecol 1992;166:1262–1270. 17 Hecher K, Spernol R, Stettner H, Szalay S: Potential for diagnosing imminent risk to appropriate- and small-for-gestational-age fetuses by Doppler sonographic examination of umbilical and cerebral arterial blood flow. Ultrasound Obstet Gynecol 1992;2:266–271. 18 Prior T, Mullins E, Bennett P, Kumar S: Prediction of intrapartum fetal compromise using the cerebroumbilical ratio: a prospective observational study. Am J Obstet Gynecol 2013;208:124.e1–e6. 19 Morales-Roselló J, Hervás-Marín D, PeralesMarín A: Proximity of term labor deepens the fall of Doppler impedance in the fetal cerebral arteries. J Matern Fetal Neonatal Med 2014; 27:283–290. 20 Siristatidis C, Salamalekis E, Kassanos D, Loghis C, Creatsas G: Evaluation of fetal intrapartum hypoxia by middle cerebral and umbilical artery Doppler velocimetry with simultaneous cardiotocography and pulse oximetry. Arch Gynecol Obstet 2004;270:265–270. 21 Severi FM, Boni C, Bruni L, Bocchi C, Aguiar RA, Reis FM, Petraglia F: The increase of blood flow in the fetal middle cerebral artery correlates with the onset of labor at term. Reprod Sci 2008;15:584–590. 22 Siristatidis C, Salamalekis E, Kassanos D, Loghis C, Creatsas G: Evaluation of fetal intrapartum hypoxia by middle cerebral and umbilical artery Doppler velocimetry with simultaneous cardiotocography and pulse oximetry. Arch Gynecol Obstet 2004; 270: 265– 270.
23 Siristatidis C, Salamalekis E, Kassanos D, Creatsas G: Alterations in Doppler velocimetry indices of the umbilical artery during fetal hypoxia in labor, in relation to cardiotocography and fetal pulse oximetry findings. Arch Gynecol Obstet 2005;272:191–196. 24 Kassanos D, Siristatidis C, Vitoratos N, Salamalekis E, Creatsas G: The clinical significance of Doppler findings in fetal middle cerebral artery during labor. Eur J Obstet Gynecol Reprod Biol 2003;109:45–50. 25 Siristatidis C, Kassanos D, Salamalekis G, Creatsa M, Chrelias C, Creatsas G: Cardiotocography alone versus cardiotocography plus Doppler evaluation of the fetal middle cerebral and umbilical artery for intrapartum fetal monitoring: a Greek prospective controlled trial. J Matern Fetal Neonatal Med 2012; 25: 1183–1187. 26 Sütterlin MW, Seelbach-Göbel B, Oehler MK, Heupel M, Dietl J: Doppler ultrasonographic evidence of intrapartum brain-sparing effect in fetuses with low oxygen saturation according to pulse oximetry. Am J Obstet Gynecol 1999;181:216–220. 27 Mula R, Savchev S, Parra M, Arranz A, Botet F, Costas-Moragas C, Gratacos E, Figueras F: Increased fetal brain perfusion and neonatal neurobehavioral performance in normally grown fetuses. Fetal Diagn Ther 2013;33:182– 188. 28 Parra-Saavedra M, Crovetto F, Triunfo S, Savchev S, Peguero A, Nadal A, Parra G, Gratacos E, Figueras F: Placental findings in late-onset SGA births without Doppler signs of placental insufficiency. Placenta 2013; 34: 1136–1141. 29 Khalil AA, Morales-Rosello J, Elsadigg M, Khan N, Papageorghiou A, Bhide A, Thilaganathan B: The association between fetal Doppler and admission to neonatal unit at term. Am J Obstet Gynecol 2014;pii:S00029378(14)01055-2. DOI: 10.1016/j.ajog.2014. 10.013. Epub ahead of print. 30 Khalil AA, Morales-Rosello J, Morlando M, Hannan H, Bhide A, Papageorghiou A, Thilaganathan B: Is fetal cerebroplacental ratio an independent predictor of intrapartum fetal compromise and neonatal unit admission? Am J Obstet Gynecol 2014;pii:S0002-9378 (14)01079-5. DOI: 10.1016/j.ajog.2014.10.024. Epub ahead of print.
Morales-Roselló et al.
Downloaded by: Lund University Libraries 130.235.136.21 - 5/12/2015 4:08:13 PM
1 Figueras F, Gratacós E: Update on the diagnosis and classification of fetal growth restriction and proposal of a stage-based management protocol. Fetal Diagn Ther 2014;36:86–98. 2 Morales-Roselló J, Khalil A, Morlando M, Papageorghiou A, Bhide A, Thilaganathan B: Changes in fetal Doppler indices as a marker of failure to reach growth potential at term. Ultrasound Obstet Gynecol 2014;43:303–310. 3 Morales-Roselló J, Khalil A, Morlando M, Bhide A, Papageorghiou A, Thilaganathan B: Poor neonatal acid-base status in term fetuses with low cerebroplacental ratios. Ultrasound Obstet Gynecol 2014. DOI: 10.1002/uog.14647. Epub ahead of print. 4 Thorp JA, Rushing RS: Umbilical cord blood gas analysis. Obstet Gynecol Clin North Am 1999;26:695–709. 5 Arduini D, Rizzo G, Romanini C, Mancuso S: Are blood flow velocity waveforms related to umbilical cord acid-base status in the human fetus? Gynecol Obstet Invest 1989;27:183–187. 6 Lindqvist PG, Molin J: Does antenatal identification of small-for-gestational age fetuses significantly improve their outcome? Ultrasound Obstet Gynecol 2005;25:258–264. 7 Acharya G, Wilsgaard T, Berntsen GK, et al: Reference ranges for serial measurements of umbilical artery Doppler indices in the second half of pregnancy. Am J Obstet Gynecol 2005;192:937–944. 8 Bahlmann F, Reinhard I, Krummenauer F, et al: Blood flow velocity waveforms of the fetal middle cerebral artery in a normal population: reference values from 18 weeks to 42 weeks of gestation. J Perinat Med 2002;30:490–501. 9 Baschat AA, Gembruch U: The cerebroplacental Doppler ratio revisited. Ultrasound Obstet Gynecol 2003;21:124–127. 10 Yudkin PL, Aboualfa M, Eyre JA, Redman CW, Wilkinson AR: New birthweight and head circumference centiles for gestational ages 24 to 42 weeks. Early Hum Dev 1987;15:45–52. 11 Wood SN: Stable and efficient multiple smoothing parameter estimation for generalized additive models. J Amer Statist Ass 2004; 99:673–686. 12 Savchev S, Sanz-Cortes M, Cruz-Martinez R, Arranz A, Botet F, Gratacos E, Figueras F: Neurodevelopmental outcome of full-term small-for-gestational-age infants with normal placental function. Ultrasound Obstet Gynecol 2013;42:201–206.
