Preterm birth in a French population: the importance of births by medical decision


*Correspondence: Dr E. Papiernik, Maternité Port-Royal, 123 Boulevard de Port Royal, Paris, France.


This analysis describes the prevalence of preterm birth by medical decision among 50,307 live births from the district of Seine-Saint-Denis in France, using a classification that distinguishes between medically decided preterm births associated with premature rupture of membranes and those for other reasons. Thirty-seven percent of singleton and 28% of twin preterm births result from labour induction or a caesarean section in the absence of labour. One-quarter of singleton indicated preterm births are associated with premature rupture of membranes. Between 28 and 31 weeks of gestation, 40% of all singleton preterm births result from a medical decision not associated with premature rupture of membranes. The high levels of indicated preterm birth must be taken into account in evaluations of preterm birth rates and trends in developed countries.


The causes of preterm birth are multiple and, in many situations, the underlying cause of the delivery remains unknown. However, a growing proportion of preterm births results from a medical decision to end a pregnancy before term for maternal or fetal reasons. A major reason for this increase is the prevention of fetal death or severe impairment as a result of growth restriction. In these cases, the indicated preterm delivery is a therapeutic intervention to prevent a fetal death or to improve the prognosis for survival without impairment.

Studies on preterm birth have documented increases in deliveries by medical decision over the past 30 years for both singleton and multiple pregnancies1–4. In some countries, this increase may have contributed to a rise in the overall preterm birth rate3. Despite the importance of this phenomenon, data on the prevalence of indicated preterm births on the population level come from only a few studies and these estimates are not comparable as medical decisions associated with preterm premature rupture of membranes are sometimes included or excluded from the group of medically decided births. Although these deliveries result from a decision to induce labour or carry out a caesarean section before labour begins, the rupture of the membranes can be considered a spontaneous onset. More data on indicated preterm births are available from hospital studies5, but they cannot be extrapolated to a population level.

Our aim in this analysis is to describe the prevalence of births by medical decision for singleton and multiple births in a population-based cohort of French births. Data are from a birth registry in Seine-Saint-Denis, a district north of Paris. The population is predominantly working class with a large immigrant population. In this district, there is a longstanding and active perinatal health policy that provides access to prenatal and obstetric care to all pregnant women regardless of their employment, insurance or immigration status.


The analysis includes 50,307 live births after 21 weeks of gestation that occurred in the 20 maternity units in Seine-Saint-Denis from 1/10/1998 to 31/12/2000. Information on each birth is recorded in the ‘certificat du 8ème jour’, a health certificate required by law for all live births in France. Data items cover obstetric history, information on the current pregnancy, the delivery and the newborn. Information on the pregnancy and delivery are extracted from medical records and filled in by the midwife after the birth. The district of Seine-Saint-Denis put into place an experimental version of this certificate in 1998 with the aim of improving the recording of items and collecting more complete information on the pregnancy.

Gestation duration is based on the obstetric assessment as recorded in the medical records. This assessment is based on clinical and, in most cases, ultrasound data. Ninety percent of women have a visit in the first trimester and 93% have a dating scan. Only 123 (0.24%) births have a missing gestational age. Mode of onset of delivery is analysed based on whether medical interventions were used to induce the delivery, either through induction of labour or by caesarean section before the onset of labour. A separate category is created for medical decisions after preterm premature rupture of membranes. Preterm premature rupture of membranes was defined as rupture of membranes 12 hours or longer before the onset of labour before 37 weeks of gestation. Data enabling us to classify births by mode of onset were missing for 8.9% of preterm births. Preterm births with missing data had the same average gestational age as others (33.9 and 33.8, respectively) and the same distribution of twin and singleton births.

The analyses are carried out on the population of live births; decisions concerning the mode of onset of delivery are different for deaths occurring before the onset of labour.


Over the study period, the preterm birth rate among live births in Seine-Saint-Denis was 6.7 per 100 births, 5.4 for singletons, 46.4 for twins. Births from multiple pregnancies constituted 3% of all live births and 21% of preterm births.

Table 1 displays the number and proportion of births resulting from medical decisions to induce labour or to carry out a caesarean section before the onset of labour by gestation duration for singleton and twin births separately. Overall, 37% of singleton preterm births are associated with a medical decision. This proportion varies with gestational age at delivery: almost all singleton births have a spontaneous onset before 26 weeks; medical decisions account for over one-third of all births at 26 and 27 weeks increasing to over half between 28 and 31 weeks. One quarter of all medical decisions are associated with premature rupture of membranes; this proportion is higher at later gestations.

Table 1.  Onset of delivery by gestational age groups for singleton and preterm births (in %).
Gestational age groupsTotal (N)Spontaneous onset1 (1)Medical decision with premature rupture of membranes (2)Medical decision—other reasons (3)Total by medical decision (2 + 3)
  1. 1Includes premature rupture of membranes followed by spontaneous delivery.

Singleton births
<26 weeks3894.
26–27 weeks5966.16.827.133.9
28–29 weeks12749.65.544.950.4
30–31 weeks17847.810.741.652.2
32–33 weeks29956.29.034.843.8
34–35 weeks72264.811.923.335.2
36 weeks97267.29.423.532.8
Total <37 weeks239563.19.827.136.9
Twin births
<26 weeks26100.
26–27 weeks1989.50.010.510.5
28–29 weeks3778.45.416.221.6
30–31 weeks5466.70.033.333.3
32–33 weeks10067.
34–35 weeks22373.13.623.326.9
36 weeks18767.92.729.432.1
Total <37 weeks64668.33.324.828.0

The proportion of births resulting from a medical decision is lower for twins than for singletons (28%). However, there are fewer medical decisions associated with premature rupture of membranes and the proportion of medically decided births for other reasons than premature rupture of membranes is close to the proportion observed for singleton births (27%). Most preterm twin births between 26 and 29 weeks are spontaneous, in contrast with early preterm singleton births.


A large fraction of preterm births in Seine-Saint-Denis are deliveries resulting from a medical decision to induce labour or carry out a caesarean section in the absence of labour. These induction practices are similar to those observed in France overall, despite differences in the population characteristics of this district. Thirty-seven percent of all singleton preterm births resulted from a medical decision to induce labour or to carry out a caesarean section before onset of labour in a national study in 19956; this proportion was 42% for births under 33 weeks of gestation in a 1997 study of nine French regions7. In Seine-Saint-Denis, about half of all births occur to women who were born outside France, mainly from northern and sub-Saharan Africa. Rates of unemployment and poverty are higher and educational levels are lower than in the rest of France. The rate of singleton preterm birth is significantly higher than the national average: 5.4% versus 4.7%8.

These results underline the importance of distinguishing medically indicated births associated with premature rupture of membranes from others. Medical decisions were associated with preterm premature rupture of membranes in about one-quarter of indicated births among singletons. Differences in practices for managing preterm premature rupture of membranes, either expectantly or by inducing delivery, could affect both the preterm birth rate overall as well as the rate of preterm birth by medical decision. Excluding induced preterm birth associated with premature rupture of membranes also reduces differences in indicated preterm birth rates between twin and singleton births: only 12% of medical decisions for twins were associated with premature rupture of membranes.

Comparative analyses must take into consideration the changing composition of preterm births. Preterm birth rates for singleton live births have declined in France over the past decades: from 7.9% in 1972 to 5.8% in 1981, to 4% in 1988/19891. Two national studies in the 1990s documented rates of 4.5% (1995) and 4.7% (1998)8. In consequence, preterm birth rates are low in France compared with many other developed countries, where singleton preterm birth rates range from 5.5% to 11%. Nonetheless, trends in indicated preterm births and twin births, which have contributed to rising preterm birth rates in other countries, are also present in France. Induced singleton births (for all reasons including premature rupture of membranes) increased from 7.3% in 1972 to 16.3% in 1981 to 37.2% in 19951,6. Twin pregnancies increased from 0.89% to 1.4% of all pregnancies from 1972 to 19989. These births thus constitute a growing fraction of all preterm births. In the population of Seine-Saint-Denis, indicated preterm births constitute over one-quarter of all preterm births and non-indicated twin births account for a further 16%. In other words, only 3.8% of the 6.7% preterm birth rate in Seine-Saint-Denis is attributable to spontaneous singleton preterm birth. For France, overall, the rate of spontaneous singleton preterm birth is probably lower as the singleton preterm birth rate in Seine-Saint-Denis is higher than the national average.

The distinction by mode of onset is also important when considering prevention strategies. For indicated deliveries of compromised fetuses, prolonging gestation may not be the best way to obtain optimal health outcomes for singletons or twins. Increased preterm birth rates in twins tend to be associated with better, not worse, prenatal care and with greater rates of indicated births10. Increases in preterm birth rates for multiples appear to be associated with declines in growth restriction and perinatal mortality4. Indicated preterm births have distinct medical risk factors from spontaneous births—in particular hypertension and fetal growth restriction—which shape the focus of prevention programmes.