A nationwide population-based study on the clinical and economic burden of anastomotic leakage in colorectal surgery

Aim Anastomotic leakage (AL) is one of the most dreaded complications in colorectal surgery. In 2013, the International Classification of Diseases code K91.83 for AL was introduced in Germany, allowing nationwide analysis of AL rates and associated parameters. The aim of this population-based study was to investigate the current incidence, risk factors, mortality, clinical management, and associated costs of AL in colorectal surgery. Methods A data query was performed based on diagnosis-related group data of all hospital cases of inpatients undergoing colon or sphincter-preserving rectal resections between 2013 and 2018 in Germany. Results A total number of 690,690 inpatient cases were included in this study. AL rates were 6.7% for colon resections and 9.2% for rectal resections in 2018. Regarding the treatment of AL, the application of endoluminal vacuum therapy increased during the studied period, while rates of relaparotomy, abdominal vacuum therapy, and terminal enterostomy remained stable. AL was associated with significantly increased in-house mortality (7.11% vs. 20.11% for colon resections and 3.52% vs. 11.33% for rectal resections in 2018) and higher socioeconomic costs (mean hospital reimbursement volume per case: 14,877€ (no AL) vs. 37,521€ (AL) for colon resections and 14,602€ (no AL) vs. 30,606€ (AL) for rectal resections in 2018). Conclusions During the studied time period, AL rates did not decrease, and associated mortality remained at a high level. Our study provides updated population-based data on the clinical and economic burden of AL in Germany. Focused research in the field of AL is still urgently necessary to develop targeted strategies to prevent AL, improve patient care, and decrease socioeconomic costs. Supplementary Information The online version contains supplementary material available at 10.1007/s00423-023-02809-4.


Introduction
A common yet dreaded postoperative complication in colorectal surgery is anastomotic leakage (AL), which is associated with longer hospitalization, a higher rate of reoperation, and higher overall morbidity and mortality [1,2].AL not only leads to a high clinical burden for the patients affected but causes significantly higher costs for hospitals and national health care systems [3].Thus, research in the field of AL has increased over recent years with a number of records in the PubMed database for "anastomotic leakage" of 423 in the year 2010 and 1097 in 2020.Preoperative, tumor-associated, intraoperative, and other risk factors for AL have been identified so far [4,5].Research in the field thus focuses on identifying biomarkers for AL as well as finding optimal surgical techniques, biomaterials, and targeted drugs to reduce the risk of AL after gastrointestinal surgery; however, no treatment option except for diverting enterostomy exists so far to reliably prevent AL [6].AL rates after lower gastrointestinal surgery are reported in the literature to occur in 1-19% of operations; however, reported leakage rates vary largely across studies [4,[7][8][9].
In 2013, the ICD (International Statistical Classification of Diseases and Related Health Problems) code K91.83 for postoperative gastrointestinal AL has been introduced to the 55 Page 2 of 11 German diagnosis-related group (DRG) system.DRG statistics data from all inpatients in German acute care hospitals are collected by the German Federal Statistical Office (DESTA-TIS).Microdata of the DRG statistics can be retrieved by researchers through the Research Data Centers associated with DESTATIS.These prerequisites make it possible to perform a retrospective population-based study analyzing AL rates and the resulting clinical and economic burden.
The aim of this study was to delineate current trends of AL rates in colorectal surgery in Germany by examining all inpatient cases from 2013 to 2018 based on DRG data sets.Furthermore, outcomes of patient care were assessed by studying therapeutic modalities for the clinical management of AL, mortality, hospital length of stay, and socioeconomic costs.

Data query and inclusion criteria
A data query through the Federal Statistical Office (DESTATIS) was performed for all inpatients undergoing colon resections (OPS 5-455) and sphincter-preserving rectal resections (OPS 5-484) from 2013 to 2018 in German acute care hospitals.Parameters retrieved were patient age and sex, main diagnosis, secondary diagnoses, postoperative complications and postoperative AL, morbidity scores, in-house mortality, therapeutic management of AL, length of hospital stay, and hospital reimbursement volume (Table S1).The Strausberg Comorbidity Score and weighed Elixhauser Score were used for the comparison of general comorbidity between patients with and without AL [10][11][12].The code for the data query was written in SAS programming language according to the DESTATIS requirements.Data were retrieved through remote-controlled data processing and provided as raw data by DESTATIS [13].The detailed methods and underlying regulations for reporting of inpatient cases in German hospitals have been previously described in detail [14][15][16][17].In summary, all acute care hospitals in Germany are required by law to document and report every inpatient case with all relevant procedures and diagnoses, mainly for financial hospital reimbursement.The data are monitored for correctness by the medical service of the health insurance funds and stored by DESTATIS.The following data items per in-house hospital case are included in the DESTATIS database and can be queried for research purposes: main diagnosis (ICD), secondary diagnoses (ICD), procedures (OPS), age, year of birth, reason and type of admission, reason and type of discharge including in-hospital death, length of hospital stay, specialist department, Case Mix, Case Mix hospital reimbursement volume in EURO, hospital location (federal state, district, municipality, postal code), and patient residence (federal state, district, municipality, postal code).No temporal information regarding the sequence of procedures or diagnosis within one hospital case and no patient follow-up data can be retrieved from the database.Raw data from data queries are provided as pooled data (number of cases for defined combinations of ICD and OPS codes).For secondary data analysis used in this study, no ethics committee statement is required [18].For data protection purposes, case numbers ≤ 2 are blinded by DESTATIS and not available to the authors.

Statistics
GraphPad Prism Version 9.1.2(GraphPad Software, CA, USA) was used for statistical testing and data visualization.Fisher's exact test, chi-square test, chi-square test for trend, and odds ratio were calculated.T-and Wilcoxon-signed rank tests were performed within the query code.Continuous parameters and variables are presented as mean with single standard deviation.Data were analyzed descriptively for each year and presented either as absolute numbers or relative rates.This study was conducted and reported using the STROBE Statement checklist [19].

Baseline characteristics
A total of 690,690 cases were registered by DESTATIS from 2013 to 2018 and included in this study, 513,951 cases with colon resections, and 176,739 with sphincter-preserving rectal resections.The total number of colon resections was 87,853 in 2013 and 85,760 in 2018 and sphincter-preserving rectal resections were performed 31,195 times in 2013 and 28,834 times in 2018, decreasing slightly over the years (Fig. 1A).

Anastomotic leakage rates
An increase in reported AL rates for both types of surgery was seen in the first 3 years after the introduction of the ICD-code K91.83 for postoperative AL.Reported relative AL rates for the total number of colon resections were 5.1% in 2013 and 6.7% in 2018 and for rectal resections 7.7% in 2013 and 9.2% in 2018 (Fig. 1B).The mean AL rate (from 2013 to 2018) was 6.2% for colon resections and 8.8% for rectal resections (Table 1).

Other postoperative complications
Postoperative abscess/surgical site infection rates following colon resections were 8.8% in 2013 and 7.7% in 2018 and for rectal resections 7.7% in 2013 and 6.6% in 2018.Wound dehiscence occurred with a rate of 6.5% in 2013 and 7.0% in 2018 for colon resections and with a rate of 5.6% in 2013 and 5.6% in 2018 for rectal resections.The rate of postoperative fistula formation was 5.6% in 2013 and 5.2% in 2018 for colon resections and 5.1% in 2013 and 4.6% in 2018 for rectal resections (Fig. S1).

Indication for surgery
Concerning the primary indication (main diagnosis) for colon and rectal resections, relevant differences in AL rates could be detected.For colon resections, patients with diverticulosis showed a significantly lower than average leakage  rate (4.6%, OR 0.68).For patients with Crohn's disease, the leakage rate was above average for colon resections (6.9%, OR 1.12).Patients with colorectal cancer showed a leakage rate of 6.3% which was not significantly different from the average leakage rate for colon resections of 6.2% (OR 1.02).For rectal resections, patients with diverticulosis showed a significantly lower than average leakage rate (6.6%, OR 0.69).For patients with Crohn's disease (14.5%, OR 1.76) and patients with colorectal cancer (10.1%,OR 1.40), the leakage rate was higher than average for rectal resections (Table 1).

Management of anastomotic leakage
44.4% of patients with AL after colon resections and 32.9% of patients with AL after rectal resections underwent relaparotomy (2013).Relaparotomy rates for cases with AL only decreased for rectal resections over time (Fig. 2A).Abdominal vacuum therapy was performed in 16.6% of cases with AL after colon resections in 2013.For cases with AL after rectal resection, abdominal vacuum therapy was performed in 9.1% in 2013 (Fig. 2A).Regarding endorectal vacuum therapy, a significant increase over time could be detected.In 2013, endorectal vacuum therapy was performed in 3.5% of cases with colon resections and 17.8% of cases with rectal resections and postoperative AL.In 2018, endorectal vacuum therapy was performed in 7.1% of cases with colon resections and 30.0% of cases with rectal resections and AL (Fig. 2A).Terminal enterostomy was performed in 10.2% of cases after colon resections and AL and 6.4% of cases after rectal resections and AL in 2013.Rates for terminal enterostomy did not change significantly over time (Fig. 2A).

Mortality
The in-house mortality for patients undergoing colon resections without AL was 7.6% in 2013 and 7.1% in 2018.Mortality for patients with AL after colon resections was 22.2% in 2013 and 20.1% in 2018.A slight negative trend in mortality rates could be detected (Fig. 2B).The in-house mortality for patients undergoing rectal resections without AL was 4.6% in 2013 and 3.5% in 2018.Mortality for patients with AL after rectal resections was 11.8% in 2013 and 11.3% in 2018.Here, only a negative trend in mortality rates could be detected in patients with rectal resections without AL (Fig. 2C).

Length of hospital stay and hospital reimbursement
The occurrence of AL had a significant influence on the length of hospital stay in both colon and rectal resections.In 80% of cases with colon resections and AL, the length of hospital exceeded 20 days while in cases without AL, 28% of patients stayed in the hospital for more than 20 days, most likely due to other complications.In 80% of cases with rectal resections and AL, the length of hospital stay was longer than 20 days while in cases without AL, 25% of patients stayed in the hospital for more than 20 days (Fig. 3A, C).
The mean hospital reimbursement sum for colon resections was 12,603€ without AL and 28,616€ with AL in 2013 and 14,876€ without versus 37,521€ with AL in 2018, showing an increase in the mean hospital reimbursement sum over time.Regarding rectal resections, the mean hospital reimbursement was 12,889€ without and 23,488€ with AL in 2013 and 14,602€ versus 30,606€ in 2018 (Fig. 3B,  D; Table 2).To estimate the potential saving that could be achieved if AL could be prevented in all cases, we calculated a hypothetical sum from the mean hospital reimbursement rates of patients with and without AL (Table 2).

Discussion
With more than 690,000 inpatient cases undergoing colon resections and sphincter-preserving rectal resections, our study is currently the largest nation-wide population-based study to analyze AL rates after surgery of the lower gastrointestinal tract.Despite increasing research in the field of anastomotic healing and improvement of surgical techniques, our data show no decrease in leakage rates from 2013 to 2018 with a mean AL rate for colon resections of 6.2% and rectal resections of 8.8%.Regarding individual risk factors for AL, known risk factors such as male gender, diabetes, hypertension, obesity, and chronic kidney disease could be confirmed by our data [4,7,8].Interestingly, cachexia showed the highest odds ratio for AL of 1.94 in cases with colon resections and 1.66 in cases with rectal resections (Table 1).When looking at the management of AL, there is a significant increase in endoscopic therapy in terms of endoluminal vacuum therapy over the years leading to a rate of 30% for AL after rectal resections in 2018; however, relaparotomy rates only slightly decreased in the studied period (Fig. 2A).Two potential factors could explain this phenomenon.Firstly, for an effective endoluminal vacuum therapy, the creation of a diverting enterostomy might be necessary thus requiring relaparotomy.Secondly, relaparotomy for peritoneal lavage might be required for patients with AL before or in combination with endorectal vacuum therapy thus not leading to a significant reduction in relaparotomy rates.Hence, our data suggests that although endoluminal vacuum therapy for AL after colorectal surgery is increasingly applied, it does not prevent revision surgery for lavage and creation of a protective enterostomy in all cases.
The AL rates that were coded increased over the observation period from 2013 to 2018, reaching a relatively stable level by 2015.The most probable cause is underreporting in the first years after the introduction of the ICD code K91.83 in 2013.The bias of under-reporting of AL in the following years is unlikely, as the hospitals would have deliberately waived a higher DRG-based reimbursement sum when treating for AL but not coding it in the case data.Overreporting of diagnoses and procedures on the other hand is strictly controlled by the medical service of the health insurance funds in Germany but could still lead to a bias in our study.Other studies have described similar leakage rates but to our knowledge, no study had nearly as many cases or patients included in their data sets.Bonström et al. describe AL in 10% of the included 6948 patients undergoing low anterior resection in a population-based study from 2019 [20].Gessler et al. describe AL rates of 7.0% for right hemicolectomy, 7.4% for left hemicolectomy, and 18.8% for rectal resection in a patient collective of 600 patients [2].In a nationwide analysis from the USA, Midura et al. however show a much lower overall leakage rate of 3.8% [7].The heterogeneity of assumed leakage rates has been reported several times recently [4,21].One confounder in most studies on AL rates is that postoperative diagnostic regimens are not standardized leading to under-diagnosis, especially of grade A leakage (according to the International Study Group of Rectal Cancer 2010) which is defined by not affecting the postoperative management [22].However, with our study, we could show that despite increasing knowledge on the risk factors for AL, there was no trend towards decreasing leakage rates in the studied time period.
Regarding the economic burden of AL, only the DRGbased hospital reimbursement volume is accessible by our type of data query.A significant increase in the hospital reimbursement sum for cases with AL compared to cases without AL can be seen.We have calculated potential savings that could be achieved if no AL would occur (130,705,439 € for colon resections and 42,235,717 € for rectal resections in 2018, Table 2).However, the real costs of AL for the individual hospital cannot be derived from the DRG data.It has been described that the real cost of AL for the individual hospital is significantly higher and is not covered by the DRG-based reimbursement system.La Regina et al. could demonstrate in a study including 95 patients undergoing colorectal cancer surgery, that the mean profit from the DRG-based reimbursement was 542€ per case without postoperative complications and the mean loss for cases with AL was 12,181€ per case for the hospital treating patients that developed AL [23].In a study from England, Ashraf et al. could also demonstrate inadequate hospital reimbursement for cases with AL after low anterior rectum resections [24].The slight increase in the overall hospital reimbursement sum is most likely due to the fact that the hospital reimbursement is calculated based on a base rate per inpatient hospital case, which increases steadily over time.
Ultimately, the question remains as to why AL rates have stagnated at such a high level.A Dutch study from 2022 investigated the impact of perioperative potentially modifiable risk factors on AL after colorectal surgery during a study period from January 2016 to December 2018 [25].They identified modifiable risk factors such as low preoperative hemoglobin, surgical site contamination, hyperglycemia, and inadequate timing of perioperative antibiotic prophylaxis.Interestingly, most of these factors were already known to increase the risk of AL, but their prevention was still not applied before and during surgery.Although we could not draw these data from the DESTATIS dataset in our study, we suspect that the Dutch data are transferable to the situation in Germany.We therefore hypothesize that despite known preventive measures to reduce AL rates, adherence is still lacking in Germany, which could at least partly explain the stagnant AL rates in our study.Furthermore, we hypothesize that even if all standards to prevent AL are met, there is a residual risk for AL that has not yet been identified.Moreover, some patient-specific risk factors cannot be modified before surgery.To date, there are no established local or systemic pharmacological therapies to prevent anastomotic complications and improve the postoperative healing process in patients undergoing colorectal surgery.Therefore, with our study, we aim to raise awareness that AL is an unresolved problem in Germany, which represents an unchanged burden for patients as well as for health care providers and insurance companies.

Conclusions
This study presents a large population-based data set on AL rates following lower gastrointestinal surgery and gives a timely overview of the current data on AL rates and associated socioeconomic costs of AL after lower gastrointestinal surgery in Germany.The data show a great need for further research in the field of AL and for better adherence to perioperative standards to minimize known risks to efficiently reduce leakage rates and thus improve patient outcomes in the future.Furthermore, treatment for AL and care for affected patients must improve to reduce the high in-house mortality associated with AL.

Fig. 1
Fig. 1 Development of surgery numbers from 2013 to 2018 for colon resection and sphincter-preserving rectal resection, anastomotic leakage rates, and risk factors.(A) The total number for colon resections was 87,853 in 2013 and 85,760 in 2018 and for sphincter-preserving rectal resections 31,195 in 2013 and 28,834 in 2018, decreasing slightly over the years.Data are absolute numbers per year.(B) The data show relative anastomotic leakage rates of 5.08% in 2013 and 6.74% in 2018 for colon resections and for sphincter-preserving rectal resections of 7.69% in 2013 and 9.15% in 2018.Data show relative

Fig. 2
Fig. 2 Management of anastomotic leakage and in-house mortality.(A) Procedures following anastomotic leakage after colon and rectal resections (relaparotomy, abdominal vacuum therapy, endorectal vacuum therapy, terminal enterostomy).Rates of procedures in cases with no anastomotic leakage (AL) are depicted for comparison.Data show relative rate per year.Chi-square test for trend.p < 0.05 = *,

Fig. 3
Fig. 3 Length of hospital stay and hospital reimbursement.(A, C) Distribution of cases to the length of hospital stay (≤ 5 days, 6-10 days, 11-20 days, ≥ 20 days).Data is depicted as percentage of total cases for colon resection ± anastomotic leakage (A) and rectal resection ± anastomotic leakage (C).A significant association between anastomotic leakage and length of hospital stay can be shown.Data are mean ± SD.Chi-square test.p ≤ 0.0001 = ****.(B, D) Mean hospital reimbursement sum per case for colon and rectal resections with and without anastomotic leakage.Data are mean reimbursement sum per year, t-test, p ≤ 0.0001 = ****.AL, anastomotic leakage

Table 1
Characteristics of study population and corresponding anastomotic leakage rates

Table 2
Hospital reimbursement and hypothetical savings * Hypothetical saving in case of no anastomotic leakage = n ("AL YES") × (mean hospital reimbursement sum "AL YES" − mean hospital reimbursement sum "AL NO") * Unpaired t-test *