Assessing complications following pancreatoduodenectomy: the Comprehensive Complication Index versus the Clavien–Dindo classification

Abstract

Background

This study aimed to compare the accuracy of the Comprehensive Complication Index (CCI) with that of the Clavien–Dindo classification in patients undergoing pancreatoduodenectomy.

Methods

A two-centre, retrospective study was undertaken that included patients who underwent pancreatoduodenectomy between 2008 and 2022. Three approaches were used to assess the two complication scores: the Spearman rank test, yielding the correlation coefficient (r), the area under the curve with 95% confidence intervals, and a mixed-effects model and a generalized mixed-effects model that yielded odds ratios and β-coefficients.

Results

A total of 596 patients were included. The CCI and Clavien–Dindo classification demonstrated no correlation with 90-day mortality (r = − 0.021, 0.618; and r = −0.003, P = 0.951) but a significant correlation with length of hospital stay (r = 0.620, P < 0.001; and r = 0.605, P < 0.001) and with 90-day readmission rate (r = 0.148, P < 0.001; and r = 0.120, P = 0.005). The accuracy of the CCI was superior to that of the Clavien–Dindo classification for length of hospital stay dichotomized at the 75th (P = 0.022) and 90th (P < 0.001) percentiles. The CCI significantly improved the effect of the Clavien–Dindo classification (random effect, P < 0.001) in the mixed-effects and generalized mixed-effects logistic regression analyses.

Conclusion

Compared with the Clavien–Dindo classification, the CCI appeared to be more accurate in terms of its association with a prolonged hospital stay and 90-day readmission rate. The CCI should complement the Clavien–Dindo classification in clinical and research settings.

This study compared the comprehensive complication index (CCI) with the Clavien–Dindo classification (CDC) in patients who underwent pancreatoduodenectomy at two large centres with extensive experience in pancreatic surgery. The findings of the study indicate that the CCI is a more robust predictor of a prolonged postoperative course (length of hospital stay at the 90th percentile) than the CDC. No significant difference was observed in the correlation between the CCI and either 30- or 90-day mortality, in comparison with the correlation between the CDC and 30- or 90-day mortality.

Introduction

Pancreatoduodenectomy (PD) is the primary treatment modality for patients with periampullary malignancies. Despite concerted efforts by pancreatic surgeons, the incidence of complications following PD is still reported to be approximately 50%^1,2. Postoperative complications have been defined differently over the past 20 years. The International Study Group of Pancreatic Surgery (ISGPS) has standardized the classification of specific complications related to the pancreas, such as postoperative pancreatic fistula (POPF)³, postpancreatectomy haemorrhage (PPH)⁴, and delayed gastric emptying (DGE)⁵. Similarly, bile leakage has been classified by the International Study Group of Liver Surgery (ISGLS)⁶.

In 2004, the Clavien–Dindo classification (CDC) was published⁷ to allow a standardized postoperative complication assessment. The CDC is a modified classification system that allows reliable assessment and consistent interpretation of complications⁸.

As the CDC is used primarily to classify only the most severe complication and all less severe complications are neglected, the Comprehensive Complication Index (CCI) was introduced in 2013⁹. The CCI is a composite measure that takes into account the burden of all complications, not just the most severe. The CCI is notable for incorporating a mathematical summation of complications classified according to the conventional CDC system. This results in a continuous score ranging from 0 (no complication) to 100 (death). There still needs to be more clarity regarding the clinical utility of the CCI in evaluating complications. Although some studies^10–13 have revealed a positive correlation between CCI scores and length of hospital stay (LOS) and clinical outcomes, others^14,15 have revealed a contrasting relationship. These inconsistencies highlight the need for further research to elucidate the actual clinical relevance of the CCI.

The aim of this study was to compare the accuracy of complication grades according to the CCI with that of the CDC in patients who have undergone PD.

Methods

Study design, patients, and inclusion criteria

This was a retrospective bicentre cohort study following the STROBE guidelines¹⁶ and the principles of the Declaration of Helsinki. Formal ethical review was approved by the respective ethical committees of the Canton of St Gallen and the Canton of Zurich (BASEC number 2024-00827).

The study included data gathered from patients who underwent all types of PD¹⁷ for both benign and malignant diseases at two Swiss tertiary referral centres specializing in pancreatic surgery between 2008 and 2022. The analysis is based on data from patients who were followed up for at least 90 days.

Outcomes

The primary outcome was the value of the CCI and the CDC in terms of association with postoperative LOS after PD. LOS was analysed as a continuous value or dichotomous variable (LOS over the 50th, 75th, and 90th percentiles). Secondary outcomes included the value of the CCI and CDC in terms of association with 90-day mortality and 90-day readmission rate.

Patient data

Data regarding co-morbidities were collected and analysed using the Charlson Co-morbidity Index and the American Society of Anesthesiologists (ASA) classification system.

Complication data

The severity of postoperative complications was initially classified prospectively according to the CDC⁷ and CCI⁹ by attending physicians in the medical ward. K.U. reviewed these data. Whenever missing data were identified, K.U. actively completed these data by reviewing the study databases and, when necessary, by contacting patients’ general practitioners by telephone. This thorough approach ensured that the data set used for analysis was ultimately complete, with no missing values in the key variables. The CDC system has five distinct levels of severity (grades I–V), with grade 0 indicating that the complication requires minimal to no treatment and grade V indicating death (Table S1). Major complications were classified as grade IIIa or higher according to the CDC system. The documentation of complications (and also the calculation of CCI score and CDC grade) refers to the time spent in hospital after surgery. For patients who were readmitted later owing to complications, the complications were also calculated for a period of 90 days after surgery, and consequently the CCI score and CDC grade were recalculated.

Pancreas-specific morbidities, such as POPF³, PPH⁴, and DGE⁵, were classified according to the ISGPS. Bile leakage was classified according to the ISGLS⁶. Surgical site infection was defined in accordance with the Centers for Disease Control and Prevention¹⁸ and characterized by sepsis (as defined by the International Guidelines for Management of Sepsis and Septic Shock¹⁹), a prolonged total intensive care unit and hospital stay, and a high 90-day mortality rate.

Calculation of CCI score

The CCI score was calculated on the basis of CDC grades for complications after surgical or interventional procedures, using the online CCI calculator (http://www.assessurgery.com)

(Table S2). The integrated complication data for each patient were entered into the calculator, and the resulting CCI values were subsequently calculated and summed for analysis. The final index is calculated on a scale of 0 to 100, where a score of 0 indicates that the complication requires minimal to no treatment and a score of 100 indicates death.

Follow-up

After discharge, the study nurses interviewed patients weekly by telephone to enquire about their postoperative status. Patients discharged with drains were consistently seen for follow-up during office hours. Each patient was followed up in the outpatient clinic 90 days after discharge.

Statistical analysis

The statistical analysis was performed using R version 4.4.2 (R Project for Statistical Computing, Vienna, Austria). P < 0.050 (2-sided) was considered to indicate statistical significance.

Categorical variables, summarized as counts and percentages, and were analysed using χ² statistics. Continuous data, expressed as median (interquartile range, i.q.r.) were analysed by Mann–Whitney U test. To ease interpretation of group differences, variance ratios were estimated for continuous data, and standardized mean differences for categorical and continuous data.

Univariable and multivariable logistic regression analyses were employed to compare differences between the CCI (score ≥ 36) and CDC (≥ grade III) in terms of confounding variables.

Three approaches were applied to further assess the CDC and CCI scoring systems: Spearman’s rank correlations and partial Spearman’s rank correlations were undertaken using the R ppcor library to examine correlations between the CDC and CCI regarding LOS, 90-day mortality, and 90-day readmission.

Receiver operating characteristic (ROC) curve analysis was conducted using the R library OptimalCutpoints to assess the diagnostic value of the CDC and CCI for predicting LOS, 90-day mortality, and 90-day readmission^20,21.

Linear mixed-effects models (LMMs) for continuous outcomes and generalized linear mixed effects models (GLMMs) for dichotomous outcomes were explored using the R lme4 library. LMMs were used to assess continuous outcomes and incorporate fixed and random effects to fit non-independent data (Supplementary material).

Results

Baseline characteristics

A total of 607 patients who had undergone PD were evaluated for eligibility. Eleven patients had to be excluded from the analysis owing to missing documentation resulting from transfer to other hospitals in the early postoperative period. For the remaining 596 patients finally included in the present analysis, at least 30 days of follow-up was available (in the absence of in-hospital death) (Fig. S1). Median age of the patients was 67.3 (i.q.r. 58.4–74.4) years, median BMI was 24.4 (21.7–27.3) kg/m², and there was a female predominance (352 women, 59.1%), The ASA grade was I in 21 patients (3.5%), II in 286 (48.0%), III in 275 (46.1%), and IV in 14 patients (2.3%). Median blood loss was 500 (i.q.r. 300–700) ml and median duration of surgery was 360 (293–460) minutes (Table S3).

Analysis of complications

A summary of postoperative complications classified according to the CDC is provided in Table 1. A total of 206 patients (34.6 (95% confidence interval (c.i.) 30.9 to 38.5)%) experienced major complications and 47 (7.9%) died within 90 days .

Table 1.

In-hospital adverse events in all patients

	Total (n = 596)	CDC grade ≥ III (n = 206)	P	CCI score ≥ 36 (n = 189)	P*
Postoperative pancreatic fistula			< 0.001*		< 0.001*
B	73 (12.2%)	40 (19.4%)		36 (19%)
C	49 (8.2%)	48 (23.3%)		48 (25.4%)
Delayed gastric emptying			< 0.001*		< 0.001*
A	94 (15.8%)	22 (10.7%)		27 (14.3%)
B	62 (10.4%)	35 (17%)		34 (18%)
C	52 (8.7%)	43 (20.9%)		37 (19.6%)
Postpancreatectomy haemorrhage			< 0.001*		< 0.001*
A	32 (5.4%)	20 (9.7%)		19 (10.1%)
B	20 (3.4%)	17 (8.3%)		14 (7.4%)
C	56 (9.4%)	56 (27.2%)		54 (28.6%)
Bile leak	39 (6.5%)	38 (18.4%)	< 0.001*	37 (19.6%)	< 0.001*
Leak of gastroenterostomy	18 (3%)	18 (8.7%)	< 0.001*	18 (9.5%)	< 0.001*
Surgical site infection
Superficial	136 (22.8%)	79 (38.3%)	< 0.001*	82 (43.4%)	< 0.001*
Deep incisional	34 (5.7%)	34 (16.5%)	< 0.001*	34 (18%)	< 0.001*
Deep organ/space	100 (16.8%)	90 (43.7%)	< 0.001*	74 (39.2%)	< 0.001*
Acute kidney injury	70 (11.7%)	63 (30.6%)	< 0.001*	64 (33.9%)	< 0.001*
Pulmonary embolism	17 (2.9%)	16 (7.8%)	< 0.001*	16 (8.5%)	< 0.001*
Pneumonia	90 (15.1%)	72 (35%)	< 0.001*	76 (40.2%)	< 0.001*
Cardiac complication	76 (12.8%)	61 (29.6%)	< 0.001*	71 (24.3%)	< 0.001*
Highest CDC grade			< 0.001†		< 0.001†
I	133 (22.3%)	0 (0%)		0 (0%)
II	246 (41.3%)	0 (0%)		23 (12.2%)
IIIa	63 (10.6%)	63 (30.6%)		28 (14.8%)
IIIb	41 (6.9%)	41 (19.9%)		36 (19%)
IVa	47 (7.9%)	47 (22.8%)		47 (24.9%)
IVb	17 (2.9%)	17 (8.3%)		17 (9%)
V	38 (6.4%)	38 (18.4%)		38 (20.1%)
Reoperation	114 (19.1%)	114 (55.3%)	< 0.001*	110 (58.2%)	< 0.001*
90-day mortality	47 (7.9%)	41 (19.9%)	< 0.001*	41 (21.7%)	< 0.001*
Length of ICU stay (hours), median (i.q.r.)	24 (19–48)	50 (24–143)	< 0.001	63 (24–168)	< 0.001
Length of hospital stay (days, median (i.q.r.)	17 (12–26)	28 (20–40.5)	< 0.001‡	28 (21–43)	< 0.001‡
90 day readmission	167 (28%)	82 (39.8%)	< 0.001*	83 (43.9%)	< 0.001*

Values are n (%) unless otherwise stated. CDC, Clavien–Dindo classification; CCI, Comprehensive Complication Index; ICU, intensive care unit; i.q.r., interquartile range. *χ2 test, except †χ2 test, Multiple Choice-simulated and ‡Mann–Whitney U test.

Eighteen patients (3%) did not experience any postoperative complication according to CDC criteria, 137 (23.7%) had a single complication, 167 (28.7%) had 2 complications, 115 patients (19.9%) had 3 complications, and there were 159 patients (26.7%) with ≥ 4 complications (Table S4).

The most prevalent complication was DGE (208 patients), followed by wound infection (136), and POPF (122). A total of 108 patients (18.1%) experienced PPH, and 114 (19.1%) required reoperation. The median postoperative LOS was 17 (i.q.r. 12–26) days for the entire patient cohort.

The median CCI score for all patients was 24.2 (i.q.r. 20.9–41.1). A median CCI score of 22.6 (8.7–24.2) was noted among patients with complications classified as grade II or lower according to the CDC; in contrast, a higher median CCI score of 50.5 (40.6–73.4) was observed among patients with major complications (P < 0.001). The distribution of CDC grades according to the CCI score is illustrated in Fig. 1.

Fig. 1.

Correlation between Clavien–Dindo grade and CCI score

The upper and lower adjacent lines represent the 90th and 25th percentiles; the line within the box the median value; and the upper and lower adjacent lines the upper and lower adjacent values. CCI, Comprehensive Complication Index. Circles are overlayed to display observed CCI values.

Patients experienced a mean(standard deviation) of 3(2) complications. Especially for higher CDC grades, the CCI score was higher than the value calculated by considering only the highest-graded complication. The CCI score increased in median value for CDC grade IIIa, reaching 9.5 (i.q.r. 2.7–15.6); for CDC grade IIIb, it increased to 11.2 (6.9–16.5); for CDC grade IVa, it increased to 14.7 (6.6–20.6); and for CDC grade IVb, it increased to 30.9 (23.9–40.7).

Risk analysis of complications in terms of CCI and CDC

A CCI score ≥ 36 was associated with blood loss (P = 0.015), complexity of PD (P = 0.049), main surgeon (P = 0.014), age at surgery (P = 0.001), and sex (P = 0.033) in univariable and multivariable logistic regression analyses. The univariable association with duration of surgery (P = 0.019), BMI (P = 0.003), and ASA grade (P = 0.030) was not confirmed in multivariable logistic regression.

CDC grade ≥ III was associated with the duration of surgery (P = 0.009), main surgeon (P = 0.018), UICC stage (P = 0.011), and BMI (P = 0.004) according to univariable and multivariable logistic regression. Multivariable logistic regression did not confirm the univariable association with estimated blood loss (P = 0.001) or PD complexity (P = 0.008). No associations were observed with age at surgery or sex (Table 2).

Table 2.

Logistic regression analysis for predictors of major complications in terms of CDC and CCI

	Highest grade of complications, CDC ≥ III				CCI score ≥ 36
	Univariable analysis*		Multivariable analysis†		Univariable analysis*		Multivariable analysis†
	Odds ratio	P	Odds ratio	P	Odds ratio	P	Odds ratio	P
Estimated blood loss per 100 ml
Continuous	1.07 (1.03, 1.12)	0.001	1.04 (1.00, 1.10)	0.068	1.07 (1.03, 1.12)	< 0.001	1.06 (1.01, 1.11)	0.015
Duration of surgery (hours)
Continuous	1.15 (1.05, 1.25)	0.002	1.19 (1.04, 1.35)	0.009	1.11 (1.02, 1.21)	0.019	1.10 (0.96, 1.25)	0.168
Extent of PD according to Mihaljevic et al.17		0.008		0.064		0.005		0.049
Type I	1.00 (reference)		1.00 (reference)		1.00 (reference)		1.00 (reference)
Type II	0.80 (0.40, 1.49)		0.77 (0.37, 1.53)		0.93 (0.47, 1.74)		0.94 (0.45, 1.88)
Type III	2.10 (1.29, 3.42)		1.86 (1.08, 3.21)		2.33 (1.43, 3.80)		2.09 (1.21, 3.63)
Type IV	2.52 (0.75, 8.88)		2.02 (0.56, 7.53)		2.05 (0.58, 6.90)		1.78 (0.48, 6.36)
Main surgeon		0.083		0.018		0.086		0.014
I	1.00 (reference)		1.00 (reference)		1.00 (reference)		1.00 (reference)
II	1.03 (0.63, 1.68)		0.69 (0.38, 1.24)		0.98 (0.59, 1.64)		0.80 (0.43, 1.47)
III	0.88 (0.52, 1.49)		0.97 (0.55, 1.69)		1.09 (0.64, 1.87)		1.15 (0.65, 2.03)
IV	0.81 (0.44, 1.46)		0.46 (0.21, 0.99)		1.07 (0.58, 1.94)		0.70 (0.32, 1.51)
Other	1.69 (1.03, 2.80)		1.46 (0.83, 2.57)		1.89 (1.14, 3.17)		1.95 (1.09, 3.50)
UICC stage		0.053		0.011		0.280		0.278
Benign	1.00 (reference)		1.00 (reference)		1.00 (reference)		1.00 (reference)
I	0.95 (0.53, 1.70)		0.82 (0.44, 1.52)		1.42 (0.78, 2.60)		1.18 (0.62, 2.25)
II	0.55 (0.34, 0.87)		0.43 (0.26, 0.72)		0.85 (0.52, 1.39)		0.67 (0.39, 1.15)
III	0.77 (0.44, 1.35)		0.74 (0.40, 1.36)		1.03 (0.58, 1.85)		0.86 (0.46, 1.63)
IV	0.97 (0.39, 2.38)		0.52 (0.19, 1.41)		1.54 (0.61, 3.80)		0.97 (0.34, 2.68)
Date of surgery
Continuous	0.97 (0.93, 1.02)	0.279	0.97 (0.92, 1.03)	0.356	1.01 (0.96, 1.06)	0.810	1.01 (0.95, 1.07)	0.785
Age at surgery (years)
Continuous	1.00 (0.99, 1.02)	0.524	1.01 (0.99, 1.03)	0.257	1.03 (1.01, 1.04)	0.001	1.03 (1.01, 1.05)	0.001
Sex		0.069		0.496		0.003		0.033
Male	1.00 (reference)		1.00 (reference)		1.00 (reference)		1.00 (reference)
Female	1.38 (0.98, 1.96)		1.14 (0.78, 1.68)		1.71 (1.20, 2.47)		1.53 (1.03, 2.28)
BMI (kg/m2)
Continuous	1.07 (1.03, 1.12)	< 0.001	1.06 (1.02, 1.11)	0.004	1.06 (1.02, 1.10)	0.003	1.04 (1.00, 1.08)	0.070
ASA fitness grade		0.220		0.565		0.030		0.477
I–II	1.00 (reference)		1.00 (reference)		1.00 (reference)		1.00 (reference)
III–IV	1.24 (0.88, 1.73)		1.12 (0.77, 1.63)		1.47 (1.04, 2.08)		1.15 (0.78, 1.69)

Odds ratios are shown with 95% confidence intervals (Wald type). A higher odds ratio indicates worse outcome. CDC, Clavien–Dindo classification; CCI, Comprehensive Complication Index; PD, pancreatoduodenectomy; UICC, Union for International Cancer Control; BMI, body mass index; ASA, American Society of Anesthesiologists. *Univariable logistic regression analysis; †multivariable logistic regression analysis. P values were calculated by likelihood ratio test.

Relationships between CCI and CDC with 90-day mortality

In analysis of 558 patients, after exclusion of 38 who died within 30 days (CDC grade V, CCI score 100), CDC and CCI were not significantly associated with 90-day mortality in the Spearman’s rank correlation (r = −0.003, 95% c.i. −0.101 to 0.098, P = 0.951; and r = 0.021, −0.113 to 0.072, P = 0.618), the partial Spearman’s rank correlation (r = 0.046, −0.035 to 0.124, P = 0.284; and r = −0.050, −0.113 to 0.026, P = 0.238), in the ROC curve analysis (area under the curve (AUC) 0.55, 95% c.i. 0.36 to 0.73; and AUC 0.51, 0.32 to 0.70), and in the GLMM (odds ratio (OR) 0.77, 95% c.i. 0.28 to 2.08, P = 0.607; and OR 0.77, 0.25 to 1.89, P = 0.597) (Table 3).

Table 3.

Generalized mixed-effects regression analysis of complications

	Predictor	n	Odds ratio	P
90-day postoperative mortality	CCI	558	0.77 (0.25, 1.89)	0.597
90-day postoperative mortality	CCI in CDC	558	0.77 (0.28, 2.08)	0.607
90-day readmission	CCI	558	1.60 (1.23, 2.09)	< 0.001
90-day readmission	CCI in CDC	558	1.65 (1.20, 2.27)	0.002

Odds ratios are shown with 95% confidence intervals (Wald type). The Comprehensive Complication Index (CCI) was incorporated as a fixed effect and assessed with and without the Clavien–Dindo classification (CDC) as a random effect. Odds ratios and P values (likelihood ratio test) are for the effect of the CCI.

Relationship between CCI and CDC with LOS

In analysis of 558 patients, after exclusion of 38 who died within 30 days (CDC grade V, CCI score 100), CDC and CCI were significantly correlated with LOS in the Spearman’s rank correlation (r = 0.605, 95% c.i. 0.543 to 0.660, P < 0.001; and r = 0.620, 0.560 to 0.673, P < 0.001) and the partial Spearman’s rank correlation (r = 0.100, 0.018 to 0.189, P = 0.018; and r = 0.199, 0.109 to 0.280, P < 0.001). For the ROC curve analysis, LOS was dichotomised at the 50th, 75th, and 90th percentiles with thresholds of 17, 26, and 37 days respectively. For the 75th and 90th percentiles, the diagnostic value of the CCI was significantly greater than that of the CDC (P = 0.022 and P < 0.001). For the 50th percentile of LOS, the diagnostic accuracies of the CCI and CDC were similar (P = 0.070) (Fig. 2). The LMM confirmed the significant association between CCI and a longer hospital stay (β = 13.74, 95% c.i. 11.09 to 16.40; P < 0.001) after adjustment for the CDC.

Fig. 2.

ROC curves examining diagnostic accuracy of CCI and CDC in association with LOS

Receiver operating characteristic (ROC) curve for a 50th percentiles of length of hospital stay (LOS) by Comprehensive Complication Index (CCI) score (area under curve (AUC) 0.79, 95% confidence interval 0.76 to 0.83), b 50th percentiles by Clavien–Dindo classification (CDC) grade (AUC 0.78, 0.74 to 0.82), c 75th percentiles by CCI score (AUC 0.85, 0.80 to 0.89), d 75th percentiles by CDC grade (AUC 0.82, 0.77 to 0.87), e 90th percentiles by CCI score (AUC 0.95, 0.90 to 0.99), f 90th percentiles by CDC grade (AUC 0.90, 0.85 to 0.96). Patients with a CDC grade of V and/or those who died within 30 days were excluded. TPR, true-positive rate; FPR, false-positive rate.

Relationship between CCI and CDC with 90-day readmission

In analysis of 558 patients, after exclusion of 38 who died within 30 days (CDC grade V, CCI score 100), CDC and CCI were significantly correlated with the 90-day readmission rate in the Spearman’s rank correlation (r = 0.120, 0.042 to 0.204, P = 0.005; and r = 0.148, 0.069 to 0.231, P < 0.001). In the partial Spearman's rank correlation analysis, CCI significantly correlated with 90-day readmission rate (r = 0.099, 0.017 to 0.187, P = 0.020) but not CDC (r = −0.046, −0.132 to 0.040, P = 0.274). ROC curve analysis revealed a low diagnostic accuracy for both the CCI and the CDC, with similar AUCs (P = 0.804) (Fig. 2). In the GLMM, CCI was significantly associated with 90-day readmission without and with the inclusion of CDC as a random effect (OR 1.60, 95% c.i. 1.23 to 2.09; and OR 1.65, 1.20 to 2.27) (Table 3).

Discussion

In the present study, the CCI was more strongly associated with a prolonged postoperative course than the CDC. Both the CCI and the CDC demonstrated low accuracy in predicting 90-day readmission. Addition of the CCI as a fixed effect resulted in more pronounced outcomes concerning 90-day readmissions, and it displayed considerable dispersion among patients with complications of CDC grade III or higher, whereas risk factors linked to the CCI demonstrated a higher degree of homogeneity.

Several parameters exist with which to validate the impact of different classifications of complications: LOS, medical costs, quality of life, 30- and 90-day mortality, and 90-day readmission²². In most studies, researchers have used LOS and medical costs^10,12,22,23. Medical costs were deliberately excluded from the comparison here because they are dependent on numerous variables. The financial outlay of using medical services in Switzerland is contingent upon many variables^24,25. LOS was selected as the primary variable for comparison between the two classifications of complications. These results indicated that the CDC grade III classification may be an oversimplification, resulting in an underestimation of complications in the CDC system. The present results are consistent with those of two previous studies^10,11 that compared the CCI with the CDC after pancreatectomy. Both studies demonstrated that the CCI and CDC were significantly associated with LOS; the cohorts included not only pancreatic resection but also other types of surgery.

Severity of the operation is a factor that could influence the outcome quality of life. The more complex and complicated the operation is, the less the influence of complications on quality of life^26,27. A study²⁸ of patients who had undergone pylorus-preserving PD, in particular, revealed that severe complications did not impair patients’ quality of life.

This is the first study to show that the CCI and CDC have low accuracy for association with 90-day readmission. One potential explanation is that postoperative circumstances alone were not the sole determinant of readmission, but also the condition at discharge. In a recently published study²⁹, this research group reported that the C-reactive protein level at discharge is a prognostic factor for determining whether a patient will require readmission to hospital.

A broader range of CCI scores was observed in patients with severe complications (CDC grades III–IV). This could be attributed to severe complications that frequently occur along with additional minor complications. Most CDC grades overlap between the upper and lower CCI values, with the upper CCI values correlating with the subsequent CDC grade. It could be inferred that a CDC grade IIIb complication (reoperation) may be associated with a more severe postoperative course than a CDC grade IVa complication in patients with multiple complications who have undergone reoperation. This suggests that the complication severity is not solely reflected in the treatment of the complication. The same outcomes have been documented in other studies^11,13.

The present study has certain limitations. Owing to its retrospective design, the inclusion of biased patients cannot be ruled out. Although the calculation of CCI scores is more complex and time-consuming than that of CDC grades, the CCI has been demonstrated to serve as a crucial instrument in assessment of the burden of complications in patients who have undergone PD.

The more consistent correlation between the CCI and extended LOS, and the correlation with mortality and 90-day readmission compared with the CDC, support the claim that the CCI should represent a more accurate method for assessing the severity of postoperative complications than the CDC. The postoperative outcome of patients with, for example, a complication with a maximum CDC grade of IIIb and only one complication, that is a CCI score of 33.7, may not be the same as that of a patient whose highest complication severity is also CDC grade IIIb, but who has, for example, a further CDC grade IIIa complication and two CDC grade II complications, which corresponds to a CCI score of 52. Patients with raised CCI values may be less likely to receive timely adjuvant therapy, potentially leading to poorer outcomes³⁰. Whether the CCI score should serve as an independent prognostic factor for survival in patients with periampullary cancer remains to be determined in future studies. Evidence from other oncological research has already underscored the utility of the CCI^31–34. One study³⁵ identified CCI score as an independent predictor of poor prognosis in patients with gastric cancer. In an era characterized by persistent escalations in healthcare expenditure, the representation of all billing-relevant complications should assume paramount significance. A recommendation for clinical practice can be derived from these findings that the CCI should be used in addition to the CDC for the purpose of reporting postoperative complications.

Funding

This study did not receive any specific grant from any funding agency in the public, commercial, or not-for-profit sectors. The Cancer League of Eastern Switzerland (KLOCH) funded part of this research, namely publication and editing costs. T.S. holds a corresponding grant.

Authors’ contributions

Kristjan Ukegjini (CRediT contribution not specified), José Oberholzer (Writing—review & editing), Philip Müller (CRediT contribution not specified), Rene Warschkow (CRediT contribution not specified), Ignazio Tarantino (CRediT contribution not specified), Jan P. Jonas (CRediT contribution not specified), Marie Klein (Writing—review & editing), Henrik Petrowsky (CRediT contribution not specified), Bruno M Schmied (CRediT contribution not specified), and Thomas Steffen (CRediT contribution not specified)

Disclosure

The authors declare no conflict of interest.

Supplementary material

Supplementary material is available at BJS Open online.

Data availability

The authors confirm that the data supporting the findings of this study are available within the article and its supplementary materials.