This study identified factors associated with ED after elective colectomy using a large, national database. There were several notable findings that emerged from this analysis. First, we found that the mean LOS for colectomy patients has decreased over time. We observed a simultaneous decrease in readmissions for the total population during this same interval. This indicates a change in practice pattern, with ED becoming more accepted as a LOS target in United States over the period of study. Patients with ED were also found to have lower rates of bounceback readmission, overall complication and complication post-discharge, in comparison to non-ED patients. These findings suggest that, in this study population, ED was not associated with a “tradeoff” of adverse outcomes. ED patients were healthier in comparison to the non-ED cohort (younger age, lower 5-mFI score, reduced rate of preoperative steroid usage, etc). The variable most robustly associated with ED was robotic technique. Patients who underwent robotic surgery had 14 times higher odds of ED compared to those who had open surgery, and more than double the odds of ED compared to those who had laparoscopic colectomy. As the healthcare system in the United States continues to struggle with high inpatient censuses and critical bed shortages, the findings in this study can help surgeons better identify patients with strong odds of ED and low resource utilization.
A key finding in this analysis was the strong association seen between robotic surgery and ED. These results are in agreement with a consistent observation in the existing literature, that patients who undergo robotic colorectal surgery have reduced LOS in comparison to those who had laparoscopic colectomy (decreased LOS by a range of 0.6-3.5 days).(20–30) The underlying mechanisms for this difference are not known, and because the studies to date are retrospective and observational, it is not possible to decipher whether the difference in LOS stems from patient selection, surgeon-specific factors (for example, do those who perform robotic surgery more likely practice ERAS in a structured program?), or truly from technical factors related to the different MIS platforms. Investigations to better understand the driving force behind this recurrent finding would be a worthy subject for future study. Regardless of the underlying cause, our findings demonstrate that patients who underwent robotic colectomy were more likely than those who had laparoscopic surgery to stay “on pathway” following surgery, achieving a discharge within 48 hours and a low likelihood of readmission. These data add to the previously published findings, by linking robotic surgery to a discrete target for successful ED, rather than simply comparing averages from a continuous variable (LOS).
The other associated characteristics that were found to be related to ED in this study indicate that the patients were individuals at lower-risk of perioperative complications compared to those with non-EDs. One possible explanation for this finding is patient selection, as surgeons may be more likely to authorize an ED if they perceive the patient to be at low risk for subsequent complications. This makes intuitive sense, and reflects the importance of clinical judgment in selecting an appropriate discharge time for each individual patient. This observation is also similar to previously published studies that evaluated postoperative LOS in populations of colorectal patients. In a study looking at the impact of chronologic age on LOS, researchers divided patients into age-related subgroups 51-70 and >70, and found that LOS increased by 3.9 days in older patients.(31) Patient frailty has also been shown to be associated with longer postoperative hospitalization. In a study that examined a cohort of 20,966 patients, frail patients (5-mFI ≥2) were more likely to experience a prolonged LOS (≥5 days) compared to pre-frail (5-mFI =1) or non-frail (5-mFI =0) patients.(32) A study of the National Readmissions Database (NRD) found that in comparison to non-frail patients, frail patients had a LOS that was 4.4 days longer (95% 4.1-4.6, p<0.001).(33) We believe that the results in this study expand on these previously published findings, by describing a more comprehensive list of characteristics for patients who are and are not likely to achieve a discharge within 48 hours. Understanding these profiles may help surgeons more easily recognize healthy patients who are veering away from expectations, and thereby detect subtle signs of complications at earlier time-points. It may also allow more flexible timelines for discharge for higher risk individuals within existing ERAS pathways, recognizing the limitations of fast-track goals in high risk cohorts.
Regarding the relationships between post-operative outcomes and LOS, three recently published studies provide in-depth evaluations and fresh insights on the subject and warrant special mention. In the first, Moon et al, performed a detailed analysis of postoperative readmissions after colorectal cancer resection using the NRD between 2010 and 2017.(19) ED patients (defined as discharge within 3 postoperative days) during this interval had a readmission rate of 7.3%, and though lower than the rate of readmission for non-ED patients (15.2%), this rate did not improve over time. ED was associated with increased rates of “bounceback” readmissions (within 7 days), as well as increased resource utilization and cost for both the index admission and the bounceback readmission. Bounceback readmissions involved high rates of anastomotic leak compared to complications incurred during the index operation (38.1 vs. 27.5%). As the authors discuss, these data point to the limitations for fast-track / ERAS pathways, and the need for more individualized decision-making rather than over-reliance on templated post-operative discharge targets, especially for patients with higher baseline comorbidity. We observed in our study that patients with ED had lower rates of overall readmission, and of bounceback readmission, in comparison to non-ED patients. The discrepancy between the studies could be related to differences in the underlying datasets, in the definitions of “ED” itself (3 days in Moon et al vs. 2 days in our study), or in the inclusion criteria (Moon et al includes rectal resection, and only included patients with cancer, whereas our study includes all indications for colectomy and does not include proctectomy). It is important to note that there is no standard definition of ED in the literature. We chose our definition based upon prior work cited in our methods, as well as our own clinical experience, which is that average-risk patients recovering from MIS colectomy without complications are ready for discharge on POD2.
In the second study of special note, Li et al evaluated LOS, readmission, and postoperative complications using NSQIP, the same database we used in our investigation.(34) Here, the authors separated the patients according to timing of post-operative complications. During the study period (2012-2018), there was a decrease in overall LOS (5 to 4 days) and 30-day complication rate (25.8 to 19.1%), but an increase in the incidence of four different complications occurring after discharge (superficial SSI, overall SSI, wound disruption, UTI). The authors argue that these trends suggest a tradeoff between ED and post-discharge complications, one that could be mitigated with improved home monitoring or avoiding ED in select patients. However, we noted that the authors do not include LOS as an independent variable in the regression modeling for predictors of post-discharge complications. It is possible therefore, that increases in post-discharge complications are driven by patients with prolonged LOS who developed complications after an already complicated admission. We observed in our study that ED patients had lower rates of post-discharge complications compared to non-ED patients. We believe this suggests, in contrast to what was proposed by Li et al, that ED is not associated with a tradeoff in higher post-discharge complication rates.
In this regard, our study is in agreement with the third paper of note on this topic, by Balas et al.(35) The authors in this study examined early discharge (ED) after uncomplicated elective colectomy using NSQIP data from 2012 through 2018, finding that ED is associated with lower odds of postdischarge complications for minimally invasive approaches and no increased risk for the open approach. Despite the comprehensive analysis by Balas, et al, our study adds to their findings in several important ways. First, our study extends the data period to 2021, offering a more contemporary dataset than any of the three aforementioned studies.(19,34,35) Given the acceleration of ERAS initiatives around the country during this time, including proposals for same-day discharge, as well as the heightened pressure to reduce LOS in the wake of the COVID-19 pandemic, we believe these more recent years add new value to this topic of study. Additionally, Balas et al. did not examine readmission as a discrete outcome. One major critique of ED is the potential for increased readmission rates, which our study addresses by replicating the "bounceback readmission" analysis from Moon et al., demonstrating that ED patients had lower overall and bounceback readmission rates. Our focus on ED as the primary outcome offers a statistical perspective that can be easily recalled by surgeons, administrators, and patients, showing that patients undergoing robotic colectomy have 14 times the odds of ED compared to open surgery and more than twice the odds compared to laparoscopic surgery. Overall, despite methodological differences, our findings are most congruent with those of Balas et al., reinforcing the robustness of their conclusions and suggesting that early discharge, even with stringent criteria, can be safely implemented in clinical practice.
This study does have several noteworthy limitations. There is a possibility of selection bias, as always, as it is a retrospective study of a large administrative database. While the use of ACS-NSQIP and the associated targeted colectomy data allowed for a relatively large sample size, the patients in this study may not adequately reflect the general population. There are some built-in limitations to the ACS NSQIP database itself which include the fact that it is a large observational database that could be prone to entry error. Although data entries into the source data from each institution are performed by highly trained individuals, this study cannot account for coding errors. It is also possible that the results are confounded by unmeasured variables with an influence on discharge timing. These include patient-related factors (physical frailty, patient resistance to ED), surgeon-related factors (surgical training as it pertains to expected LOS, level of experience, surgical volume), and hospital-related variables (availability of a structured ERAS program, hospital volume, hospital size). Lastly, although we defined ED as ≤2 days after colectomy based on prior publications and based on our own surgical experience, there is no consensus on the definition of ED in the literature.