Introduction

Surgery remains an important approach to treat inflammatory bowel disease (IBD) despite the great advances in surgical techniques and postoperative care. Roughly, 80% of Crohn’s disease (CD) patients and 20% of ulcerative colitis (UC) patients will require surgical intervention in their lifetime.1,2,3 The surgical technique has significantly evolved over the last two decades,4 shifting the focus from mortality toward other endpoints such as morbidity, quality of life, and medical costs. Therefore, it is important to find an objective and reproducible tool to achieve quality control and to assess the postoperative morbidity.

Currently, the most commonly used grading system of postoperative complications is the Clavien-Dindo classification (CDC) first described in 20045 and reassessed in 2009.6 This system was designed to capture postoperative complications in five ranks according to the invasiveness of the treatment required.5,6 It is widely used for its reproducible interpretation of postoperative complications.7,8,9 However, about 80% of the literature describing complications failed to provide information about their severity10 because the CDC system only reports the highest grade of complication in most studies, thereby underestimating the disease burden in many cases.

The comprehensive complication index (CCI) is a newly developed approach adopting methods from operation-risk-index analysis in marketing research, which takes all postoperative experience into considerations.11 Based on the CDC system, the CCI uses a mathematical formula to integrate all complications graded in CDC system, summarizing the postoperative course with a new scale ranging from 0 (no complication) to 100 (death). For example, if one patient developed postoperative anastomotic leak requiring drainage, this patient is considered to have grade IIIa complication according to CDC and the corresponding CCI value is 26.2. The CCI has already been proven to be a more sensitive endpoint to detect differences of treatment effect in three randomized trials.12 Another recent study revealed CCI is more strongly related to postoperative hospital stay than CDC in gastric cancer surgery.13

Up to 21% of CD patients and 9–65% UC patients who underwent surgery develop early postoperative complications.14,15 Also, postoperative morbidity is significantly higher after IBD surgery compared to other colorectal procedures due to malnutrition, hypoalbuminemia, preoperative inflammatory status, previous laparotomy, prolonged glucocorticoid, and biologics usage.16,17 More importantly, it is common for IBD patient to develop multiple complications.18,19 The application of CCI in patients undergoing colectomy has been described in UC patients,20,21 however, the comparison of it with CDC system in IBD surgery is lacking. Therefore, we hypothesized that the CCI system was more applicable in IBD patients undergoing surgery than the CDC system. The aim of this study was to compare the CCI with the conventional CDC at our IBD center and longitudinally monitor complications using the CCI system.

Methods

Patients

The protocol of this study was approved by the Ethics Committee of Jinling Hospital. Consecutive patients who underwent surgery in Jinling Hospital from Sep 1, 2015 to Aug 31, 2017 for IBD were screened from the prospectively maintained IBD database. Any bowel resection or strictureplasty were included and abscess drainage without bowel resection, reversal of ileostomy or colostomy, bypass surgery, reoperation for postoperative complications, as well as those who underwent surgery for isolated perianal CD lesions were excluded from the study. Details of surgical technique were described as previously.22

Demographic data of gender, age, age at diagnosis, body mass index (BMI), diagnosis, operation method (open, laparoscopic), surgical procedure (small bowel resection, ileocecal/right colectomy, transverse colectomy, left colectomy, sigmoid/rectal resection, (sub) total colectomy), emergent procedure (yes/no), disease phenotype and behavior at time of surgery (Montreal classification),23 history of previous abdominal surgery, intraoperative parameters including duration of surgery (in minutes), and estimated blood loss (in mL) were reviewed. Preoperative 5-ASA, immunomodulators and steroids usage was defined as 5-ASA, azathioprine, 6-mercaptopurine and corticosteroids within 4 weeks before surgery. Preoperative biologics usage was defined as infliximab administered within 12 weeks prior to surgery. Inpatient records were reviewed for total medical cost and postoperative length of stay (LOS). Comorbidity data were collected and converted to Charlson comorbidity index, which is the most widely used scale to quantify the overall burden of comorbidities.24 Total medical cost was collected from the nursing records. Briefly, total medical cost included expenditures for medication, operation, examination, laboratory tests, and use of consumptive material. Missing or incomplete data were collected by reviewing medical records from the hospital and noted accordingly in the results if unavailable.

Postoperative Management

An enhanced recovery after surgery (ERAS) protocol was applied. Briefly, early enteral nutrition was started from postoperative day (POD) 0 and adjusted depending on the patient’s physical status and type of surgery. Daily volume and speed were added until full capacity if the patient displayed no signs of intolerance such as abdominal distension. Urinary catheter was removed on POD 1. Postoperative pain control was performed using patient-controlled analgesia comprising sufentanil for maximally 48 h. A semi-fluid diet was added followed by a semi-blended diet according to patients’ tolerance. Serum laboratory tests were scheduled as clinically needed including erythrocyte count, white blood cell count, platelet count, hemoglobin concentration, neutrophil to lymphocyte ratio, electrolytes, albumin, C-reactive protein (CRP), interlukin-6 (IL-6), and procalcitonin (PCT) level.

Criteria for Discharge

The attending surgeon reviewed laboratory tests and vital signs were measured on a daily base. Discharge was postponed for patients who had elevated body temperature (> 37.8 °C). In the absence of complications, patients were discharged provided that they tolerated a regular diet or full dosage of enteral nutrition, had passed stool, weaned off intravenous fluids, and were fully ambulant with oral analgesics.

Complications

Complication data were extracted from the database where detailed clinical course and events were described according to daily ward rounds. Clinical events were converted to the CDC grade with delicate interpretation of the five special clinical scenarios described by Clavien et al.25 For example, intra-abdominal abscess treated with both antibiotics and subsequent drainage was counted as separate complications. Postoperative ileus requiring kinetics, fever ≥ 38.5 °C treated with antipyretics, postoperative retention of urine treated with diuretics, electrolyte disturbance, and wound infection opened at beside were considered as grade I complications. Surgical site or remote infection treated with antibiotics, non-operated bowel obstruction, gastrointestinal hemorrhage treated with transfusion, deep vein thrombosis treated with anticoagulants, and anastomotic leakage treated with total parenteral nutrition for ≥2 weeks was considered as grade II complications. Intra-abdominal abscess or anastomotic leakage treated with percutaneous drainage, endoscopically treated gastrointestinal hemorrhage, wound dehiscence requiring re-suture, urinary retention, and intra-cavity fluid collection treated with percutaneous drainage was considered as grade IIIa complications. Anastomotic leakage, anastomotic stricture, stoma ischemia, stoma stricture, intra-abdominal abscess re-operated upon were considered as grade IIIb complications. Single- and multi-organ dysfunction were considered as grade IVa and IVb complications, respectively. Death was considered as grade V complication. Complications following the reoperation were integrated into the primary operation. All complications that occurred within 30 days after surgery or prior to discharge were considered relevant to the surgery.

Comprehensive Complication Index

The calculation of the CCI has been described previously in details by Slankamenac K et al.11 Generally, this index integrates the perspective of patients and physicians by investigating the median reference value of patients (MRVpat) and that of physicians (MRVphys). After multiplication and summary of the MRVs, the square root of the raw CCI was divided by two.

$$ \mathrm{CCI}=\sqrt{\sum \left(\mathrm{MRVphys}\times \mathrm{MRVpat}\right)/2} $$

The calculator is accessible on https://www.assessurgery.com. For this study, the CDC complication data for each patient were calculated and the final CCI were acquired for analysis.

Continuous Monitoring for the Complication Index of Individual Surgeons

The cumulative summation (CUSUM) technique is a tool for performance monitoring in surgery.26 The cumulative sum is \( {\sum}_{i=1}^n\left( Xi-\mu \right) \), in which Xi represents the CCI for each operation and μ is the target value. This method continuously calculates the accumulated difference between the event and target value. Implementation of the CCI in CUSUM model provides delicate visualization of surgical quality which would otherwise be crude in the scale of the CDC. Previous study has described this method in gastric cancer.13 In our center, surgeries for CD were performed by surgeon A while surgeries for UC were performed by surgeon B. The case-event CUSUM chart of the second year was drawn for surgeon A and B with the target value set for the average CCI of the first year (8.5 vs. 16.9).

Statistics

Data for demographic characteristics were expressed as mean ± SD or median (range). Kolmogorov-Smirnov method was used to test the normality of the variables. Continuous variables were analyzed using Student t test or Mann-Whitney U test as appropriate. Categorical variables were compared using χ2 test.

Univariate analysis was conducted to detect any risk factors of developing postoperative complications. Significant factors were further analyzed for between-group difference of the CCI value and the CDC grade using χ2 test. Spearman’s rank test was performed to compare the correlation of each complication scale with postoperative LOS and medical cost. Correlation coefficients were calculated for comparison and scatterplots were drawn. Univariate analysis was performed to compare difference of prolonged LOS in various factors. Significant factors were assembled for binary logistic regression model with each complication system. Receiver operating characteristic curve analysis was conducted to investigate the relationship between the CCI and prolonged LOS.

Statistical Package for Social Science version 20.0 (SPSS, Chicago, IL, USA) was used for all analyses. A two-sided p value of < 0.05 were considered as statistically significant.

Results

Patient Characteristics

Five hundred and seventy-five patients who underwent surgery for IBD from September 2015 to August 2017 were screened. One hundred and forty-nine were excluded due to stoma closure (n = 121), bypass surgery (n = 12), reoperation for complications as initial surgery (n = 9), abscess drainage (n = 5), and surgery for isolated perianal disease (n = 2). Finally, 426 patients were enrolled in the study, including 331 CD and 95 UC patients.

Demographic and baseline characteristics were shown in Table 1. Age, gender, BMI, disease classification, and preoperative medication including 5-ASA, immunomodulators, biologics were similar between patients who developed postoperative complications and those did not. In contrast, ASA score, Charlson comorbidity index, preoperative glucocorticoids usage showed significant difference between groups. Preoperative serum albumin level was lower (34.8 ± 5.8 vs. 38.2 ± 4.4 g/L) and serum CRP level was higher (30.9 ± 47.1 vs. 13.9 ± 27.0 mg/L) in patients with complications than those without. Complicated patients had longer postoperative LOS (19.7 ± 8.6 vs. 8.2 ± 3.3 d) and higher medical cost (97,410 ± 40,201 vs. 56,183 ± 22,297 CNY) compared with non-complicated patients.

Table 1 Patients’ characteristics
Full size table

Perioperative parameters were shown in Table 2. There was no difference in operation method, operation time, estimated blood loss and intraoperative transfusion between the two groups. However, emergent procedure (p < 0.001) and (sub) total colectomy/IPAA (p = 0.001) were related to higher incidence of complications.

Table 2 Perioperative parameters and surgical procedure
Full size table

Complications

Totally, 297 complications occurred in 144 (33.8%) patients. The rate of severe complications (CDC grade ≥ III) was 12.9% and mean CCI was 9.7 ± 14.5. The overall complication rate was 28.1% for CD and 53.7% for UC, respectively. One hundred and three of the 144 patients developed multiple complications and 41 patients developed a single complication. According to the Clavien-Dindo classification, 18 patients had grade I complications, 69 patients had grade II complications, 40 patients had grade IIIa complications, 10 patients had grade IIIb complications, 5 patients had grade IVa complications, 1 patient had grade IVb complication and 1 patient died of pulmonary and renal failure after septic shock.

The most frequent complication was wound complications (13.1%), followed by ileus (12.0%), intra-abdominal abscess (7.7%), and urinary complications (6.8%). Eight patients underwent reoperation because of anastomotic leak (4/8), intra-abdominal sepsis (2/8), stricture of the stoma (1/8), and necrosis of the stoma (1/8). (For details of recorded complications, see Supplementary Table A and B).

The distribution of CCI and CDC was shown in Fig. 1. The most frequent CDC complications were grade II and IIIa. Similarly, the CCI value of most cases was in the range of 20.9–26.1 and 26.2–33.6, since 20.9 and 26.2 represented the CCI value of single grade II and IIIa Clavien-Dindo complication, respectively. In patients with complications, the mean CCI was 28.9 ± 12.8. The rate of severe complications (≥ III) was 39.5%, however, 81 (56.2%) cases demonstrated a CCI value of ≥ 26.2.

Fig. 1
figure 1

Overall distribution of complication in different scales. Most cases were graded as II or IIIa according to CDC. Similarly, the CCI of most cases clustered in the values that represented CDC grade II and IIIa

Full size image

Comparison of CDC and CCI

Significant variables in the univariate analysis were further analyzed using χ2 test or student t test to compare the sensitivity of CDC and CCI. The results were shown in Table 3. ASA score ≥ 3, preoperative hypoalbuminemia, elevated CRP level, emergent surgery, and total colectomy were related to higher incidence of severe complications and higher CCI value. Charlson score index, preoperative glucocorticoids usage, and previous abdominal surgery were not associated with higher incidence of severe complications (p = 0.682, p = 0.117, p = 0.177, respectively), however, they were associated with higher CCI value (p = 0.009, p = 0.002, p = 0.006, respectively).

Table 3 Comparison of the CCI and the CDC regarding factors significantly related with complication rate
Full size table

Correlation Analysis

The capability of CDC and CCI to reflect postoperative morbidity was evaluated by their correlation with postoperative LOS as well as medical cost. Notably, in patients with multiple complications (N = 103), the CCI showed a stronger positive correlation (ρ = 0.604, p < 0.001) with postoperative LOS than CDC (ρ = 0.508, p < 0.001). Scatterplots and correlation coefficient were shown in Fig. 2. In patients with prolonged LOS (> 30 days, N = 20), the CCI displayed a significant correlation (ρ = 0.481, p = 0.032) with postoperative LOS while the CDC did not (ρ = 0.342, p = 0.140). Both CCI and CDC were correlated with medical cost (ρ = 0.554 vs. ρ = 0.552, p < 0.001 vs. p < 0.001).

Fig. 2
figure 2

Scatterplots demonstrating the relationship between each complication scale and LOS in patients with multiple complications. Spearman’s rank test revealed the CCI had a stronger relationship with postoperative LOS than CDC (ρ = 0.604, p < 0.001 vs. ρ = 0.508, p < 0.001); LOS (d), length of stay (days)

Full size image

Logistic Regression

Apart from the CCI value and CDC grade, univariate analysis showed preoperative glucocorticoids usage (p = 0.050) and CRP level > 8 mg/L (p = 0.002) were related to prolonged LOS. Multivariate logistic regression analysis was performed and the results were summarized in Table 4. Neither preoperative glucocorticoids nor elevated CRP level could predict prolonged LOS. However, higher CCI value (p < 0.001, OR 1.161, 95% CI 1.093–1.234) and CDC grade (p < 0.001, OR 3.811, 95% CI 2.283–6.362) both conferred to an increased risk of prolonged LOS. Notably, the CCI model had a higher R2 than the CDC model (0.568 vs. 0.467).

Table 4 Logistic regression of factors associated with prolonged postoperative length of stay (LOS) (> 30 days)
Full size table

Monitoring the Complication Rate of Individual Surgeon

CCI-CUSUM charts were drawn to monitor postoperative complication outcome of each surgeon. Target value was set for the average CCI of the first year. For surgeon A, 331 cases were depicted with the target value set for 8.5. In Fig. 3a, the CUSUM score showed a constant fluctuation around base line during the first year, with the highest value (124.9) peaking at case 26. In the second year, the CUSUM score showed a gradual decrease with a final score of − 176.2 at case 331, indicating improvement of the complication outcome. For surgeon B, 95 cases were depicted with the target value set for 16.9. CUSUM score in Fig. 3b peaked at case 5 (61.0), followed by a gradual decrease. The final score was − 111.0, indicating improvement of the complication outcome.

Fig. 3
figure 3

Monitoring surgical complications. a CUSUM score for surgeon A was derived from target value of average CCI (8.5) in the first year (2015.9–2016.8). The final CUSUM score was − 176.2, indicating a decreased CCI compared to the previous year. b CUCUM score for surgeon B was derived from a target value of average CCI (16.9) in the first year. The final CUSUM score was − 111.0, indicating a decreased CCI compared to the previous year. c Change of complication rate over time for CD surgery with the final rate being 28.1% at case 331. d Change of complication rate over time for UC surgery with the final rate being 53.7% at case 95

Full size image

Receiver Operating Characteristic Curve Analysis

Receiver operating characteristic curve analysis demonstrated that the CCI value is predictive of prolonged postoperative length of stay (> 30 days) with an optimal cutoff value of 24.2. The area under curve was 0.951 (95% CI 0.925 to 0.970, p < 0.001). The sensitivity was 94.1% and the specificity was 85.0%, respectively. (Fig. 4).

Fig. 4
figure 4

Receiver operating characteristic curve analysis demonstrated that the CCI value is predictive of prolonged postoperative length of stay. The optimal cutoff value was 24.2 and the area under curve was 0.951 (95% CI 0.925 to 0.970, p < 0.001). The sensitivity was 94.1% and the specificity was 85.0%, respectively

Full size image

Discussion

The complication rate after IBD surgery is higher than other intestinal diseases due to preoperative poor nutrition status, preoperative medication and inflammatory status.14,17,27 To date, this is the first study comparing CCI and CDC system after IBD surgery. The results revealed that CCI had a better reflection of the postoperative morbidity in IBD patients who underwent surgery compared to the conventional CDC system. Our data revealed the CD and UC patients who underwent surgery had a mean CCI of 8.0 and 15.8, respectively. These results proved CCI as a qualified marker to reflect postoperative surgical morbidity in IBD patients.

It is well known that the CDC focuses on the complication of highest grade while neglecting the concomitant ones. A significant proportion of IBD patients (71.5%) in this study developed two or more different complications. Therefore, failure to capture the number and the severity of minor complications will result in insufficient report of the postoperative morbidity in IBD patients. The CCI integrates all complications into one formula and depicts the entire spectrum of complications, thereby comprehensively demonstrating the postoperative morbidity. Additionally, comparison between multiple minor complications and single severe complication are often cumbersome under the CDC system. However, the CCI scale allows direct comparison of morbidity and identification of more severe patients, thus aiding in the clinical decision for IBD patients.

Due to its superiority, CCI instead of CDC has been used in many publications. For example, Kalisvaart et al.28 reported liver transplantation had a median CCI of 37.1 at discharge and 47.6 after 6 months. Another study showed laparoscopic gastrectomy had a lower CCI of 4.9 vs. conventional open gastrectomy (9.0).13 Kim et al. also revealed that CCI is more strongly correlated with postoperative hospital stay than is the conventional CDC after gastric cancer surgery. After externally analyzing the CCI value, Slankamenac et al.12 found a more significant inter-group difference in postoperative morbidity with p value of 0.009 compared to the original study (p = 0.035).29 Therefore, the CCI was able to compare morbidity after different surgery, to monitor the increase of disease burden over time, and was more sensitive to grade complications compared to the conventional grading system.

In this study, patients with preoperative glucocorticoids usage had a higher CCI (12.5 vs. 8.3, p = 0.002) and a prolonged postoperative LOS (p = 0.050) according to univariate analysis. However, there was no difference of severe complications (CDC ≥ Gr III) in patients with preoperative glucocorticoids administration and those without (17.2 vs. 11.7%, p = 0.117). As indirect comparison of effect sizes, the between-group differences for the CCI presented lowest p value (p = 0.002), compared to the differences for the specific outcome “prolonged postoperative LOS” (p = 0.050) and the differences for the “rate of severe complication” according to CDC (p = 0.117). These findings suggest that the CCI has a higher sensitivity and may serve as a more effective endpoint for assessing clinical outcomes after IBD surgery.

Postoperative morbidity is measured in different ways. Medical cost and postoperative LOS are both good indicators of postoperative complications. However, the economic status of the patients and type of reimbursement in Chinese hospital may affect the medical cost to a great deal. On the contrary, our center had an established procedure for IBD patients and criteria for discharge. Therefore, our results showed that medical cost was only moderately related to the CCI value (ρ = 0.554, p < 0.001) while postoperative LOS was strongly related to the CCI value (ρ = 0.756, p < 0.001). In addition, after excluding patients with one or no complication, the correlation coefficient of CCI appeared significantly higher (0.604 vs. 0.508). In cases of prolonged hospital stay (> 30 days), CCI showed a significant correlation while CDC did not. This reflected the major drawback of reporting the most severe complication only in the CDC system and these results indicated that the CCI was more capable of grasping the magnitude and number of complications especially in multi-complicated patients.

The CCI value is a continuous scale ranging from 0 to 100 while most of the cases of our study clustered between 20.9 and 33.6. This is because the complications recorded in this study were primarily surgical site infection, which were treated with antibiotics or percutaneous drainage under local anesthesia. Ileus was another frequent complication but it was usually accompanied by complications of higher grades. Thus, complication grades that occurred most frequently were II and IIIa under the CDC system.

When the CCI and CDC were assembled for multivariate analysis, both complication scales were risk factors for prolonged LOS (> 30 days). However, the CCI model had a higher R2 value. Since the R2 value of each model represents the proportion of variation explained,30 the CCI model had a greater goodness of fit compared to the CDC model. This, from another different perspective, reflected the advantage of the CCI system over the conventional CDC grading system.

In terms of monitoring surgical outcomes for individual surgeon, the CCI can be implemented in the CUSUM model. Considering the capability of CCI to reflect the overall magnitude and number of complications, continuous monitoring of the CCI provides more detailed information about surgical performance than monitoring the complication rate alone. Thus, any deviation or lapses in surgical performance will be detected and corrected promptly.

Several limitations must be considered when interpreting the results of our work. First, this is a retrospective design carried out in a single center. Second, continuous monitoring the CCI for a single patient can provide information about changes in the postoperative course over time. This study did not include follow-up of the patients and evaluation of the long-term CCI, leaving the potential advantage of monitoring the CCI unexplored. Third, the CCI has certain drawbacks. For example, reoperation for complications has higher CCI value than non-operative treatment but the decision of surgical intervention is largely affected by the practice of individual surgeon. Therefore, consensus of treatment must be reached regarding different complications. Finally, in some cases, length of stay might be prolonged in patients awaiting rehabilitation or home services, leading to potential bias.

In conclusion, the CCI is applicable in IBD patients who undergo surgery and is more sensitive in reflecting both the number and severity of postoperative complications than the conventional CDC grading system. The CCI is more strongly correlated to postoperative LOS than CDC in multi-complicated patients. Surgeons can in advance identify patients with higher CCI value and initiate radical therapies instead of conservative treatment. In addition, continuous monitoring of the CCI provides surgical quality feedback to individual surgeons.