Clinical and economic burden of surgical complications during hospitalization for digestive cancer surgery in France

Abstract Background Surgical complications and particularly infections after digestive cancer surgery remain a major health and economic problem and its burden in France is not well documented. Aims The aim of this study was to analyse recent data regarding surgical complications in patients undergoing major digestive cancer surgery, and to estimate its burden for the French society. Methods and Results Using the 2018 French hospital discharge database and 2017 National CostStudy we studied hospital stays for surgical resection in patients withdigestive cancer. The population was divided into three groups based onpostoperative outcomes: no complications (NC), related infectious complications (RIC) and other complications. The main analysis compared the length and cost per stay between RIC and NC. Forty‐Four thousand one hundred and twenty‐three stays following a digestive cancer resection were identified. Lower gastro‐intestinal cancers were the most prevalent representing 74.8% of stays, the rate of malnutrition was 32.8% and 15.8% of patients presented RIC. Mean (SD) length of stay varied from 11,7 (9.0) days for NC to 25,5 days (19.5) for RIC (p < 0.01). The mean cost per patients' stay (SD) varied from €10 641 (€ 5897) for the NC to €18 720 (€7905) for RIC (p < .01). Conclusion The risk of RIC after digestive cancer resection remains high (>15%) and was associated with significantly longer length of stay and higher cost per stay. Although important prevention plans have been implemented in recent years, care strategies are still needed to alleviate the burden on patients and the healthcare system.

the length of hospital stay, [3][4][5][6] a complicated drug treatment, repeated surgical interventions, an increased risk of mortality 7 and issues related to antimicrobial resistance. 8The 2017 annual epidemiological report of the European Centre for Disease Prevention and Control indicated that the surgical site infection (SSI) percentage was influenced by the type of procedure, with the highest rate of 10.1% SSIs reported for colon surgeries. 9Consistently, a study conducted in UK hospitals from 2013 to 2018 reported that colon surgery presented the highest cumulative incidence of SSIs with a percentage of 8.7%. 10 A French study carried out in 2006 in patients (N = 330) hospitalized for digestive surgery highlighted a prevalence of 5.9% of SSIs for digestive surgery, and 15.9% for digestive cancer surgery, 11 while a study conducted in Japan showed a prevalence of 10.7% of postoperative infections in patients who underwent digestive surgery. 12other French study analyzing patients identified with ICD10 codes of cancer and digestive surgery act-procedures from 2012 to 2016 from the national health database named "Echantillon généraliste des bénéficiaires" reported that 46% of patients have infectious complications. 135][16][17][18][19] In France, a national nosocomial infection control program was established in 1995 and updated over the years. 20,21spite all groups, programs, and recommendations, the prevalence of infections remains high.Recent data from the UK and France even suggest increased rates in recent years up to 60% after large and small bowel and digestive surgeries. 2,22Studies are available on the total economic burden 23 however, none has assessed the incidence of infections and its economic consequences after surgery according to the location of digestive cancers.
As many protocols have been put in place to improve recovery and reduce complications after digestive cancer surgery in France, the aim of this study was to provide updated figures on their burden.

| Data source
This study was a retrospective analysis using the 2018 French National hospital discharge database for medicine, surgery, and obstetrics (Programme de Médicalisation des Systèmes d'Information-Médecine, Chirurgie et Obstétrique, PMSI-MCO).This medicoadministrative database gathers medico-administrative information for all patients admitted to public and private hospitals in France until their discharge.The primary aim of this database is to finance hospitals with a classification system as American hospitals do with their diagnosis-related group (DRG); it summaries discharge information like number of patients, number and duration of stays per patient.The International Classification of Diseases, 10th revision (ICD-10) is used for the medical information.Medical procedures using the common classification of medical acts (CCAM) which is a classification of all procedures performed by medical doctors in hospitals, are coded for all stays.The severity of stays is also coded using a medico-economic model which considers the healthcare resource utilization for each DRG.Severity ranges from no severity (0) to high severity (4).
A quality control of hospital coding is audited annually by health authorities for a sample of hospitals.

| Study population
Forty-Four thousand one hundred and twenty-three stays have been analyzed.Mean (SD) age was 69.2 (12.4) years old and the men proportion was 57% (Table 1).The study was performed on an extraction of this database which is all the hospital stays of patients who were hospitalized in 2018 for surgery for digestive cancer using ICD-10 codes for cancers (from C15 to C26 as Principal or Related Diagnosis) and CCAM codes for digestive surgery procedures: esophagectomy, gastrectomy, segmental resection of duodenum, pancreatectomy, hepatectomy, splenectomy, and biliary tract surgery; and segmental resection of the small bowel, colectomy and proctectomy.No exclusion criteria were applied.
Subgroups were defined depending on the digestive location of the cancer: upper gastro-intestinal (GI), hepato-pancreato-biliary, and lower GI.The CCAM codes and subgroups used for the inclusion criteria and for the subgroups were reviewed by a digestive surgeon (GP).
Then, three groups of hospital stays were defined according to the complication after surgery: stays with related infectious complication (RIC), stays with other complication (OC) meaning a noninfectious complication, and stays without complication (NC) using diagnosis of additional complication classification to compare them.
The RIC group was defined by the ICD-10 code T81.4 "Infection following a procedure, not elsewhere classified", or by a diagnosis of additional complication with ICD-10 related to infectious pathologies (see Additional file 1: List of 48 pathologies) combined with a complication code (T81.8,T81.9, T81.7).The OC group was defined by stays with codes related to a complication and already selected in the RIC groups (details of coding in the Appendix).Finally, the NC group included patients not part of the two previous groups.A sub-group of RIC has been defined: "The confirmed-related infectious complication group (C-RIC)" was a subgroup of RIC including only the ICD-10 code T81.4.
Coding at hospital was done under the supervision of a medical doctor expert in coding.As we were interested in infections directly related to surgery, this study did not include re-admission for infection.Of note, PMSI database do not collect infections occurring at home after hospital discharge.

| Costs calculation
For each DRG, a cost is associated by the French agency in charge of hospital funding (ATIH).Costs of the patients groups of interest were calculated using the 2017 National Cost study method (Echelle Nationale de Coûts à Méthodologie Commune, ENCC), and not using the tariff method, as per the French National Authority for Health (Haute Autorité de Santé) recommendation. 24In the ENCC, the cost is established using a representative sample of about 135 hospitals (public and private) which   represents 17% and 8% of stays in public and private hospitals in France, respectively, yet with correct to good statistical estimates according to ATIH as the hospital sample represent a majority of hospital stays.In the ENCC model all the costs per stay are included, e.g.staff resource used, equipment, drug, and support services.All these costs are consolidated into Diagnosis related groups (DRG) costs.

| Statistical analysis
Descriptive analysis was provided with mean, standard deviation for all quantitative variables (age, LOS, cost).Percentages were calculated for qualitative variables (sex, severity, malnutrition, cancer location, transfer to ICU).Average cost per stay and length of stay comparison  between groups were performed using unpaired Wilcoxon test because the assumptions for using the t-test were not met as the standard deviations of costs and length of stay by patient group or by patient group and digestive site were not similar.Analyses were performed with R Studio software 3.6.1.The dispersion of costs showing quartiles was represented using boxplot.

| RESULTS
Overall, 44 123 stays for digestive cancer surgery were included (Figure 1).The rate of stays was: RIC: 15.8% (C-RIC: 11.2%), OC: 9%, and NC: 75.2% (Table 1).Surgery of tumor located on the lower GI tract was the most frequent and concerned 74.8% of the stays ( The mean length of stay and the mean cost per stay was significantly greater than NC, regardless of type of complication (p < .01,Tables 3 and 4).The length of stay was increased by a factor of 2.2 T A B L E 3 Hospital cost and length of stay by patients group.3).
With regards to the dispersion of cost per stay, 75% of the stays of the NC group cost lower than 75% of the RIC and C-RIC groups, as the third quartile (Q3) of NC stay cost (€ 11 406) was similar to the first quartile (Q1) of RIC and C-RIC groups (Figure 2).Furthermore, 50% (median) of stays with OCs costed more than €11 406.
The significant increase in the length of stay and the cost per stay in RIC (confirmed in C-RIC) and OC groups versus NC was observed regardless of the digestive location, except for length of stay in lower GI when comparing the OC group versus the NC group (p = 0.136).In every digestive site, there was at least a two-fold increase in length of stay and in cost between the NC and RIC groups (confirmed in C-RIC) (Table 4).

| DISCUSSION
This study provides updated numbers of incidence of complications, including infections of the surgery for digestive cancer and new numbers on their associated economic burden.The analysis was performed on the PMSI database which gathered all French hospitalizations occurring during the year 2018 providing a full picture of the rate of infection post digestive surgery which has been coded.
In this study the incidence appears to be stable compare with a previous study suggesting a plateau effect in reducing these complications: 15.8% of RIC observed in this study is similar to 15.9% observed by Minchella et al., in 2008. 11 The difference observed with the study carried out by Challine et al., in 2019 (46%) 13 can be explained by the fact that only infectious complications related to digestive oncological surgery were considered in our study, while all infectious complications were considered in the study by Challine et al.
Similarly, our results on the duration of hospitalization are of the same order of magnitude as those previously published.In the literature, an increase of length of stay related to SSI has been reported, varying from 10 to 16.8 days according to the type of surgery that was considered. 5,13,25Accordingly in the current study, the mean additional length of stay was 13.8 days for RIC (15 days for C-RIC).Furthermore, the increase in mean hospital cost per stay was € 8079 for RIC (and € 7887 for C-RIC), in line with previous studies reporting extra cost from € 6100 to € 18 300. 5,25is study highlights the high incremental cost of € 56 318 709, (mean cost for NC 18720-mean cost for RIC 10641)* number of RIC stays i.e. 6971) and confirms that this burden ranging from €8079 (costs of RIC vs. NC) to €3995 (cost of OC vs. NC) is high regardless of the location (Table 3).
A French study published in 2013 by Lamarsalle et al. 26 reported a threefold increase of length of stay due to HAIs after surgery, while a two-fold increase in this study was observed.However, the length of stay observed with HAIs was similar: 22 and 25.5 days.Lamarsalle et al. also estimated the additional annual cost to be € 57 892 715 related to infection which was similar to the additional cost observed for RIC in our study (€ 56 318 709).
The advantage of a retrospective analysis based on PMSI database is that it summarizes the information related to healthcare resource used for all public and private hospital stays, hence eliminating the sampling errors.No linkage with external datasets is required to provide the data, hence eliminating linkage difficulties.Coding errors are expected to be rare, as this database constitutes the prospective financial system based on hospital activity and financial resources are directly related to the coded information.However, as a common rule, the quality of the coding increases when the code related to a major feature of the stay (e.g.surgical procedures, digestive cancer in this study) or to a severity characteristic (e.g.infections) because it increases the funding of the stay, which is the case for most of the codes used to define RIC and C-RIC groups.Furthermore, to prevent miscoding and potential underestimation of SSI at hospital, stays with infection but not confirmed as "related infection due to surgery" were included in the stud.
F I G U R E 2 Boxplot of cost per hospital stay by patient group.
This study focused on SSIs only reported during the hospital stay with digestive cancer surgery.Consequently, the two SSIs reported after hospital discharge and readmission were excluded which is a limitation of this study.According to a systematic review investigating the proportion of SSI occurring after hospital discharge, 9.9% of surgeries leading to SSI and 60.1% of SSI appeared after discharge. 27A U.S. study 28 concluded that patients with post-operative infections where about five times more likely to be readmitted to hospital than those with no complication.Outpatients diagnosed with SSI have more frequent readmissions than unaffected patients (27.8% vs 6.8% respectively), and these readmissions are on average 2 days longer in duration than readmissions for other causes. 29,30Hence, proportions of SSI and associated additional cost have certainly been underestimated in this study.
18,19 The ERAS ® Society (Enhanced Recovery After Surgery) was created in 2001 to develop perioperative care and to improve recovery through research, education, audit, and implementation of evidence-based practice.,19 Monitoring systems and control program 20,21 have contributed to a decrease of the prevalence of healthcare-associated infections in France between 1996 and 2012.Based on the first French Prevalence Survey Study conducted in 1996 31 6.7% of hospitalized patients had at least one nosocomial infection against 5.1% in the 2012 study. 32The proportion of SSI among HAI has increased from 10.6% in 1996 to 13.5% in 2012 31,32 and now, it seems that a threshold has been reached.Despite the existing international recommendations and French guidelines, nosocomial infections and particularly SSIs remain a major health problem in France.The prevalence of infected patients has remained stable since 2012, with 4.98% of patients with at least one HAI per year; nevertheless, the proportion of SSIs has increased by 2.4 point in 5 years to reach 15.9% of HAIs in 2017. 2 In view of this progression and given the high clinical and economic burden of infectious complications in patients with digestive surgery, training programs for healthcare professionals and implementation of international recommendations in the current practice should be continued to reduce the SSI rates and their detrimental consequences.Given the significant differences between the non-complication and the complications patients groups, it is relevant to contemplate enhanced perioperative recovery programs and immunonutrition when they apply.Enhanced perioperative recovery programs and immunonutrition could alleviate clinical and economic burdens of SSIs associated with digestive cancer surgery and its consequences. 15,16,19Impact of prehabilitations programs have been proposed but with no current validate evidence at present. 33Additionally, institution of integrated practice units, which are practices that organize care delivery through multidisciplinary teams consisting of surgeons, nurses, case managers, technicians, dieticians, psychologists, and others can have a large impact on patients outcomes.Such teams can include infection control specialists to monitor patient status during and after the surgery. 34

| CONCLUSIONS
This study confirmed that the risk of related infectious complications after digestive cancer resection remains high in France (>15%) and showed that those complications are associated with significantly longer length of stay and higher cost per stay.Although important prevention plans have been implemented in the past years, there is still a need for care strategies to alleviate the burden on the patients and healthcare system.

1
Number of stays per patient group.T A B L E 2 Description of location of malignant neoplasm in study population.

Table 2 )
Malnutrition was present in 32.8% of stays and was more to intensive care during the stay compared with 54.7% in the NC group ( p < .01),and the death rate was at least twice higher in groups with complications (RIC and OC) compared with NC ( p < .01).
Hospital cost and length of stay by patients group and digestive site.
a p versus NC.T A B L E 4 a p versus NC.PIESSEN ET AL.