A Model for Predicting Severe Intra-Abdominal Adhesions following Prior Cesarean Sections

Postoperative adhesions are common complications of abdominal surgeries [1]. Cesarean sections (CSs) are considered a leading cause of intra-abdominal adhesions during childbearing years [2]. With CS rates steadily rising and currently standing at over 30% of deliveries in the USA [3, 4], intra-abdominal adhesions following CS have become a major concern. The incidence of adhesions is known to correlate with the number of prior CS, which is considered a major risk factor for adhesion formation. Adhesion formation occurs after the first CS in about 25–46% of cases, and rates increase to 43–75% after the second CS, and up to 83% after the third CS [2, 5, 6]. There are also several additional known risk factors that influence the formation of adhesions, such as smoking, history of chronic illness (e.g., diabetes), previous postoperative infection, extensive peritoneal trauma, and the presence of blood and clots following surgery [7‒10]. In contrast, peritoneal and rectus muscle closure was also suggested as a preventive measure against the formation of post-CS adhesions [11, 12].

The rate of adhesion formation varies according to different types of surgeries [13] and individual risk factors in women, such as prior medical history and surgical approach [14], with distinct differences between open and laparoscopic surgeries. The literature describes numerous complications associated with intra-abdominal and pelvic adhesions [15, 16], underscoring the critical importance of early detection and efforts to prevent adhesions to reduce the risk of severe postoperative complications.

The importance of identifying and understanding post-CS adhesions lies in their deleterious maternal and obstetrical implications. Intra-abdominal adhesions can lead to bowel obstruction [2, 13] as well as to bladder and urinary tract injury [2, 17, 18]. Furthermore, they comprise a known risk factor for gynecological complications, such as infertility, increased risk of ectopic pregnancy [2, 19, 20], chronic pain [19], and significantly more re-hospitalizations [21]. Obstetrical complications include delay in subsequent CS delivery time in correlation with adhesion severity [6, 22, 23]. Therefore, prediction of the likelihood of complications by the assessment of adhesion risk factors is essential for surgical decision-making. The aim of this study was to examine whether it is possible to predict severe intra-abdominal adhesions following CS based upon the presence of risk factors.

We designed a prospective cohort study to identify risk factors of severe post-CS adhesions and to predict their presence.

Women who had previously at least one CS and underwent CS between January and July 2021 were eligible for study. Women for whom there was no information regarding their previous surgery were excluded from the study.

This study protocol was reviewed and approved by the Tel Aviv Sourasky Medical Center Committee, approval number 0579-21 TLV. The need for informed consent was waived by the committee. Written informed consent to participate statement was not required.

Adhesions were assessed by questionnaires that were filled out by the surgeons immediately after each CS. This questionnaire was described and validated in our previous study [24] and based on previous classifications [25], with the aim of ensuring that there would be no differences between raters, regardless of their level of seniority and experience. The questionnaire addressed 4 main anatomical sites: 1. skin-to-fascia, 2. uterus-to-bladder, 3. abdomen-to-uterus, and 4. other (Fig. 1). The term “other” site refers to adhesions in the ovarian area, omentum, intestine, or any other intra-abdominal site. The surgeons were asked to categorize the adhesions into none (score = 0), filmy (score = 1), or dense (score = 2). Filmy adhesions are those that are weak and therefore easily removed without using a device, whereas the removal of dense adhesions requires a device (scissors, scalpel, diathermy). The sum of the 4 sites ranges between 0–8, and a score of ≥5 represents severe adhesions since it requires that there be at least one site with dense adhesion (rated as 2) and additional adhesions at other sites (to complete a minimum score of 5). Severe adhesions (score ≥5) were defined as the primary outcome measure.

Details of any prior CS were obtained from the electronic medical records, including urgency (elective or non-elective), surgery duration (in minutes), complications (including post-CS fever >38.0°C or need for blood transfusion) and newborn birth weight. A history of maternal morbidity including gestational hypertension, pre-eclampsia, and gestational diabetes was also obtained. Other study variables were extracted from the electronic medical records, including current maternal age, pre-gestational body mass index (BMI), smoking status, number of previous CS, time between CS (in months), chronic diseases (pre-gestational hypertension and pre-gestational diabetes), newborn birth weight, and the use of assisted reproductive technologies (ART). Skin tone (light, medium, or dark) and scar hypertrophy (hypertrophied or keloid) were also noted.

Descriptive statistics were used to assess the distribution of variables. Continuous variables were summarized as mean values with standard deviations, and categorical variables were summarized as counts and percentages. First, a simple logistic regression was applied to each of the variables. Variables with p values of < 0.1 (maternal age, BMI, duration of prior CS, time between CS, birth weights of newborns of each CS, smoking status, number of CS (1, 2, 3+), urgency to perform a CS, current chronic illnesses, and the use of ART) were chosen for multivariable logistic regression to calculate the odds ratios (OR). Finally, a prediction model was built using a backward variable selection based upon the significant risk factors that were found in the multivariable logistic regression (p < 0.05). Additionally, the contribution of non-significant variables (p > 0.05) to model significance was tested by log likelihood ratio (LRT), beginning with the most non-significant variable. Each variable with a non-significant contribution was removed until all variables were significant in the multivariable logistic regression or in the LRT. This analysis was conducted for the prediction of non-severe adhesions (score ≤4) versus severe (≥5) by a logit model. The analyses were carried out using Python version 3.73. Significance was defined as a p value of > 0.05.

Overall, during the study period, 1,562 CS were conducted at our center, and questionnaires assessing adhesions were filled in on 1,071 of them. A final total of 341 CS for which questionnaires had been filled in and essential information was available on prior CS were included in this study, yielding 92 cases (26.9%) of severe adhesions (score 5–8) and 249 cases (73.1%) without severe adhesions (score 0–4) (Fig. 2).

The women with severe post-CS adhesions had significantly higher BMI (26.1 ± 5.1 vs. 24.1 ± 4.8, respectively, p < 0.001), and more prior CS (≥3 in 13% vs. <3 in 6.4%, p < 0.001 for both). The development of severe post-CS adhesions correlated with longer CS time (38.8 ± 15.3 vs. 32.3 ± 12.8 min, p < 0.001) and higher rate of maternal morbidity (diabetes, hypertensive disorders) at the time of previous CS (30.4% vs. 17.3%, p = 0.012). Out of the study population, 156 women underwent non-elective surgery, with 115 (46.2%) vs. 41 (44.6%) women with non-severe and severe adhesions, respectively (Table 1).

Figure 3 presents the trend of increasing adhesion prevalence with a higher number of CS. A significant difference was observed between women with no prior cesarean sections and those with one, two, or three or more cesareans. Although there is a visible trend toward higher adhesion scores from the first to the second and then to three or more surgeries, this difference is not statistically significant.

A logistic regression analysis for severe post-CS adhesions as the dependent variable revealed several parameters as having significant positive ORs (Table 2). All those significant parameters then underwent multivariable logistic regression analysis (Table 3). Significant positive ORs for an event of severe post-CS adhesions were found for the number of previous CS (2 and ≥3: OR 3.22, confidence interval [CI] 1.84–5.63 and OR 2.85, CI 1.18–6.91, respectively, p < 0.001 and p < 0.02, respectively), maternal BMI (OR 1.06, CI 1.00–1.11, p = 0.02), maternal morbidity at the time of previous CS (OR 2.05, CI 1.07–3.89, p = 0.03), and previous CS surgery times (OR 1.032, CI 1.01–1.05, p < 0.001). None of the non-significant variables was found to have a positive significant contribution toward the prediction model by the LRT calculation. The final model (Table 4) had a 79.1% accuracy predicting post-CS adhesions and whether or not the adhesions were likely to be severe, with a positive predictive value (PPV) of 68.4% and a negative predictive value (NPV) of 79.5% for severe adhesions.

This study demonstrated that the significant risk factors for the presence of severe adhesions in CS were repeated CS, longer time to perform previous CS, the pregestational BMI, and existence of maternal morbidity at the time of previous CS. It was possible to predict the presence or absence of severe post-CS adhesions based upon these risk factors with 79.1% accuracy, with a PPV of 68.4% and an NPV of 79.5%.

Several studies have shown that repeated CS led to abdominal adhesions in correlation with the number of CS [2, 6, 23]. This study demonstrates a clear trend of increasing adhesion scores with a higher number of prior cesarean sections. A significant difference was found between women with no prior surgeries compared to those with one, two, and three or more cesareans. However, differences in adhesion scores were not statistically significant among women with one, two, or three or more prior surgeries, despite the observed trend, likely due to the wide range of adhesion scores among women with previous cesareans. This trend might have reached statistical significance with a larger sample size. Furthermore, in this study, we addressed severe adhesions by aggregating different sites into an overall score. Upon examining the contribution of various adhesion sites, it was found that there was no significant difference between skin-to-fascia, uterus-to-bladder, and abdomen-to-uterus, while the site labeled as “Other” had a lesser contribution to adhesions.

The literature describes a correlation of any post-CS adhesions with several risk factors, some of which are addressed in this study: repeat CS [2, 6, 23], previous CS duration [1], and BMI [12, 26]. We also examined potential risk factors, such as the use of ART, smoking status, prior blood transfusion, post-CS infection, skin tone, scar hypertrophy, and others. Most of those potential risk factors emerged as being non-significant, possibly because we examined not only post-CS adhesion rates but also their severity by a scoring system as described above. That finding may also be due to the relatively small sample sizes.

Post-CS adhesions are a major concern due to potential maternal, surgical, and obstetrical complications [17‒21, 27]. Due to the difficulties surgeons encounter in women with severe post-CS adhesions, it is important to predict the severity of adhesions during the pre-CS evaluation, and a method of doing so would also aid in pre-CS planning and counseling. We describe such a method that uses risk factors to correctly predict most cases of severe or non-severe adhesions (79.1%). Interestingly, it emerged that our predictive model even more accurately predicted scenarios without severe adhesions than those with severe adhesions (PPV of 68.4%, NPV of 79.5%).

There are several notable advantages of this study. To the best of our knowledge, this is the first study that attempted to predict the severity of post-CS adhesions. Second, we used a validated scoring system for post-CS adhesions [24]. Furthermore, the assessment included uni- and multivariable analyses based upon the use of several potential risk factors and confounders, including data about current and previous CS.

This study also has a few limitations. First, some known risk factors for adhesions were not available to us, such as surgical history other than CS [28]. Other risk factors that were not available to us were endometriosis, pelvic inflammatory disease, and genetic factors, as well as aspects of prior CS, such as peritoneal and rectus muscle closure [11, 12, 29]. Another limitation is the study’s sample size. While it did comprise 341 women, a larger sample might have provided evidence of additional risk factors and allowed for the development of a more precise predictive model.

In conclusion, his study demonstrates that post-CS adhesions are mostly predictable. Our study may improve pre-CS planning, as well as enhance counseling of women with regard to the operative risks of adhesions. Additional studies are warranted to further characterize risk factors and to construct advanced and more precise predictive models.

The study was approved by the Tel Aviv Sourasky Medical Center Committee, Approval No. 0579-21 TLV. The need for informed consent was waived by the committee. A consent to participate statement was not required.

The authors report no conflict of interest.

This study was not supported by any funding sources.

Shai Ram – design, planning, conduct, data analysis, and manuscript writing. Hila Shalev – Ram, Yariv Yogev, and Roza Berkovitz-Shperling – design, planning, and manuscript writing. Shira Alon – planning, conduct, and data analysis. Ziv Shapira – design, planning, and conduct. Margaret Johansson-Lipinski – design, conduct, and manuscript writing. Ariel Many – design, conduct, planning, and manuscript writing.

The data that support the findings of this study are not publicly available due to confidentiality of information that could compromise the privacy of research participants but are available from the corresponding author, Shai Ram, M.D., upon reasonable request.