Extreme Preterm Birth: Onset of Delivery and Its Effect on Infant Survival and Morbidity

Johanzon, Madelene BSc; Ödesjö, Helena MSc; Jacobsson, Bo MD, PhD; Sandberg, Kenneth MD, PhD; Wennerholm, Ulla-Britt MD, PhD

Obstetrics & Gynecology:
doi: 10.1097/01.AOG.0000295866.97499.35
Original Research

OBJECTIVE: To investigate whether correlations could be found between the onset of preterm delivery and infant outcome, that is, survival and major morbidity.

METHODS: The study was a retrospective, hospital-based cohort study. All women with a live fetus on admission, giving birth at 22+0 to 27+6 weeks of gestation between 1998 and 2003 were included. The deliveries were subdivided into those that began with either preterm labor, preterm premature rupture of membranes (PROM), or iatrogenic preterm delivery. These groups were compared for survival and survival without major morbidity (intraventricular hemorrhage grade 3–4, periventricular leukomalacia, retinopathy of prematurity grade 3–4, bronchopulmonary dysplasia, or necrotizing enterocolitis) at discharge.

RESULTS: The cause of the preterm birth was preterm labor in 154 of 288 (53%), preterm PROM 83 of 288 (29%), and iatrogenic preterm delivery 51 of 288 (18%). There were 83% liveborn children, and 67% survived until discharge. Survival was lower for preterm PROM (54%) than for preterm labor (75%) and iatrogenic preterm delivery (67%). Multivariable analyses showed that survival was positively associated with gestational age and antenatal steroid treatment. Negative associations concerning survival were found for preterm PROM and being small for gestational age. Survival without major morbidity did not differ significantly between the groups and was positively associated with gestational age and negatively associated with being small for gestational age.

CONCLUSION: Infant survival was significantly lower when the onset of preterm delivery was preterm PROM as compared with preterm labor and iatrogenic delivery. For surviving infants there was no significant difference in major morbidity between the groups.


In Brief

Survival rates are lower when the type of onset of delivery is preterm premature rupture of membranes than in other types of preterm delivery.

Author Information

From the Perinatal Center, Department of Obstetrics and Gynecology, the Institute of Clinical Sciences, Sahlgrenska University Hospital, Göteborg, Sweden.

Supported by a Swedish government grant (grant no. ALFGBG-5173) given to researchers in public health service and the Mary von Sydow Foundation (decision no. B432 616/06).

Corresponding author: Helena Ödesjö, Perinatal Center, Department of Obstetrics and Gynecology, the Institute of Clinical Sciences, Sahlgrenska University Hospital/Östra, SE - 416 85 Göteborg, Sweden; e-mail: helena.odesjo@gmail.com.

Financial Disclosure The authors have no potential conflicts of interest to disclose.

Article Outline

Preterm delivery, a major perinatal challenge today, is an area of great interest to obstetricians and neonatologists all over the world. Infants born extremely preterm have particularly high mortality and morbidity rates, although survival at lower gestational weeks and improvements in outcome have been observed in recent decades.1–5 The improvements are attributable to new technologies, caring methods, and organizational structures.6–8

Several reports have been published on the short- and long-term outcomes for infants born extremely preterm. However, comparisons between studies are problematic, owing to variations in study population, gestational age, whether fetal deaths are included, and the definitions of survival and morbidity used.4 Few studies also include maternal and antenatal factors in their analyses.1 One of the factors that may influence the outcome for the child is the onset of delivery, for example, spontaneous (preterm labor leading to preterm birth or preterm premature rupture of membranes) or iatrogenic preterm delivery. In previous studies using the Swedish National Birth Register, we have found that the subgroups of preterm birth in a Swedish population seem to be similar to those in other populations9 and that the child outcomes differ depending on the reason for onset of preterm delivery.10 To further investigate the details we performed an in-depth, hospital-based study of extremely preterm births. The aim of this study was to investigate whether correlations could be found between the onset of delivery and the outcome, that is, survival and major morbidity, of the extremely preterm born infants.

Back to Top | Article Outline


This study was carried out as a retrospective, hospital-based cohort study at Sahlgrenska University Hospital, Göteborg, Sweden. The Department of Obstetrics at Sahlgrenska University Hospital is a tertiary-level regional referral center serving Western Sweden, which has a population of 1.7 million inhabitants. This study was approved by the Regional Ethical Review Board in Göteborg.

Routinely, women in gestational weeks 23+0 to 27+0 with imminent delivery are transferred to Sahlgrenska University Hospital from regional county hospitals in Western Sweden. After 27+0 weeks of gestation, the infants are delivered at their regional county hospital.

All women with a live fetus on admission, giving birth at 22+0 to 27+6 weeks of gestation, between January 1, 1998, and December 31, 2003, were included. Multiple gestations, in utero transfers, and infants with congenital malformations were included. Pregnancy terminations were excluded. Infants delivered at another hospital and then admitted to neonatal care at Sahlgrenska University Hospital were not included.

Maternal and neonatal records were reviewed for live births and stillbirths. Maternal data included demographic information, antenatal history, pregnancy complications, medical treatment, perinatal data, and onset of delivery.

The deliveries were classified into three groups according to the onset of preterm delivery: preterm labor, preterm premature rupture of membranes (PROM), and iatrogenic preterm delivery. When there was more than one reason for preterm delivery, the classification was based on the reason considered to be the principal one according to the chronologic order in which they occurred.

In 242 of 248 (98%) of the pregnancies, duration of gestation was based on routine ultrasound examination performed before the end of the 19th completed gestational week. For the remaining pregnancies, duration of gestation was based on last menstrual period.

Antenatal steroids (betamethasone) were administered to women with a high risk of preterm delivery within the next 7 days after 23 completed gestational weeks. Before 1999, two doses 12 hours apart were considered a complete course. A second course was sometimes administered if delivery did not take place within 7 days. In the latter part of the time period, a course consisted of two doses separated by 24 hours, and a second course was no longer given. In this study a woman was considered as having received steroid treatment if at least one dose of steroids was administered. After 23 weeks of gestation, tocolytic therapy (indomethacin, diclofenac, terbutaline, or atosiban) was given to women in preterm labor. Antibiotics were administered when indicated, for example, by clinical signs of infection, or as a prophylactic treatment during cesarean delivery. Cardiotocography was not routinely used before 25 completed gestational weeks, and cesarean delivery was not performed on fetal indication when the gestational age was less than 25 completed weeks of gestation.

Fetal or infant information comprised birth data such as single or multiple birth, mode of delivery, birth weight, and gender. For all liveborn children, data regarding mortality and morbidity were reviewed. Live birth was defined as having heartbeats at delivery and survival as being alive at discharge. Small for gestational age (SGA) was defined as an infant with a birth weight more than 2 standard deviations below the mean (less than –22%).11 Morbidity included intraventricular hemorrhage, periventricular leukomalacia (PVL), retinopathy of prematurity, infant respiratory distress syndrome, bronchopulmonary dysplasia, sepsis (culture proved or clinically highly suspected), necrotizing enterocolitis, and persistent ductus arteriosus. Intraventricular hemorrhage and PVL were diagnosed by cranial ultrasonography. Intraventricular hemorrhage was graded with the Papile scale from one to four. Periventricular leukomalacia was defined as cyst formation. Retinoscopy was performed by a pediatric ophthalmologist, and retinopathy of prematurity was scored from stage 1 to stage 4. Necrotizing enterocolitis was defined as radiologic findings of either intramural gas or free intraabdominal gas or operative findings of intestinal perforation. Bronchopulmonary dysplasia was defined as need of supplemental oxygen and abnormal pulmonary X-ray findings at a corrected gestational age of 36 weeks. Survival without major morbidity was defined as survival without suffering or having suffered from intraventricular hemorrhage grade 3–4, PVL, retinopathy of prematurity grade 3–4, bronchopulmonary dysplasia, or necrotizing enterocolitis at discharge from the neonatal intensive care unit.

For children transferred to other regional county hospitals for neonatal care, records have been acquired. For three infants, follow-up data were not available concerning morbidity. Survival information for these three infants has been retrieved from the Swedish population register.

All analyses were conducted with SPSS 13.0 (SPSS Inc., Chicago, IL) and SAS/STAT 8.2 (SAS Institute Inc., Cary, NC). Fisher exact test and exact χ2 test were used for comparisons between two and three groups, respectively, regarding dichotomous variables from mother and single birth. Corresponding analyses with continuous variables were performed by using Mann-Whitney U test for two-group comparisons and Kruskal-Wallis test for three-group comparisons. Normality of distribution was evaluated with the Kolmogorov-Smirnov and Shapiro-Wilk tests. For data on multiple births, a univariable logistic regression using the method of generalized estimating equations was used to adjust for the dependence between individuals. A compound symmetry covariance pattern was used to model the dependence within the mothers. To select independent predictors to survival and survival without major morbidity, a stepwise multiple logistic regression was performed with univariable significant predictors. These results were then adjusted and validated with multiple logistic regression using the generalized estimating equations method. Odds ratios with 95% confidence intervals (CIs) were calculated. All tests were two-tailed, and a P<.05 or a 95% CI not including 1.0 was considered statistically significant. Given 83 patients in the preterm PROM group and 154 patients in the preterm labor group, we get a power of 90% to detect a clinically relevant difference between 60% and 80%, respectively, for live birth, given alpha=0.05.

Back to Top | Article Outline


Our study population comprised 248 women with 288 fetuses alive at admission, of which 238 were live births. Singleton pregnancies were 206 of 248 (83%), twin pregnancies 41 of 248 (16%), and triplet pregnancies 1 of 248 (0.4%). In three twin pregnancies, only one fetus was alive at admission. According to the Swedish Medical Birth Registry,12 92,942 children were born in Western Sweden during the study period. Of these, 264 of 92,942 (0.25%) were born before 28 weeks of gestation; consequently, 238 of 264 (90%) were delivered at Sahlgrenska University Hospital. Of the children born before 28 weeks of gestation, 208 of 264 (79%) survived until 1 year of age, compared with 194 of 238 (82%) of the children in our study.

Demographics are presented in Table 1. Of the 248 mothers included, 116 (47%) were at least para 1, and 223 (90%) were without a history of previous preterm birth. The only variable that differed significantly in the three groups was multiple gestation (P=.018).

In our study, 218 of 248 (88%) women had one or more of the following complications: vaginal bleeding during the first or second trimester, cervical insufficiency, hypertensive disorder, placental abruption, clinical or histologic chorioamnionitis, umbilical cord complication, and breech presentation (Table 2). Significant differences were found between the groups in terms of first-trimester vaginal bleeding (P=.007), second-trimester vaginal bleeding (P<.001), hypertensive disorder (P<.001), clinical chorioamnionitis (P=.037), and histologic chorioamnionitis (P<.001). Table 2 also presents the data regarding mothers who received selected medical treatment. Significant differences among the three groups were found between tocolytic treatment (P<.001) and antibiotic treatment before (P<.001) and antibiotic treatment during delivery (P=.027).

Fetal and infant characteristics are presented in Table 3. Liveborn children were 238 of 288 (83%). Stillbirths occurred in 50 of 288 (17%), 28 of 288 (9.7%) died during their first week, and 9 of 288 (3.1%) during the period 8–28 days. The cause of the preterm birth was preterm labor in 154 of 288 (53%), preterm PROM in 83 of 288 (29%), and iatrogenic preterm delivery in 51 of 288 (18%). Significant differences for single births among the groups were found concerning the proportion of cesarean deliveries, gestational age, SGA infants, birth weight, and neonatal survival at 7 days, at 28 days, and at discharge.

Data on survival and survival without major morbidity per gestational week are shown in Table 4. The number of survivors at discharge was 194 of 288 (67%) for the whole group, 115 of 154 (75%) for preterm labor, 45 of 83 (54%) for preterm PROM, and 34 of 51 (67%) for iatrogenic preterm delivery. Significant difference for single births was found between the preterm labor group and the preterm PROM and iatrogenic groups (P=.007). Survival without major morbidity did not differ significantly among the groups. Survival rates, as well as survival without major morbidity, increase significantly with gestational age (odds ratio [OR] per gestational week 2.12, 95% CI 1.70–2.65, and 1.74, 95% CI 1.30–2.31, respectively). For the preterm PROM group, the median time from rupture of the membranes to delivery (latency time) was 8.5 days in 38 of 83 nonsurvivors and 6.0 days in 45 of 83. A multivariable analysis using the generalized estimating equations model for the preterm PROM group, with adjustment for gestational week, antenatal steroid treatment, and gender, showed that latency time was associated with a significant reduction of survival OR 0.94 (95% CI 0.91–0.98) per day (P=.008).

In Table 5 infant morbidity is presented according to the onset of delivery. A significant difference for single births between the groups was found for infant respiratory distress syndrome, with fewer infants with infant respiratory distress syndrome in the preterm PROM group compared with the iatrogenic group (P=.038). Bronchopulmonary dysplasia was significantly more common for single births in the iatrogenic group compared with the preterm labor and preterm PROM groups (P=.022). Ten of 191 (5.2%) infants did not have any of these outcomes.

Results of analyses of potential risk factors in relation to survival at discharge and survival without major morbidity are presented in Table 6. In univariable analyses, increased survival was associated with preterm labor, antenatal steroid treatment, tocolytic treatment, antibiotic treatment during delivery, cesarean delivery, female sex, and gestational age. Obstetric complications, chorioamnionitis, preterm PROM, and SGA were adversely associated with survival. In multivariable analysis, antenatal steroid treatment and gestational age were associated with increased survival. Preterm PROM and SGA were associated with reduced survival (preterm PROM: OR 0.24, 95% CI 0.12–0.48).

Gestational age was associated with increased survival without major morbidity in both univariable and multivariable analyses, and female sex was associated with increased survival without major morbidity in multivariable analysis (Table 6).

Back to Top | Article Outline


The main finding of this study is that survival is significantly lower when the onset of delivery is preterm PROM than when it is preterm labor or iatrogenic preterm delivery. For the infants who survive until discharge from the neonatal intensive care unit, there is no significant difference in major morbidity among the three groups.

We have created an overview of an extremely preterm born population from a tertiary level regional referral center. Wanting to have the complete population, and as opposed to many other studies, we included stillborn fetuses. The extremely preterm birth group is a heterogeneous population, and other studies have chosen other inclusion criteria, making it difficult to compare overall findings. In the present hospital-based study, the cause of preterm birth was preterm labor in 154 of 288 (53%), preterm PROM 83 of 288 (29%), and iatrogenic preterm delivery 51 of 288 (18%). A meta-analysis of the onset of preterm delivery shows wide variation between different studies, and the findings are influenced by population characteristics and inconsistencies in terminology.13 However, in a population-based study from Sweden, where spontaneous preterm singleton live births at less than 28 weeks of gestation constituted 50% and iatrogenic preterm births 17%,9 the proportion of iatrogenic preterm deliveries is comparable to ours. Comparisons are exacerbated, owing to the use of different subcategories in different studies. The remaining 33% in the referred study are categorized as intrauterine deaths, congenital malformations, multiple births, or unknown type of onset. Because in our study the majority of stillbirths, multiple births, and/or infants with congenital malformations are categorized as spontaneous births (88%), the same might apply to the referred study, which suggests that the incidences of iatrogenic and spontaneous preterm births are similar in the two studies and that the present hospital-based study has the same composition as the corresponding population-based study. Overall survival for all fetuses included in this study was 194 of 288 (67%), ranging from 1 of 20 (5%) in week 22 to 58 of 74 (78%) in week 25 and 41 of 47 (87%) in week 27.

Interestingly, in our study the chance of survival was significantly lower after preterm birth secondary to preterm PROM than for preterm labor and iatrogenic preterm delivery. To our knowledge, no previous studies have found this association, although Ananth et al14 found, in a register-based study, that preterm birth secondary to preterm PROM was associated with higher crude perinatal mortality than preterm labor, but not in relation to iatrogenic preterm delivery. In contrast, other studies have reported comparable outcomes according to birth by preterm labor, preterm PROM, and iatrogenic preterm delivery.17–19 There are various possible explanations for lower survival after preterm birth secondary to preterm PROM. Infections, cord accidents, and abruptio placentae may play a role. In our study, chorioamnionitis was found to be adversely related to survival, and antibiotic treatment during delivery was associated with higher survival rates. In our study, although antibiotic treatment was not routinely administered unless there were clinical signs of infection, 62% of the women in the preterm PROM group were treated with antibiotics before delivery. This level of antibiotic usage is in agreement with a recent survey of current practice in Australia.15 Histologic evidence of chorioamnionitis was present in 76% in the preterm PROM group in our study. In the most recent meta-analysis of studies concerning prophylactic antibiotic treatment in patients with preterm PROM, investigators found a statistically significant delay in delivery and a reduction in maternal and neonatal morbidity if patients with preterm PROM were given adjunctive antibiotics.16 No difference in perinatal death was seen. The largest study included in the meta-analysis has been criticized for not having any significant difference in the primary outcome (defined as a composite of neonatal death, chronic lung disease, or major cerebral abnormality on ultrasonography before discharge from hospital).17 However, a significant difference was found in a post hoc subgroup analysis of only singleton births where erythromycin treatment was associated with a reduction in the composite primary outcome. On the other hand, co-amoxiclav treatment was associated with a significantly higher rate of neonatal necrotizing enterocolitis than no co-amoxiclav treatment. In many clinical guidelines today, antibiotic therapy is recommended during expectant management of preterm PROM.18 Although it is possible that more aggressive adjunctive antibiotic therapy might have improved the survival rate in our preterm PROM group, it could also have led to higher morbidity levels for the surviving infants. In any case, in accordance with the discussion above, we did not find any data supporting differences in mortality with more frequent use of antibiotics.

The reduced rate of cerebral abnormality on ultrasonography before discharge in singletons treated with erythromycin in the ORACLE study17 is a novel finding, and we look forward to seeing the results of the ongoing long-term follow-up study from the UK MRC ORACLE group. In our study there was no significant difference in major morbidity between the groups except for the increased risk of infant respiratory distress syndrome in iatrogenic preterm delivery, which concurs with the findings in a previous register-based study,10 as well as the findings in other studies.18,19

In another Swedish study, the same inclusion criteria as in ours were used, but the study group was limited to 23–25 weeks of gestation.1–3 These papers concluded that active perinatal management in this group virtually eliminates stillbirths and deaths in the delivery room and that morbidity among these children is not higher than in comparable studies.4,19,20

Our results are of interest in that we have the same survival rates as the referred Swedish study in spite of the fact that we did not use the same active perinatal approach in pregnancies of less than 25 weeks of gestation during the study periods. In the referred study, survival at discharge was 37% in week 23, 61% in week 24, and 74% in week 25. The survival figures for the same weeks in our material are 30%, 64%, and 78%. With more active management, mortality seems to be mainly within the first week and, with a more expectant attitude within the first 24 hours, while the overall survival does not seem to be affected.

There are few randomized controlled trials evaluating the optimal care for women with pregnancy complications at the border of viability, for example, at 22–25 weeks of gestation. Most decisions are based on empirical and practical experience rather than on evidence-based data. For example, the optimal mode of delivery (vaginal or cesarean delivery) for the fetus and the mother in the periviable period is not known. In general, cesarean delivery for fetal indications is not recommended before 24+0, it is suggested after 25+0, but there is still a great deal of uncertainty about 24+0 to 25+0 weeks of gestation.21 Our practice has been to not perform cesarean delivery before 25 weeks of gestation on the basis of fetal indications. This implies that we do not perform cardiotocography before 25 weeks of gestation. We have chosen more active management concerning corticosteroid therapy, which was given routinely from 23+0 weeks of gestation to women with threatening preterm birth. In summary, we have found infant survival to be significantly lower when the onset of delivery was preterm PROM than in other types of preterm delivery.

Back to Top | Article Outline


1. Serenius F, Ewald U, Farooqi A, Holmgren PA, Hakansson S, Sedin G. Short-term outcome after active perinatal management at 23–25 weeks of gestation: a study from two Swedish tertiary care centres. Part 1: maternal and obstetric factors. Acta Paediatr 2004;93:945–53.
2. Serenius F, Ewald U, Farooqi A, Holmgren PA, Hakansson S, Sedin G. Short-term outcome after active perinatal management at 23–25 weeks of gestation: a study from two Swedish tertiary care centres. Part 2: infant survival. Acta Paediatr 2004;93:1081–9.
3. Serenius F, Ewald U, Farooqi A, Holmgren PA, Hakansson S, Sedin G. Short-term outcome after active perinatal management at 23–25 weeks of gestation: a study from two Swedish perinatal centres. Part 3: neonatal morbidity. Acta Paediatr 2004;93:1090–7.
4. Hack M, Fanaroff AA. Outcomes of children of extremely low birthweight and gestational age in the 1990s. Semin Neonatol 2000;5:89–106.
5. Marlow N. Neurocognitive outcome after very preterm birth. Arch Dis Child Fetal Neonatal Ed 2004;89:F224–8.
6. Sandberg K, Bokemark T, Wennergren M, Olegard R. Decreased mortality and reduced morbidity of premature infants: an effect of the organizational changes in Gothenburg [in Swedish]. Lakartidningen 2002;99:1700–3.
7. Chan K, Ohlsson A, Synnes A, Lee DS, Chien LY, Lee SK. Survival, morbidity, and resource use of infants of 25 weeks’ gestational age or less. Am J Obstet Gynecol 2001;185:220–6.
8. Johansson S, Montgomery SM, Ekbom A, Olausson PO, Granath F, Norman M, et al. Preterm delivery, level of care, and infant death in Sweden: a population-based study. Pediatrics 2004;113:1230–5.
9. Morken NH, Kallen K, Hagberg H, Jacobsson B. Preterm birth in Sweden 1973–2001: rate, subgroups, and effect of changing patterns in multiple births, maternal age, and smoking. Acta Obstet Gynecol Scand 2005;84:558–65.
10. Morken NH, Kallen K, Jacobsson B. Outcomes of preterm children according to type of delivery onset: a nationwide population-based study. Paediatr Perinat Epidemiol 2007;21:458–64.
11. Marsal K, Persson PH, Larsen T, Lilja H, Selbing A, Sultan B. Intrauterine growth curves based on ultrasonically estimated foetal weights. Acta Paediatr 1996;85:843–8.
12. Swedish Medical Birth Registry. The National Board of Health and Welfare. Available at: http://www.socialstyrelsen.se/Om_Sos/organisation/Epidemiologiskt_Centrum/index.htm. Retrieved October 22, 2007.
13. Savitz DA, Blackmore CA, Thorp JM. Epidemiologic characteristics of preterm delivery: etiologic heterogeneity. Am J Obstet Gynecol 1991;164:467–71.
14. Ananth CV, Joseph KS, Oyelese Y, Demissie K, Vintzileos AM. Trends in preterm birth and perinatal mortality among singletons: United States, 1989 through 2000. Obstet Gynecol 2005;105:1084–91.
15. Buchanan S, Crowther C, Morris J. Preterm prelabour rupture of the membranes: a survey of current practice. Aust N Z J Obstet Gynaecol 2004;44:400–3.
16. Kenyon S, Boulvain M, Neilson J. Antibiotics for preterm rupture of membranes. Cochrane Database Syst Rev 2003;(2):CD001058.
17. Kenyon SL, Taylor DJ, Tarnow-Mordi W. Broad-spectrum antibiotics for preterm, prelabour rupture of fetal membranes: the ORACLE I randomised trial. ORACLE Collaborative Group [published erratum appears in Lancet 2001;358:156]. Lancet 2001;357:979–88.
18. Premature rupture of membranes. ACOG Practice Bulletin No. 80. Clinical management guidelines for obstetrician-gynecologists. Obstet Gynecol 2007;109:1007–19.
19. Batton DG, DeWitte DB, Espinosa R, Swails TL. The impact of fetal compromise on outcome at the border of viability. Am J Obstet Gynecol 1998;178:909–15.
20. Battin M, Ling EW, Whitfield MF, Mackinnon M, Effer SB. Has the outcome for extremely low gestational age (ELGA) infants improved following recent advances in neonatal intensive care? Am J Obstet Gynecol 1998;15:469–77.
21. Higgins RD, Delivoria-Papadopoulos M, Raju TN. Executive summary of the workshop on the border of viability. Pediatrics 2005;115:1392–6.
© 2008 by The American College of Obstetricians and Gynecologists. Published by Wolters Kluwer Health, Inc. All rights reserved.