Abstract
Background
Emergency general surgery (EGS) patients who undergo interfacility transfer (IFT) experience higher rates of complications and mortality compared to those directly admitted (DA) to a hospital. However, their failure-to-rescue (FTR) rates—defined as mortality following a major complication—remain less studied. Given the increased burden of adverse outcomes in this population, we hypothesized that IFT patients would have higher risk-adjusted FTR rates than DA patients.
Methods
We performed a 5-year (2016-2020) retrospective analysis using the National Surgical Quality Improvement Program (NSQIP) database, focusing on patients aged 18 years and older undergoing high-risk EGS procedures, including enterectomy, colectomy, peptic ulcer surgery, and laparotomy. To assess the impact of IFT, we employed multivariable logistic regression models, adjusting for demographic factors, comorbidities, and procedure type.
Results
Among 70 028 patients (52% female, 66% white, median age 66), 15 032 (21.4%) underwent IFT. After risk adjustment, IFT patients demonstrated significantly higher odds of major complications (OR 1.09, 95% CI 1.04-1.14), mortality (OR 1.23, 95% CI 1.16-1.31), and FTR (OR 1.12, 95% CI 1.04-1.19), suggesting that transferred patients are at a distinct disadvantage compared to DA patients.
Key Takeaways
• Interfacility transfer is linked to higher failure-to-rescue rates, with transferred patients facing greater mortality after major complications. • Transferred patients have higher risks of major complications, even after adjusting for demographics and comorbidities. • Outcome disparities persist across various emergency surgical procedures, highlighting the need for targeted interventions.
Introduction
Emergency general surgery (EGS) is a subset of acute care surgery (ACS) which provides services to any patient requiring an emergent evaluation within the realm of a general surgeon. 1 Emergency general surgery accounts for over 3 million inpatient admissions and 11% of surgical admissions annually.2,3 Although only accounting for a 10th of the surgical admissions, EGS is associated with nearly 50% of surgical mortalities. 2 Emergency general surgery care can prove to be problematic due to a variety of reasons, including technical requirements of the performing hospital, need for a knowledgeable and available workforce, as well as patient complexity.
The burden of EGS poses a larger challenge to facilities who may not have the necessary resources to provide safe and effective care. As a result, patients with EGS conditions at these facilities are frequently transferred to more equipped tertiary hospitals with dedicated ACS services. Interfacility transfer (IFT) patients make up 2%-7% of the admitted EGS patients.4-6 These cases are a unique threat to surgical quality care as studies have shown complications are more frequent in EGS patients who were transferred compared to those directly admitted.5,7-11 These studies focus on complications, mortality, and morbidity, but there are limited studies evaluating failure-to-rescue (FTR) rates in this patient population.
Failure-to-rescue within surgery is a quality care metric initially described in elective surgical cohorts to measure rates of death after postoperative complicaitons. 12 The aim of this study was to evaluate the relationship between transfer status and FTR rates in emergency general surgery patients. We hypothesized that in EGS patients with complex conditions, FTR rates are higher in patients who were transferred than in those who were not.
Methods
Patient Selection
Current Procedural Terminology Codes Use by Procedure
Patients were then stratified into either direct admission (DA) or interfacility transfer (IFT). A case was defined as IFT if the admission source within the NSQIP data set was from an outside emergency department, acute care hospital inpatient setting, nursing home (chronic care and intermediate care), or transfer from another facility. Direct admits were defined as having an admission source as admitted from home. If the patient’s admission status was unknown, the case was removed from analysis.
Outcome Variables
The first outcome measured was the occurrence of a major complication, including wound complications (deep organ space surgical site infection, deep surgical wound infection, and wound dehiscence), pulmonary complications (pneumonia, unplanned intubation, pulmonary embolism (PE), deep venous thrombosis (DVT), and mechanical ventilation >48 hours), renal complications (acute kidney injury or dialysis), central nervous system (CNS) complications (postoperative stroke), cardiac complications (myocardial infarction or cardiac arrest), sepsis (septic shock or bloodborne sepsis), and bleeding complications (return to operating room, postoperative bleeding) using a modification of the methodology of Ferraris 13 in which only returns to the operating room events associated with CPT codes associated with bleeding were counted as bleeding complications (35 840, exploration for postoperative hemorrhage; 49 020, drainage of peritoneal hematoma; 10 140, I&D of hematoma; 44 604, control of hemorrhage, small intestine). We additionally measured mortality within 30 days post-surgery. Our primary outcome measurement was FTR, defined here as a death after a major complication as described above. To ascribe FTR cases to complication groups, we used the first recorded postoperative complication.14-16
Statistical Analysis
Major complication, mortality, and FTR rates were compared between the DA and IFT groups using chi2 test for categorical variables and Kruskal-Wallis test for continuous variables. Multivariate logistic regression modeling was used to evaluate the association between IFT and outcomes of interest while adjusting for patient characteristics and comorbidities. Those factors found to be associated with the outcomes of interest (the occurrence of a major complication, mortality or FTR) in univariate logistic regression modeling with a P-value of <0.2 were included in our multivariate logistic regression models. Characteristics and comorbidities analyzed included age, sex, race, weight, weight loss >10% within previous 6 months, diabetes requiring treatment, functional status, chronic obstructive pulmonary disease (COPD), hypertension, current ventilator use, disseminated cancer, previous wound infection, smoking within a year of surgery, steroid use preoperatively, ascites within 30 days prior to surgery, congestive heart failure, presence of pneumonia, sepsis and/or septic shock, bleeding disorder, preoperative transfusion, American Society of Anesthesiologists (ASA) class, renal failure within 24 hours of operation, and dialysis within 2 weeks of surgery. All analyses were performed using STATA v17.0.
Results
Patient Demographics
Transfer Status by Surgery Type. Abbreviations: PUD, Peptic Ulcer Disease
Patient Comorbidities
Comparing Patient Demographics and Comorbidities in DA and IHT Patients. P Values are for chi2 Test for Categorical Variables and Kruskal-Wallis Test for Continuous Variables. Abbreviations: COPD, Chronic Obstructive Pulmonary Disease; CHF, Congestive Heart Failure; PNA, Pneumonia; SIRS, Systemic Inflammatory Response; ASA, American Society of Anesthesiologists
Unadjusted Complication, Mortality and FTR Rates
Mortality Rates After Specific Complication Types, or Fractional Contribution to the Overall Observed FTR Rate
Outcomes by Procedure Type
Effect of Transfer Status on Complications by Type
Adjusted Complication, Mortality, and FTR Rates
In multivariable logistic regression modeling to control for the baseline differences in patient characteristics, our analyses continued to show an increased likelihood for complications, mortality, and FTR following the forementioned EGS procedures for interfacility patients vs those directly admitted. After risk adjustment, IFT patients still had higher odds of major complications (OR 1.09, 95% CI 1.04-1.14), mortality (OR 1.23, 95% CI 1.16-1.31), and FTR (OR 1.12, 95% CI 1.04-1.19). This finding of an increased risk-adjusted likelihood of failure-to-rescue in transfer patients is similar when analyzed by procedure type (Figure 1). Risk-Adjusted Probability of FTR by Transfer Status
Discussion
Our retrospective cohort study using the NSQIP data set shows transfer status is associated with increased rates of complications, mortality, and failure-to-rescue occurring in patients undergoing operation for peptic ulcer disease, small bowel resections, colectomies, and exploratory laparotomies. This continued to hold true after controlling for individual patient comorbidities and characteristics. Moreover, when looking at failure-to-rescue rates for each surgery, the IFT cohort had similar findings of increased rates.
Previous published studies have shown complication and mortality rates to be higher in transferred patients, including emergency general surgery patients.5,7-11 These studies report complication rates ranging from 6.7%-31% for DA patients and 18.9%-48% for IFT patients. Moreover, mortality rates ranged from 1.6%-9.4% and 4.4%-24.7% in directly admitted and IFT patients, respectively. In our study we found complication rates of 51.6% vs 62.8% and mortality rates of 11.9% vs 20.4% for DA and IFT patients, respectively. Our increased complication rates likely reflect the fact that we limited our cohort to 4 procedures noted to have the highest complication and mortality rates 3 while others, such as Coimbra et al, 8 used the same NSQIP data set but included 8 emergency general surgeries, several known to have lower complication and mortality rates. 3 Their results differed from ours with reported complication rates of 6.7% vs 18.9% and mortality rates of 9.4% vs 10.4% for direct admitted and IFT patients, respectively. Allen et al 11 reported findings that were closest to ours with complication rates of 31% vs 48% and mortality rates of 7% and 14% for DA and IFT patients, respectively. This study included a broader array of EGS conditions (including hernia repairs and pancreatitis) and also included any postoperative complication for their analyses rather than focusing on major complications as we did. More recently, a 2021 network analysis found that while EGS transfers do concentrate patients at larger, higher-volume centers with more ICU resources, no difference in risk-adjusted mortality or FTR rates were observed. 17 However, risk adjustment in this study was limited in that it only used administrative data from the Healthcare Cost and Utilization Project, included patients with any EGS diagnosis, and included any cases in any EGS diagnosis existed, rather than limiting to those cases where the EGS diagnosis is the primary diagnosis, as in our study. Given the heterogeneity in included procedures and outcomes in the existing literature, the differences between our findings and those of prior studies are not unexpected.
One consistent finding in ours and previous studies is that transferred EGS patients tend to be more complex than directly admitted patients, more often having worse physiological status and multiple patient comorbidities. Even after using the available demographic, physiologic, and comorbidity variables in the NSQIP data set to risk adjust our findings, striking differences in complication, mortality and FTR rates persist. While as an observational study we are not able to make definitive statements on causality, we suspect that there is an element of confounding by indication at play—that is, sicker, more complicated patients tend to be transferred and the degree of this complexity is not always well-captured by even databases as rigorous as the NSQIP. As EGS cases are to some degree inherently time-sensitive, our findings could represent the effects of differential delays in definitive treatment of the underlying conditions that required surgical intervention in the IFT group referent to the DA group. Unfortunately, due to limitations of the data set we are not able to account for what occurred at the transferring center, including what procedures may have occurred prior to transfer or the timeline of transfers.
The relationship between IFT and FTR may also depend on the context in which it is measured. For instance, although we found that IFT was associated with a higher rate of FTR referent to DA patient at the centers included in our study, it may be that at smaller medical centers with less intensive staffing models and resource allocation, patients who are transferred out actually have lower FTR rates than those that remain. Unfortunately, the NSQIP data set does not contain center level variables nor longitudinal data that would be necessary to test this hypothesis.
We examined our outcome rates both by including and excluding patients with preoperative sepsis or septic shock who also had postoperative sepsis or septic shock in our analysis. Since FTR was defined in elective surgical populations, by definition this was not an issue that needed to be addressed at the time the original FTR metric was described. FTR metrics described in non-elective cohorts, such as trauma and EGS, have therefore been constructed differently to account for the fact that patients that require emergent surgical intervention often present with conditions like shock, acute kidney injury, respiratory failure, and sepsis that would be considered complications if they occurred after surgery but since they precede the surgery do not meet this definition. The way such conditions are treated methodologically becomes a reflection of what one’s perception of what FTR is meant to measure: the ability to rescue patients after postoperative complications (the original definition as per Silber et al 12 ) or the ability to rescue complex surgical patients (a more inclusive definition based on the work of Peitzman and others 18 ). We chose the latter approach based on what we believe the FTR metric should represent for EGS patients and for the pragmatic reason that sepsis and septic shock are extremely common presentations of EGS conditions (accounting for ∼45% of patients in our cohort), and excluding these patients from our FTR analysis could render our findings non-generalizable to typical EGS populations. While the exclusionary approach did reduce the number of patients coded as having sepsis or septic shock as a first complication, it did not meaningfully change our complication, death, or FTR rates between IFT and DA patients overall, suggesting that our findings are robust to methodological differences with respect to this issue in this cohort.
Although the difference in outcomes we describe adds to the existing literature surrounding IFT in EGS patients, many questions remain as to the proximal causes and mitigation strategies for these findings. One strategy could focus on improving the timeliness and processes of care for EGS patients who require a higher level of care under the assumption that delays in transfer may account for differences in outcomes. Alternatively, it could be argued that outcomes for complex EGS patients could be improved through the benefits of the ACS model of care19-23 at transferring hospitals. The best approach may well involve interventions at both transferring and receiving hospitals but until such time as those strategies are developed, our work serves to highlight a critically vulnerable population that can be easily recognized by the practicing clinician as at increased risk for complications and FTR after those complications.
Limitations
Several limitations of our work must be acknowledged. First, as a retrospective study, our work is susceptible to known biases, such as selection bias. Second, our pool of patient data is limited to the NSQIP data set which is a nonrandom sample voluntarily abstracted and shared by hospitals. While the quality of this data set is well known, there are inherent limitations in the variables collected as well as their definitions. For example, because variables containing information about the transferring center are not collected, we were not able to ascertain what if any procedures patients may have undergone or complications they may have incurred prior to transfer. The absence of center identifiers also prohibits the use of clustering by institution in our statistical analyses. Along similar lines, although FTR rates have been demonstrated to be associated with hospital structures of care including anesthesiologist and surgeon board certification rates12,24 we do not have the ability to account for the differences in these factors between transferring and receiving hospitals. Due to limitations in the coding of reasons for return to the operating room, utilizing only those codes that strictly map to bleeding indications may underestimate the true incidence of bleeding complications. Similarly, although the NSQIP is arguably the highest-quality data set available for surgical outcomes research, use of CPT codes to indicate what occurred in the OR has inherent limitations—for instance, our surgical group “exploratory laparotomy” may have had only this procedure (surgical exploration through a laparotomy incision) or potentially other uncoded procedures performed, but this cannot be reliably resolved from the variables included. Regardless of what actually may be included in this grouping, it has previously been demonstrated to be strongly associated with mortality. 25
As with any large data set, it is not possible to verify the accuracy or completeness of the data collected, including complications. However, as the NSQIP utilizes dedicated personnel to abstract and manage data, we believe this data to be best-in-class for surgical outcomes research. On a similar note, it is not possible to ascertain which patients as recorded as having died without a complication may have actually had an unrecorded complication. Finally, FTR events and the complications that permit them are not always preventable, 26 and as such may not always reflect quality of care but it is not possible to ascertain the preventability of the FTR events in this analysis.
In conclusion, our study showed interfacility transfer patients undergoing emergency general surgery have an increased risk of complications, mortality, and failure-to-rescue. This continued to hold true after controlling for patient characteristics and comorbidities. These findings reinforce the concept that transferred patients represent a particularly vulnerable subset of EGS patients at risk for negative outcomes.
Footnotes
Author Contributions
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.