Repeat low order caesarean delivery, risk factors for complications: A retrospective, longitudinal study.

In the past few decades there has been a sharp increase in the rate of caesarean deliveries (CDs) worldwide, with a global average of 21% [ 1 ]. This trend is most noticeable in middle to high income countries, reaching up to 50% of all deliveries [ 1 – 3 ]. Approximately one-third of CDs are repeat surgeries [ 4 ]. Large observational studies have repetitively shown that high order repeat CDs (≥4 CD), as well as placenta accreta spectrum, are associated with maternal and fetal complications [ 5 – 8 ]. These include severe maternal hemorrhage, need for blood transfusions, damage to adjacent organs, thromboembolic events, uterine rupture, peripartum hysterectomy, and neonatal morbidity and mortality [ 5 – 8 ]. However, only 8% of repeat CDs are high order CDs and even less are complicated by the placenta accreta spectrum (0.5%) [ 8 ]. Hence, low order, second (CD2) and third (CD3) caesarean deliveries, comprise most of the repeat CDs (92%) [ 8 ]. The main challenge of lower order CD, assuming normal placental adherence, is abdominal-pelvic adhesions. Damage to the peritoneum, the protective surface layer of various abdominal-pelvic organs, during surgery initiates a physiological cascade that could lead to adhesion formation. This cascade is based on cytokines and other inflammatory mediators, released by the damaged mesothelial cells and macrophages, that stimulate mesothelial migration to the damaged area, thus, initiating the process of re-epithelialisation. As part of the inflammatory cascade, fibrin is deposited and provides a matrix that facilitates repair. During this process, two damaged adjacent peritoneal surfaces may join to form an adhesion. Normally during the healing process, tissue plasminogen activator (tPA) and urokinase plasminogen activator (uPA), found in mesothelial cells, enable the conversion of plasminogen to plasmin, which breaks the fibrin to fibrin-degradation products. Any reaction that interferes with this process, such as the excess of plasminogen activating inhibitor (PAI)-1 and PAI-2, can lead to adhesion formation [ 9 ]. Infection, hypoxia, tissue manipulation and chemical irritation are all potential factors for the activation of the abdominal-pelvic adhesions cascade leading to fibrous band-like structures that could potentially cause chronic pelvic pain, bowel obstruction, infertility, and complications in future surgeries [ 9 , 10 ]. During the process of adhesiolysis, in a subsequent surgery, damage to the adjacent muscles and structures such as, the ureter, bladder, and colon may occur, with a potential for massive bleeding/blood transfusion, peritonitis, infection, and septic shock. These complications can result in acute and chronic renal failure, need for a relaparotomy, and admission to the intensive care unit, resulting in prolonged maternal hospitalization [ 5 – 10 ]. Adhesiolysis increases the length of the operation and was hypothesized to contribute to the high rate of perioperative maternal complications [ 11 – 13 ]. Prolonged operative time was shown to be a risk marker for perioperative maternal complications and is associated with the presence of abdominal-pelvic adhesions [ 11 , 12 ]. Previous studies defined a complicated maternal caesarean delivery by a composite outcome including, excessive bleeding requiring blood transfusion, relaparotomy, uterine rupture/dehiscence, hysterectomy, injury to bladder/ureter or bowel, wound infection, endometritis/sepsis or prolonged operative time [ 10 – 13 ]. The purpose of this study was to examine the risk factors for a complicated maternal caesarean delivery (CD) among women undergoing a repeat low-order CD, and to develop a predictive model for women at risk. Specifically, we aimed to study if women who experienced a composite complicated maternal CD, unrelated to abnormal placentation, during the second caesarean delivery (CD2), were at increased risk for perioperative maternal complications in the subsequent CD (CD3), using a matched group of women. As data was collected from the same woman at two time points, each served as her own control.

During the study period, there were 159011 deliveries, 18695 (12%) underwent CD, 3974 (21%) were a recurrent CD of which 3305 (83%) were low order CD2 and CD3. We identified 1331 women eligible for inclusion in the study ( Fig 1 ). Women presenting for CD3 compared to their CD2 were an average of three years older, had a longer interpregnancy interval (7 months) and were less likely to attempt TOLAC (8% vs 35%, p<0.001). The operative time for CD3 was longer than CD2 by a mean of 3 minutes, with 142 (11%) of operations exceeding the 90 th percentile (62 minutes), compared to 99 (7%) in CD2, p = 0.002. As shown in Table 1 , of the 1331 women in the cohort, 159 women (11.9%) had a composite complicated maternal CD2 compared to 226 (16.9%) of women during CD3, an increase of 67 cases (42%) p<0.001. Rate of maternal complications including uterine scar dehiscence/rupture, need for relaparotomy and blood transfusion, were similar for both groups. None of the women were admitted to the intensive care unit for postpartum observation or treatment at either time. * Composite complicated maternal CD either one of the following: Blood transfusion > 1 packed cells, uterine rupture, dehiscence, relaparotomy or duration of operation > 90 percentile. As shown in Table 2 , the univariate analysis identified six factors significantly associated with an increased risk for a composite complicated maternal CD3. Factors related to CD2 included prolonged operative time (≥90 th percentile, OR = 2.4 p<0.001), dehiscence or rupture or relaparotomy (OR = 2.1 p = 0.013), while Sub-Saharan African origin (OR = 3.2 p<0.001), gestational age (OR = 0.8 p<0.001), birth weight (OR = 0.9 P<0.001) and TOLAC (OR = 2.1 p<0.001) were related to CD3. A composite maternal complication at CD2 was consistent with an increased risk for a complicated CD3 (OR = 2.3 P<0.001). Composite complicated maternal CD either prolonged operative time, blood transfusion, dehiscence/uterine rupture or relaparotomy. *1.006(0.979;1.034). **0.978(0.935;1.023). #1.007(0.946;1.072). ##1.023 (1.012–1.033). ###0.999 (0.999;1.000). ^1.007(0.997;1.017). A binomial logistic regression was performed to determine the effect of a maternal complication in previous CD2 together with potential risk factors at CD3 on the likelihood of composite maternal complication in CD3. The model included; maternal age, parity, preterm delivery, TOLAC and birthweight of newborn, a Sub-Saharan African origin and a composite maternal complication in previous CD2. The model included 1328 cases, was statistically significant, χ2(7) = 50.760, p <0.001, explained 6.3% of the variance of composite complicated maternal CD3 and correctly classified 82.9% of cases ( Table 3 ). Included in the model 1328 women, χ2(7) = 50.760, p < .001. Nagelkerke R Square = 0.063, Hosmer and Lemeshow = 0.427. Sensitivity of the model was 1.8%, specificity was 99.5%, PPV and NPV were 44.4% and 83.1% respectively. Of the seven predictor variables, only three were statistically significant; composite complicated maternal CD2 aOR = 2.2 p<0.001, preterm delivery aOR = 1.7 p = 0.010 and Sub-Saharan African origin aOR = 3.7 p<0.001. The area under the ROC curve was 0.639, 95% CI [0.597 0.681]. A composite complicated maternal CD2 was found to be associated with maternal complications in CD3, p < 0.001. The potential use of a "composite complicated maternal CD" as a "diagnostic tool" for maternal complications in CD3 was assessed. As noted in Table 4 , the Sn, Sp, PPV, and NPV were 21%, 90%, 30% and 85% respectively. Person Chi-Square 1 packed cell, uterine rupture, dehiscence, relaparotomy or duration of operation > 90 percentile.

Based on the results of this study, we suggest that a composite complicated maternal CD2 together with a Sub-Saharan African origin and a preterm delivery in the subsequent CD3 are all risk factors for maternal complications in CD3. Although it is hard to predict which women will experience complications, it is easier to predict which woman will undergo an uneventful CD3 given the SP of 99.5% and the NPV of 83% of the model. Identifying women who are not at risk for complication may enable us to provide better counselling for women undergoing a repeated low order CD.

A retrospective longitudinal study of all women who underwent two low-order repeat CDs (CD2 and CD3) at a single large university-affiliated, tertiary medical centre, Shaare Zedek Medical Center (SZMC), Jerusalem, between July 2005 and December 2016. Obstetric surgical reports were extracted from the electronic medical record containing both mandatory fields and free text summary notes, continuously updated during labor and delivery. Maternal variables extracted from the mandatory fields included age, BMI, previous medical history, parity, number of previous vaginal and CDs, noting if performed in another facility, gestational age, indication for CD, duration of CD, length of interpregnancy interval, trial of labor, uterine rupture, uterine dehiscence, blood transfusion, gestational diabetes or hypertension. Fetal variables extracted included number of fetuses, gender, birth weight, Apgar score, and admission to the neonatal intensive care unit. Surgical technique for all CDs performed at our facility was comprised of skin opening by Pfannenstiel’s incision and blunt extension of uterine cut. However, among senior surgeons there existed slight variations with respect to fascia opening (blunt/sharp/electrosurgery), uterine closure (one versus two layers), closure versus non-closure of peritoneum and muscles and skin closure (staples/monocryl/nylon). The vast majority of repeat CDs (greater than 95%) were performed under regional anesthesia. Unique to our medical center, women with two documented prior low segment transverse CDs (LSTCD) were eligible for a trial of labor (TOLAC) under a strict protocol that includes the absence of previous uterine dehiscence or rupture, estimated fetal weight < 4000, vertex presentation, previous vaginal delivery and spontaneous progress of labor [ 14 ]. As such, in both CD2 and CD3 women may have had a trial of labor. The operative time of the CD was measured from the of the initial incision to skin closure and was a mandatory field in the electronic medical record, documented in all CD [ 10 , 11 ]. Prolonged operative time was defined as a CD that exceeded the 90 th percentile of the operative time of the study group [ 10 ]. The outcome, composite maternal complicated CD, was comprised of at least one of the following: (1) a clinically significant hemorrhage necessitating blood transfusion, (2) a uterine rupture or dehiscence, (3) a relaparotomy for bleeding control, (4) an admission to the intensive care unit or (5) a prolonged operative time ≥ 90th percentile. Relaparotomy and uterine rupture/dehiscence were identified using International Classification of Diseases, 9th Revision, Clinical Modification Diagnosis Codes. Women with placenta accreta spectrum and those who underwent a hysterectomy were excluded from the study, since the intent was to focus on complications not related to abnormal placentation. The primary outcome of the study was to evaluate if a complicated CD2 is associated with complications in the subsequent CD (CD3) and if such, to calculate the sensitivity (Sn), specificity (Sp), positive and negative predictive value (PPV, NPV) of a complicated CD2 on maternal complications in CD3. Secondary outcomes were to (1) identify risk factors at CD3 associated with a composite complicated maternal CD3, and to (2) build a statistical model including factors from CD2 and CD3 to predict women at risk for a composite maternal complication during CD3. Data was validated by defining distributions and quantifying missing values. Obstetric characteristics comparing CD2 and CD3 were presented as proportion, median or mean with interquartile range respectively or standard deviation, depending on the variable characteristics; categorical, ordinal or continuous, respectively. Statistical significance was defined by a two-sided p value ≤ 0.05 using the Chi-square test or Fisher Exact test for categorical variables. Wilcoxon Signed Rank Test was employed for ordinal or continuous variables with non-Gaussian distribution and the Student Paired t-test was utilized for continuous variables with a normal distribution. A univariate analysis was performed followed by the development of a multivariate logistic regression model to evaluate if a composite complicated maternal CD2 is an independent risk factor for complications in the subsequent CD (CD3). The association between a composite complicated maternal CD2 and CD3 was evaluated by the Chi-square test. The Sn, Sp, PPV and NPV of a complicated CD2, for detecting complications in the subsequent CD3 was calculated. Based on previous studies [ 9 ] the difference between CD2 and CD3 in respect to the prevalence of severe adhesions was 28% and 41% respectively. The sample size needed to find such a difference with a power of 80% was 418 (209 in each group). In an effort to decrease selection bias, the entire cohort that met inclusion criteria was analyzed. As the annual delivery volume is approximately 14,500 with a CD rate of 12%, we assumed an 12 year time frame would contain enough women needed for the study. The study was performed in accordance with the ethical standards of the Declaration of Helsinki and its later amendments and was approved by the Institutional Review Board of the Shaare Zedek Medical Center (SZMC IRB 0260-16-SZMC). The manuscript is presented according to the STROBE guidelines [ 15 ].

(SAV) Click here for additional data file.