Enhanced Recovery After Surgery for Noncolorectal Surgery?

A Systematic Review and Meta-analysis of Major Abdominal Surgery

Anthony Visioni, MD; Rupen Shah, MD; Emmanuel Gabriel, MD, PhD; Kristopher Attwood, PhD; Moshim Kukar, MD; Steven Nurkin, MD

Disclosures

Annals of Surgery. 2018;267(1):57-65. 

In This Article

Results

Study Identification and Characteristics

39 studies were identified that met inclusion and exclusion criteria (Figure 1). Fourteen of these studies were randomized trials.[4,12–24] Two of the randomized trials included 4 arms; 1 reporting on laparoscopic and open gastrectomy separately,[4] and another reporting on patients aged 45 to 74 years and 75 to 89 years, separately.[12] For the purpose of this meta-analysis, these trials were each considered 2 independent studies. The remaining 25 trials were cohort studies.[2,3,5,25–46] These typically relied on consecutive patients undergoing enhanced recovery pathways compared with recent consecutive historical controls. One of these studies evaluated gastrectomy and esophagectomy patients separately[44] and, therefore, was considered 2 separate studies in this meta-analysis.

Figure 1.

PRISMA flow diagram.

This meta-analysis included 6511 patients of which 3055 were control patients and 3456 were study patients. For the primary endpoints, 68.8% of randomized studies reported LOS data in a format appropriate for meta-analysis and 75% reported complication rates. Only 34.6% of cohort studies reported LOS data in a usable format, however, 84.6% reported complication rates. The median number of ERAS interventions was 10 (range 5–16) as shown in Table 1.[2–5,12–45]

The average number of interventions was 10.4 (SD: 2.9) for the randomized trials and 9.8 (SD: 3.3) for the cohort studies; this was not statistically significant (P = 0.53). The mean LOS for all studies was 7.5 days (SE +/- 0.7) in the ERAS groups and 10.1 days (SE +/- 0.9) for the control group.

The 2 most common ERAS interventions in the studies were early initiation of enteral nutrition and early mobilization (92.5% each). This was followed by eliminating prolonged preoperative fasting (80%), avoidance of opioid analgesics (72.5%), organized preoperative counseling (67.5%), and preoperative carbohydrate loading (67.5%). It should be noted that almost all studies included appropriate antibiotic prophylaxis and DVT prophylaxis as part of their ERAS care. However, only 1 study specifically stated that DVT prophylaxis was changed as part of the ERAS pathway,[35] and 2 studies specifically stated that antibiotic prophylaxis was different in the ERAS pathway.[13,32] These aspects of the ERAS pathway were counted toward the total interventions in only these studies.

Cohort studies were evaluated for bias based on the Newcastle-Ottawa Scale (Figure 2). Only 2 studies achieved the maximum of 9 stars.[3,46] Of the remaining 23 studies, 16 achieved 7 stars, and the remainder had 6 stars. The majority of bias was found in the comparability of cohorts. Although many of these studies showed that there was no statistically significant differences between the cohorts only the Li et al[3] and Williamsson et al[46] studies used multivariate analysis to show that outcome differences were related to ERAS pathways. Randomized trials were evaluated for bias using the Cochrane Handbook (Figure 3). All 14 randomized trials were fairly similar in their risk of bias. The biggest source of bias was from lack of blinding, both by the trial participants and in the assessment of outcomes. None of the 14 studies was able to blind the physicians or patients to the intervention. Eleven of the 14 studies did not blind investigators assessing outcomes; however, 3 studies[4,13,15] did have blinded investigators assessing outcomes. Importantly, in 5 studies there were no definitive statements on the method of random sequence generation or allocation concealment.

Figure 2.

Assessment of bias in cohort studies. + denotes low risk of bias, − denotes high risk of bias.

Figure 3.

Assessment of bias in randomized trials. + (green) denotes low risk of bias, − (red) denotes high risk of bias, ? (yellow) denotes uncertain risk of bias.

Funnel plots for both primary outcomes (LOS and complication rate); demonstrate minimal risk of publication bias (supplemental figure, http://links.lww.com/SLA/B217). Funnel plots for readmissions and hospital cost also were consistent with minimal publication bias. However, the funnel plot for time to first flatus does show the possibility of publication bias.

Primary Outcome Measures

There were 20 studies with the appropriate LOS data available (Figure 4). The estimated mean difference for the meta-analysis of these studies was –2.5 days (95% CI: –3.2 to –1.8, P < 0.001), indicating a significant reduction in the mean LOS for the ERAS patients as compared with the control group. The Q (433.4, P < 0.001) and I2 (95.6%) statistic indicates that there was a high degree of heterogeneity in the study outcomes, which was demonstrated in the forest plot. Of these 20 studies, there were 11 randomized trials. The estimated mean difference for the meta-analysis of randomized trials was –2.6 days (95% CI: –3.5 to –1.7, P < 0.001), indicating a significant reduction in the mean LOS for the ERAS as compared with the control. The Q (227.5, P < 0.001) and I2 (95.6%) statistic indicated that there was, again, a high degree of heterogeneity in the study outcomes.

Figure 4.

Length of stay (LOS) of all studies and randomized trial subgroup. LOS was significantly decreased in ERAS patients in both all studies and the randomized trial subgroup (P < 0.001 for both).

There were 34 studies with appropriate complication data available (Figure 5). The estimated mean OR for the meta-analysis of these studies was 0.70 (95% CI: 0.56–0.86, P = 0.001), indicating a significant reduction in the odds of a complication for the ERAS as compared with the control group. The Q (58.1, P = 0.04) and I2 (43.2%) statistic indicated that there was a high degree of heterogeneity in the study outcomes, as noted in the forest plot. 12 of these studies were randomized trials. The estimated mean odds ratio for the meta-analysis of these studies was 0.68 (95% CI: 0.43–1.10, P = 0.12), indicating no significant reduction in the odds of a complication for the ERAS as compared with the control group. The Q (26.1, P = 0.006) and I2 (57.9%) statistic indicated that there was a high degree of heterogeneity in the study outcomes.

Figure 5.

Complication rate in all studies and randomized trial subgroup. Although complications were significantly decreased when evaluating all studies (P = 0.001), this was not statistically significant for the randomized trial subgroup (P = 0.12).

Secondary Outcome Measures

There were 25 studies with appropriate readmission data available (Figure 6A). The estimated mean OR for the meta-analysis of these studies was 1.03 (95% CI: 0.84–1.26, P = 0.80), indicating no significant increase in the odds of a readmission for the ERAS group as compared with the control group. The Q (23.6, P = 0.48) and I2 (0.0%) statistic indicated that there was little heterogeneity in the study outcomes. Of the 25 studies, 7 were randomized trials with appropriate readmission data available. The estimated mean OR for the meta-analysis of these trials was 1.60 (95% CI: 0.87–2.96, P = 0.13), indicating no significant increase in the odds of a readmission for the ERAS group as compared with the control group. Once again, the Q (5.4, P = 0.49) and I2 (0%) statistic indicated that there was little heterogeneity in the study outcomes.

Figure 6.

Secondary outcomes. A, Readmission rate was not significantly different between control and ERAS groups (P = 0.80). B, Time to passage of first flatus was significantly shorter in ERAS groups (P < 0.001). C, Hospital costs were significantly decreased in ERAS patients (P < 0.001).

There were 9 studies with the appropriate time to first flatus data available (Figure 6B). The estimated mean difference for the meta-analysis of these studies was –0.8 days (95% CI: –1.1 to –0.4, P < 0.001), indicating a significant reduction in the mean days to flatus for the ERAS group as compared with the control group. The Q (112.1, P < 0.001) and I2 (92.9%) statistic indicated that there was a high degree of heterogeneity in the study outcomes. Of the 9 studies that reported time to first flatus, 8 were randomized trials. Meta-analysis of these trials revealed a similar significant reduction in time to first flatus (–0.7 days, P = 0.002).

There were 10 studies with the appropriate cost data available (Figure 6C), all of which were randomized trials. The estimated mean difference for the meta-analysis of these studies was –$5109.10 (95% CI: –$5852.40 to –$4365.80, P < 0.001), indicating a significant reduction in the mean adjusted cost for the ERAS group as compared with the control group. The Q (15.2, P = 0.09) and I2 (40.9%) statistic indicated that there was a moderate degree of heterogeneity in the study outcomes.

Comments

3090D553-9492-4563-8681-AD288FA52ACE

processing....