Clinical impact of developing better practices at the institutional level on surgical outcomes after distal pancreatectomy in 1515 patients: Domestic audit of the Japanese Society of Pancreatic Surgery

Abstract Background and Aim Institutional standardization in the perioperative management of distal pancreatectomy (DP) has not been evaluated in a multicenter setting. The aim of the present study was to assess the influence of institutional standardization on the development of postoperative complications after DP. Methods Data were collected from 1515 patients who underwent DP in 2006, 2010, and 2014 at 53 institutions in Japan. A standardized institution (SI) was defined as one that implemented ≥6 of 11 quality initiatives according to departmental policy. There were 541 patients in the SI group and 974 in the non‐SI group. Clinical parameters were compared between groups. Risk factors for morbidity and mortality were assessed by logistic regression analysis with a mixed‐effects model. Results Proportion of patients who underwent DP in SI increased from 16.5% in 2006 to 46.4% in 2014. The SI group experienced an improved process of care and a lower frequency of severe complications vs the non‐SI group (grade III/IV Clavien‐Dindo; 22% vs 29%, respectively, clinically relevant postoperative pancreatic fistula; 22% vs 31%, respectively, P < .05 for both). Duration of in‐hospital stay in the SI group was significantly shorter than that in the non‐SI group (16 [5‐183] vs 20 postoperative days [5‐204], respectively; P = .002). Multivariate analysis with a mixed‐effects model showed that soft pancreas, late drain removal, excess blood loss and long surgical time were risk factors for post‐DP complications (P < .05). Pancreatic texture, drain management and surgical factors, but not standardization of care, were associated with a lower incidence of post‐DP complications.


| PATIENTS AND METHODS
The questionnaire audits consisted of two parts. The first determined institutional characteristics, and the second was the perioperative data of 1515 patients who underwent DP in 2006, 2010 and 2014 at a total of 53 institutions in the Japanese Society of Pancreatic Surgery. The audit for PD was done simultaneously and is already published. 9 The first part of the questionnaire audit consisted of clinical questions concerning hospital volume, surgeon volume, and the 11 quality initiatives defined for the current study according to departmental policy at an institutional level, as shown in Table 1. Implementation of the quality initiatives was ranked according to levels of decision-making authority from A to C (A, full dependence on departmental policy; B, surgeon's decision in part; C, surgeon's decision). Based on this ranking, a standardized institution was defined as one in which ≥6 of 11 quality initiatives were ranked as "A" in each year (2006, 2010 and 2014). Quality initiatives in perioperative management were determined in accordance with the presence or lack of institutional criteria for perioperative management. Hospital volume was defined as low (0-24 PD per year), intermediate  PD per year), and high (50 or more PD per year). 9 Surgeon volume (number of PD/year per surgeon) was defined as low (0-11 PD in a year) and high (12 or more PD in a year). 9 The second part of the questionnaire audit comprised data col-

| Statistical analysis
The database was investigated by biostatisticians at Statcom Co. Ltd (Tokyo, Japan), as already reported. 9 The first questionnaire audit was common, as the data were previously reported. 9 Continuous variables were expressed as median and range. Nominal data were

| Trends of DP in 2006, 2010, and 2014
Proportion of patients who underwent DP at a SI increased from 17% in 2006 to 37% in 2010 to 46% in 2014. 9 As shown in Table 2

| Standardized group vs non-standardized group
Distal pancreatectomy was carried out for 541 patients in the SI group and for 974 patients in the non-SI group. As shown in Table 3, the SI group contained a higher proportion of high-surgeon volume centers relative to the non-SI group (38% vs 26%, respectively; P < .001). In terms of drain management, a higher rate of closed suction drainage use was found in the SI group relative to the non-SI group (71% vs 52%, respectively; P < .001). Moreover, the median time to drain removal in the SI group (POD-5) was shorter than that in the non-SI group (POD-7, P < .001). In comparisons of postoperative complications, a lower incidence of overall complications (54% vs 64%), grade III/IV/V Clavien-Dindo classification (22% vs 29%), CR-POPF (22% vs 31%), and SSI (incisional, 2.6% vs 5.0%; organ/space, 17% vs 23%) was found in the SI group relative to the non-SI group, respectively (P < .05 for all). Median duration of hospital stay in the SI group was also shorter than that in the non-SI group (POD-16 vs POD-20, P = .002). Several authors have reported that high-volume and specialized centers achieve better surgical outcomes after pancreatectomy. [13][14][15] However, Riall et al suggested that there is still significant variability in the outcomes of pancreatic resection. 16 Lucas and Pawlik have

|
proposed that quality improvement efforts should focus not only on who is operating or where the operation occurs (surgeon or hospital volume), but also on how the process occurs. 17 Which measures beyond morbidity and mortality may better reflect quality in DP? These measures include traditional clinical   outcomes, as well as processes of care and structural elements of care.
Among them, the "process of care" can be under the control of surgeons and the medical staff. Vollmer et al proposed that improved process management can mitigate the impact of preoperative risk and effectively deliver quality advances, despite traditional outcomes that may already meet or exceed benchmark outcomes for a given major surgical procedure. 18 Implementation of a clinical pathway as a tool for introducing a well-established process of care has been reported to be associated favorably with short-term outcomes after DP, including length of hospital stay in single institutional studies. [19][20][21][22] Recently, we reported that the standardized adoption of a wellorganized process of care for PD at the institutional level, but not hospital/surgeon volumes, was associated with a reduction in post-PD complications in a multicenter setting. 9 PD is a complicated surgery which consists of multi-organ resection with at least three anastomoses, and it is associated with high morbidity and mortality. Therefore, standardization of the surgical technique and perioperative management is greatly required, and can be a critical indicator for assessing the clinical outcomes of PD. In contrast, the surgical procedure of DP can be categorized as technically simple relative to PD.
Perioperative management of patients who undergo DP is also simple in terms of the absence of pancreatico-enteric anastomosis. In this study, standardization of the perioperative care process at the institutional level did not affect the occurrence of post-DP complications including CR-POPF. The international multi-institutional distal pancre- institution and to use the definition of SI consistently, other important indicators or methods for assessing "standardization" might exist. Second, we assessed the fact that a conscious attempt was made to improve the process of care in each institution, but we could not evaluate that the actual processes were applied more frequently or more regularly. Thus, perioperative management strategies varied across institutions. However, this provides a realistic picture, reflecting inherent variability in the clinical practice of DP.
Third, institutions participating in this study are specialized centers for pancreatectomy (or include at least one surgeon certified by the Japanese Society of Hepatobiliary Pancreatic Surgery) and, therefore, the findings may not be generalizable to all hospitals.

| CONCLUSIONS
Standardized adoption of a well-organized process of care for DP at the institutional level did not reduce post-DP complications. Traditional factors such as pancreatic texture, drain management and surgical factors were associated with a lower incidence of post-DP complications. Sustainable efforts will be required to reduce post-DP complications.

ACKNOWLEDGEMENTS
We would like to express our sincere appreciation to the surgeons and institutions that participated in this study, and to Drs S Yamaki, S Hirooka, and H Yanagimoto for their significant contribution to this study. Names of the institutions are listed in Table S1.