Enhanced recovery programs (ERPs) incorporate multiple evidence-based interventions aiming to reduce the metabolic stress occurring during surgery, but also to organize care for patients undergoing a particular procedure by limiting unwanted variability in care processes [1]. Guidelines from the ERAS® society recommend the implementation of more than 20 perioperative care components for patients undergoing colorectal surgery [2]. However, compared to traditional care, ERPs have been shown to consistently improve postoperative outcomes regardless of the number, the type, the combination, or the level of evidence of the elements used [3]. Furthermore, each intervention adds complexity to the pathway and may require additional resources.

Monitoring care process to evaluate the quality of perioperative care may be especially relevant for procedures in which serious postoperative events such as reoperation or mortality are rare, as in case for colectomy [4]. In the context of enhanced recovery, the impact of overall adherence to care processes, namely ERP elements, and the relative contribution of each intervention to patient recovery are still questioned. To date, there is preliminary evidence that adherence to care processes is associated with improved patient recovery [5–9]. However, existing reports mainly dealt with early phases of protocol implementation in which overall adherence was low [7], or failed to take into account important confounding factors such as postoperative morbidity [9]. The McGill University Health Centre first implemented an ERP for colorectal patients in 2006 [10] and is currently running a well-established care pathway incorporating a large number of perioperative interventions, which reduces hospital stay and societal costs compared to usual care [11].

The objective of this study was to estimate, in patients enrolled in an established ERP, to what extent the degree of adherence with ERP interventions impacts on recovery after surgery, and to identify which elements are associated with postoperative outcomes.

Methods

This study was designed and reported following the STROBE guidelines for the conducting and reporting of observational cohort studies [12].

Study design

This is a retrospective review of a prospectively collected database including patients undergoing elective bowel surgery treated within an ERP in a university-affiliated tertiary teaching institution. Study approval was granted by the institutional review board (14-170-SDR). Patients who underwent scheduled bowel resection between September 2012 and December 2014 were identified from the institutional operating room database and cross-checked with a prospective database specifically designed to capture adherence to ERP care processes and postoperative outcomes. Patients not followed prospectively by a clinical auditor were excluded from the study.

Enhanced recovery program

In 2006, an ERP for selected patients undergoing laparoscopic colorectal resection was first introduced in our institution [10]. In August 2010, this ERP was modified to include a total of 23 perioperative care components (Table 1) and was extended to all patients undergoing elective bowel resection [13].

Table 1 Perioperative care ERP interventions and definition of adherence
Full size table

Patients were admitted to hospital on the day of surgery. All cases were performed by one of three fellowship-trained colorectal surgeons (BS, ASL, PC). Discharge was planned for postoperative day (POD) 3 or earlier if patients achieved the following discharge criteria: tolerance of oral intake, recovery of gastrointestinal function (i.e., passage of flatus), adequate pain control with oral analgesia, ability to mobilize and self-care, and no evidence of complication or untreated medical problems. Final decision on discharge remained at the individual surgeon’s discretion.

Data collection

From September 2012 a dedicated, trained clinical auditor prospectively collected patient data up until 30 days after surgery. Data were entered into the ERAS Interactive Audit System (http://www.erassociety.org, ENCARE, Kista, Sweden), an international web-based registry that was specifically designed for interactive audit and research. Each patient data field contained approximately 140 different variables including preoperative patient characteristics, operative data, adherence to perioperative care processes, and postoperative outcomes. Every 3–4 months, the auditor reviewed adherence and outcome data with the ERP team and discussed any new trend or issue arising from the latest time period. A clinical researcher unaware of patient’s adherence to ERP components verified postoperative outcome data for each patient by reviewing medical charts and the electronic medical record. Additional information was retrieved from medical records to compute Charlson comorbidity index [14], CR-POSSUM score [15], Apfel postoperative nausea and vomit (PONV) risk score [16], and the comprehensive complication index (CCI) [17], which were not originally included in the ERAS audit system.

Adherence and outcome measures

Adherence to each ERP component was defined as the successful completion of a planned intervention (e.g., a patient planned to have regular food on POD 1 actually receives a meal and consumes it). Definitions for each measure of adherence can be found in Table 1. Adherence to intraoperative intravenous infusions followed recent recommendations [18], taking into consideration the patient ideal body weight, use of mechanical bowel preparation (MBP), surgical approach, duration of surgery and blood loss. Cutoffs for adherence were set at <3 ml/kg/h for laparoscopic surgery and <5 ml/kg/h for open surgery. Adherence to PONV prophylaxis considered patients Apfel score in accordance with consensus guidelines [19]. Overall adherence to the ERP was also calculated for each patient.

Primary outcome measure of the study was successful recovery, a composite end point defined as discharge by POD 4, no occurrence of complications or hospital readmission within 30 days of operation. Secondary outcomes of the study included length of primary hospital stay (LOS), 30-day postoperative morbidity, and the comprehensive complication index (CCI). LOS was defined as the number of nights spent in hospital during the primary stay from the day of admission to the day of discharge. Intraoperative and postoperative complications were defined a priori (Supplementary Material 1). The severity of each complication was graded according to the Dindo–Clavien classification [20], and a CCI was then generated for each patient. This is a validated measure summarizing the complete spectrum of complications occurred and their severity in a single score ranging from 0 to 100 [17].

Statistical analysis

We performed a complete case analysis including patients with data for all ERP adherence variables and outcome measures. Relevant characteristics of the included patients and those excluded from the study were compared using Chi-square test or Fisher’s exact test for categorical data, and Student’s t test or Mann–Whitney U test for continuous data, as appropriate.

Univariate analysis (i.e., logistic regression for binary outcomes and linear regression for continuous outcomes) was performed to assess the association of patient or procedure-related variables and ERP interventions with outcomes. To identify ERP elements independently associated with the outcomes of interest, we used binary logistic regression (for the outcomes successful recovery and occurrence of 30-day complications) and linear regression (for log-transformed LOS and log-transformed 30-day CCI). Variables significant at p < 0.10 were retained in the final model. Only early postoperative elements (up to POD1) were considered in the analysis because it is difficult to distinguish whether adherence to late care processes are indicators or predictors of poor recovery outcomes. In example, delayed transition to oral analgesia can be a consequence of adverse events leading to poor recovery, and not a cause. In the same way, patient mobilization on late postoperative days represents an outcome rather than indicate the adherence to a care process. In fact, both ability to mobilize and pain control with oral analgesia represent discharge criteria. To evaluate the impact of overall adherence to the pathway (i.e., the overall number of ERP elements to which a patient was adherent) on outcomes, we also ran multivariate analyses adjusted for confounding factors.

All multivariate models were adjusted for relevant prognostic factors known to affect patient adherence and postoperative outcomes such as age, gender, preoperative American Society of Anesthesiologists (ASA) score, pelvic surgery and inflammatory bowel disease [21].

According to previous research, between 5 and 10 events per independent variable are required to obtain reliable regression coefficients in multivariate logistic regression models [22], and a minimum of 15 subjects per variable are required for linear regression [23]. As ERP elements may be highly correlated with one another, the risk of multicollinearity was assessed by inspecting correlation matrices between independent variables and by computing the variance inflation factor (VIF). VIF values exceeding 10 indicate serious multicollinearity, and values greater than 4 may be a cause for concern [24]. The discriminative power of the logistic model equations was determined by running a receiver operating characteristic (ROC) curve, and by calculating the relative area under the curve (AUC). The goodness-of-fit of the predictive models was assessed through the Hosmer–Lemeshow test.

Because of the confounding effect of intra- and postoperative complications on adherence to ERP components, we also performed a sensitivity analysis excluding patients who experienced intraoperative or early postoperative complications (before POD 2).

Descriptive data are reported as mean (95 % confidence interval, CI), or median (interquartile range, IQR), otherwise specified. Statistical analysis was performed using STATA® version 13.1 software (StataCorp, College Station, TX, USA). All statistical tests were 2-sided, a “p” value < 0.05 was considered to indicate statistical significance.

Results

Four hundred and forty patients underwent scheduled bowel resection during the study period. Ninety-three (21 %) patients were excluded from the analysis because they were not followed prospectively through the ERAS interactive audit system. The remaining 347 patients were included in this study. Included and excluded patients had similar demographic and clinical features (Supplementary Material 2).

Table 2 shows baseline characteristics of patients included in the study. One hundred twelve (32 %) patients had a low preoperative physical status as measured by ASA score. Two hundred twenty-eight (66 %) patients were operated for malignancy, 125 of which (55 %) had localized colorectal cancer (TNM stage 0–II).

Table 2 Patient characteristics
Full size table

Table 3 includes patients’ operative characteristics. One hundred eighty-two (52 %) patients underwent segmental colonic resection, while 121 (35 %) patients underwent pelvic surgery including low anterior resection, abdominoperineal resection and total proctocolectomy. Two hundred seventy-eight (80 %) patients were approached with laparoscopy, 28 (8 %) required conversion to an open procedure. Intraoperative complications occurred in 25 (7 %) patients.

Table 3 Operative characteristics
Full size table

Table 4 shows adherence rates for single ERP elements. Overall, patients were adherent to median 18 (16–20) elements. Median adherence was similar for preoperative, intraoperative, and postoperative elements (75, 78, and 80 percent, respectively). The only elements with poor adherence (<50 %) were intraoperative balanced intravenous infusions and intake of oral nutritional supplementation on the day of surgery postoperatively.

Table 4 Patient compliance to enhanced recovery program elements
Full size table

Table 5 reports postoperative outcomes. One hundred fifty-six (45 %) patients had a successful recovery with early discharge, no readmission and no complications. Median length of primary hospital stay was 4 days (3–7); 135 (39 %) patients were discharged within the third day after surgery. One hundred seventy-five (50 %) patients experienced at least one complication within 30 days after surgery, and 56 (16 %) patients were treated for a complication after discharge. Hospital readmissions at 30-days occurred in 44 (13 %) patients. Most common reasons for readmission were intraperitoneal or pelvic abscess (n = 10, 23 %), bowel obstruction (n = 7, 16 %), anastomotic leak (n = 6, 14 %), and other gastrointestinal complications (n = 6, 14 %).

Table 5 Postoperative recovery outcomes
Full size table

The impact of overall adherence on outcomes

Figure 1 shows the significant association between overall adherence to ERP elements and recovery outcomes. There was a positive association between adherence and successful recovery (adjusted OR 1.39 (95 % CI 1.24–1.57) for every additional element, p < 0.001). An inverse relationship was found between adherence and LOS (11 % reduction, 95 % CI −14 to −8 %, p < 0.001), 30-day postoperative morbidity (OR 0.78, 95 % CI 0.70–0.87, p < 0.001), and the CCI (17 % reduction, 95 % CI −23 to −10 %, p < 0.001). There was no association between overall adherence to ERP and hospital readmission (adjusted OR 1.03 (95 % CI 0.89–1.18) for every additional element, p = 0.726).

Fig. 1
figure 1

Relationship between overall adherence to enhanced recovery pathway elements, successful recovery and 30-day complications

Full size image

Predictors of recovery outcomes

Table 6 reports univariate logistic regression for predictors of successful recovery including patient- and procedure-related confounding factors such as age, gender, ASA score, type of surgery, and intraoperative complications. At multivariate analysis (Table 7), laparoscopy (OR 4.32, 95 % CI 2.26–8.27, p < 0.001), early mobilization out of bed (OR 2.25, 95 % CI 1.13–4.47, p = 0.021), and early termination of IV fluid infusion (OR 2.00, 95 % CI 1.16–3.45, p = 0.013) were significantly associated with successful recovery. These factors were also key elements associated with reduction in LOS, 30-day postoperative morbidity and complication severity index as measured by the CCI. In addition, intraoperative balanced IV fluids were associated with a 21 % reduction in LOS, while epidural anesthesia prolonged LOS by 19 %.

Table 6 Univariate analysis for predictors of successful recovery defined as discharge home by postoperative day 4, no 30-day morbidity, and no hospital readmission
Full size table
Table 7 Multivariate regression models for independent predictors of successful recovery, length of primary hospital stay, 30-day morbidity, and 30-day comprehensive complication index
Full size table

Sensitivity analysis

Excluding patients who experienced intraoperative or early complications (n = 71), the positive association between overall adherence to the pathway and successful recovery remained significant (OR 1.37 (95 % CI 1.20–1.57) for every additional element, p < 0.001). At multivariate analysis, laparoscopy (OR 4.61, 95 % CI 2.28–9.33, p < 0.001) and early mobilization (OR 2.22, 95 % CI 1.06–4.66, p = 0.035) were the only factors significantly associated with successful recovery.

Discussion

This study relying on prospectively collected data in a tertiary Canadian university hospital showed that increased adherence to ERP interventions is independently associated with successful recovery, reduced LOS, and complications following bowel surgery. In this large series of patients treated between 2012 and 2014, where mean adherence to the pathway was high, laparoscopic surgery, early mobilization out of bed, and perioperative fluid management resulted as key elements correlating with improved postoperative outcomes.

When compared to traditional perioperative care, ERPs in colorectal surgery have been shown to improve postoperative outcomes in terms of accelerating recovery, reducing LOS, medical complications [25], and societal costs [11]. Positive results for ERPs seem to be achieved regardless of the number, type, or the combination of interventions implemented in different series [3]. This may suggest that it is not the effect of single ERP elements or the number of elements, but simply the coordination of care into an organized pathway that reduces variability and is sufficient to improve outcomes. In our study, we assessed the relationship between adherence to ERP elements and postoperative results and identified key processes facilitating recovery in order to consider whether additional resources should be allocated to increase adherence to those elements (e.g., dedicated personnel for patient mobilization).

In previous series, adherence was usually lower for postoperative elements [7, 8, 26]. In fact, compliance with postoperative interventions may be influenced by patient symptoms and by the occurrence of complications. In the current series, adherence was high and similar throughout the different perioperative phases reflecting a well-organized and established clinical pathway. In addition, only early postoperative interventions (i.e., within the first 24 h) were considered in the analysis to limit the potential influence of adverse events on adherence. Low adherence was found for intraoperative fluid infusions and intake of oral nutritional supplements on the day of surgery postoperatively. The former may be related to the lack of a specifically dedicated anesthesia team for colorectal surgery but also to the adoption of a strict definition of adherence following recently published guidelines [27]. The latter represents an organizational problem (e.g., product availability, patient arriving to the surgical ward late in the day). Both elements represent targets for improvement.

Gustafsson and colleagues [7] previously demonstrated that adherence to more than 70 % of the planned care processes was associated with lower morbidity and shorter LOS compared to lower adherence. However, they used arbitrary adherence thresholds and did not adjust for the confounding effect of postoperative complications. In our series, patients with higher adherence to the pathway had a greater chance of a successful recovery. In fact, we found a strong association between adherence and outcomes even when patients experiencing intraoperative or early morbidity where excluded from the analysis. For every additional element to which patients were compliant, LOS and 30-day morbidity decreased, confirming results from a large UK study [9] that found a significant inverse relationship between the mean adherence to the ERP and LOS in multiple surgical subspecialties. In our analysis, we chose a composite end point as a proxy for successful postoperative recovery, taking into account not only primary hospital stay and morbidity but also hospital readmission, which is a key post-discharge outcome associated with delayed patient recovery and increased healthcare costs [28]. The use of LOS as measure of postoperative recovery is debated as it may be influenced by non-clinical factors such as surgeon’s preference and hospital tradition [29]. An alternative and valid measure of short-term postoperative recovery could have been the time to readiness for discharge (i.e., the time to achieve standardized discharge criteria) [30], but this variable was not part of our database and could not be reliably collected retrospectively.

It is still unclear whether there is an ideal combination of perioperative interventions, or whether there are single items with a greater impact on recovery within an ERP. In a study including mostly open colonic surgery [5], early mobilization, urinary catheter removal, and early oral diet were found as independent predictors of early discharge. However, the lengthy study period including a phase of implementation and the lack of controlling for confounding factors and postoperative complications is a potential source of bias. In a recent multicenter study by the ERAS compliance group [6], where only preoperative and intraoperative items were considered, carbohydrate loading, laparoscopy and restrictive intravenous fluids were found as independent factors associated with reduced LOS and morbidity, while epidural analgesia delayed discharge. In our cohort, laparoscopic approach, early mobilization and discontinuation of intravenous fluids were key elements associated with successful recovery and reduced 30-day morbidity and complication severity. Epidural analgesia was the only ERP element delaying discharge, corroborating recent evidence from a randomized controlled trial [31]. For laparoscopic colorectal surgery within an ERP, the use of transversus abdominis plane blocks has been proposed as a new opioid-sparing strategy to prevent epidural catheter management issues that may delay discharge [32].

Research shows that laparoscopy in the context of ERPs is associated with reduced pain, ileus [33], and attenuated inflammatory response [34] compared to open surgery. Evidence supports that the combination of laparoscopy and enhanced recovery is associated with improved outcomes. However, ERPs significantly reduce the gap in postoperative results between open and minimally invasive surgery [35]. Early mobilization after surgery is considered a key component of ERPs. It is well known that staying in bed leads to deconditioning that can largely be prevented by physical activity [36]. However, there is little evidence that the implementation of specific interventions to increase mobilization improves outcomes [37]. In our analysis, simply mobilizing out of bed once on the first day after surgery was a significant predictor of early discharge and reduced morbidity. Future studies are warranted to clarify the impact of mobilization on postoperative recovery and verify whether there is a dose–effect relationship.

Adherence to postoperative elements of the pathway may be difficult to interpret as it is confounded by the patient’s recovery status. For example, a patient symptomatic for postoperative ileus is less likely to start early oral nutrition and feel like ambulating. Thus, in our regression models, we only considered early postoperative elements (up to 24 h after surgery) that are potentially less influenced by the development of complications and also performed a sensitivity analysis excluding patients with intraoperative or early postoperative morbidity as previously reported by Larson et al. [8]. In this analysis, laparoscopy and early mobilization remained significant factors associated with successful recovery.

Notably, post-discharge complications occurred in 16 % of our cohort resulting in a considerable number of patients returning to the emergency department (ED) (20 %) and eventually being readmitted (13 %). These results corroborate with recent enhanced recovery colorectal surgery series where hospital readmissions ranged between 9 and 13 % [35, 38, 39]. In a previous study, compliance to ERP greater than 93 % was found as a protective factor for readmission [38]. In our series, adherence to ERP had no effect on post-discharge outcomes. More than half of emergency visits resulted in hospital readmission, mostly because of late infectious and surgical complications requiring in hospital management. Patients who returned to the ED but were not readmitted represent a target for quality improvement, as they could have been managed in an outpatient setting or through the local community services or family physicians. Patient information regarding post-discharge care can certainly be improved. In addition, establishing a dedicated follow-up service may also be considered in order to anticipate any serious clinical issue or to prevent unnecessary ED visits.

Strengths and limitations

Although data were prospectively collected and patients were followed by an independent auditor not involved in clinical activities, this analysis carries intrinsic limitations of all observational studies. Sample size may be considered insufficient for the regression analysis we carried out because of the high number of predictors included. However, we limited our series to two consecutive years to avoid time-related bias since ERPs are constantly evolving through time, and new elements are added or removed according to the evidence. Around 20 % of patients who underwent colorectal resection in our institution in the target period were not included in this series as their data were not prospectively collected. These patients were randomly missed because of personnel unavailable to record data. While retrospective chart review can provide reliable postoperative outcome data, it does not provide patient-reported information regarding adherence. Thus, it was decided to exclude these patients from the study to ensure a high quality of the data. Notably, no difference in preoperative characteristics and postoperative outcomes was found between included and excluded patients.

Main strengths of this study were that it followed recent recommendations for reporting and used clear definitions of adherence and outcome measures [40]. In fact, most of the existing works on enhanced recovery fail to thoroughly report the definitions of adherence to each implemented interventions. Other strengths and unique features of this study include the use of sensitivity analysis, and the choice of a composite end point encompassing different clinically relevant recovery outcomes allowing us to better ascertain the role of adherence to ERP elements and its association with postoperative results. Furthermore, all multivariate models were adjusted for relevant factors influencing adherence and postoperative outcomes including patient comorbidities and procedural factors as previously described [21, 41].

The findings of this study confirm the positive impact of ERP interventions on recovery and suggest that auditing adherence to the pathway is a key metric to assess the effectiveness of an ERP. Our results should encourage the allocation of resources toward strategies likely to increase adherence. For example, adopting a structured pathway for all bowel resection patients with standardized sets of perioperative clinical orders minimizes the variability of patient care favoring adherence to care processes. In addition, reinforcing perioperative education and patient engagement in the pathway may prove effective in improving adherence to elements requiring a high degree of self-management such as mobilization and diet.

Conclusions

In this retrospective study relying on prospectively collected data within an established ERP, increased adherence to enhanced recovery interventions was associated with successful early recovery and a reduction in postoperative morbidity and complication severity. Laparoscopic approach, perioperative fluid management, and patient mobilization were key elements associated with improved outcomes. Our findings suggest that further measures should be implemented to increase adherence to ERP interventions.