Article Text

Download PDFPDF

Neonatal mortality and morbidity in extremely preterm small for gestational age infants: a population based study
  1. S H Westby Wold1,
  2. K Sommerfelt1,2,
  3. H Reigstad2,
  4. A Rønnestad3,
  5. S Medbø4,
  6. T Farstad5,
  7. P I Kaaresen6,
  8. R Støen7,
  9. K T Leversen1,2,
  10. L M Irgens8,
  11. T Markestad1,2
  1. 1
    Department of Clinical Medicine, University of Bergen, Bergen, Norway
  2. 2
    Department of Pediatrics, Haukeland University Hospital, Bergen, Norway
  3. 3
    Department of Pediatrics, Rikshospitalet University Hospital, Oslo, Norway
  4. 4
    Department of Pediatrics, Ullevål University Hospital, Oslo, Norway
  5. 5
    Department of Pediatrics, Akershus University Hospital, Lørenskog, Norway
  6. 6
    Department of Pediatrics, University Hospital of Northern Norway, Tromsø, Norway
  7. 7
    Department of Pediatrics, St. Olav University Hospital, Trondheim, Norway
  8. 8
    Medical Birth Registry of Norway, Locus of Registry-Based Epidemiology, Department of Public Health and Primary Health Care, University of Bergen, Bergen, Norway
  1. Correspondence to S H Westby Wold, Department of Pediatrics, Barneklinikken, 5021 Haukeland University Hospital, Bergen, Norway; helenwold{at}


Aim: To assess if growth restricted (small for gestational age, SGA) extremely preterm infants have excess neonatal mortality and morbidity.

Methods: This was a cohort study of all infants born alive at 22–27 weeks’ post menstrual age in Norway during 1999–2000. Outcomes were compared between those who were SGA, defined as a birth weight less than the fifth percentile for post menstrual age, and those who had weights at or above the fifth percentile.

Results: Of 365 infants with a post menstrual age of <28 weeks, 31 (8%) were SGA. Among infants with a post menstrual age of <28 weeks, only chronic lung disease was associated with SGA status (OR 2.7, 95% CI 1.0 to 7.2). SGA infants with a post menstrual age of 26–27 weeks had excess neonatal mortality (OR 3.8, 95% CI 1.3 to 11), chronic lung disease and a significantly higher mean number of days (age) before tolerating full enteral nutrition. SGA infants with a post menstrual age of 22–25 weeks had an excess risk of necrotising enterocolitis.

Conclusion: Extremely preterm SGA infants had excess neonatal mortality and morbidity in terms of necrotising enterocolitis and chronic lung disease.

View Full Text

Statistics from

Whereas the neonatal and long-term outcomes of extremely preterm births in general have been extensively studied, the clinical significance of simultaneous sub-normal intrauterine growth is less well understood.1 2 Previous studies have often used birth weight cut-off inclusion criteria instead of post menstrual age, and have often not appropriately adjusted for the confounding effect of post menstrual age. This has led to some apparently contradictory findings, especially regarding neonatal morbidity including chronic lung disease (CLD).3 However, even in recent studies using appropriate study design and analysis, the effect of being small for gestational age (SGA) on the neonatal outcome of extremely preterm infants is unclear,1 2 although SGA seems to be associated with increased neonatal mortality.4 5 The importance of this issue for optimal timing of delivery has increased with the improving possibility of evaluating prenatal growth and feto-placental circulation using ultrasound and Doppler techniques.6 7

What this study adds

  • Intrauterine growth restriction in extremely preterm infants is associated with excess mortality and excess risk of necrotising enterocolitis in newborn infants.

The aim of the present study was to compare neonatal mortality and morbidity in extremely SGA and appropriate for gestational age (AGA) infants in a national, population based cohort born after the introduction of modern neonatal intensive care.


The data were derived from a cohort study comprising all births in Norway of 22 (from 22 weeks, 0 days) to 27 (maximum 27 weeks, 6 days) weeks of gestation or a birth weight of 500–999 g born between 1 January 1999 and 31 December 2000.8 In the present study we only included infants with a post menstrual age of <28 weeks in our main analysis. Virtually all pregnant women comply with the national standardised pregnancy follow-up program. Ultrasound screening is offered at a post menstrual age of 17–18 weeks. The expected date of delivery is based on this ultrasound examination if available. Post menstrual age at birth was calculated as completed weeks based on this ultrasound determination. If ultrasound examinations were unavailable (6%), post menstrual age was based on the date of last menstrual period.8

The infants were admitted to 15 different neonatal intensive care units (NICUs), and 95% of the infants received their initial treatment at the hospital where they were born. Routines regarding treatment and examinations of the newborns were left to the discretion of each neonatal unit. Routine cerebral ultrasound evaluations were, with few exceptions, performed at ⩽1 week of age for all surviving infants, and repeated at least once at ⩾3 weeks of age for all but a few infants. During the hospital stay, the infants were regularly monitored by an ophthalmologist.

The study was coordinated by the Medical Birth Registry of Norway (MBRN), which receives compulsory notification of all stillbirths and live births after 12 weeks of gestation.9 Data on maternal health, pregnancy, delivery and the newborn period until death of the infant or discharge home, were collected consecutively by local obstetricians and neonatologists. Completed forms were returned to the MBRN, and the data were verified and expanded through linkage to the routinely recorded MBRN data.8


SGA was defined as a birth weight less than the fifth percentile for post menstrual age, and AGA as birth weight greater than or equal to the fifth percentile for post menstrual age according to recently revised sex-specific standards for birth weight by post menstrual age in Norway. The fifth percentile was calculated from the published data using the following formula: Mean birth weight for each post menstrual age minus 1.645 times the standard deviation for that post menstrual age separately for each sex.10 Growth restriction at birth has been variously defined, ranging from the third to the tenth percentile in previous studies. Since the main aim of the present study was to focus on infants whose growth restriction was maternal (placental) and not due to “normal” genetic factors (such as small parents), the fifth percentile was chosen as a reasonable compromise.

Preeclampsia (including HELLP syndrome and eclampsia) corresponded to maternal systolic blood pressure ⩾140 mm Hg and/or diastolic blood pressure ⩾90 mm Hg and proteinuria. Prolonged rupture of membranes (PROM) was defined as leakage of amniotic fluid for more than 6 days before delivery. Prenatal steroids were recorded if given at least 24 h before delivery, or at least as two doses. Chorioamnionitis was recorded if clinically evident but not histologically verified.

Neonatal mortality was defined as death after admission to the NICU and before discharge home. Neonatal morbidity comprised severe cerebral haemorrhage (SCH), periventricular leucomalacia (PVL), necrotising entercolitis (NEC), retinopathy of prematurity (ROP), persistent ductus arteriosus (PDA), CLD and sepsis. Duration of parenteral feeds was measured as well as Apgar score.

SCH was defined as grade 3–4 according to Papile et al,11 and PVL as more than one or two cysts limited to one side. NEC was defined as being treated for either proven or suspected disease. For a diagnosis of proven NEC, intramural air or perforation was required. ROP was graded according to the Committee for the Classification of Retinopathy of Prematurity.12 All grades of ROP were analysed together, and as treated (with cryo- or laser therapy) or not. PDA included infants where PDA was suspected clinically and/or confirmed by echocardiography. Treated PDA included fluid restriction, non-specific medication such as digoxin and diuretics, prostaglandin inhibitors, and surgery. CLD was defined as needing oxygen or assisted ventilation at or beyond 36 weeks’ post menstrual age. Sepsis was defined as positive blood cultures in conjunction with clinical signs of systemic infection. In episodes with a monomicrobial coagulase-negative staphylococcus (CONS) isolate, or with polymicrobial isolates, an increase in C-reactive protein levels was required for classification as sepsis.13

Statistical analysis

Data were analysed using the SPSS v 15.0 statistical program for Windows. To compare the outcome of SGA and AGA infants, Fisher’s exact and χ2 tests were used for categorical variables and the two-sided Student`s t test for continuous variables. When cells in 2×2 tables were empty, only p values (Fisher’s exact test) were used. In addition to comparing outcome for SGA against AGA infants with a post menstrual age of <28 weeks, we stratified the analysis according to two post menstrual age groups: ⩽25 weeks and 26–27 weeks.

Multivariable logistic regression analysis was used when adjusting for the confounding effect of gestational age. It was performed to allow comparisons of results with previous studies using birthweight inclusion criteria.2 3

Rates are presented as percentages and odds ratios (OR) with 95% confidence intervals (CI). A two-tailed p value below 0.05 was considered significant. Rates of PVL were only analysed for infants surviving at least 28 days, and rates of ROP and CLD only for those surviving until at least 36 weeks’ post menstrual age.

The study was approved by the Regional Committee on Medical Research Ethics and the Norwegian Data Inspectorate. Written informed consent for participation was obtained from at least one of the parents for 99% of all infants transferred to an NICU but was not requested for infants who were stillborn or died in the delivery room.


Of a total of 119 611 births, 638 had a post menstrual age of less than 28 weeks or a birth weight of less than 1000 g. Of these, 174 (27%) were stillborn or died in the delivery room, and three were excluded from analysis because they had lethal malformations (one diaphragmatic hernia, one trisomy 18, and one trisomy 13). The parents of two infants refused participation, leaving 459 (99%) in the study. Of these, 365 infants had a post menstrual age of <28 weeks and were therefore included in the main analyses. Thirty one (8%) of these infants were SGA. Altogether, 94 infants with a post menstrual age of ⩾28 weeks had a birth weight of <1000 g. Of these infants, 55 (59%) were SGA with a mean post menstrual age of 29 weeks (28–32 weeks), and 39 (41%) were AGA with a mean post menstrual age of 28 weeks (28–29 weeks).

Among mothers with an SGA infant, 19% were smokers compared to 21% of AGA mothers, and thus no effect was observed of maternal smoking on SGA infants below 28 weeks (OR 1.9; table 1). However, preeclampsia was more frequent in SGA pregnancies (52%) than in AGA pregnancies (13%) with a strong association between preeclampsia and SGA (OR 7.4), and particularly at a post menstrual age of 22–25 weeks (OR 17.0). On the other hand, chorioamnionitis was less frequently reported in SGA cases (3%) than in AGA cases (25%) (OR 0.1; table 1). PROM was borderline less frequently in SGA cases (3%) compared to AGA cases (16%), however this was not statistically significant (OR 0.2, 95% CI 0.0 to 1.3), and abruptio placentae was not associated with SGA (table 1). Caesarean section was more frequent in SGA infants (90%) than in AGA infants (50%) (OR 9.3; table 1), while prenatal use of steroids occurred to the same extent in both groups, at 71% and 68%, respectively (OR 1.2).

Table 1

Risk factors for and perinatal outcome of SGA infants observed in a cohort born with a post menstrual age of <28 weeks and admitted to an NICU in Norway in 1999–2000

Altogether, 75 (21%) (10 SGA and 65 AGA infants) of the 365 infants with a post menstrual age of <28 weeks (21%) died (table 2). Mortality was significantly higher only for the SGA infants with a post menstrual age of 26–27 weeks (OR 3.8, 95% CI 1.3 to 11; p = 0.02). For the 10 SGA infants, causes of death (not mutually exclusive) were cerebral (four), respiratory (five), acute complication (three), sepsis (one) and NEC (one). For the 65 AGA infants, causes were respiratory (29), cerebral (20), sepsis (12), acute complication (seven) and NEC (five), or combinations of these.

Table 2

Neonatal outcome of SGA infants observed in a cohort born with a post menstrual age of <28 weeks admitted to an NICU in Norway in 1999–2000*

Neonatal morbidity

Among SGA infants, excess neonatal morbidity was observed for CLD, but not for SCH, PVL, ROP, PDA or sepsis (table 2). In the subgroup of infants with a post menstrual age of 22–25 weeks, NEC was significantly more common for the SGA infants (OR 7.2, 95% CI 1.5 to 34; p = 0.03). The mean number of days (age) until able to tolerate full enteral nutrition was significantly higher for the SGA compared to the AGA infants among the subgroup of infants with a post menstrual age of 26–27 weeks (table 2).

In the multivariable regression analyses, which included the 94 infants with a post menstrual age of ⩾28 weeks, controlling for the effect of post menstrual age, SGA status was associated with an almost fourfold increased mortality (OR 3.9, 95%, CI 1.8 to 8.4; p = 0.00), a threefold risk of NEC (OR 2.9, 95% CI 1.1 to 7.7; p = 0.03), a threefold risk of CLD (OR 3.3, 95% CI 1.7 to 6.5; p = 0.00), a twofold risk of sepsis (OR 1.9, 95% CI 1.0 to 3.6; p = 0.04) and a higher mean number of days (age) before tolerating full enteral nutrition (B 9.1, 95% CI 0.05 to 18; p = 0.04) compared to AGA status where B is the unstandardised regression coefficient, corresponding to difference in mean number of days. In analyses not controlling for post menstrual age, SGA status was not associated with increased mortality (OR 0.9, 95% CI 0.5 to 1.7; p = 1.00), NEC (OR 0.6, 95% CI 0.1 to 2.7; p = 0.75), CLD (OR 1.0, 95% CI 0.6 to 1.6; p = 1.00) or sepsis. However, in the same analyses not controlling for post menstrual age, SGA status was associated with increased risk of PDA (OR 0.4, 95% CI 0.2 to 0.7; p = 0.00), while no significant association was found when post menstrual age was controlled for.


In this national cohort of infants, with a post menstrual age of less than 28 weeks admitted to an NICU, SGA infants had excess neonatal mortality and excess risk of CLD and NEC.

Strengths of the present study were the prospective and population based design, and high participation rate in a country with generally high quality, and uniform pre-, peri- and neonatal care. Use of recently revised national birth weight by post menstrual age reference standards when defining SGA status strengthens the study.

In the assessment of risks in infants born after 27 weeks’ post menstrual age, the results may be inaccurate since only infants with birth weights of less than 1000 g were included, that is, a disproportionately high percentage of infants were SGA or moderately growth restricted within the AGA range.14 The consequence of this over-representation of moderately growth restricted infants in the AGA group may result in an underestimation of risks of mortality and morbidity associated with being SGA in this gestational age range. This is supported by findings of somewhat higher mortality and morbidity risks associated with SGA in this gestational age range in the large population based study of Regev et al where an appropriate AGA reference group was available.5

In agreement with previous studies and as expected, we found that preeclampsia and caesarean section were associated with SGA status, and that PROM and chorioamnionitis were associated with AGA status.5 7 15 A likely explanation is that perceived risks associated with preeclampsia and intrauterine growth restriction may have resulted in an active approach to early delivery, while expectation or active prolongation of pregnancy may have been chosen in cases of spontaneous labour or rupture of membranes. In addition, infection may in itself be a cause of spontaneous labour, extremely early birth and rupture of membranes.5

Excess mortality in the newborn period among infants born SGA in our study is in agreement with other post-surfactant studies when infants were included based on a post menstrual age5 16 or birth weight using appropriate control for post menstrual age in the analysis.4 In a recent large population based study, the mortality rate was approximately doubled for SGA infants when post menstrual age was less than 32 weeks.5 When we recalculated odds ratios for SGA associated mortality in that study according to the post menstrual age groups used in the present study, we found the following ORs: 2.3 (95% CI 1.1 to 4.7) for a post menstrual age of 24–25 weeks, 3.4 (2.1 to 5.3) for a post menstrual age of 26–27 weeks, 3.9 (2.5 to 6.0) for a post menstrual age of 28–29 weeks, and 1.7 (0.8 to 3.7) for a post menstrual age of 30–31 weeks. This pattern, with the highest SGA associated mortality in the gestational age range somewhat above what most perinatologists consider the threshold of viability, is similar to our findings. The explanation may be that obstetricians refrain from actively terminating pregnancy at the border of viability because of poor infant prognosis due to immaturity per se, while an active approach to delivery is chosen more often with increasing post menstrual age since intrauterine growth retardation (IUGR) then may be considered a greater risk than prematurity itself. The GRIT study gives some support to this notion in that obstetricians seemed to correctly balance these risks.7

In the present study, SGA was not associated with severe SCH or PVL. This is in accordance with most recent studies with post menstrual age inclusion criteria,2 5 17 but not all.1 This finding may suggest that cerebral vascular circulation is relatively spared in IUGR. However, PVL occurs rarely despite cerebral damage when post menstrual age is less than approximately 28 weeks18 and may therefore not be a good marker of brain pathology in the gestational age range in the present study.

Our finding that NEC occurred more frequently in the SGA infants agrees with some19 but not with other previous studies with post menstrual age inclusion criteria.5 20 The increased risk of NEC due to IUGR might be explained by intestinal ischaemia occurring before birth. With fetal circulatory compromise due to placental insufficiency, blood flow to the brain and some other organs is preserved at the expense of the supply to other areas, such as the intestines.21 22 Although plausible, sub-analyses performed in the present study did not reliably support this hypothesis since NEC was as common among SGA infants born of pregnancies with or without preeclampsia. This is in line with another study using fetal Doppler evaluation of fetal circulation.23

The findings of the present study support most previous studies with inclusion based on post menstrual age in that SGA was associated with CLD.5 16 24 25 However, the result in the present study should be interpreted as of borderline statistical significance since the lower limit of 95% CI is close to including 1.0. Some studies have suggested that SGA status might protect the lungs against respiratory distress syndrome (RDS) and CLD by inducing fetal lung maturation.2 3 26 Most likely this is the result of inappropriate control for the confounding effect of post menstrual age when using birth weight inclusion criteria,26 or selection bias due to lack of population based designs.3 Although RDS may predispose to CLD, recent findings indicate that CLD may develop in the absence of preceding RDS.16 This suggests that unknown factors associated with impaired intrauterine growth might be causative.

Previous findings of an association between SGA and neonatal sepsis are hampered by lack of population based designs.20 24 Thus, the finding in the present study of an almost doubled risk of sepsis among the SGA infants was only observed in the sample where a birth weight inclusion criterion was added and post menstrual age adjusted for in a less reliable multivariable regression analysis. Some previous studies with inclusion based on gestational age have found a similar association.20 24 These studies were, however, not population based leaving the finding uncertain. A larger, appropriately designed study is needed to adequately address this question.

We conclude that in extremely preterm infants, those born SGA with a post menstrual age of 26–27 weeks had an excess neonatal mortality and, to a lesser extent, neonatal morbidity compared to infants born AGA. These risks need to be taken into account when counselling parents and planning clinical interventions. The results underscore the need to further clarify the mechanisms involved in these excess risks. Such knowledge is important when choosing the optimal time of delivery in growth restricted fetuses.


Other members of the Norwegian Extreme Prematurity Study were as follows: Pediatrics: Inger E Silberg, Østfold Central Hospital, Fredrikstad; Theresa Farstad, Akershus Central Hospital, Lørenskog; Jørgen Hurum, Oppland Central Hospital, Lillehammer; Rugmini Palat, Hedmark Central Hospital, Elverum; Per A Tølløfsrud, Buskerud Central Hospital, Drammen; Alf Meberg, Vestfold Central Hospital, Tønsberg; Sveinung Slinde, Telemark Central Hospital, Skien; Marianne Skreden, Aust-Agder Central Hospital, Arendal; Kaare Danielsen, Vest-Agder Central Hospital, Kristiansand; Lars Stjernberg, Haugesund County Hospital, Haugesund; Jens Terum, Sogn and Fjordane Central Hospital, Førde; Steinar Spangen, Møre and Romsdal Central Hospital, Aalesund; Bjørn Myklebust, Innherred Hospital, Levanger; Ingebjørg Fagerli, Nordland Central Hospital, Bodø; Pal Ivan, Hammerfest Hospital, Hammerfest; Obstetrics: Lillian N Berge, Ulleval University Hospital, Oslo; Per E Børdahl, Rikshospitalet University Hospital, Oslo; Bjørg Lorentzen, Aker Hospital, Oslo; Ditlev Fossen, Østfold Central Hospital, Fredrikstad; Aasle-Marit Ullern, Akershus Central Hospital, Lørenskog; Jacob Nakling, Oppland Central Hospital, Lillehammer; Harry Aronsen, Oppland Central Hospital, Gjøvik; Turid Skjaeret Pedersen, Hedmark Central Hospital, Elverum; Erik Hovland, Kongsvinger Hospital, Kongsvinger; Liv Ellingsen, Buskerud Central Hospital, Drammen; Ulf Jan Andersen, Ringerike Hospital, Hønefoss; Halfdan Sundt, Vestfold Central Hospital, Tønsberg; Arne Christensen, Telemark Central Hospital, Skien; Arild Kloster Jensen, Aust- Agder Central Hospital, Arendal; Eli Smedvig, Rogaland Central Hospital, Stavanger; Torunn Eikeland, Haugesund County Hospital, Haugesund; Odd Harald Rognerud Jensen and Ingrid Borthen, Haukeland University Hospital, Bergen; Peer E Bjørgo, Voss County Hospital, Voss; Bjørg Ladehaug, Sogn and Fjordane Central Hospital, Førde; Paal Wølner-Hanssen, Laerdal Hospital, Laerdal; Arna Jaernbart, Nordfjordeid County Hospital, Nordfjordeid; Jørg Kessler, Møre and Romsdal Central Hospital, Aalesund; Ottar Rekkedal, Volda County Hospital, Volda; Runa Heimstad, St Olav’s University Hospital, Trondheim; Oddbjørn J Bolaas, Innherred Hospital, Levanger; Helge Brobak, Namdal Hospital, Namsos; Bjørn Holdø, Nordland Central Hospital, Bodø; Odd Andersen, Vefsn Hospital, Mosjøen; Margit Steinholt, Sandnessjoen Hospital, Sandnessjøen; Einar D Johansen, Harstad Hospital, Harstad; Ingar N Vold, Lofoten Hospital, Gravdal; Kristen Olav Lind, Stokmarknes Hospital, Stokmarknes; and Ingard Nilsen, University Hospital of Northern Norway, Tromsø.


View Abstract


  • Funding The study was funded by the Norwegian Foundation for Health and Rehabilitation through The Unexpected Child Death Society of Norway, the Research Council of Norway and Helse Vest Hospital Trust.

  • Competing interests None.

  • Ethics approval The study was approved by the Regional Committee on Medical Research Ethics and the Norwegian Data Inspectorate.

  • Patient consent Parental consent obtained.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.