The impact of preoperative nutritional screening, ERAS protocol, and mini-invasive surgery in surgical oncology: A multi-institutional SEM analysis of patients with digestive cancer

Abstract

Background: Mini-invasive surgery (MIS), ERAS, and preoperative nutritional screening are currently used to reduce complications and the length of hospital stay (LOS); however, inter-variable correlations have seldom been explored. This research aimed to define inter-variable correlations in a large series of patients with gastrointestinal cancer and their impact on outcomes.

Methods: Patients with consecutive cancer who underwent radical gastrointestinal surgery between 2019 and 2020 were analyzed. Age, BMI, comorbidities, ERAS, nutritional screening, and MIS were evaluated to determine their impact on 30-day complications and LOS. Inter-variable correlations were measured, and a latent variable was computed to define the patients' performance status using nutritional screening and comorbidity. Analyses were conducted using structural equation modeling (SEM).

Results: Of the 1,968 eligible patients, 1,648 were analyzed. Univariable analyses documented the benefit of nutritional screening for LOS and MIS and ERAS (≥7 items) for LOS and complications; conversely, being male and comorbidities correlated with complications, while increased age and BMI correlated with worse outcomes. SEM analysis revealed that (a) the latent variable is explained by the use of nutritional screening (p0·004); (b) the variables were correlated (age-comorbidity, ERAS-MIS, and ERAS-nutritional screening, p < 0·001); and (c) their impact on the outcomes was based on direct effects (complications: sex, p0·001), indirect effects (LOS: MIS-ERAS-nutritional screening, p < 0·001; complications: MIS-ERAS, p0·001), and regression-based effects (LOS: ERAS, MIS, p < 0·001, nutritional screening, p0·021; complications: ERAS, MIS, p < 0·001, sex, p0·001). Finally, LOS and complications were correlated (p < 0·001).

Conclusion: Enhanced recovery after surgery (ERAS), MIS, and nutritional screening are beneficial in surgical oncology; however, the inter-variable correlation is reliable, underlying the importance of the multidisciplinary approach.

Keywords: digestive cancers; mini-invasive surgery; nutritional screening; structural equation modeling (SEM); surgical oncology.

Figure 1

STROBE flowchart of the study population showing identification of the cohort, exclusion of patients, and the final selection of patients analyzed.

Figure 2

Partial correlations of endogenous variables. On the right side, the legend presents bubble dimensions and colors to express strong positive (blue) and negative (red) correlations among endogenous variables (mini-invasive surgery, ERAS, nutritional screening, BMI, age, comorbidity, and sex).

Figure 3

Steps for structural equation modeling (SEM) analysis: the process is described starting from the top as illustrated in the process diagram depicted on the left side of the picture. Steps are presented in the same sequential order as the statistical analyses. First step: univariable analyses and partial correlations. In the first part, the results of statistical value reported in Table 1 are depicted, with the positive (blue) or negative (red) correlation of variables with the outcomes [i.e., increased age correlated with increased length of hospital stay (LOS), blue line; whereas the use of mini-invasive surgery negatively correlated with the same outcome, red line]; the correlation diagram is the same as presented in Figure 2. Second step: definition of the latent variable according to factor analysis and based on those variables presenting greater uniqueness criteria: comorbidity and nutritional screening. Third step: finally, the final stage is depicted at the bottom of the figure (the definition of direct–yellow and indirect–pink effects on the outcomes).

Figure 4

Graphical representation of the structural equation modeling (SEM) analysis. The blue boxes represent endogenous variables and the blue arrows are depicted to link inter-variables correlations of statistical value (p < 0·05, see Table 4); the red boxes are for the outcomes of interest (LOS, length of postoperative hospital stay; 30-day complications, 30-day postoperative complications); the green box is for the latent variable to represent patients' performance status, and the green arrow is to link this latent factor with the endogenous variable of statistical value (p < 0·05, see Table 4); the red arrows show regression-based effects with statistical significance (p < 0·05, see Table 4); the yellow arrows show direct effects of statistical value (p < 0·05, see Table 4); the purple arrows show the indirect effect with statistical significance (p < 0·05, see Table 4). The bottom line provides an explanation of the SEM analysis.