A Breakthrough Improvement Collaborative Significantly Reduces Hospital Stay After Elective Colectomy for Cancer Across a Healthcare System

Abstract

Objective: 

This multicenter study aimed to assess (1) the effect of an improvement collaborative on enhanced recovery after surgery (ERAS) protocol adherence after elective colectomy and (2) the association between adherence and patient outcomes.

Background: 

ERAS pathways provide a framework to standardize care processes and improve postoperative outcomes in patients after colon surgery. Despite growing evidence of its effectiveness, adherence to these guidelines remains a challenge.

Methods: 

This prospective, multicenter collaborative was initiated throughout 11 hospitals in Flanders, Belgium. A structured audit tool was used to study patient outcomes and adherence to 12 ERAS components, defined by the collaborative. Three retrospective audits (based on patient record analysis) were conducted in 2017, 2019, and 2021, respectively.

Results: 

Overall, 740 patients were included (45.4% female; mean±SD age, 71±12 years). The overall adherence increased from 42.8% in 2017 to 58.4% in 2019 and 69.2% in 2021. Compared with low adherence, length of stay was increasingly reduced by 1.3 days for medium [95% confidence interval (95% CI): −2.5; 0.0], 3.6 days for high (95% CI: −4.9; −2.2), and up to 4.4 days for very high adherence (95% CI: −6.1; −2.7). Corresponding odds ratios for postoperative complications were 0.62 (95% CI: 0.33; 1.17), 0.19 (95% CI: 0.09; 0.43), and 0.14 (95% CI: 0.05; 0.39), respectively. No increase in 30-day readmissions was observed.

Conclusions: 

A peer-constructed improvement collaborative effectively increases adherence to an ERAS protocol in individual hospitals. Across time, length of stay and postoperative complications decreased significantly, and a dose-response relationship was observed.

Colorectal cancer (CRC) is the most common type of cancer in Europe. Approximately 1 in 20 men and 1 in 35 women will develop CRC at some point in their life.¹ Oncological resection remains the mainstay of the curative treatment for colonic cancer.² A high variability in both postoperative morbidity and implementation of minimally invasive surgery (MIS) persists.³ Over the last decade, surgical care for patients with CRC has become more standardized using enhanced recovery after surgery (ERAS) protocols. ERAS protocols, enhanced recovery pathways (ERPs), and care pathways are structured care methodologies, that facilitate aadherence to these protocols protocols and stimulate multidisciplinary collaboration. This multimodal approach was introduced by Kehlet in 1997.⁴ In 2005, the ERAS Group provided a structured enhanced recovery approach based on literature review.⁵ The goal of these pathways is to reduce the patient’s surgical stress response, preserve physiological function, and facilitate postoperative recovery.^6,7 Despite increasing evidence in favor of the use of these standardized protocols, adherence to and implementation of ERAS protocols remains challenging in daily practice.^7–12 A previous study noted only a 44% adherence rate across 12 European hospitals.³ One approach to successfully implement a complex multidisciplinary quality program such as ERAS is the Breakthrough Improvement Collaborative (BIC) methodology.^13–16 In short, a BIC consists of 4 learning sessions and 3 action periods supported by a faculty of experts.¹³ The care process is optimized based on evidence, feedback, and knowledge sharing. Collaboratives have been implemented in different countries with diverse healthcare systems to achieve quality improvement across organizations.^14,17 The efficacy of improvement collaboratives has been demonstrated in cardiac surgery,^18,19 bariatric surgery,^20,21 trauma surgery,²² general and vascular surgery,²³ chronic heart failure,²⁴ and neonatal intensive care units.²⁵

The aims of the present study were: (1) to evaluate the effect of an improvement collaborative on adherence to ERAS components after elective colectomy; and (2) to assess the relationship between protocol adherence and patient outcomes: length of stay and 30-day postoperative readmissions, and 30-day postoperative complications.

METHODS

Study Population

The BIC was launched in 11 hospitals within the Flemish region of Belgium. Included hospitals were a mix of academic/teaching, nonteaching small and large volume hospitals. This provides a cross sectional view on how colorectal surgical care is provided in our healthcare system. The annual caseload of colorectal resection for cancer of these hospitals varied between 30 and 300.

Adult patients admitted for elective nonmetastatic colon cancer surgery (ICD-10-CM Diagnosis Code C18) were included. Exclusion criteria were (1) emergency surgery; (2) severe dementia (DSM IV) or severe concomitant disease that might affect short-term outcomes; (3) rectal cancer surgery.

BIC

The BIC methodology is a structured improvement model, and consists of 4 learning sessions and 3 action periods with accompanying measurement occasions at the end of each action period.¹³ During the learning sessions, the focus is providing an overview of evidence and guidelines, feedback on clinical performance (adherence to ERAS components), and sharing (inter)national expertise, knowledge and best practices. During the action periods, improvement strategies are implemented at a hospital level, every hospital could have a different focus for improvement and another strategy to achieve that goal. Finally, after each action period a retrospective assessment is performed (record analysis of 25 consecutive patients) to record the effect of the newly introduced care program. During the entire project there was a continued support of the research team: (1) hospital-specific on-site training activities and learning sessions; (2) peer instructions and support in group meetings; and (3) interaction by e-mail, motivational conversations, and phone/Skype meetings.¹³

Per participating center, the collaborative consisted of a surgeon, an involved anaesthesiologist, a nurse (specialist) and a quality assessor. There were 3 assessment periods, with a baseline in 2017, a second assessment in 2019, and a third in 2021. After each assessment, hospitals received peer-review and an anonymous feedback report with a benchmark for their adherence to ERAS components. A local study coordinator performed retrospective data collection. For the first assessment period, the last 25 consecutive patients treated in 2017 were included. For the second and third assessment periods, the research team randomly selected 25 patients per hospital who were admitted over the last 6 months. Participants were blinded for the assessment periods.

Protocol Development of 12 Components

The ERAS components were listed by a selected group of “experts,” based on previous research.³ For the present study, the focus was on components that were defined as clinically relevant (high level of evidence and/or high grade of recommendation), but that showed low adherence in clinical practice. Afterward, these components were discussed under peers of the collaborative to come to a final 12-component ERAS program. Some aspects of ERAS guidelines were not part of our 12-component program. However, perioperative fluid management certainly warrants attention as it is a major determinant in bowel and general recovery. A structured audit tool was used for the evaluation of patient characteristics and the 12 predefined ERAS components: postoperative nausea and vomiting screening and related medical therapy, avoidance of preoperative sedative medication, antibiotics prophylaxis, carbohydrate loading, avoidance of urinary catheterization, avoidance of abdominal drain, avoidance of nasogastric tube, avoidance of epidural anesthesia, MIS, early postoperative nutrition, avoidance of postoperative opioid use, and early postoperative mobilization (Table 1). The following 3 outcomes were evaluated: length of stay (LOS), 30-day postoperative complications, and 30-day unplanned readmission. A patient was considered to have a complication in case of one (or more) of the following adverse events: surgical site infection (superficial and deep), postoperative bleeding, more than 4 days of ileus and the occurrence of an anastomotic leakage.³⁰

TABLE 1 - Predefined ERAS Components

ERAS Component	Recommendation	Quality of Evidence
PONV Screening	Multimodal postoperative nausea and vomiting (PONV) prophylaxis*	High
Sedative medication	Avoid routine sedative medication*	Moderate
Antibiotic prophylaxis	Intravenous antibiotic prophylaxis should be given within 60 min before incision as a single-dose administration*	High
Carbohydrate loading	In elective colorectal surgery in patients without delayed gastric emptying; 6-h fasting for solids and 2-h for clear fluids including carbohydrate drinks*,† ‡	High
Urinary catheterization	Early removal recommended. Patients at low risk should have routine removal of catheter on the first day after surgery*,§	High
Drain administration	Drainage of the peritoneal cavity and pelvis should not be performed routinely*,§	High
Nasogastric intubation	Postoperative nasogastric tubes should not be used routinely; if inserted during surgery, they should be removed before reversal of anesthesia*	High
Epidural anesthesia	Spinal anesthesia with low-dose opioids gives good analgesic effects. Early removal is recommended in order not to compromise recovery*,§	Moderate
Minimally invasive surgery	A minimally invasive approach (laparoscopic/robotic) is recommended*,§	High
Postoperative nutrition	Most patients can and should be offered food and oral nutritional supplements from the day of surgery*,§	Moderate
Postoperative opioid use	Avoid opioids and apply multimodal analgesia*	Moderate
Postoperative mobilization	Early mobilization*,§	Moderate

^*Gustafsson et al.²⁶

^†Kielhorn et al²⁷

^‡Gianotti et al.²⁸

^§Carmichael et al.²⁹

This study was approved by the ethics committee of the University Hospitals Leuven (S60907) as the coordinating center. All participating centers provided consent for data sharing and analysis.

Statistical Analysis

ERAS adherence at the patient-level was arbitrarily categorized as low (<6 components adhered to), medium (6–7 components), high (8–9 components), and very high (10–12 components). Patient characteristics and adherence to ERAS components are presented as the mean±SD. Differences in patient characteristics and adherence to ERAS components between periods were assessed using the Student t test for continuous variables and the Pearson χ² test for categorical variables. Statistical models investigating the association between outcomes (LOS, 30-day complications and 30-day unplanned readmissions) and period as well as between outcomes and adherence to ERAS components (in one and the same model) were adjusted for patient age, sex, ASA classification, and type of surgery (open versus laparoscopic/robotic). Models for LOS and readmissions were additionally adjusted for complications. Linear mixed models were used for continuous outcomes (LOS) and generalized estimation equations for binary outcomes (complications and 30-day readmissions). Statistical analyses were performed using SAS version 9.4, SAS Institute Inc., Cary, North Carolina.

RESULTS

Descriptives

A total of 740 patients were included, with 219, 252, and 269 in the first, second, and third period, respectively. Patient characteristics according to the period are summarized in Table 2. The mean age (SD) was 71 (12) years, and 45.4% of patients were female. The proportion of patients who underwent MIS increased from 73.5% in the first period to 89.6% in the third period, and adherence to ≥8 ERAS components increased from 16.0% to 68.8%. Mean LOS (SD) decreased significantly from 9.4 (7.4) days in the first period to 6.3 (5.2) days in the third period (P<0.001). Complications and unplanned readmissions did not differ significantly between periods.

TABLE 2 - Descriptives

	Mean (SD) or N (%)
Characteristic	Period 1 (N=219)	Period 2 (N=252)	Period 3 (N=269)	Overall P *	P Period 2 vs. 1*	P Period 3 vs. 2*	P Period 3 vs. 1*
Age (y)	71.7 (11.2)	70.6 (12.4)	70.2 (12.8)	0.371	0.323	0.679	0.163
Sex
Male	123 (56.2)	137 (54.4)	144 (53.5)	0.841	0.695	0.849	0.561
Female	96 (43.8)	115 (45.6)	125 (46.5)
BMI†	26.0 (4.4)	26.5 (4.5)	26.2 (4.2)	0.539	0.277	0.526	0.609
ASA
I	11 (5.0)	21 (8.3)	18 (6.7)	<0.001	<0.001	0.019	0.001
II	68 (31.1)	123 (48.8)	104 (38.7)
III	45 (20.6)	64 (25.4)	76 (28.3)
IV	11 (5.0)	6 (2.4)	3 (1.1)
Missing	84 (38.4)	38 (15.1)	68 (25.3)
Type of surgery
Open	37 (16.9)	29 (11.5)	21 (7.8)	0.001	0.094	0.048	<0.001
Laparoscopic or robot	161 (73.5)	208 (82.5)	241 (89.6)
Conversion to laparotomy	20 (9.1)	15 (6.0)	7 (2.6)
Missing	1 (0.5)
Overall ERAS performance (number of guidelines adhered to)
Low (<6)	122 (55.7)	72 (28.6)	27 (10.0)	<0.001	<0.001	<0.001	<0.001
Medium (6–7)	62 (28.3)	62 (24.6)	57 (21.2)
High (8–9)	24 (11.0)	76 (30.2)	99 (36.8)
Very high (10–12)	11 (5.0)	42 (16.7)	86 (32.0)
Length of stay	9.4 (7.4)	8.6 (7.6)	6.3 (5.2)	<0.001	0.249	<0.001	<0.001
Complications	21 (9.6)	29 (11.5)	23 (8.6)	0.520	0.500	0.260	0.690
Clavien-Dindo grade I–II	14 (6.4)	24 (9.5)	16 (5.9)
Clavien-Dindo grade III–IV	7 (3.2)	5 (2.0)	7 (2.7)
Readmissions	16 (7.3)	12 (4.8)	9 (3.4)	0.133	0.244	0.412	0.048

Bold represents statistical significance (after Bonferroni correction for pairwise period comparisons).

^*Analysis of variance for continuous and the Pearson χ² test for categorical variables.

^†Because of missing data, results presented are based on N=211, N=244, and N=251 patients in period 1, 2 and 3, respectively.

Improvement in Adherence to Predefined ERAS Components

The variation in adherence to ERAS components between hospitals is shown in Figure 1. Overall, the lowest adherence was observed for carbohydrate loading, opioid-free analgesia, and early mobilization after surgery, the highest for avoidance of sedative medication, antibiotic prophylaxis, and use of MIS. A significant increase in adherence to each of the 12 ERAS components from the first to the third period was observed (P<0.001) (Table 3). The largest improvement in adherence was observed for avoidance or early removal of the epidural anesthesia, avoidance or early removal of urinary catheters and early postoperative mobilization (with an improvement of 46%, 39%, and 37%, respectively). The use of abdominal drains and nasogastric tubes reduced from 66% to 41%, and from 47% to 17%, respectively.

FIGURE 1: Heat map presenting the hospital-specific adherence (%) to each of the 12 ERAS components, with red indicating low adherence and green high adherence. This shows the improvement in adherence to the agreed ERAS components from the first measurement period to the third one. Both components and hospitals are ordered according to overall adherence during the first period.

TABLE 3 - Description of and Adherence to Each of the 12 ERAS Components

	N (%)*
	Period 1 (N=219)		Period 2 (N=252)		Period 3 (N=269)		Overall P †	P Period 2 vs. 1†	P Period 3 vs. 2†	P Period 3 vs. 1†
PONV screening	91	41.6	138	54.8	203	75.5	<0.001	0.004	<0.001	<0.001
No sedative medication	157	71.7	223	88.5	234	87.0	<0.001	<0.001	0.601	<0.001
Antibiotics prophylaxis	202	92.2	219	86.9	267	99.3	<0.001	0.061	<0.001	<0.001
Carbohydrate loading	34	15.5	55	21.8	105	39.0	<0.001	0.082	<0.001	<0.001
No epidural (laparoscopic/robotic) or removed the day after surgery (open/conversion)	77	35.2	171	67.9	219	81.4	<0.001	<0.001	<0.001	<0.001
Minimally invasive surgery (laparoscopic/robotic)	161	73.5	208	82.5	241	89.6	<0.001	0.018	0.020	<0.001
No drain	75	34.3	142	56.4	160	59.5	<0.001	<0.001	0.470	<0.001
No nasogastric (NG) tube or NG removed on day of surgery	116	53.0	186	73.8	224	83.3	<0.001	<0.001	0.008	<0.001
No opioids postoperatively	41	18.7	51	20.2	87	32.3	<0.001	0.679	0.002	<0.001
No urinary catheter or catheter removed the day after surgery	56	25.6	138	54.8	175	65.1	<0.001	<0.001	0.017	<0.001
Fluid nutrition on day of surgery or day after when surgery after 12 am	94	42.9	143	56.8	194	72.1	<0.001	0.003	<0.001	<0.001
Walking 5 m day after surgery	22	10.1	93	36.9	126	46.8	<0.001	<0.001	0.022	<0.001

Bold represents statistical significance (after Bonferroni correction for pairwise period comparisons).

^*Number of patients (%) for which the guideline was adhered to.

^†Pearson χ² test.

Dose Relation Between Adherence to ERAS Components and Patient Outcomes

Across the periods, LOS and postoperative complications were inversely related to ERAS protocol adherence (Table 4). Compared with the patient group with low protocol adherence (<6 components), LOS was 1.3 days [95% confidence interval (95% CI): −2.5; 0.0], 3.6 days (95% CI: −4.9; −2.2), and 4.4 days (95% CI: −6.1; −2.7) shorter in the medium (6–7 ERAS components), high (8–9 ERAS components), and very high (10–12 ERAS components) adherence group, respectively. There was no increase in the 30-day readmissions, even in the group with the shortest LOS. Estimated odds ratios for postoperative complications were 0.62 (95% CI: 0.33–1.17), 0.19 (95% CI: 0.09–0.43), and 0.14 (95% CI: 0.05–0.39) for medium, high, and very high protocol adherence.

TABLE 4 - Relation of Period and ERAS Performance With Patient Outcomes

	LOS			30-Day Complications			30-Day Readmissions
	Mean±SD	Estimate* (95% CI)	P	N (%)	OR* (95% CI)	P	N (%)	OR* (95% CI)	P
Period
First	9.4±7.4	Reference	0.0302	21 (9.6)	Reference	0.121	16 (7.3)	Reference	0.3748
Second	8.6±7.6	0.8 (−0.4; 2.0)		36 (14.3)	1.76 (0.89; 3.45)		12 (4.8)	0.63 (0.25; 1.60)
Third	6.3±5.2	−0.6 (−1.8; 0.7)		28 (10.4)	2.14 (1.01; 4.53)		9 (3.4)	0.46 (0.15; 1.41)
ERAS performance (number of guidelines adhered to)
Low (<6)	11.6±8.2	Reference	<.0001	34 (15.4)	Reference	<.0001	17 (7.7)	Reference	0.4616
Medium (6–7)	9.0±7.3	−1.3 (−2.5; 0.0)		30 (16.6)	0.62 (0.33; 1.17)		7 (3.9)	0.51 (0.17; 1.53)
High (8–9)	5.6±4.6	−3.6 (−4.9; −2.2)		14 (7.0)	0.19 (0.09; 0.43)		9 (4.5)	1.27 (0.42; 3.86)
Very high (10–12)	4.5±1.9	−4.4 (−6.1; −2.7)		7 (5.0)	0.14 (0.05; 0.39)		4 (2.9)	0.95 (0.23; 4.00)

^*Estimates (with 95% confidence intervals) represent the difference in number of days for LOS and OR for complications and 30-day readmission and are adjusted for age, sex, ASA classification, and type of surgery. Estimates for LOS and 30-day readmissions are additionally adjusted for complications.

DISCUSSION

This multicenter study, over 5 years, showed that higher adherence to ERAS components significantly improved the postoperative outcomes. A progressive and significant increase in adherence to ERAS components was observed over the different assessment periods and across the different institutions.

At baseline, an important variation in adherence to ERAS components was observed. This finding is rather unexpected as participating hospitals are located in a distinct geographic area with the same financial incentives and participating physicians share the same training background. At the time of the start of the study only 2 hospitals had an ERAS protocol running, which could be explained by the absence of (financial) incentives to reduce postoperative hospital stay within our healthcare system.

This further emphasizes the role of a peer organized BIC to implement a complex pathway as ERAS. Despite CRC surgery not being centralized, improvements were achieved across all institutions, with low-volume centers improving equally.

A major finding of this study was the significant increase in overall and component-specific adherence over the different assessment periods. The components: avoidance of epidural anesthesia, urinary catheters, and early postoperative mobilization showed the largest improvements. The use of abdominal drains dropped over time, but are still placed in 40% of the cases, despite evidence.²⁶ Furthermore, there was a limited decrease in opioid use (81%–68%), indicating the difficulty of creating an opioid-free environment. non steroïdal anti inflammatory drugs (NSAIDs are a key opioid-sparing component in multimodal analgesia.²⁶ Current evidence shows that the use of NSAIDs in the early postoperative setting after surgical resection for CRC is safe.³¹ However, at the time of development of our ERAS protocol there was ongoing debate on the safety of the use of NSAIDs in the postoperative setting (risk for bleeding, increased risk for anastomotic leakage). This explains why only 28% of patients received NSAIDs postoperatively, and the persistent need for the use of opioids. This finding emphasizes the importance of refining a multimodal postoperative pain protocol according to newest insights.

Overall, there was a gradual increase in adherence to ERAS components from 42.8% in 2017 to 58.4% in 2019 and 69.2% in 2021. Participating hospitals differed in size and patient caseload. Despite those differences, implementation of and increased adherence to ERAS components was achieved at the different sites.

The effect of an ERAS protocol on postoperative hospital stay is well established. MIS is a key component of ERAS. However, the increase in MIS cannot explain as a sole reason the obtained reduction in LOS. Despite the progressive and near total implementation of MIS in all Belgian hospitals, the “acceptable” LOS, what reflects the mean LOS within our system remained high (10.6 days in 2017 and 8.8 days in 2021, according to federal data).

Our data corroborate the data on a dose relation between protocol adherence and postoperative outcomes. Increased adherence to ERAS components was associated with decreased incidence of postoperative complications and a shorter LOS. These findings are consistent with previous studies showing that increased adherence is associated with better short-term clinical outcomes and shorter length of stay.^32–34 A recent systematic review showed that reduced compliance to ERAS components increases LOS, morbidity and readmission. This underscores the importance of an ERAS bundle rather than the implementation of a limited number of components (“cherry picking”).

One of the strengths of this study was peer-review of a predefined protocol in a multicenter setting. Moreover, the adherence to a bundle of 12 predefined ERAS components that are relevant to the care process of colonic cancer surgery was assessed over different time periods. The results show that adherence to these predefined components was not uniform. Components with the highest adherence rates were antibiotic prophylaxis, MIS, and avoidance of preoperative sedative medication. In contrast, the ERAS components with the lowest adherence across the three measurement periods were early postoperative mobilization, carbohydrate loading, and opioid avoidance. This is in line with previous research showing that adherence to ERAS guidelines remains challenging and results in persistent high variation in delivered care across hospitals.³

This study has several important limitations. One could criticize the retrospective nature of the assessments. However, this reflects better the actual practice of an institution. There was no specific focus possible to the protocol as caregivers were not aware of the “sampling” periods, and data recording was a requirement to participate. The BIC intends to continue its monitoring and reflects on the use of prospective data recording of all consecutive patients in the future.

In this study, no control group was assessed. However, the “acceptable” LOS for elective colectomy reflects current practice in Belgium, and this could substitute a control arm.

Third, there was no formal assessment of the reasons for nonadherence. All hospitals, except one, improved significantly, but reasons for the lack of change for some aspects could not be retrieved. In addition, there was a higher adherence to some ERAS components, but no additional analyses were performed to elucidate this observation.

The peer collaborative will review the ERP components and adjust where needed according to new insights (eg, the role of oral antibiotics in the prevention of surgical site infection; the use of NSAIDs in the multimodal postoperative pain control) and further allow other institutions to participate.

The group further explores the possibility to prospective data recording of all consecutive cases to assess difference between high-volume and low-volume hospitals.

Currently, a registration tool has been developed to continuously follow-up care and to evaluate the possible retention of improvement initiatives. Therefore, a timely critical evaluation of ERAS components is necessary, to keep pace with actual insights.

CONCLUSIONS

This BIC shows that timely feedback and peer knowledge sharing results in an improved adherence to ERAS components. This improvement can be obtained across a healthcare system. There is a dose-response effect between adherence to ERAS components and length of stay.