Anastomotic leaks continue to be one of the most detrimental complications specific to gastrointestinal surgeries. [
1‐
3] Clinical manifestations of anastomotic leaks include fever, abscess, septicemia, peritonitis, and/or organ failure. This complication and its associated sequela of events following anastomotic leak result in increased length of stay (LOS), reoperation, readmission rates, an overall reduction in quality of life, and mortality [
3‐
6].
Colorectal surgeries often involve resecting a part of the diseased colon and/or the rectum and then reconnecting the remaining segments by creating a surgical anastomosis or connection [
7]. Bariatric surgeries such as laparoscopic Roux-en-Y gastric bypass (LRYGB), laparoscopic sleeve gastrectomy (LSG), and biliopancreatic diversion with duodenal switch (BPD) are highly effective procedures for weight loss, and commonly used in the USA [
5].
Unfortunately, given the complexity of these procedures, adverse events such as leaks ensue, and the reported leak rates for colorectal and bariatric surgeries vary widely, depending on the definition and the type of resection that is performed [
8]. Existing literature reports the index anastomotic leak rates (at the time of the initial procedure) for colorectal surgeries at 3.3% [
7]. Furthermore, the incidence of anastomotic leaks after bariatric surgeries ranges from 0 to 5.2% [
9].
In addition to adverse events at index (inpatient hospital stay), patients also experience postoperative readmissions due to these events. Payers have increasingly emphasized the use of readmissions as a quality and efficiency indicator. In fact, the CMS physician quality reporting system (PQRS) has included the reporting of anastomotic leak interventions that occur after colorectal and gastric bypass surgery. Hospitals have also been interested in looking at anastomotic leak reduction as an important quality improvement initiative, due to the high risk of infection [
5,
10]. In an increasingly value-based environment, quality performance benchmarks such as readmission rates are also critical to both reimbursement and quality of care.
Previous studies have found that 30-day all-cause readmissions for colorectal surgeries reportedly range from 7.5 to 14.9% [
7,
11]. A retrospective multi-institutional study found the readmission rate due to anastomotic leak following bariatric surgery to be 0.28% [
5]. The overall or all-cause readmission rate of anastomotic leaks in the bariatric population has not yet been evaluated. These findings represent a deficiency in United States-based studies that explore the economic impact of anastomotic leaks.
The current variation in incidence rates and gaps in associated economic evidence suggests that further research is necessary. This retrospective claims analysis was conducted to identify the full episode-of-care burden due to anastomotic leaks specific to both colorectal and bariatric populations. The analysis seeks to assess the incidence of bariatric and colorectal anastomotic leaks, associated costs, LOS, and readmissions using real-world longitudinal data. The purpose of this study is to provide an improved understanding into the impact of leaks on readmissions, LOS, and overall costs of anastomotic leaks due to bariatric and colorectal surgery care episodes.
Methods
Study design and data sources
The analysis examined the 30-day hospital inpatient episode-of-care costs related to anastomotic leak compared to cases without anastomotic leak. Costs in this analysis were defined as public (CMS) or private (commercial) reimbursement paid to the hospital, which can be used as a proxy for overall hospital costs given that reimbursement often considers cost. A retrospective claims analysis identified the frequency, cost, and LOS of anastomotic leaks among Medicare Fee-For-Service and commercial members that underwent bariatric and colorectal surgical procedures (
Appendix). Medicare Fee-for-Service (FFS) claims data from the Medicare Inpatient Standard Analytical File (IPSAF) compiled by the Center for Medicare and Medicaid Services (CMS) was analyzed. This data is representative of medical services provided to nearly 37 million Medicare FFS beneficiaries. Commercial claims data from the OptumInsight Inc. (Eden Prairie, Minnesota) database was also examined; this data is comprised of medical claims representing roughly 25 million members from a large US-based health plan. The sample included members who underwent bariatric and colorectal procedures between January 1, 2013 and August 31, 2015. IRB approval and informed consent were not required for this study as the data utilized were de-identified and not considered human subject research.
The outcomes of this analysis were collected at two points in time: during the index inpatient hospital stay (referred to as the index procedure) and for the 30 days after the procedure (post-index period). During this 30-day period, readmissions were evaluated and the rate of readmission was observed for both all-cause readmission(s) and readmission(s) due to anastomotic leak. This analysis included cases discharged alive for the index period admission and also those that displayed a record of continuous health plan coverage during the analysis timeframe (index + 30 days).
Study population identification
As there are no specific medical codes to identify anastomotic leaks, the occurrence of anastomotic leaks was approximated through the presence of at least one of the ICD-9 diagnosis codes pertaining to infection, peritonitis, septicemia, or abscess (see
Appendix). These cases also needed to show the presence of at least one of the following procedures that are indicative of the anastomotic leak intervention: (1) laparotomy, (2) incision of abdominal wall, and (3) requirement of drainage during the index stay. For further details on the coding, refer to
Appendix.
Statistical analysis
Data was analyzed based on the categorical or continuous nature of the data. Due to the distribution of the data, including the distribution specifically of healthcare cost data, all costs were reported as medians. An outlier analysis based on index cost was conducted for both cohorts. Statistical comparisons were performed comparing the cost, utilization, and LOS of the index stay and the inpatient readmissions within 30 days of the index event. A gamma distribution with a log link function was used to analyze cost distributions among those who accrued costs. Test differences among dichotomous and categorical variables were analyzed using Pearson’s Chi-squared test and those among continuous variables were analyzed using the Mann–Whitney U test. All statistical analyses were performed using the SAS Enterprise Guide 7.1 software (SAS Institute INC., Cary, NC).
Results
Colorectal Medicare and commercial analysis
The colorectal Medicare analysis identified 239,350 patients undergoing colorectal surgery, 41.99% were male and 60.93% of patients were between the ages of 65 and 80 (Table
1). Overall, 3.6% of cases presented with a leak on index and 1.7% presented with a leak 30-day post-discharge amounting to an anastomotic leak rate occurrence of 5.0%. A similar incidence rate of anastomotic leak occurrence was seen for the commercial population (Table
2).
Table 1
Descriptives by payer and procedure type
Commercial | Age group |
< 19 | 1.87 | 0.35 |
19–30 | 6.02 | 10.48 |
31–40 | 11.08 | 26.77 |
41–50 | 22.60 | 31.48 |
51–64 | 58.41 | 30.92 |
> 64 | 0.01 | 0.00 |
Male | 50.40 | 24.33 |
Medicare | Age group |
< 65 | 12.87 | 54.01 |
65–69 | 22.17 | 24.72 |
70–74 | 20.72 | 9.92 |
75–79 | 18.04 | 5.23 |
80–84 | 13.75 | 3.42 |
> 84 | 12.46 | 2.69 |
Male | 41.99 | 31.02 |
Table 2
Occurrence of anastomotic leaks by procedure and payer population
Colorectal commercial | 19,985 | 3.6% (721/19,985) | 1.7% (335/19,985) | 5.1% (1012/19,985) |
Colorectal Medicare | 239,350 | 3.6% (8564/239,350) | 1.7% (3964/239,350) | 5.0% (11,966/239,350) |
Bariatric commercial | 18,932 | 0.2% (39/18,932) | 0.4% (80/18,932) | 0.6% (114/18,932) |
Bariatric Medicare | 62,292 | 0.7% (444/62,292) | 1.0% (612/62,292) | 1.7% (1056/62,292) |
Medicare analysis revealed that anastomotic leaks are drivers of resource use. The index admission costs were $30,670 (or 167.5%) greater for those with a leak compared to those without a leak ($48,982 vs. $18,312;
p < 0.0001). The index LOS was 12 days longer for a patient with a leak compared to a patient without a leak (19 vs. 7 days;
p < 0.0001). The overall readmission costs were $8,755 (or 105.5%) higher for patients with a leak versus patients without a leak ($17,055 vs. $8300;
p < 0.0001). The LOS for readmission was 4 days longer for patients with a leak compared to those without (9 vs. 5 days;
p < 0.0001) (Table
3).
Table 3
Colorectal inpatient episode cost and length of stay (LOS) by payer population
Index costs ($) | $46,760 | $24,813 | < 0.0001 | $48,982 | $18,312 | < 0.0001 |
Index LOS (days) | 12 | 5 | < 0.0001 | 19 | 7 | < 0.0001 |
Readmission costs ($) | $23,319 | $12,224 | < 0.0001 | $17,055 | $8300 | < 0.0001 |
Readmission LOS (days) | 7 | 5 | < 0.0001 | 9 | 5 | < 0.0001 |
The colorectal analysis of the commercial data identified 19,985 cases undergoing colorectal surgery, 50.40% were male and 81.01% of patients were between the ages of 41 and 64 (Table
1). Trending similarly to the Medicare data, the colorectal commercial population exhibited highly significant results, with nearly double the index admission and readmission costs and an index LOS that was 7 days longer, when comparing those with a leak on versus those without a leak on index (Tables
2,
3).
Bariatric Medicare and commercial analysis
There were 62,292 patients included in the bariatric analysis of the Medicare FFS data, 31.02% were male and 39.87% of patients were between the ages of 65 and 80 (Table
1). Overall, 0.7% of cases presented with a leak at index and 1.0% presented with a leak within 30-day post-discharge leading to a combined anastomotic leak rate of 1.7% (Table
2).
Although the incidence of anastomotic leaks was lower than colorectal, the economic consequences were similar. The analysis of the Medicare cohort revealed an increase in cost for cases that had an anastomotic leak. Costs during the index admission were $30,885 (or 237.0%) higher for those with a leak compared to patients without a leak ($43,918 vs. $13,033;
p < 0.0001). The index LOS at index was 15 days longer for a patient with a leak compared to those without a leak (17 vs. 2 days;
p < 0.0001). Readmissions resulted in increased costs of $14,348 (or 183.0% higher) for those with a leak vs no leak ($22,190 vs. $7842;
p < 0.0001). The LOS for readmissions was 7 days more for patients with a leak versus those without (11 vs. 4 days;
p < 0.0001) (Table
4).
Table 4
Bariatric inpatient episode costs and length of stay (LOS) by payer population
Index costs ($) | $85,673 | $19,215 | < 0.0001 | $43,918 | $13,033 | < 0.0001 |
Index LOS (days) | 15 | 2 | < 0.0001 | 17 | 2 | < 0.0001 |
Readmission costs ($) | $17,990 | $11,690 | < 0.0001 | $22,190 | $7842 | < 0.0001 |
Readmission LOS (days) | 9 | 4 | < 0.0001 | 11 | 4 | < 0.0001 |
Congruent with the Medicare data, the commercial bariatric population (
N = 18,932) also exhibited similar incidence rates, index admission costs that were over four times higher, index LOS that was 13 days longer, and readmission costs that were 1.5 times higher, when comparing patients with a leak on index versus patients without a leak on index (Table
4). Within the commercial bariatric population, 24.33% were male and 62.40% were between the ages of 41 and 64 (Table
1).
Association with index anastomotic leak and readmissions risk
For all four anastomotic leak cohorts, the occurrence of an index anastomotic leak was associated with a higher risk of all-cause readmission and readmissions related to anastomotic leak compared to those without anastomotic leak and was highly significant for all cohorts (
p < 0.0001). This study found a 4.2% difference in anastomotic leak readmission risk for colorectal commercial cases and a difference of 12.4% risk of readmission for bariatric commercial cases in those who had an anastomotic leak at index (Table
5). Similarly, comparisons of the all-cause readmission rates for the colorectal commercial cases found a difference of 4.4% and a difference of 21.0% for the bariatric commercial cases (Table
6). This finding further confirms the need to identify anastomotic leaks at an early stage to prevent future morbidity.
Table 5
Occurrence of readmissions due to anastomotic leak by procedure and payer population
Colorectal commercial | 5.7% (41/721) | 1.5% (294/19,985) | < 0.0001 |
Colorectal Medicare | 3.7% (317/8564) | 1.52% (3647/239,350) | < 0.0001 |
Bariatric commercial | 12.8% (5/39) | 0.4% (80/18,932) | < 0.0001 |
Bariatric Medicare | 5.2% (23/444) | 0.99% (612/62,292) | < 0.0001 |
Table 6
Occurrence of all-cause readmissions by procedure and payer population
Colorectal commercial | 12.90% (93/721) | 8.5% (1689/19,985) | < 0.0001 |
Colorectal Medicare | 21.59% (1849/8564) | 14.1% (33,769/239,350) | < 0.0001 |
Bariatric commercial | 25.64% (10/39) | 4.61% (873/18,932) | < 0.0001 |
Bariatric Medicare | 29.95% (133/444) | 12.41% (7731/62,292) | < 0.0001 |
Discussion
Anastomotic leaks have been shown to have a significant impact on hospital reimbursement and cost, length of stay, readmission rates, morbidity, and mortality. A commercial payer claims analysis of patients who underwent surgery with an anastomosis showed that 30-day mortality was significantly higher in patients with an anastomotic leak versus patients without a leak, in both the colonic (29% vs. 4.2%;
p < 0.0001) and rectal groups (10% vs. 2%;
p = 0.014) [
12]. The only study based in the USA observed that patients who incur anastomotic leaks after colorectal surgery experience LOS that is 7.3 days longer and costs over $24,000 greater than patients without leaks [
6]. Unfortunately, comparable information is absent from current literature on the economic impact of anastomotic leaks after bariatric surgeries in the USA. Therefore, this is the first study of its kind to evaluate the economic impact of bariatric anastomotic leaks including the entire episode cost reflective of both the index inpatient hospital stay and readmission related costs.
The rate of anastomotic leakage after colorectal surgery and bariatric surgery are consistent with previously reported results in the literature. Furthermore, this study found that patients who develop leakage after bariatric and colorectal surgeries have significantly longer LOS and costs, both at the initial stay and within 30 days following the procedure. In fact, this colorectal analysis of the commercial population showed that during the index admission, the cost for a patient with a leak was 1.88 times greater than the cost for a patient without a leak ($46,760 vs. $24,813; p < 0.0001). Similarly, the LOS at index was 2.4 times longer for those with a leak compared to those without a leak (12 vs. 5; p < 0.0001).
The cost of readmissions was also significantly greater for both populations; the bariatric Medicare population incurred costs that were 2.83 times higher for those with a leak compared to those without a leak ($22,190 vs. $7842; p < 0.0001). It is possible that the associated costs are even higher than reported, as they do not account for patients who required a reoperation or readmission past the 30-day timeframe. For example, a patient who needed re-exploration and underwent a protective stoma will require reoperation for stoma reversal 3 months later. Furthermore, this study did not account for the cost associated with other complications such as incisional hernia or small bowel obstruction, which may require additional surgeries or readmissions. These may present much later, spanning months to years, require multiple subsequent treatments, and result in a much greater cost burden than shown in this study.
Another crucial outcome of our analysis exhibited that the rate of readmissions due to anastomotic leak was significantly higher for patients who previously had a leak on index. For example, within our colorectal commercial population, the rate of readmission due to anastomotic leak was 3.8 times greater compared to patients who did not have a leak on index. This result was consistent in both the colorectal and bariatric analyses and among the Medicare and commercial populations.
The causes of anastomotic leaks are likely multifactorial. In addition to technical and wound healing related causes, increasing evidence suggest that microbiomes may play a significant role [
13]. Given the elevated cost of anastomotic leaks, several preventative steps prior to the procedure have been discussed. Prior to the procedure, oral antibiotics, preoperative antibiotics, and bowel prep may be beneficial to reduce anastomotic leakage. Even with the routine application of the above measures, anastomotic leakage is unavoidable. In colorectal surgery, protective stoma may reduce clinically significant leakage but routine use is not practical due to an associated stigma, morbidity, and additional cost associated with stoma maintenance and reversal. This method of treatment is also not possible with bariatric surgery.
Anastomotic leaks that occur during the procedure is one of the greatest risk factors for postoperative complications. Despite this risk, the ability to detect anastomotic leaks visually during an operation is limited [
14]. Because of this, preventative approaches to reduce anastomotic leakage including techniques during surgery have also been recognized. Imaging modalities such as intraoperative endoscopy (IOE) and indocyanine green (ICG) and technologies such as intraoperative anastomotic air-testing are effective approaches to decrease the risk of anastomotic leaks in colorectal surgery. In addition, patients who underwent routine anastomotic air leak testing of left-sided colorectal anastomosis have been shown to develop a significantly lower rate of leakage [
15‐
17]. Haddad investigated the use of IOE during laparoscopic RYGB with IOE performed on patients following anastomotic closure [
9]. Use of IOE identified 3.5% air leak cases during the operative procedure leading to further assessment and reinforcement of suture line. Following IOE assisted procedure, postoperative anastomotic leaks were detected in 0.2% of cases. Intraoperative endoscopy can therefore play a fundamental role in visualization and prevention of adverse surgical outcomes [
18].
This study is certainly not without limitations, particularly those associated with a retrospective study design. ICD-9 procedure codes used to identify bariatric and colorectal procedures are imprecise in the description and do not always provide the detail regarding the type of surgical approach or location of surgery. Furthermore, since medical coding (CPT codes) is not available for directly identifying inpatient anastomotic leaks within claims data, a case identification logic was developed using relevant ICD-9 diagnostic and procedure coding. Although clinical experts have evaluated all of the coding logic, this could have impacted the study by not capturing all of the associated leaks and led to the underreporting of leak incidence rates. The design of this study is descriptive in nature and controlling for patient comorbidities was not in scope of this study but warrants consideration in further research. Nevertheless, this study contributes several important highlights of the clinical and economic impact of anastomotic leaks, a subject that is sparsely explored in the current literature.
Conclusion
The results of this study show that anastomotic leaks are associated with increased costs and LOS during the initial procedure. They are also associated with a significantly greater readmission rate, LOS, and overall costs when comparing leak versus no leak readmissions 30 days after the initial stay. Both the index and readmission measures indicate a large economic burden to the hospital facility and the payer. Moreover, these study results prove a strong correlation between the index anastomotic leak and a subsequent readmission risk for anastomotic leak. With readmission rates at the forefront of assessing quality of care and reimbursement metrics connected to readmission rates, any approaches that can help decrease readmission rates and LOS can have the dual benefit of managing hospital costs and potentially increasing reimbursement.
Prevention of leaks is critical to generate favorable economic consequences associated with the index procedure and related readmissions. Due to these reasons, hospitals should certainly consider making more prudent decisions and implementing a quality assurance program aimed at early identification and prevention of anastomotic leaks to avoid downstream impact of these occurrences. The measurement of anastomotic leak intervention in PQRS is certainly a strong indicator of the importance of leak prevention, their associated burden on hospital facilities, and the need for appropriate quality assurance methods to optimize care. In order to decrease cost, promote healthier outcomes (decrease morbidity), and increase patient satisfaction, it is vital to further explore the aforementioned measures and the ability to prevent the onset of leaks during colorectal and bariatric surgeries.
Acknowledgements
The authors would like thank Mr. Keith Needham, Manager with Baker Tilly, and Mr. Christopher Adams, MPH, Senior Consultant with Baker Tilly for their analytical support on this important piece of economic research.
Compliance with ethical standards
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.