Most obstetric clinics have a program for the identification of small-for-gestational age (SGA) fetuses because of the increased risk of fetal complications that they present. We have a structured model for the identification and follow-up of SGA pregnancies. We aimed to determine whether the recognition of SGA antepartum improves fetal outcome.
All pregnancies at Malmö University Hospital from 1990 to 1998 (n = 26 968) were reviewed. SGA fetuses identified prior to delivery (n = 681) were compared with those not identified (n = 573). Also, all pregnancies with SGA fetuses were compared with those appropriate-for-gestational age (AGA) (n = 24 585). The risk of serious fetal complications (hypoxic encephalopathy grade 2 or 3, intracranial hemorrhage, Apgar score <4 at 5 min, neonatal convulsions, umbilical pH <7.0, cerebral palsy, mental retardation, stillbirth, intrapartum or infant death) was assessed with cross-tabulation and logistic regression analysis, adjusted for gestational age and degree of SGA.
When compared with SGA fetuses identified before delivery (54%), SGA fetuses not identified before delivery were characterized by a four-fold increased risk of adverse fetal outcome (odds ratio, 4.1; 95% CI, 2.5–6.8). Similarly, compared with AGA fetuses, SGA fetuses were associated with a four-fold increased risk of serious fetal complications.
Intrauterine growth restriction (IUGR) is a complication of pregnancy occurring during the second half of 3–5% of all pregnancies1. Fetuses that are growth restricted are more prone to serious outcome complications such as severe fetal distress, cerebral damage, long-term neurological sequelae and fetal death2–4. A large proportion of small-for-gestational age (SGA) pregnancies are IUGR and the major proportion of IUGR pregnancies are also SGA5. Therefore, most obstetric units have a program for identifying SGA pregnancies before delivery.
In a randomized study of fetuses identified with suspected SGA, surveillance with Doppler velocimetry of the umbilical artery at specified intervals depending on blood flow classes was shown to allow antenatal monitoring and obstetric intervention to be aimed more precisely compared with cardiotocography (CTG)6. The authors used surrogate parameters of fetal outcome such as induction of labor, emergency Cesarean delivery for fetal distress and admission to a neonatal intensive care unit. No randomized study has compared neonatal outcome between identified and unidentified cases of SGA. In addition, data are still lacking as to whether the antepartum recognition of SGA is effective in lowering serious outcome complications.
This study, therefore, was performed as a critical investigation of whether the recognition of SGA before delivery lessens the risk of adverse fetal outcome.
All pregnancies at Malmö University Hospital without chromosome aberrations during the 9-year period from 1990 to 1998 were included in this study (n = 26 935). (There were 33 cases of chromosomal aberrations during the study period that were not included.) The study was approved by the ethics committee of Lund University, and consent for inclusion was obtained from a newspaper inquiry approved by the ethics committee. Two women declined and 1057 were large-for-gestational age, leaving a study population of 25 876. Of these 24 585 pregnancies were appropriate-for-gestational age (AGA) and 1291 were SGA. For 37 women with an SGA pregnancy for whom we had full outcome information, the medical records could not be found; it could therefore not be determined whether these were identified antepartum, and these 37 were excluded in the comparisons in Tables 2 and 3.
In Malmö, gestational age is routinely estimated by ultrasound (at around 19 weeks of gestation) in 96% of all cases and by last menstruation date in the remaining 4%7. Gestational age was classified as extremely preterm (i.e. gestational age <224 days), preterm (≥224 but <259 days), term (≥259 but <294 days), and post-term (≥294 days).
We considered SGA to be fetuses SGA at birth. As a measurement of SGA, we calculated the birth-weight deviation, (weight at birth − expected birth weight by gestational age)/expected birth weight, expressed as a percentage1. Intrauterine expected fetal weight was based on the model by Marsál et al.1:
where f(x) denotes fetal weight (in grams) at a given gestational age (x in days)1. SGA was defined as birth-weight deviation ≤−22%, i.e. below −2 SD of a Swedish term reference population1. In this context, this fell close to the lowest 5th percentile of our pregnant population due to the increased prevalence of SGA in preterm pregnancies1. We defined moderate SGA as ≤−22% to >−27% in birth-weight deviation (used as a reference group for analysis of SGA pregnancies only); severe SGA as ≤−27% to >−33%; and extreme SGA as ≤−33%. We defined AGA as >−22% to < + 22% in birth-weight deviation.
The spectrum of blood velocity in the uterine artery was analyzed for uterine artery score according to Sekizuka et al.8, defined as: 0 or normal = normal pulsatility index (PI) and no signs of a notch in either uterine artery; 1 = either increased PI or a notch in one of the arteries; 2 = two parameters abnormal; 3 = three parameters abnormal; 4 = bilateral notch and increased PI in both uterine arteries.
Stillborn was defined as intrauterine death at ≥24 weeks of gestation, but prior to labor and before hospital admission. Intrapartum death was defined as death occurring after admission and during labor. Infant death was defined as death occurring after delivery, but before 1 year of age.
Methods for prenatal identification of SGA
All pregnant women were scheduled for late pregnancy ultrasound fetometry free of charge at 32 weeks of gestation. On this occasion, the fetus was measured for biparietal diameter (BPD), abdominal diameter (AD), and femur length (FL). Fetal weight was calculated according to the formula developed by Persson and Weldner9:
Intrauterine-estimated SGA (eSGA) included those fetuses ≤−22% estimated fetal weight deviation (i.e. true SGA and false-positive SGA). Women with eSGA fetuses were followed as described later. The remaining 20%, those with the lowest weight deviation (i.e. from ≤−6% to >−22% estimated fetal weight deviation), were considered at risk for SGA and were scheduled for re-examination at 37 weeks of gestation. Women who reached 42 gestational weeks were offered a third fetometric examination, including amniotic fluid index measurement.
Women were identified as SGA if they had an SGA infant at birth and:
•at least one ultrasound fetometry had indicated eSGA (i.e. ≤−22% estimated fetal weight deviation), or
•there had been a fall of 10% in fetal weight deviation units between examinations at 32 and 37 weeks of gestation, or
•a pathological Doppler examination of the umbilical artery indicated placental insufficiency, and
•they were identified at least 12 h before delivery. Thus, cases found to be SGA who were delivered immediately (≤12 h) were considered to have been identified at, rather than before, delivery.
Thus, cases in which there was a fall in symphysis-fundus curve (or other suspicion of small fetus) that was not considered an indication for ultrasound fetometry or a Doppler examination, were not considered identified antenatally. The above-mentioned variables are routinely checked at the delivery unit.
Management of SGA pregnancies
A Doppler velocimetry model of umbilical artery and uterine arteries was used for the surveillance of pregnancies identified as eSGA. The Doppler measurements were planned every 14 days up to three times a week depending on the severity of both growth restriction and Doppler flow changes, as previously described in detail6. Women with normal Doppler flow velocimetry in the umbilical and uterine arteries, and whose fetus was moderately SGA, were examined by Doppler velocimetry every 2 weeks until delivery. Cases of severe SGA were examined weekly, and those with extreme SGA twice-weekly, in addition to being admitted to hospital. Patients in blood flow class I of the umbilical artery10 (PI > mean + 2 SD) and those in class II (absence of end-diastolic flow) were examined at least twice a week. A finding of blood flow class III (absence of blood flow velocity throughout diastole or reversed diastolic flow) was regarded as an indication for delivery. However, in the case of extremely preterm pregnancies, the timing of delivery was decided individually. In addition to the Doppler velocimetry examinations, women with eSGA fetuses were scheduled for additional ultrasound fetometry examinations every second week. Among identified SGA cases, this program has been shown to be effective in lowering the incidence of inductions, antenatal maternal admissions and emergency Cesarean sections for fetal distress6. Furthermore, CTG (Sonicaid System 8000, Oxford Instruments, Surrey, UK) was usually done at each visit to the specialized maternal health care unit. In addition to the previously published program6, women with a uterine artery score of 2 were usually scheduled for a re-examination after 2 weeks, and those with a uterine artery score of 3 or more were seen weekly for repeated blood flow measurements, due to their increased risk11. This information was used by senior obstetricians for decision on mode and timing of the delivery. Cases of SGA were liberally monitored by continuous CTG throughout delivery.
Management of uncomplicated pregnancies (those assumed to be AGA)
Women at low risk were scheduled for six to 10 visits to their midwife during pregnancy. CTG was monitored at the time of admittance to the delivery ward (for 10–20 min), at 2–4-h intervals during the first stage of labor, and continuously throughout the third stage of labor.
Construction of the study groups and details on retrospective assignment
We followed a predetermined sequence in our investigation. First, the local registry was searched (the computerized delivery system, the manual delivery report, and the records at our neonatal ward). All relevant pregnancies were scrutinized and missing data regarding birth weight, Apgar score and gestational age at birth were entered. As far as possible, this work was conducted blind with respect to adverse fetal outcome. Second, the expected birth weight and birth-weight deviation were calculated using the above-mentioned formula. Third, our local data on fetal death were compared with those in the national birth registry, where over 99% of all deliveries in Sweden are recorded9. (In Sweden, every patient who has been hospitalized, even overnight, is registered in the national patient registry.) All children with a diagnosis of cerebral palsy or mental retardation born in Sweden within the study period were identified and cross-matched, by means of their unique Swedish personal registration number, with those delivered in Malmö. In this manner, all children with cerebral palsy or mental retardation who had been born in Malmö and were subsequently hospitalized anywhere in Sweden were identified. In order to determine children who died after delivery, we searched the local registry at the delivery and neonatal units. In addition, we also searched the national death registry, where all deaths in Sweden are registered, by birth date within the study period. In this way, all infants who died during the first year of life were identified and cross-matched with those born during the study period. Finally, when all other data were established, we determined whether the SGA pregnancy had or had not been identified before delivery.
Serious fetal complications
The International Classification of Diseases, 9th revision (ICD9) diagnosis numbers or the corresponding ICD10 classification as serious fetal complications were cerebral palsy (G80* and G81*), mental retardation (F79*, F82*) interventricular hemorrhage (772B, 772C), subdural hemorrhage (767A), neonatal convulsions (779A), hypoxic ischemic encephalopathy grade II or III (779B or C), or stillbirth (656E, 768A, 768B). Details of maternal age, state of parity and smoking habits were those obtained on a woman's first visit to the Maternal Health Service, usually at around 12 weeks of gestation. Data regarding Apgar score, umbilical pH, mode of delivery and gestational age at delivery were routinely obtained at birth, i.e. blinded to the knowledge of identification as SGA.
Relative risks were estimated by odds ratios (ORs). Bivariate ORs were calculated with cross-tabulation and 95% CIs. Adjusted ORs were calculated with logistic regression analysis using serious fetal complications as a dependent variable and the identification of SGA before delivery (yes/no) as an independent dichotomous variable. In addition, gestational age at birth and birth-weight deviation were also included as categorized independent variables, in order to adjust for these variables. AGA was used as the reference in comparing SGA with AGA pregnancies, and moderate SGA was the reference in comparing SGA pregnancies alone. Term pregnancy was used as the reference point regarding gestational age. For the purpose of presentation in Figure 1, the AGA group was subdivided into: normal group < + 22% to ≥−11%, 20th percentile group (<−11% to ≥−16.5%), and 10th percentile group (<−16.5% to >−22%). The calculation of difference between the normal group and the various subgroups, and the difference between identified and unidentified SGA pregnancies, was done in the adjusted logistic regression model. The data were analyzed using SPSS software 11.5 (Statistical Package Social Sciences (SPSS) Inc., Chicago, IL, USA). P-values <0.05 were considered statistically significant.
We assumed the rate of serious complication to be 10% among SGA pregnancies. A two-sided χ2 test with a 0.05 significance level applied to two equal groups of 600 pregnancies would yield a power of 87% to detect a halved incidence of serious complications in the group with antepartum recognition of SGA.
The number and risk of serious fetal complications among those with SGA and AGA are given in Table 1. Compared with AGA fetuses, SGA fetuses were at a four-fold increased risk (OR, 4.1; 95% CI, 3.2–5.0) for serious fetal complications, and about 8% of SGA fetuses suffered adverse fetal outcome. However, SGA fetuses were at an approximately seven-fold increased risk for death (either intrauterine or infant death). Compared with AGA fetuses, fetuses with moderate SGA, severe SGA, and extreme SGA were at a 4.2- (95% CI, 2.9–6.2), 7.0- (95% CI, 4.2–11.6), and 35.8- (95% CI, 20.4–62.8) fold increased risk of serious fetal complications, respectively. Similarly, compared with women delivering at term, those delivering extremely preterm, preterm and post-term were at a 24.7- (95% CI, 18.1–33.6), 2.9- (95% CI, 2.2–3.8), and 1.4- (95% CI, 1.0–1.9) fold increased risk of serious fetal complications (adjusted for SGA ‘yes/no’).
Table 1. Fetal outcome for pregnancies with small-for-gestational age (SGA) and appropriate-for-gestational age (AGA) fetuses
SGA (n = 1291) (n (%))
AGA (n = 24 585) (n (%))
Bivariate OR (95% CI)
Adverse outcome was defined as at least one of the presented variables.
The same infant might appear in more than one subgroup.
pH and APGAR at 5 min were not recorded for those stillborn. HIE 2-3, hypoxic ischemic encephalopathy grade 2 or 3; NA, not applicable.
Table 2 presents adverse fetal outcome among SGA fetuses, divided into those either ‘identified’ or ‘not identified’ before delivery. The risk is presented both as a bivariate OR and an adjusted OR for adverse fetal outcome. SGA fetuses not identified before delivery, when compared with those identified, were characterized by a four-fold increased risk of serious complications (OR, 4.1; 95% CI, 2.5–6.8). Fifty-four percent of all SGA pregnancies were identified as SGA in advance of delivery. Pregnancies with moderate SGA were identified in 44% (297/683) of all cases, and the proportions of severe SGA and extreme SGA were 62% (221/357) and 76% (163/214), respectively. Similarly, women who delivered extremely preterm were identified in 50% (30/60) of all pregnancies; the proportions for those delivering preterm, term and post-term were 73% (159/219), 49% (439/896), and 67% (53/79), respectively. Of the total number of SGA pregnancies, 92% (1185/1291) had at least one late ultrasound fetometry screening, as scheduled. Eighty of the women with SGA fetuses considered at risk for SGA at 32 weeks were not re-examined as scheduled: 32 because of premature delivery and 48 due to missed appointments. The levels and number of non-identified SGA pregnancies close to the −22% cut-off limit were as follows: −21% (n = 24); −20% (n = 24); −19% (n = 35); −18% (n = 30); −17% (n = 38); −16% (n = 21); −15% (n = 35).
Table 2. Fetal outcome for pregnancies identified or not identified as small-for-gestational age (SGA) before delivery*
In Figure 1 we present the main results in graphic form. The identified SGA pregnancies were at significantly lower risk of adverse fetal outcome, compared with those not identified, in all three SGA subgroups. Those with identified moderate SGA, identified severe SGA, and the 20th percentile subgroup were not at significantly increased risk compared with the normal group (P = 0.2, P = 0.3 and P = 0.1, respectively). However, the 10th percentile group and the identified extreme SGA group were both at significantly increased risk for adverse fetal outcome.
In Table 3 we present mode and timing of delivery versus fetal size, categorized as follows: identified as SGA, not identified as SGA and AGA. A large, expected difference between identified and unidentified SGA pregnancies was the increased prevalence of elective Cesarean delivery (13% vs. 1%, P < 0.001). Of interest in Table 3 is that almost half of the identified cases were delivered by Cesarean section, compared with 19% of those not identified and 8.8% of those AGA. Surprisingly, there was a significant difference in the incidence of instrumental vaginal delivery in pregnancies identified as SGA, compared with both the unidentified and AGA categories (3.4% vs. 7.5% and 5.9%, respectively).
Table 3. Mode of delivery among pregnancies identified or not identified as small-for-gestational age (SGA)
Not identified as SGA (n = 573) (n (%))
Identified as SGA (n = 681) (n (%))
AGA (n = 24 585) (n (%))
Mode of delivery
Normal vaginal delivery
20 960 (85.3)
Instrumental vaginal delivery
Cesarean delivery elective
Cesarean delivery acute
Timing of delivery
21 624 (88.0)
Around 50% of the cases with severe outcome among both identified and unidentified SGA pregnancies were preterm. The mean absolute risk of having an adverse outcome in an unidentified SGA pregnancy was 11.6%, ranging from 63% among extremely preterm cases to 7–8% among those at least at term. Thus, on average the numbers needing treatment to avoid one case of adverse fetal outcome might be estimated to be 11. In the group of excluded SGA pregnancies (n = 37), there were two cases with adverse outcome.
We found that an awareness of SGA before delivery, in combination with a structured program of surveillance for those identified as SGA, was related to a four-fold lowered risk of adverse fetal outcome. All three identified SGA subgroups were at significantly lower risk of adverse fetal outcome, compared with unidentified SGA pregnancies. In addition, all three identified SGA subgroups were at a risk roughly similar to that of the 10th percentile group, while the risk in unidentified SGA pregnancies rose exponentially with the severity of SGA (Figure 1).
Compared with AGA fetuses, SGA fetuses were shown to have a four-fold increased risk of adverse fetal outcome. Although no previous study has used the same index of adverse fetal outcome, the results are in agreement with previous reports; Thornberg et al. found SGA fetuses to be at a five-fold increased risk for birth asphyxia12. Previous reference data are lacking for comparison of identified and unidentified SGA fetuses.
We used a Doppler follow-up program to distinguish various causes of SGA. Small fetuses of small mothers and those small due to chromosomal aberration usually have normal Doppler tracings of umbilical and uterine arteries5. The rationale for using Doppler velocimetry studies for fetal surveillance of SGA is as follows: SGA due to placental insufficiency causes the fetus to preserve energy—a physiological adaptation to fetal starvation. It does this by centralizing the circulation to the fetal heart, brain and adrenals, which causes difficulties in monitoring. CTG monitors variations in the fetal heart rate, which in turn mirrors the cerebral control of the fetal heart. Since the fetal cerebral circulation is favored, the CTG might be normal in a non-stressed situation. However, during the stress of contractions, ominous late decelerations may be seen, showing the vulnerability of a severely compromised fetus. Doppler velocimetry examination of the umbilical arteries may measure the change in fetal circulation, and examination of the uterine arteries should measure changed resistance in the placental circulation. Pregnancies with signs of placental insufficiency should be handled in accordance with the recommendations of Almström et al6.
SGA pregnancies may be identified by several means; it remains to be established which method of identification is preferable. The symphysis–fundus curve is widely used, but is not sufficiently sensitive13. The model for identification of SGA in Malmö is based mainly on ultrasound fetometry screening in the 32nd week. In the present study of our non-selected pregnant population, 54% of all SGA pregnancies were identified antepartum. A large proportion of the unidentified SGA pregnancies were just below our cut-off point for intensified surveillance (i.e. −22% in weight deviation), and thus were not included in our program. If we had used the same cut-off point as Laurin and Persson (≤−15%), we would have had exactly the same identification percentage (71%)14. The cut-off limit was changed in the late 1980s from −15% to −22% in order to reduce the number of false positives. For one to increase the identification rate from 54% to 71%, almost double the number of pregnancies would have to be included in the Doppler program. To do so would require financial backing, and our Doppler program has been partly underwritten by research funding. We were surprised by the comparatively large number of post-term pregnancies not identified despite two scheduled ultrasound fetometries (n = 26). The reason was attributable mainly to missed examinations, which accounted for 16 of the 26 unidentified women.
Another common definition criterion for SGA is the lowest 10th percentile. In comparison with this, our definition of SGA would be considered as severe SGA. The lowest 10th percentile corresponds to a weight deviation of <−16.5%. Therefore, for the purpose of presentation, we subdivided the AGA group into normal, 20th percentile and 10th percentile groups in Figure 11, 15. The cut-off limits for moderate, severe and extreme SGA fetuses were −22%, −27% and −33%. In Malmö, we use an ultrasound-based method of calculating expected fetal weight, which takes into account the higher percentage of IUGR and pre-eclampsia among preterm pregnancies1. The high prevalence of SGA found among those who delivered extremely preterm (25%) and preterm (14%) explains the high prevalence of SGA in our study, 4.6% compared with 2.5% in a normal curve based on birth weights1, 8.
A major weakness of our study is its retrospective observational nature. However, it would be both practically and ethically difficult to perform a randomized controlled trial. Nevertheless, the major part of the management routine was highly structured and published in 1992 as a randomized controlled trial of identified SGA cases6. The study size had to be large enough to enable the use of poor fetal outcome parameters instead of surrogate variables. We initiated this study to evaluate critically whether our surveillance of identified SGA cases improved outcome. If we had found no differences in adverse fetal outcome among identified SGA cases in comparison with those not identified, we would have discontinued allocating resources to 32-week ultrasound fetometry screening. As the correct determination of gestational age is the first step in prenatal screening of SGA fetuses, one strength of our study was that we worked with a high proportion of pregnancies dated by ultrasound in early pregnancy8. A possible drawback of the study is that all infants with neurological handicaps might not have been identified because such a diagnosis is seldom made during the first years of life. In addition, mild cases may not have been hospitalized at all. However, we do not believe that any missing data is biased.
There is a also risk of introducing ascertainment bias. In order to minimize this, however, we tried to follow a predetermined sequence in our investigation. Since the risk of adverse fetal outcome is associated with the degree of SGA, and severely growth-restricted fetuses are more likely to be identified, we adjusted for weight deviation. Similarly, pregnancies with shorter gestational ages are at increased risk for serious fetal complications. Therefore, we adjusted for gestational age as well. It should be emphasized that in this study we dealt only with SGA (i.e. only the fetal size), used as a proxy for IUGR. Twenty-five women were included as identified when a pathological Doppler had been obtained as a result of other indications, but they were delivered before undergoing ultrasound fetometry. These women were handled at the delivery ward as if SGA and were therefore included. It should also be noted that estimation of fetal size was done with a model using abdominal diameter, biparietal diameter, and femur length, and not just abdominal circumference.
Most obstetric units have strategies for identifying SGA fetuses before delivery. Since the risk of adverse fetal outcome diminishes with antepartum awareness of SGA, a high proportion of antepartum-identified SGA pregnancies, in combination with a surveillance program for those identified, must be a goal of all obstetric units. The data indicate that deviation from the population-specific growth potential is a predictor of adverse outcome, which is in agreement with the theory of customized growth curves of Gardosi et al.16.
Our findings indicate that, compared with the risk for those SGA fetuses not identified antepartum, a structured antenatal program of identified SGA fetuses results in a reduced risk of adverse fetal outcome.
We are grateful for the discussions with Dr P.H. Persson and Dr Jana Brodzky. This study was made possible by ‘ALF-medel’ from the South Swedish Regional Health Authorities (Region Skåne).