The study protocol was approved by the Ethics Committee of the Medical Faculty of Lund University, Sweden. Informed consent was obtained from all participants, after the nature of the procedures had been fully explained. We aimed at recruiting 100 primipara who had undergone a non-instrumental vaginal delivery, 100 women who had undergone one Cesarean section, and 50 women who had undergone more than one Cesarean section. The names and addresses of women who had delivered at our institution within the preceding 5–8 months and fulfilled our eligibility criteria (at least 18 years old, primipara with uncomplicated vaginal delivery or woman delivered by Cesarean section at least once) were obtained from the labor ward database of our institution. The women so identified were sent a letter of invitation to participate in the study, and those who accepted the invitation were booked for an ultrasound examination. Our exclusion criteria were: previous surgery on the uterus (other than cone biopsy, loop electrosurgical excision procedure, dilatation and curettage, or dilatation and evacuation), pregnancy at the ultrasound examination or no clear information about earlier uterine surgery.
The ultrasound examinations were carried out 6–9 months after the latest delivery. Immediately before the examination, a secretary took a history following a standardized research protocol (parity, medication, contraceptives, breast feeding, day of menstrual cycle, earlier deliveries and gynecological operations) and noted the information on a paper form. The ultrasound examiner was blinded to this information when performing the ultrasound examination as well as when evaluating the images afterwards.
All ultrasound examinations were performed by the first author. The abdomen of the woman to be examined was covered with a towel to hide any abdominal scar, and the women had been instructed to reveal nothing about their obstetric history to the ultrasound examiner. In this way, the ultrasound examiner was blinded to patient history. The examination was carried out transvaginally with the woman in the lithotomy position and with an empty bladder. The uterus was scrutinized for the presence of Cesarean section scars and scar defects. The ultrasound images were evaluated during the ultrasound examination but, in addition, representative images of both longitudinal and transverse sections through the uterus were stored on our digital image storing system, Siemens Syngo® Dynamics, version 5.0 (Siemens Medical Solutions Health Services, Malvern, PA, USA). Any visible defect or indentation in the scar, however small, was classified as a defect. On the basis of subjective evaluation, scar defects were classified as large or not large by the ultrasound examiner. If the ultrasound examiner perceived a scar defect to be large, the woman was informed that the clinical relevance of large defects was unknown, but that we offered a transvaginal ultrasound examination early in a subsequent pregnancy to exclude Cesarean scar pregnancy. Ultrasound images of a scar classified subjectively by the ultrasound examiner as an intact scar, a scar with a small defect and a scar with a large defect are shown in Figure 1.
Figure 1. Ultrasound images of an intact Cesarean section scar (a), a scar with a small defect (b) and a scar with a large defect (c).
Download figure to PowerPoint
Immediately after the ultrasound examination, with the obstetric history of the woman still unknown to the examiner, the stored ultrasound images were scrutinized offline and measurements were taken. The following ultrasound features were noted: anteflexion or retroflexion of the uterus, number of visible Cesarean section scars in the anterior wall of the uterus (categorized as clearly visible or difficult to detect), presence of a scar defect (yes or no), shape of a scar defect (triangular, round, oval, or total defect with no remaining myometrium over the defect), and location of the scar defect (right, left, central or other). If more than one scar was seen, the scar lowest in the uterus was called Scar 1, and scars located closer to the fundus uteri than this were numbered 2, 3, etc., with the scar located closest to the fundus uteri being assigned the highest number. The following measurements were taken: the myometrial thickness of the isthmus uteri at the level of the internal cervical os, the distance between an intact scar and the internal cervical os, and the distance between a scar with a defect and the inner cervical os. We defined the level of the internal cervical os as the level at which there is a slight narrowing of the uterus between the corpus and the cervix at the lower boundary of the urinary bladder.
The size of a scar defect was measured both on a longitudinal section through the uterus (length and height of the defect, thickness of the remaining myometrium over the defect, and thickness of the myometrium adjacent to and fundal to the defect; the latter measurement corresponds to measurement ‘m’ in Figure 2c), and on a transverse section through the uterus (width of the defect). If there was a scar defect, the ratio (expressed as a percentage) between the thickness of the remaining myometrium over the scar defect and the thickness of the myometrium adjacent to and fundal to the defect was calculated. The measurement technique is described in Figure 2.
Figure 2. The distance (d1) between the inner cervical os and an intact scar was measured as shown in (a), i.e. an imaginary (dotted) line was drawn from the internal cervical os (io) to the surface of the anterior cervical lip perpendicular to the cervical canal (this imaginary dotted line represents the thickness of the myometrium in the isthmus), and the distance was measured from the top of this imaginary line to the top of the scar. The distance (d2) between the inner cervical os and a defect scar was measured in the same manner (b). The length (L) and height (h) of the defect, the thickness of the remaining myometrium over the defect (r) and the thickness of the myometrium close to and fundal to the defect (m) were measured as shown in (c). The ratio between r and m (r/m) was calculated. The gray shaded triangular areas in (b) and (c) represent a scar defect.
Download figure to PowerPoint
All ultrasound examinations were performed using a GE Voluson 730 Expert ultrasound system (General Electric, Zipf, Austria) equipped with a 2.8–10-MHz transvaginal transducer. Not until all ultrasound examinations had been performed and all images evaluated and measurements taken, was information from the obstetric records of the women obtained.
Statistical calculations were performed using SPSS version 12.02 (SPSS Inc., Chicago, IL, USA). The statistical significance of differences in categorical data was determined using the chi-square test or Fisher's exact test, as appropriate, and the statistical significance of differences in continuous data using the Mann–Whitney U-test or Kruskal–Wallis test. P < 0.05 was considered statistically significant.
To determine which measurements best predicted whether a scar defect in the lowest scar was subjectively perceived to be large by the ultrasound examiner, receiver–operating characteristics (ROC) curves were drawn separately for women who had undergone one and those who had undergone two Cesarean sections. The area under the ROC curve was calculated with its 95% CI. The measurement was considered to have discriminatory potential if the lower limit of the CI for the area under the ROC curve exceeded 0.5. The measurement with the largest area under the ROC curve was considered to be the best predictor of a defect being perceived to be large by the ultrasound examiner. The ROC curves were also used to determine the best cut-off value mathematically for predicting whether a defect would be perceived to be large by the ultrasound examiner. The best cut-off value mathematically was defined as that corresponding to the point situated furthest from the reference line.