Overview

This population-based registry study was complemented with information from corresponding medical records. The Regional Ethics Committee (2010/1609-4) and the Data Inspectorate of Norway approved the study.

We used 2 independent data sources to identify possible cases of uterine rupture. First, we searched the Medical Birth Registry of Norway (MBRN), established in 1967. This national registry contains information on all births in Norway after 16 weeks of gestation regarding maternal health, information about delivery and complications, and information about the newborn. The midwives attending a birth complete and send a standardized MBRN form within 7 days after delivery.

Second, we searched the Patient Administration System (PAS), which is a local registry at each maternity unit. These registries maintain records of all diagnoses for in-patients since 1970. We requested permission to perform a PAS registry search from all maternity units (n = 48) in Norway, and 21 units agreed to participate. These 21 units were distributed throughout Norway, and they exhibited a wide range of delivery rates, from <500/year to ≥3000/year. The target population included maternities recorded in these 21 units during 1967–2008 (n = 1,443,271), which represented 59.81% of the pregnant population in Norway. We excluded individuals with missing registration numbers in the MBRN (missing = 2,716) or missing gestational age (missing = 8303). Hence, the sample included 1,432,252 maternities.

In this study, we examined risk factors for complete uterine rupture after starting labor. Thus, we excluded all individuals who gave birth through a CD before starting labor. The final sample included 1,375,826 maternities. We studied risk factors for complete uterine rupture separately for women without a previous CD (all parities) (n = 1,317,967) and those with a previous CD (parities ≥1) (n = 57,859). In general, in Norway, pregnant mothers with 1 previous lower-segment CD are offered the opportunity to attempt a labor. The majority in this study had 1 low transverse segment incision. Few had low vertical incision or classical vertical incision. Single- or double-layer suturing of the uterus at a previous CD is not registered in the MBRN.

For mothers undergoing a second delivery with a history of CD in a first delivery (parity = 1; n = 34,550 mothers), we constructed a data set, in which the first 2 deliveries for these mothers were linked. This data set was used to calculate the risk of complete uterine rupture in the second delivery, based on the previous obstetric history.

After identifying potential cases of uterine rupture in the MBRN and in the PAS, we validated the diagnosis of complete uterine rupture with corresponding medical records, which were considered the gold standard. The first author of the present study (I.A.-Z.) identified these cases and studied the medical records of mothers diagnosed with uterine rupture by visiting 16 maternity units and reading posted copies of records from 5 additional units. The medical records included detailed information about the mothers and their newborns.

Measures

The outcome measure was complete uterine rupture, defined as a tear through all layers of the uterine wall, including the serosa and amniotic membranes. A partial uterine rupture was defined as a tear in the muscular layers, with intact serosa or amniotic membranes. The diagnosis of uterine rupture was reported in plain text on the MBRN registration forms by the birth attendant, and the appropriate code was recorded in the electronic file by MBRN personnel.

Before 1999, the code used was 71; starting in 1999, the International Classification of Diseases (ICD), 10th revision, diagnostic codes were used (O710, O711). Uterine rupture was identified in the PAS by the ICD, eighth revision, code: 956 (1967–1978); ICD 9 codes: 6650 and 6651 (1979–1998); and ICD, 10th revision, codes: O710 and O711 (1999–2008). These codes did not specify rupture type. The type of rupture, whether complete or partial, was identified in the medical records. In this study we included data only from mothers with complete ruptures.

The risk factors were identified, after being recorded in registration forms in plain text or by marking prespecified tick boxes. They included demographic factors, previous miscarriage, pregnancy factors, and obstetric factors. Demographic factors included maternal age, grouped as <35 years or ≥35 years; parity, grouped as <3 or ≥3; the mother’s country of birth, grouped as Western (Europe, North America, and Australia) or non-Western; and the decade of delivery, grouped by the 4 decades included in the study: first decade (1967–1977), second decade (1978–1988), third decade (1989-1999), or fourth and most recent decade (2000–2008); the fourth decade was taken as the reference.

Maternal age and parity were analyzed first as continuous variables, but there was no significant difference seen before the cutoff level mentioned in the previous text, and these were therefore grouped into categorical variables. A previous miscarriage was defined as 1 or more miscarriages that occurred in the first trimester.

Pregnancy factors included gestational age in weeks, grouped as 37–40 (reference), ≥41, or 24–36 weeks; fetal presentation, grouped as occipitoanterior vertex (reference), non–occipitoanterior, or a breech or transverse lie; antepartum fetal death, and birthweight, grouped as <4000 or ≥4000 g.

Obstetric factors included labor start, grouped as spontaneous or induced labor; augmentation of labor (defined as augmentation of contractions after an established spontaneous or induced labor), grouped as no oxytocin use or oxytocin use; and induction methods, grouped as spontaneous labor, induction with prostaglandins or oxytocin (each used with or without amniotomy or other methods), other methods (sweeping of membranes, transcervical balloon catheters, or unspecified methods), and sequential induction (prostaglandins and oxytocin combined with or without amniotomy or other methods).

The prostaglandins administered were mainly vaginal dinoprostone (prostaglandin E2); however, misoprostol (prostaglandin E1) was increasingly used in unscarred uteri, starting in 2004; also, gemeprost was used occasionally for the termination of pregnancy for those treated in the first 3 decades of the study. The maximum and total doses and the duration of prostaglandins or oxytocin administered are not registered in the MBRN form. Other labor factors included breech extraction and manual removal of the placenta after vaginal delivery.

Among mothers with a CD in a first delivery (Para1), the combined data set of the first and second deliveries was examined for risk factors that were present at the previous CD. These risk factors included maternal age, gestational age, an interdelivery interval <16 months, the type of CD (elective or emergency CD), prolonged labor, and severe postpartum hemorrhage (defined as blood loss ≥1500 mL or a need for blood transfusion, regardless of the amount of blood loss).

Data analysis

We computed the overall incidence and the incidences of complete and partial uterine rupture among groups without and with previous CD. Cross-tabulations and crude logistic regressions were used for each risk factor to calculate the risk and observed odds ratios (ORs) with the corresponding 95% confidence intervals (CIs) of complete uterine rupture. Only those that were statistically significant were included further in multivariable regression analysis. We used multivariable logistic regression in different models to calculate adjusted odds ratios (AORs). Demographic factors were adjusted first to each other. Other significant risk factors were adjusted for confounding relevant factors in separate models. Relevant cofactors preceded analyzed risk factors in time. The level of significance was set to P < .05. Data analyses were performed with SPSS, version 21 (SPSS Inc, Chicago, IL).