Introduction

Enhanced recovery after surgery (ERAS), often-labeled fast-track surgery, is a multidisciplinary, multimodal, and evidence-based approach to perioperative management [1]. It represents a paradigm shift from traditional surgical care delivery; one that relies heavily on continuous audits and the dissemination of clinical pathways [1]. Initially reported in 1997 by Kehlet et al [2], ERAS guidelines have since been successfully applied to many major surgical procedures with promising results [3,4,5,6,7,8,9,10]. In particular, reduced post-operative morbidity and shortened hospital length of stay (LOS) have been consistently reported [11, 12].

In 2017, there were approximately 106,000 breast reconstructions performed in the United States; this represents a 35% increase from 2000 [13]. In a health care climate increasingly focused on optimizing care quality, patient experience, and containing costs, it comes as no surprise that health systems leverage ERAS pathways as a quality improvement initiative in breast reconstruction. Several studies have examined the safety and benefits of ERAS pathways in patients undergoing breast reconstruction, particularly free flaps [14,15,16]. The ERAS society recently performed an audit of care processes associated with high-quality outcomes in breast reconstruction and put forward recommendations for optimal perioperative management [17].

The aim of this review and meta-analysis is to appraise the existing literature for ERAS in breast reconstruction, specifically (a) the current state of ERAS implementation (i.e., to what extent are the core elements of published guidelines adopted in clinical practice), and (b) safety and efficacy of ERAS on reducing length of stay (LOS) and opioid use.

Methods

Data sources and search strategy

This study was submitted to the PROSPERO registry a priori (CRD42018085433), and performed in accordance with the Preferred Reporting Items for Systematic reviews and Meta-analyses (PRISMA) guidelines. A comprehensive electronic literature search of EMBASE, Cochrane library, and PubMed databases was performed on January 22, 2018 (see Supplemental data file 1 for search terms). All articles from 1975 to present were included without language or further study design restrictions.

Inclusion and exclusion criteria

Criteria included full-length reports of studies involving ERAS implementation among females at least 18 years old, undergoing breast reconstruction. Reported outcomes included at least one of length of stay (LOS), opioid use, or major complications. Studies involving aesthetic or mastectomy alone procedures, or male patients were excluded. Editorials, opinion letters, review articles, purely technical descriptions, and meeting abstracts were excluded. To reduce heterogeneity in statistical analysis, results abstracted from “transition groups” wherein there was partial or incomplete implementation of ERAS pathway elements were excluded [15, 18].

Study selection and data extraction

Two independent reviewers (CG and SB) performed the initial search and abstract screen in duplicated. Discrepancies were settled by reviewing the full article. All included abstracts were then subject to a full article review in duplicate. Discrepancies at this stage were settled by a third reviewer (ACO).

Data collection for the articles was performed using a standardized data extraction form tested by four independent reviewers (ACO, CG, SB, and CJC). Variables included lead author, publication year, study design, patient characteristics, breast reconstruction modality, ERAS protocol details, and post-operative complications. Variables collected for each ERAS protocol included preoperative consult, preoperative diet, preoperative medications, preoperative opioids, intraoperative antibiotics, intraoperative medications, post-operative medications, and goals of care.

Outcome measures

In line with prior studies, our primary outcomes are length of stay (standardized to days) and opioid use reported as milligrams (mg) of oral morphine equivalents (OME) while in hospital. OME values, an objective measure of post-operative pain, are calculated using a web-based conversion calculator (available at: http://clincalc.com/opioids/default.aspx). Secondary outcomes of interest include patient-reported pain scores using a numerical rating scale (0–10) and complication rates. Complications include overall short-term (i.e., 30 day) major complications, and specifically readmission, hematoma, and infection. Missing or inconsistent outcomes data were addressed by contacting the respective study authors (first and senior/corresponding) via email for clarification, which occurred once [15]. The authors were successfully contacted and data outcomes were confirmed directly [15].

Data synthesis

The Downs and Black Checklist was used to scrutinize study quality and risk of bias [19]. Quality reporting of included studies is derived from the total Downs and Black score: excellent (\(\ge\) 26), good (20–25), fair (15–19), and poor (\(\le\) 14) [20, 21].

Statistical analyses were performed using Review Manager 5.3 (The Nordic Cochrane Centre, The Cochrane Collaboration, Copenhagen, Denmark 2011). Where applicable, prosthetic and autologous breast reconstruction cases are reported separately. Mean differences are reported with 95% confidence intervals. When necessary, interquartile ranges are used to estimate standard deviations according to methods published in the Cochrane Handbook [22]. The minimal clinically important difference (MCID) for length of stay is considered one day. The MCID for total opioid use is considered a 20% reduction. Results are also interpreted with Cohen’s effect size, where a small effect size is > 0.2, medium > 0.5, and large > 0.8 [23]. Inter-study heterogeneity is quantified using the I2 measurement as follows: I2 < 50% low, 50–75% medium, and > 75% high [24]. For both primary outcomes, a random effects model (REM) is used to address the heterogeneity (methodological and clinical) across and within the included studies.

Results

Search results

Our initial search strategy resulted in 260 citations. Following title and abstract screening, 13 articles underwent full-text review and nine articles (3.46%) satisfied inclusion criteria [14,15,16, 18, 25,26,27,28,29] (Fig. 1). Among these are reports of five retrospective cohort studies, three case series, and one prospective cohort study. All studies were performed in a single-institution, tertiary-care setting. Four were conducted in Canada, three in the United States, and two in Denmark. Methods of breast reconstruction are implants in one study [18], pedicled flaps (latissimus dorsi and pedicled transverse abdominis muscle) in two studies [27, 28] and free flaps (deep inferior epigastric perforator and muscle sparing transverse rectus abdominis muscle) in five studies [14,15,16, 25, 26] (Table 1). Two studies include patients who received either pedicled or free flap reconstruction [14, 29]. A total of 1191 patients are included in our potential analytic sample, 566 in the ERAS group, and 625 in the control group. Study characteristics are outlined in Table 1. The predominant reconstruction timing is delayed rather than immediate, and most cases are bilateral.

Fig. 1
figure 1

Flow diagram of search and study selection

Full size image
Table 1 Characteristics of included studies
Full size table

Risk of bias and quality score

There are no randomized controlled trials included in our analysis. Most studies are characterized as having “fair” methodological quality according to our instrument. This is reflected in a median score of 16 points (range 14–18). Weaknesses in study design included insufficient blinding of subjects (9/9 studies), non-contemporaneous recruiting of subjects (7/9 studies), retrospective bias (8/9 studies), and lack of pre-hoc power analysis (9/9 studies).

Current state of ERAS in clinical practice

Although a number of interventions have been formally proposed by the ERAS Society, variable implementation is observed [17]. Using categories consistent with society guidelines [17], the common themes are preoperative consultation, multimodal analgesia, nausea and vomiting prophylaxis, venous thrombosis prophylaxis, intraoperative analgesia, and targeted discharge planning (Table 2). The latter varies according to the type of reconstruction. Less commonly used ERAS elements include early feedings, perforator mapping in free flaps, minimization of hypothermia, and a priori wound management strategy. Detailed descriptions of the ERAS pathway elements implemented across studies are provided in supplemental data file 2.

Table 2 Components of implemented ERAS protocols
Full size table

Impact on opioid consumption

Opioid use while in hospital is reported in five studies, and outcomes for the respective 623 patients are pooled [14, 15, 26, 28, 29] (Table 3). ERAS pathways significantly reduce total opioid use, mean difference (MD) − 183.96, 95% CI − 340.27 to − 27.64, p = 0.02 (Fig. 2). This is a medium effect size, given standardized mean difference (SMD) − 0.68, 95% CI − 1.11 to − 0.25. However, there is considerable heterogeneity between studies (I2 = 95%), mitigated via subgroup analysis according to the dominant reconstructive procedure (pedicled versus free flaps). This sub-grouping is justified because pain outcomes for these procedures are likely to be different. Repeat analysis including predominantly free flap reconstruction demonstrates a greater opioid reduction with ERAS, MD -248.13, 95% CI − 387.95 to − 108.32, p < 0.001, I2 = 69% (Fig. 2b). This subgroup still represents a medium effect size, and it is clinically significant at the lower limit of the confidence interval, a reduction of over 100 morphine equivalent doses. The Davidge et al. study included only pedicled reconstructions, demonstrating a significant reduction in opioid use among ERAS patient [28]. The lower limit of the confidence interval is 15.94; this is likely not clinically significant. There is a significant difference between subgroups, p = 0.002.

Table 3 Post-operative pain
Full size table
Fig. 2
figure 2

a Total opioid use for ERAS versus traditional recovery pathway. b Subgroup analysis of total opioid use for ERAS versus traditional recovery pathway

Full size image

Impact on length of stay

Length of stay is reported in eight studies, and outcomes for 1151 patients are pooled [14,15,16, 18, 25, 26, 28, 29] (Table 4). Results are consistent between studies. ERAS protocol significantly reduces length of stay, MD -1.58, 95% CI − 1.99 to − 1.18, p < 0.00001, I2 = 65% (Fig. 3). This is a large effect, SMD − 0.95, 95% CI − 1.29 to − 0.62. It is also clinically significant, because the lower limit of the confidence interval represents a length of stay one day shorter versus standard protocol. Moderate heterogeneity is again present and explored with subgroup analysis of patient comorbidities influencing length of stay. The 2016 Bonde study [25] was removed from this analysis for the following reasons: (a) absence of a control group and (b) patients were already included in a preceding 2015 article [16] (“double counting”). Repeat analysis demonstrated a greater reduction in LOS, MD − 1.78, 95% CI − 2.07 to − 1.48, p < 0.001, I2 = 6% (Fig. 3b). This represents approximately a day and half at the upper end of the confidence interval. There is a significant difference between subgroups, p = 0.006.

Table 4 Length of stay
Full size table
Fig. 3
figure 3

a Length of stay for ERAS versus traditional recovery pathway. b Subgroup analysis of length of stay for ERAS versus traditional recovery pathway

Full size image

Post-operative pain scores

Studies cannot be pooled given methodological heterogeneity in the timing of pain assessment, and high subjectivity inherent with numerical rating scales (Table 3). Afonso et al. report lower maximum pain scores on POD1-2 in the traditional recovery cohort compared to ERAS recipients [29]. Maximum pain scores after POD2 are not significantly different between the two groups. On the other hand, Batdorf et al. report lower pain scores in the ERAS group at 24 h post-operatively compared to the traditional recovery group [14]. Astanehe et al. [15] also demonstrate consistently reduced mean pain scores with ERAS in patients who underwent autologous microvascular reconstruction compared to the traditional recovery groups (Table 3). Armstrong et al. show low mean pain scores, over time, after implementing ERAS pathway; however, control group scores are not available for comparison [27]. Multiple regression analysis identifies preoperative pain, type of surgery, and current use of anxiolytic and/or antidepressant medications as independent predictors of post-operative pain.

Patient safety—complications

Most studies evaluated post-operative complications in order to assess the safety of enhanced recovery pathways in patient undergoing breast reconstruction (Table 5). No important differences are identified between ERAS and traditional recovery groups. Analysis includes overall major complications, readmission, hematoma, and infection. Major complications are reported at 30–45 days post-operatively for 1086 patients from six studies [15, 16, 18, 26, 29]. There is no difference in overall major complications between patients managed with ERAS versus traditional pathways, OR 0.86, 95% CI 0.55–1.34, p = 0.51, I2 = 31% (Fig. 4). Unplanned readmissions are reported in four studies [14, 15, 26, 29] entailing 532 patients; there is no difference between ERAS and traditional pathways, OR 1.17, 95% CI 0.52–2.62, p = 0.71, I2 = 0% (Fig. 5). Post-operative hematomas are reported in six studies [14,15,16, 18, 26, 29], entailing 1086 patients; there is no difference between ERAS and traditional pathways, OR 1.12, 95% CI 0.67–1.86, p = 0.68, I2 = 0% (Fig. 6). Post-operative infections are reported by six studies [14,15,16, 18, 26, 29], with 1086 respective patients and no difference between ERAS and traditional pathways, OR 1.27, 95% CI 0.67–2.42, p = 0.46, I2 = 14% (Fig. 7).

Table 5 Post-operative complications
Full size table
Fig. 4
figure 4

Overall major complications for ERAS versus traditional recovery pathway

Full size image
Fig. 5
figure 5

Readmissions for ERAS versus traditional recovery pathway

Full size image
Fig. 6
figure 6

Hematomas for ERAS versus traditional recovery pathway

Full size image
Fig. 7
figure 7

Infections for ERAS versus traditional recovery pathway

Full size image

In the Batdorf et al. study, 30 days post-operative complication rates are similar for the ERAS and traditional recovery groups, except for a higher rate of skin flap cellulitis in the ERAS group (20% vs 6%; p = 0.03) [14]. Additionally, there are three partial and two complete flap losses reported in the ERAS group compared to 0 and 1 in the traditional pathways group, respectively; this is not significantly different. Finally, Kaoutzanis et al. demonstrate no difference in major and minor complications among enhanced and traditional recovery groups, with the exception of delayed wound healing in the ERAS group (36% vs. 16%, p = 0.02) [26].

Discussion

In this meta-analysis, we demonstrate statistically significant reductions in post-operative LOS and opioid consumption following ERAS implementation in recipients of breast reconstruction. Of equal importance, patient safety is not shown to be compromised, as there are no meaningful differences in complications. This added value of the ERAS program is consistent with published meta-analysis findings in other surgical specialties such as liver surgery [30], colorectal surgery [11], pancreatic surgery [31], orthopedic surgery [32], and bariatric surgery [33]. Our results fill a gap in the extant literature because of the dearth of published systematic reviews analyzing the utility of ERAS in breast reconstruction [34].

ERAS is a systematic, protocol-driven approach to the management of surgical patients. Core tenets of the ERAS pathway have been outlined previously and are positioned along the entire surgical care continuum: pre-, intra-, and post-operative phases [1]. Examples include but are not limited to the following: preoperative counseling, goal-directed fluid resuscitation, opioid-sparing regimens for pain control and early ambulation [1, 17]. The bulk of the available literature investigating the safety and efficacy of ERAS pathways in breast reconstruction is centered on free flap recipients. The resource-intensive nature, longer inpatient time horizon, and higher cost structure, relative to implant-based reconstruction, make this population uniquely suited for the continuous audit and protocol-driven care processes inherent to ERAS.

We believe that our study findings will be of great interest to the many stakeholders involved with the delivery of surgical care: patient groups, payers, hospital management, and policy makers. Length of stay reduction, a reliable proxy for hospital resource utilization and costs [35, 36], is a performance metric that is germane to health care delivery today. An isolated single-center analysis involving 50 ERAS cases and 50 traditional pathway cases projected the savings from ERAS directed-care to be $279,258 per annum for free flap breast reconstruction [26]. An estimated $4400 savings per ERAS patient was determined due to a reduction in LOS by 1.7 days resulting in a total reduction of 108 inpatient days. With health-related expenditures rising uncontrollably in the US (17.8% of the 2016 gross domestic product) [37], the main themes of current discourse are how to reimburse for the quality and efficiency of services delivered (“value-based care”) in lieu of volume and intensity of services (“fee-for-service”).

Our reduced LOS among ERAS recipients is commensurate with the lower OME consumption observed in this group. Intuitively, this makes sense because inadequate post-operative pain management is an established driver of prolonged hospital stay following breast reconstruction [38]. Furthermore, the ongoing US prescription opioid crisis adds a powerful public health dimension to the potential value of ERAS implementation. In 2015, there were over 33,000 deaths attributable to an opioid overdose [39]. Accompanying these alarming data is robust evidence showing a positive correlation between average daily OME and both opioid-related mortality and dependence [40]. The use of multimodal perioperative analgesia, characteristic of ERAS, might be a viable strategy for lowering systemic opioid use post-operatively and better physician stewardship of opioid prescriptions [39].

This potential for ERAS to optimize perioperative resource allocation while generating savings and maintaining quality, makes a compelling case for it to become the standard of care in free flap reconstruction. However, despite mounting evidence of the benefits of ERAS, it is clear that gaps in education and execution still persist [1]. Inconsistent and incomplete implementation of all elements of an archetypal breast reconstruction ERAS program [17] was observed across the nine included studies (Table 2). Study-specific pathways were instead observed [34]. This might reflect a need for greater institutional support for these programs in order to ensure their sustainability and optimal performance.

Regardless of the level of hospital administration or local leadership “buy-in,” updating perioperative practice is a notoriously arduous process [41]. However, there is evidence to show that adopting a structured ERAS implementation plan in step with provider education might generate improved and sustained front-line compliance with all components of the ERAS pathway [1]. Additionally, level 1 evidence by way of randomized controlled clinical trials is also lacking and might present a barrier to widespread, complete dissemination. This is because their absence may call to question the veracity of the evidence base with which we are updating current clinical practice. A systematic examination of the economic benefits of ERAS pathways, i.e., healthcare-related cost savings, would also be a worthwhile area of future study. Lastly, the medium- and long-term outcomes of ERAS-facilitated surgical recovery are still largely unknown: cancer-specific 5 year mortality, patient satisfaction, aesthetic outcomes, and long-term complications.

Our review has several limitations that should be acknowledged. Across the included studies, we encountered considerable variations in surgical technique, unbalanced patient comorbidities between analytic groups, varied reconstruction techniques, retrospective study designs, and small sample sizes. Furthermore, despite our attempts to identify as many qualified studies as possible, only nine studies fulfilled inclusion criteria. There are numerous sources of clinical heterogeneity identified between studies: ERAS protocol elements, type of reconstruction (pedicled versus free tissue), and patient comorbidities. While heterogeneity was incorporated into random effects model throughout analyses, these factors likely explain the wide confidence intervals observed.

Conclusion

The implementation of ERAS programs is associated with improvements in the outcomes of post-mastectomy reconstruction. Subsequent implications for clinical practice are manifold. The shortened recovery times, with no increased risk of readmissions, will positively impact cost efficiency and institutional productivity given the fact that breast reconstruction, especially that involving autologous tissue, is a resource-intensive enterprise [42]. Reduced LOS will also improve patient access to breast reconstruction [42]. Finally, the demonstrated superiority of ERAS protocols in reducing post-operative opioid consumption points to a sustainable path for improved surgeon stewardship of prescription narcotics.

Our results can also inform future research efforts, as there are little data outlining the operational challenges associated with ERAS implementation. Future studies are still needed to elucidate the long-term clinical outcomes and economic benefits associated with ERAS implementation in this patient population.