Early elective versus elective sigmoid resection in diverticular disease: not only timing matters—a single institutional retrospective review of 133 patients

All patients with an acute or chronic sigmoid diverticular disease undergoing laparoscopic sigmoidectomy at our department from January 2004 to July 2021 were retrospectively analyzed. Relevant data on demographics, comorbidities, recent attacks of diverticulitis, clinical and radiological features at the time of presentation, abscess-drainage placement, intraoperative, and postoperative course, including morbidity and mortality after sigmoidectomy, were extracted from our prospectively maintained medical database. These data included age, gender, body mass index (BMI), American Society of Anesthesiologists (ASA) score, relevant comorbidities (e.g., cardiovascular or metabolic diseases) and immunosuppression, preoperative laboratory parameters including inflammatory markers and hemoglobin, preoperative medical course, time since the last attack to surgery, type and duration of surgery, conversion rate, necessity of an ostomy creation, intraoperative transfusion, all documented postoperative minor and major complications until the date of hospital discharge, and the total hospitalization time. The postoperative morbidities were categorized according to the Clavien-Dindo classification [12]. A critical review of clinical, radiographic, and intraoperative data enabled an accurate classification of the disease stage in each patient. This was categorized using the classification of diverticular disease (CDD) [6]. In addition, we divided our collective into 2 separate groups depending on the time of surgery: group A represents the early elective cohort with patients who underwent a laparoscopic sigmoidectomy within 6 weeks of the last attack of acute or complicated diverticular disease. Group B comprises patients who received surgical therapy in the non-inflammatory interval at least 6 weeks after the last attack and documented relief of symptoms associated with diverticular disease (elective group). Patients representing with large intraabdominal abscess formation were initially considered for interventional drainage placement. Our standard operation procedure (S.O.P) for early elective and elective sigmoidectomy was as follows: Upon presentation in our department, every patient with the suspected diverticular disease received after clinical examination abdominal computed tomography (CT) imaging beside clinical and laboratory tests. Cases with free perforation (CDD 2c) were directly scheduled for emergency surgery. Patients with acute or chronic symptomatic diverticular disease were admitted. Based on the imaging results, larger mesocolic or pelvic abscess formations were drained interventionally if applicable. Our standard i.v. antibiotic regimen included ceftriaxon 2 g (once daily) and metronidazol 500 mg (thrice daily) or in severe cases piperacillin 4 g/tazobactam 0.5 g every 6 h for at least 5–7 days beside initial parenteral feeding and gradual oral nutrition intake in the latter course. Patients who responded to the conservative treatment by means of complete symptom relief and normalizing laboratory parameters were either discharged and referred to our outpatient clinic for a close follow-up or given the opportunity for an early elective resection based on their personal preference and quality of life burden in close interaction with the responsible surgeon. Thus, the decision towards early elective and elective resection was based on the surgeons’ personal experience and preference alongside the clinical presentation and success of the medical therapy taking into account the current German guidelines of diverticular disease [6] and each patient individual decision as well as quality of life expectations and restrictions. However, patients with persistent symptoms, apparent clinical deterioration despite medical therapy, or known risk factors (e.g., immunosuppression or relevant comorbidities) for a more complicated disease course underwent an early elective resection during index hospitalization although not initially intended. The surgical strategy did not differ between the early elective and the elective cases in our study. Note that we excluded patients with free perforation (CDD type 2c) or sepsis requiring emergency surgery, as well as patients with primarily open procedures from our analysis. Only specialized surgeons (board-certified visceral surgeons according to the requirements of the Federal German Medical Council) performed the procedures. If necessary, especially in the case of decision-making for conversion to laparotomy, a senior consultant or the head of the department was involved during the operation. All surgeons were trained and skilled in open and laparoscopic colorectal surgery. Each surgeons’ personal surgical statistics are recorded in our database. Briefly, after placement of 4 ports, lateral to medial mobilization, splenic flexure mobilization and low tie ligation were usually performed. The specimen was removed through a Pfannenstiel incision of approximately 5 cm in length. Restoration of the gastrointestinal passage was achieved through an end-to-end anastomosis using a 31-mm circular stapler. Intraoperative findings and the surgeon’s decision determined the operative course including conversion via median laparotomy and ostomy creation. Standard postoperative care with routine laboratory evaluation, mobilization, and gradual return to a normal diet after a documented bowel movement was administered. Patients with postoperative prolonged bowel paralysis were put on additional total parenteral nutrition (TPN). After discharge from the hospital, all patients were seen on regular basis for clinical follow-up at our department.

An ethical approval was obtained prior to the study commencement by the institutional ethical board of the Medical Faculty, Heinrich-Heine University Duesseldorf, Germany (study-no.: 2021–1346) and all reported procedures and steps were in accordance with the current ethical standards and requirements in the latest version of the Declaration of Helsinki.

Statistical analysis was performed using G*Power [13], SPSS 23.0 software (IBM Corp., Armonk, NY), and R version 4.1.1 with packages MICE [14], Hmisc, readxl, rms, ResourceSelection, and MASS. The normal distribution of continuous variables was assessed by the Kolmogorov–Smirnov and Anderson–Darling test. For descriptive data analysis, continuous variables were compared using the Mann–Whitney U or t-test. Categorical variables were compared using Fisher’s exact or chi-square test. Post hoc statistical power regarding the primary outcomes of the study, conversion rate depending on the timing of surgery or CDD, was found to be 80.9% and 98.5% respectively, assuming a significance level of 0.05. For missing data, we used the MICE package which imputes incomplete multivariate data by chained equations. The binary logistic regression model was used when the dependent variable had two categories. For this purpose, continuous variables remained unchanged when normally distributed (age) or were transformed by grouping either according to clinically relevant classifications (BMI, CDD), median (operation time), or by using the log₁₀ function (CRP, leucocytes). To establish a clinical selection model, we first screened for independent variables based on a stepwise backward selection in which the model with the lowest Akaike information criterion (AIC) is selected. Next, for cross-validation, we used the bootstrap method by resampling 100 times with replacement to assess the consistency of the predictors selected with the stepwise selection. Finally, we performed the logistic regression model that was suggested from the bootstrap method. The independent influencing factors were utilized to draw a nomogram. Model discrimination ability was assessed by C-statistics. A C-index of 1 indicates that the model perfectly predicts the outcome, over 0.8 and over 0.7 indicates a strong and good model, respectively, and a value of 0.5 means that the model is not predicting better than random chance. The goodness-of-fit (GOF) was evaluated by the Hosmer–Lemeshow test in which p-values < 0.05 indicate poor fit. Due to the small sample size, we validated the reproducibility of our model internally by the bootstrap method by resampling 100 times and assessing calibration curves. A linear regression model was performed using stepwise backward selection when there was a dependent continuous variable. In all analyses, a p-value of < 0.05 indicated statistical significance.

During the study period, a total of 133 patients underwent laparoscopic sigmoid colon resection due to diverticular disease at our department. Based on the cutoff value of 6 weeks since the onset of symptoms and the surgical procedure, 88 patients (66.2%) were enrolled in the early elective group A and 45 patients (33.8%) in the elective group B. Patient characteristics are summarized in Table 1. There were 69 men (51.9%) and 64 women (48.1%) with a median age of 55 years at the time of surgery. Baseline demographic data including sex, age, BMI, and ASA score did not differ between the two groups. Comorbidities were equally distributed apart from the incidence of diabetes which was significantly lower in the early elective group (p = 0.009). A review of the preoperative laboratory parameters revealed significantly higher inflammatory values (white blood cell, WBC count, and C-reactive protein, CRP) in group A vs. group B (p < 0.001) as well as decreased sodium levels (p = 0.003) prior to surgery. Patients in both groups suffered from a median of two diverticulitis attacks before surgery (p = 0.843). In the early elective group, 7 patients (8%) underwent abscess drainage in comparison to 2 patients (4.4%) in the elective group (p = 0.446). The median time interval between the last episode of diverticulitis and sigmoid resection was significantly shorter with 12 days (± 10.2 days) in group A vs. 62 days (± 35.5 days) in group B (p < 0.001). In both groups, an equal distribution of diverticulitis stages according to CDD was observed (p = 0.304).

The median operation time was comparable within both groups (group A: 295 ± 65.5 min vs. group B: 279 ± 84.3 min; p = 0.553). In group A sigmoid resection with a primary anastomosis was performed in 94.3% (n = 83) and in 4.5% (n = 4), a protective diversion ostomy was additionally constructed. A Hartmann procedure was necessary for one patient (1.1%). In contrast, all patients in group B underwent primary resection with anastomosis (100%) and without an ostomy. Regarding the need for intraoperative transfusion of blood units, crystalloid or colloidal infusions no significant differences between the groups became evident. A summary of available operative data is depicted in Table 2. However, conversion rates to laparotomy were significantly higher with 26.1% (23 of 88 patients) in group A as opposed to only 6.7% (3 of 44 patients) in group B (p = 0.007). The main reasons for conversion were severe inflammatory adhesions (n = 17), inflammatory conglomerate (n = 6), or uncontrollable bleeding (n = 3). To further elucidate a possible relationship between disease stage and conversion rate, we graphically displayed conversion rates depending on the CDD for the entire study group (Supplementary Fig. 1). Accordingly, in patients undergoing laparoscopy for diverticular disease, we observed higher conversion rates to laparotomy for type 2b (50%) and 3c (32%) disease. Next, we analyzed the disease stage-dependent conversion rates according to the time point of surgery. Interestingly, conversion rates were only significantly higher (p = 0.038) in group A patients suffering from CDD 3c when compared to patients having elective surgery for diverticular disease (Fig. 1). Of note, in the group of patients with CDD type 2b who were operated in the early elective period, the majority (14 out of 19 patients) demonstrated a complete resolution of the acute attack after medical treatment and preferred early elective surgery after information about the therapeutic strategies (early elective vs. elective surgery). Out of the 14 patients with positive medical response, conversion to laparotomy was necessary for 7 patients (50%). Of the remaining patients with CDD type 2b diverticulitis refractory to conservative treatment who underwent early elective surgery (n = 5), conversion to laparotomy was performed in 3 cases. Although statistical analysis showed no significance when comparing the conversion rate with the response rate (p = 0.56), due to the small number of cases, these observations should be interpreted with caution.

The most common inflammatory manifestation of the chronic complicated diverticulitis in the CDD type 3c subgroup was a luminal obstruction (early elective n = 11 vs. elective n = 9). However, none of these patients experienced an acute mechanical bowel obstruction requiring emergent surgical intervention. One patient with recurrent diverticulitis and a colo-vesical fistula underwent early resection. In each group, 2 patients with concomitant luminal stenosis and conglomerate mass were included. The reasons for an early elective approach in the CDD type 3c subgroup (n = 14) were as follows: colo-vesical fistula with bacteremia (n = 1), clinical deterioration with aggravating abdominal symptoms and subtotal stenosis (n = 3), and the remaining 10 patients displayed a positive response to the medical therapy and preferred an earlier surgical intervention. Among them, 2 patients experienced a history of ≥ 6 episodes of a diverticulitis of whom one patient had immunosuppressive medication after kidney transplantation. Three patients were aged 50 or younger at the time of early elective resection in the CDD type 3c subgroup after complete recovery from the acute attack. Note, in the early elective surgery group of patients with CDD type 3c, a conversion to laparotomy was indicated in 4 patients (40%) with a complete relief from symptoms and in 3 patients (75%) with persistent or increasing clinical symptoms (p = 0.28).

Thus, our data gave a first indication that both the CDD stage and the time of surgery may predict surgical conversion rates. To test this hypothesis, we performed binary logistic regression analysis. Therefore, we included variables that might be clinically relevant such as age, sex, BMI, preoperative WBC and CRP, and CDD stage and timing of operation and conducted independent variable selection by applying a stepwise backward selection and bootstrap resampling. Using this approach, we identified CDD and timing of surgery as two factors that were associated with surgical conversion. Both covariates were entered in a final regression model and a nomogram was constructed (Table 3, Fig. 2a) in which from both predictors a vertical line can be drawn to the points axis. The points obtained by this way are added up to give the total points. Finally, a vertical line can be drawn from the total point axis to the “probability (conversion)” axis, which reflects the probability of conversion to laparotomy. For example, a patient undergoing early elective laparoscopic sigmoid resection (25 points) for CDD 2b (100 points) achieves a total number of 125 points which corresponds to a probability of approximately 55% for intraoperative conversion to laparotomy. Of note, a C-index of 0.798 indicated a good model. In addition, the Hosmer–Lemeshow test confirmed the GOF (p = 0.927) and the calibration curve was almost congruent with the ideal line, indicating a well-calibrated model (Fig. 2b).

An uneventful postoperative course was recorded in 91 patients (68.4%). When comparing the rate of complications in group A and group B patients according to Clavien-Dindo [12], no statistically significant difference became evident (p = 0.540) (Table 4).

To identify predictors for an eventful postoperative course, which we defined as at least grade 1 complication according to the Clavien-Dindo classification, we again performed binary logistic regression analysis. The following variables were entered into a binary regression model performing a stepwise backward selection: Age, sex, WBC count, CRP, conversion to laparotomy, operation time, CDD, and time of surgery. Using bootstrap resampling, logistic regression analysis identified conversion to laparotomy and higher age as predictive variables for an eventful postoperative course (Table 5). Again, based on these results, we designed a nomogram to predict the probability of an eventful postoperative course (Supplementary Fig. 2a). The Hosmer–Lemeshow test (p = 0.339) as well as the calibration curve (Supplementary Fig. 2b) underlined the goodness of fit.

Next, we were interested in identifying factors which affect hospital stay in patients who underwent laparoscopic surgery for diverticular disease. Hence, we initiated a backward stepwise linear regression by including age, sex, BMI, WBC count, CRP, timing of operation, operation time, conversion to laparotomy, and the postoperative course. Interestingly, our model found elevated preoperative CRP, conversion to open surgery, and an electively planned surgery to influence the length of hospital stay (Table 6). For example, a patient with a preoperative CRP of 1 mg/dl, receiving elective laparoscopic sigmoid resection without conversion and having an uneventful postoperative course will most likely be discharged on postoperative day 9. In contrast, for a patient with a preoperative CRP of 15 mg/dl who receives early elective laparoscopic surgery, but needs to be converted to laparotomy and experiences a complicated postoperative course, a hospital stay of 26 days was predicted.