Risk Factors for Readmission and Mortality Following Colonic Surgery: A Consecutive Retrospective Series of More Than 2500 Cases

Introduction

Colonic resection is a common surgical intervention with a variety of indications, including: neoplasms, inflammatory bowel disease, diverticulitis, ischemia, and infectious diseases. In the United States, more than 300,000 surgical procedures involving the colon are performed annually.^1,2

Postoperative complications following colonic resection can carry a significant burden on the patient and the health care system. Postoperative complications following colectomy include among others wound infections, anastomotic leak, ileus, and bleeding. Although the majority of postoperative complications are transitory and can be treated with simple measures, some can be life threatening and require aggressive interventions to ensure a safe outcome. ³ As complications may have late presentation, some may occur after the patient had been discharged, and require readmission. ⁴

Readmission rate is considered one of the most important measures of treatment quality, as it serves as a good surrogate of delayed clinically important complication rate and has a significant impact on treatment cost. It is estimated that unplanned 30-day readmissions occur in up to 19.6% of Medicare beneficiaries, with an annual cost of more than $17 billion. ⁵ In colorectal surgery, unplanned 30-day readmission rates are fairly variable and range from 6% to 25%. ⁶ It is therefore essential to identify possible pre- and intraoperative factors that might determine the risk for patient readmission following surgery.

In this study, we focused on patients undergoing various types of colonic resections, to evaluate which factors can predict an increased risk of readmission, thus presenting the contemporary status of readmission after colectomy and risk factors for readmission to assist physicians to improve the treatment given to patients and to develop and apply strategy to avoid unwanted postoperative outcomes.

Materials and Methods

We conducted a retrospective single-center study, reviewing the data of all adult patients who underwent colonic resection, between January 2008 and December 2020, in a large tertiary medical center. Data were collected from patients' medical records, using MDClone^© (Beer-Sheba, Israel) software, a data extraction and synthesis Web-based platform connected to medical records of patients treated in our facility, and provides patient-level data around an index event (http://www.mdclone.com), which was defined for this study as any type of colonic resection. The data extraction criteria were predefined by the authors based on relevance to study outcomes. Data from medical charts of patients who were readmitted 30 days or less following the surgical intervention, were manually extracted for an additional analysis. Ethics approval was granted by the hospital's Institutional Review Board. Informed consent for this study was waived by the Institutional Review Board (SMC-6018-19).

Data collection included demographic (age, gender, body mass index, smoking and immigration status), clinical (comorbidities, prior abdominal surgery, inflammatory bowel diseases), and perioperative data (surgical approach, stoma formation, postoperative complications, 30-day readmissions, and mortality). Anastomotic leak was identified using computerized tomography (with extravasation of contrast from the anastomosis) or during surgery as a perforation at the anastomotic site. Intraabdominal abscess included all patients with a well-defined collection of pus or infected fluid surrounded by inflamed tissue as identified on imaging.

The main outcome was overall postoperative 30-day readmission rate following colonic resections. Secondary outcomes were overall mortality, rate of minimally invasive surgery, and length of stay (LOS).

Statistical analysis was performed using SPSS (Version 25, IBM Corp, Armonk, NY, USA). We performed univariate analysis using liner regression and multivariable analysis using binary logistic regression and results were presented with odds ratio (OR). The analysis included two logistic regression models: one for 30-day readmissions and the other for 30-day mortality. Variables that were found significantly associated in univariate analysis were included in multivariable analysis. P < 0.05 was considered significant. Data are presented as mean and standard deviation or median and interquartile range.

Results

Overall, 2547 patients underwent colonic resection during the study period. There were 1338 (52.6%) females with a median patient age of 68.3 years (range 18.1–98.2). Patients' basic demographic and clinical data are listed in Table 1. The most common type of colonic resection was right colectomy (1428 patients, 54.5%), followed by left colectomy (835 patients, 31.9%) and sub/total colectomy (283 patients, 10.8%). Laparoscopic resection was performed in 974 patients (37.2%). Elective surgery was done in 2112 patients (82.9%) and 435 (17.1%) underwent an emergency operation.

Overall, the 30-day readmission rate was 8.3% (218 patients). Thirty-day mortality rate was 1.65% (42 patients). The mean LOS after surgery is estimated to be 9.42 ± 10.23. Other outcomes of index operations and hospitalization are detailed in Table 2. Among the 218 patients who were readmitted, we analyzed the complications that led to readmission. The most common postoperative complications were intraabdominal abscess (n = 34, 15.6%), surgical site infection (n = 33, 15.1%), anastomotic leak (n = 27, 12.4%), and other gastrointestinal complication (n = 34, 15.6%) as detailed in Table 3.

Univariate analysis showed correlation between 30-day readmission and cognitive decline (P = .015, OR = 3.10), cirrhosis (P = .018, OR = 5.40), preoperative albumin (P = .045, OR = 1.18), and blood transfusion during hospitalization (P < .001, OR = 2.00). Age (P = .005, OR = 0.98), LOS before surgery (P < .001, OR = 0.74), and laparoscopic surgery and postoperative LOS (P = .013, OR = 0.55) were protective factors of readmission. In the univariate analysis for 30 days mortality, the following were found to be correlated: age (P < .001, OR = 1.12), immigration status (P < .001, OR = 7.78), hypertension (P = .005, OR = 2.40), chronic renal failure (P < .001,OR = 5.55), ischemic heart disease (P < .001, OR = 3.15), cognitive decline (P < .001, OR = 15.24), congestive heart failure (P < .001, OR = 5.24), preoperative hyperglycemia (P = .003, OR = 1.00), year of operation (P < .001, OR = 1.08), and type of colectomy (P = .002, OR = 1.86). In addition, we found a strong correlation between patients presenting with preoperative hyperglycemia and a history of diabetes mellitus (46.8% in diabetic patients versus 13.7% in nondiabetic patients; P < .001).

On the other hand, laparoscopic surgery (P < .001, OR = 0.07), ostomy formation (P < .001, OR = 4.14), and elective surgery (P < .001, OR = 0.16) were protective factors of 30-day mortality. Univariate analysis of demographics, clinical and surgical characteristics correlated with 30-day readmission and 30-day mortality, are detailed in Table 4.

Multivariable analysis demonstrated independent association between 30-day readmission and blood transfusion during surgery (OR = 2.14, P < .001) as well as year of operation (P = .024, OR = 1.05). On the other hand, laparoscopic surgery (P = .006, OR = 0.62), LOS, and older age (P = .004, OR = 0.96) (P = .001, OR = 0.98) were demonstrated to be independent protective factors for 30-day readmission, as detailed in Table 5. To assess whether older age was truly protective for readmission, we performed a linear regression analysis of the correlation between overall LOS and patients' age, demonstrating that older patients were hospitalized for longer periods following surgery (R = 0.1, P < .001), perhaps contributing to the lower rate of readmissions in this patient population (Fig. 1).

OPEN IN VIEWER

FIG. 1.

Correlation between patients age and length of hospital stay. Older age was associated with an increased mean hospital stay following colon surgery (Pearson's R = 0.1, P < .001).

Multivariate analysis demonstrated independent association between 30-day mortality and older age (P < .001, OR = 1.10), immigration status (P = .006, OR = 3.38), cognitive decline (P < .001, OR = 13.44), and hyperglycemia (P = .013, OR = 1.00). Conversely, hospitalization days before operation (P = .007, OR = 0.75), elective surgical intervention (P = .017, OR = 0.35), and laparoscopic colectomy (P = .022, OR = 0.08) were shown to be independently protective for 30 days' mortality. Multivariable analysis of demographics, clinical and surgical characteristics, and 30-day readmission, as well as mortality are detailed in Table 6.

Discussion

In this study, we sought to define the perioperative risk factors for rehospitalization and mortality following colonic resections. We retrospectively reviewed a consecutive cohort of more than 2500 patients operated on in one tertiary academic center, over a period of 13 years. We found that the mortality rate during hospitalization following colectomy was 1.65%, similar to other studies that found 30-day mortality rate to range between 0.9% and 9.9%.^7–17 The 30-day readmission rate in our series was 8.3%, which again is in accordance with recent meta-analyses, which found that the rate of readmission ranged from 7% to 25%. ¹⁸ We found that different factors were associated with these postoperative outcomes, such as the type of surgery, age of patients, as well as multiple comorbidities and other factors.

Our analysis demonstrated that blood transfusion during colectomy was correlated with a higher odds ratio for 30-day readmission. This finding was also found as a risk factor for readmissions in the National Surgical Quality Improvement Program database of the American College of Surgeons, among patients undergoing general surgery interventions. ¹⁹ Factors correlated with a lower risk of readmission were found to be laparoscopic approach, longer length of hospital stay, and older age. Laparoscopic surgery was found to be associated with reduced length of hospital stay, ²⁰ intraoperative blood loss, ²¹ faster restoration of normal bowel function, ²² and lower rates of postoperative pulmonary and cardiac complications, ²³ all contributing to the reduction in the risk of readmission. Furthermore, laparoscopic surgery is used more often in elective cases, which are less prone to postoperative complications and readmissions as a result. ²⁴ While these factors were somewhat predictable, an interesting finding was that older age was correlated with lower risk of readmission.

Our hypothesis was that older patients were hospitalized longer, probably because they are more prone to complications and there is also an inherent tendency to keep these patients as inpatients as a precaution because of their increasing frailty. ²⁵ Indeed, our analysis showed that elderly patients were indeed admitted longer, which explains why they were readmitted less frequently and why an older age had low OR for readmission.

We also found that older age, immigrants, cognitive decline, and hyperglycemia before surgery were associated with a higher likelihood of mortality following colonic surgery, while laparoscopic approach, elective surgical intervention, and length of hospitalization before the surgery correlated with a lower risk for 30-day mortality. Patients with a preexisting diagnosis of cognitive decline had an associated risk 13 times higher for all-cause mortality 30 days after surgery, whereas being an immigrant was associated with a threefold. While some of these factors are somewhat obvious and nonmodifiable, including the association between higher mortality and older age and cognitive decline, attention should be paid to pre and perioperative glucose control to reduce the risk for postoperative mortality. As for immigration, we speculate that the higher risk for readmission stems from a lower access to supporting outpatient systems ²⁶ and linguistic disparities, ²⁷ which contribute to the higher rate of readmission in this population.

Assessing the possible risk factors for readmission and mortality following colonic resection can play a pivotal role in the decision-making process by surgeons and patients. Although some of the abovementioned factors are nonmodifiable, including age, prior medical history, and immigration status, other factors such as laparoscopic approach, glucose control before surgery, and preoperative hospitalizations are certainly within the spectra of factors that can be optimized before the surgical intervention. In addition, the existence of nonmodifiable risk factors may not alter the postoperative course, but it can aid physicians to better prepare for postoperative adverse events and complications.

In our series, some of the risk factors found to be correlated with increased rates of mortality following surgery have been described in the past. For example, Neuman et al. reviewed data from the SEER Medicare database of patients diagnosed and operated for colon cancer, focusing on patients older than 80 years of age. The authors found that older age, male gender, frailty, hospitalizations in the year before surgery, and dementia were all associated with poor survival. ²⁸ Another study from Pennsylvania published a few years ago, including all patients that underwent a colectomy in a single year and that were included in the Pennsylvania Health Care Cost Containment Council database, demonstrated that factors correlated with readmission included urgent operation, postoperative complications that occurred during index hospitalization, and low volume for the performing surgeon. Factors that were associated with lower rate of readmission were primary construction of ileostomy and laparoscopic approach. ²⁹

Another example is a study by Zheng et al., focusing on the role of laparoscopic approach for colonic resection. The authors reviewed the National Cancer Database (NCDB), demonstrating that in oncologic resections laparoscopic surgery was correlated with a lower rate of 30-day mortality and shorter length of stay. ⁷

Although our results are meaningful, several limitations should be recognized to translate these findings into clinical practice. These include the retrospective nature of the research and the possibility of residual confounders not accounted for in the multivariable analysis, such as performance status and the severity of the listed comorbidities. Unfortunately, our registry does not include routine performance status such as ECOG (Eastern Cooperative Oncology Group) or the Karnofsky performance scale that might have been helpful with better evaluation of these patients. Furthermore, since this is a single-institution experience, it is more prone to a selection bias due to lack of patient randomization and variance in follow-up.

In our study, <50% of the procedures were performed by laparoscopic approach, which shows better outcomes for readmission and lower mortality as we discussed before. The lower numbers of laparoscopic approaches can be explained by the expertise of the surgeon, surgical history of the patients, and mainly the urgency of the surgery. Patients in need of urgent surgery are sicker and more commonly require an open intervention. Although our results support the use of laparoscopic surgery, it is not always feasible and contribute to the selection bias in our study.

Nonetheless, our study is strengthened by the fact that it was conducted in a large single tertiary referral center that is part of the Israeli public health care system, which enables us to examine a wide array of possible risk factors, free from economical and insurance reporting considerations. These data provide convincing evidence that preoperative factors such as cognitive decline and operative technique such as laparoscopic approach are associated with patient outcomes in colon surgery, and that these risk and protective factors should be taken into consideration when managing patients undergoing colonic surgical interventions.

To conclude, readmission following colonic surgery is common; in our study we demonstrated several key risk factors correlated with readmission following surgery. Although some of these factors are nonmodifiable, others like improved perioperative glycemic control, optimization of patients' nutritional status, and optimization of patients with cirrhosis are well within the scope of preoperative planning. Our findings should aid in patient counseling and if possible, should also be addressed before surgery to reduce the likelihood of readmission.