Same-Hospital Re-Admission Rate Is Not Reliable for Measuring Post-Operative Infection-Related Re-Admission

Abstract

Background:

Post-operative infections cause morbidity, consume resources, and are an important quality measure in assessing and comparing hospitals. Commonly used metrics do not account for re-admission to a different hospital. The Nationwide Readmissions Database (NRD) tracks re-admissions across United States (US) hospitals. Infection-related re-admission across US hospitals has not been studied previously.

Patients and Methods:

The 2013 NRD was queried for admissions with a primary International Classification of Diseases and Related Health Problems, 9th revision, Clinical Modification code for the most frequently performed operations. Non-elective all-cause, infection-related, and different hospital 30-day re-admission rates were calculated, using All Patient Refined Diagnosis Related Groups codes. Multi-variable logistic regression identified risk factors for re-admission.

Results:

Of 826,836 surviving to discharge, 39,281 (4.8%) had an unplanned re-admission within 30 days, occurring at a different hospital 20.5% of the time. The most common reason for re-admission was infection (25.1%). Orthopedic and spinal procedures were at highest risk for all-cause and infection-related different hospital re-admission. Infection-related different hospital re-admission risk factors included: Length of stay >30 days (odds ratio [OR] 2.28 [1.62–3.21], p < 0.01), age ≥65 years (OR 1.56 [1.38–1.76], p < 0.01), and Charlson Comorbidity Index >1 (OR 1.14 [1.01–1.28], p < 0.01) and differed from predictors of same-hospital infectious re-admission. Non-elective surgical procedure (OR 0.79 [0.72–0.87], p < 0.01) and initial hospitalization at a large hospital (OR 0.66 [0.59–0.74], p < 0.01) were protective.

Conclusion:

A substantial proportion of post-operative re-admissions are missed by same-hospital re-admission data. All-cause and infection-related post-operative re-admissions to a different hospital are affected by unique patient and institution-specific factors. Re-admission reduction programs, quality metrics, and policy based on same hospital re-admission data should be updated to incorporate different hospital re-admission.

Agrowing interest in patient safety coupled with the rising cost of healthcare has spurred an interest in reforming the healthcare system. One focus is unplanned hospital re-admissions, particularly after surgical procedures. Unplanned re-admissions can be related to modifiable differences in care provided by hospitals and cost the American healthcare system more than $41 billion yearly [1 –3]. Thus, re-admission rate is an important quality metric.

Unplanned post-operative re-admission is a marker of post-operative complications [4 –7]. Post-operative infections are a major cause of post-operative complications, morbidity, and cost [2,5]. Hospitals with excess re-admission rates, as determined by the Centers for Medicare and Medicaid Services (CMS), have been penalized since 2012 with reduced Medicare payments [8]. More surgical diagnoses are added yearly to the Hospital Readmissions Reduction Program (HRRP) [9].

Given the influence of benchmarking on tracking patient outcomes and institutional viability, it is increasingly important to accurately capture post-operative re-admission rates, including for infection. It has not been possible previously to nationally track re-admissions across different hospitals, however. Thus, quality initiatives and benchmarks are based on studies limited to a single institution or state, often with an inability to track re-admissions across hospitals or differentiate between elective and non-elective re-admissions [10 –14]. To date, there are no national studies examining non-elective re-admission rates across different hospitals for post-operative infection.

The purpose of this study was to describe non-elective infection-related re-admission incidence across different hospitals after common operations and identify risk factors for infection-related and different-hospital re-admission. We hypothesized that infection is a significant cause of non-elective post-operative re-admission, and a significant portion of these re-admissions occurred at a different hospital.

Patients and Methods

The Nationwide Readmissions Database (NRD) is a part of the Healthcare Cost and Utilization Project (HCUP), curated by the Agency for Health Research and Quality (AHRQ). First released in 2015, the initial database included data from 22 states during 2013, totaling 14 million admissions, accounting for 49.1% of all United States (US) hospital admissions, and represents a national sample. Novel to the NRD is that each patient is assigned a unique identifier that allows for tracking across different hospitals within a state. To improve accuracy of data, the NRD combines multiple records on the same day with the same identifier. This prevents same-day transfers between hospitals or a discharge from and admission to two different hospitals on the same day from being logged as a re-admission.

The NRD captures data on patient demographics, institutional characteristics, admissions, and time between admissions. In addition, billed hospital charges are recorded. Because billed hospital charges can demonstrate wide, unexplained variability, HCUP provides a cost-to-charge ratio for each institution that normalizes charges between hospitals, allowing cost to be estimated across different hospitals [15]. The HCUP calculates this ratio yearly for all institutions based on CMS accounting reports [16]. This improves meaningful comparison of admission cost between hospitals.

The 2013 NRD was queried for all admissions with a principal International Classification of Diseases and Related Health Problem, (ICD), 9th revision, Clinical Modification code for the most frequently performed operations. Operations comprising more than 2% of operations overall in the NRD were included. Patients were excluded if they died during the index admission or if there were any missing data. Exclusion criteria also included cases where the identifier was comprised of a collapsed record involving transfer to a short-term hospital or if the patient was discharged to a short-term hospital. The former case precludes identifying the index hospital and thus the determination whether re-admission to a different hospital occurred. The latter case can occur if the patient transfer is across state lines or occurred overnight where it would appear that the index admission and transfer occurred on different days, thus preventing the NRD to collapse the two hospitalizations into a single identifier [17].

Patient demographic, institutional, admission, and re-admission data, including All Patient Refined Diagnosis-Related Groups (APR-DRGs) were collected. The Charlson Comorbidity Index (CCI) was calculated using the ICD Programs for Injury Categorization version 3.0 software within STATA/SE version 12.0 for Mac (StataCorp, College Station, TX) [18]. A composite variable of re-admission for infection was created using the following APR-DRGs: 137 (Major Respiratory Infections and Inflammation), 139 (Other Pneumonia), 248 (Major Gastrointestinal and Peritoneal Infections), 344 (Osteomyelitis, Septic Arthritis, and Other Musculoskeletal Infections), 383 (Cellulitis and Other Bacterial Skin Infections), 463 (Kidney and Urinary Tract Infections), 710 (Infectious and Parasitic Diseases Including HIV with Operative Procedure), 711 (Post-Operative, Post-Trauma, and Other Device Infections with Operative Procedure), 720 (Septicemia and Disseminated Infections), 721 (Post-Operative, Post-Traumatic, and Other Device Infections), and 724 (Other Infectious and Parasitic Diseases).

Hospital charge data were recorded. Using the unique cost-to-charge ratio for each hospital, estimated admission and re-admission cost was calculated for each patient [16]. Outcomes included 30-day: All-cause re-admission to any hospital, all-cause re-admission to a different hospital, infection-related re-admission to any hospital, and infection-related re-admission to a different hospital. For the last two outcomes, the subgroup of patients who were re-admitted within 30 days for infection, as determined by the composite APR-DRG variable, were analyzed.

Categoric variables were compared using the chi-square test. Continuous variables were reported as medians with interquartile ranges and compared using the Mann-Whitney U test. Univariable logistic regression of all categoric variables was performed. Significant variables were then selected as covariates for multi-variable logistic regression analyses and reported as odds ratios with 95% confidence intervals, relative to the most common operation. Statistical significance was defined as p < 0.05. Statistical analysis was performed using IBM SPSS Statistics version 24 (IBM Corp, Armonk, NY).

Results

There were 11 common operations identified based on inclusion criteria. Combined, these accounted for 94.4% of all operations performed during the study period. Of the 827,827 patients undergoing the most common operations, 826,386 (99.8%) survived to discharge. Of the survivors, 39,281 (4.8%) had a non-elective 30-day re-admission. Re-admission rates for the most common operations are reported in Table 1.

Table 1.

Incidence and Odds of All-Cause and Infection-Related Re-Admission to Index and Different Hospitals ^*

Procedure	n	Re-admission rate (%)	Proportion of re-admissions occurring at a different hospital (%)	Risk of re-admission to a different hospital (OR [95% CI])	p	Proportion of re-admissions because of infection (%)	Risk of re-admission because of infection (OR [95% CI])	p	Proportion of re-admissions to a different hospital because of infection (%)	Risk of re-admission to a different hospital because of infection (OR [95% CI])	p
Knee arthroplasty	176,808	2.9	26.7	—	—	20.8	—	—	20.5	—	—
Hip replacement	140,442	4.3	23.6	0.85 [0.78–0.93]	<0.01	21.8	1.53 [1.41–1.66]	<0.01	22.8	1.15 [0.94–1.39]	0.18
Spinal fusion	99,984	4.4	26.1	0.97 [0.88–1.06]	0.48	25.8	1.89 [1.74–2.05]	<0.01	23.3	1.18 [0.96–1.44]	0.12
Appendectomy	70,109	4.1	11.8	0.37 [0.32–0.42]	<0.01	41.5	2.85 [2.62–3.09]	<0.01	8.6	0.36 [0.28–0.47]	<0.01
Cholecystectomy	61,556	6.2	15.7	0.51 [0.46–0.57]	<0.01	17.3	1.78 [1.62–1.97]	<0.01	11.4	0.50 [0.38–0.66]	<0.01
Laminectomy	59,269	4.6	24.0	0.87 [0.78–0.96]	0.01	25.7	1.95 [1.77–2.14]	<0.01	21.6	1.07 [0.85–1.35]	0.58
Hip/femur reduction/fixation	57,982	7.7	22.1	0.78 [0.71–0.85]	<0.01	27.4	3.54 [3.26–3.85]	<0.01	22.3	1.11 [0.91–1.36]	0.30
Lysis of adhesions	50,778	8.0	14.1	0.45 [0.40–0.50]	<0.01	24.7	3.31 [3.04–3.61]	<0.01	10.8	0.47 [0.37–0.60]	<0.01
Leg reduction/fixation	45,579	4.4	22.3	0.79 [0.70–0.89]	<0.01	24.6	1.78 [1.60–1.98]	<0.01	18.1	0.86 [0.65–1.13]	0.27
Oophorectomy	36,345	3.7	17.7	0.59 [0.51–0.69]	<0.01	30.7	1.87 [1.67–2.10]	<0.01	14.8	0.67 [0.50–0.92]	0.01
Colectomy	28,975	8.1	11.5	0.36 [0.31–0.41]	<0.01	27.8	3.76 [3.41–4.15]	<0.01	11.1	0.48 [0.36–0.64]	<0.01

OR = odds ratio; CI = confidence interval.

Risk of readmission is reported as odds ratio compared with the most common operation performed, knee arthroplasty.

Re-admissions to a different hospital accounted for 20.5% of non-elective 30-day re-admissions. Patients experiencing a non-elective 30-day re-admission were older (65 y [51–77] vs. 61 y [48–71], p < 0.01) and had a higher CCI (1 [0–2] vs. 0 [0–1], p < 0.01). Females comprised 58.6% of the re-admitted group compared with 59.6% of the sample. Index admission length of stay (LOS) was longer in the re-admitted group (4 d [3 –6] vs. 3 d [2 –4], p < 0.01). A higher percentage of re-admitted patients were in the lowest quartile of median household income as determined by patient ZIP code (26.0% vs. 23.3%, p < 0.01) and receiving public insurance (66.2% vs. 51.3%, p < 0.01). Index admission cost was higher for patients who were re-admitted eventually ($14,494.43 [10,293.92–21,225.22] vs. $13,286.71 [9381.63–18,771.55], p < 0.01).

Compared with patients admitted to the same hospital within 30 days, patients admitted to a different hospital were older (68 y [55–78] vs. 65 y [51–77], p < 0.01). Females comprised 58.5% of the different hospital re-admission group compared with 58.6% of the larger sample. Index admission LOS was shorter in the different hospital re-admission group (3 d [2 –6] vs. 4 [3 –6], p < 0.01). There was no difference in the percentage of patients in the lowest quartile of median household income as determined by patient ZIP code between same- and different-hospital re-admissions (25.9% vs. 26.7%, p = 0.11). A higher percentage of patients re-admitted to a different hospital were receiving either Medicaid or Medicare insurance (70.6% vs. 65.0%, p < 0.01).

Index admission cost was higher for patients who were re-admitted eventually to a different hospital ($15,201.01 [10,878.23–22,503.07] vs. $14,328.96 [10,150.43–20,901.48], p < 0.01). There was no difference in re-admission cost between same- and different-hospital re-admissions ($7,616.77 [4,537.32–13,857.30] vs. $7,645.88 [4,628.36–13,945.50], p = 0.27).

Results of the multivariable regression analyses to identify predictors of re-admission, infection-related re-admission, and different-hospital re-admission are presented in Table 2. Predictors of re-admission included: LOS >30 days, CCI >1, non-elective surgical procedure, residing in a ZIP code with the lowest quartile of median household income, and initial hospitalization at a large or urban academic hospital. In contrast, non-elective operation or initial hospitalization at a large or urban academic hospital was protective against re-admission to a different hospital. Predictors of re-admission to a different hospital were different and included: LOS >30 days, public insurance, age ≥65 years, lowest quartile of median household income, and CCI >1.

Table 2.

Predictors of All–Cause, Infectious, and Different Hospital Re-Admission

Risk factor	Re-admission (OR [95% CI])	p	Different hospital re-admission (OR [95% CI])	p	Re-admission for infection (OR [95% CI])	p	Different hospital re-admission for infection (OR [95% CI])	p
Non-elective	1.74 [1.71–1.78]	<0.01	0.69 [0.66–0.73]	<0.01	1.10 [1.06–1.15]	<0.01	0.79 [0.72–0.87]	<0.01
Male	1.07 [1.05–1.09]	<0.01	1.05 [1.00–1.10]	0.04	1.03 [0.98–1.08]	0.22	1.02 [0.92–1.13]	0.68
Age ≥65 y	1.07 [1.04–1.09]	<0.01	1.17 [1.11–1.24]	<0.01	0.98 [0.93–1.04]	0.51	1.56 [1.38–1.76]	<0.01
Charlson Comorbidity Index >1	1.96 [1.92–2.01]	<0.01	1.07 [1.01–1.12]	0.02	1.05 [1.00–1.10]	0.05	1.14 [1.02–1.27]	0.02
Length of stay >30 d	2.29 [2.08–2.53]	<0.01	1.70 [1.39–2.09]	<0.01	1.36 [1.13–1.64]	<0.01	2.28 [1.62–3.21]	<0.01
Lowest median household income quartile	1.07 [1.04–1.09]	<0.01	1.10 [1.04–1.16]	<0.01	1.01 [0.96–1.06]	0.62	1.12 [1.00–1.26]	0.05
Public insurance	1.62 [1.58–1.66]	<0.01	1.18 [1.11–1.26]	<0.01	1.02 [0.96–1.07]	0.61	1.11 [0.98–1.27]	0.11
High–volume hospital	0.99 [0.96–1.01]	0.25	0.94 [0.89–1.00]	0.04	0.93 [0.88–0.98]	<0.01	0.94 [0.83–1.07]	0.35
Large hospital	1.09 [1.07–1.12]	<0.01	0.67 [0.63–0.70]	<0.01	1.08 [1.03–1.13]	<0.01	0.66 [0.59–0.74]	<0.01
Public hospital	0.99 [0.96–1.02]	0.51	1.00 [0.93–1.08]	0.98	1.06 [0.99–1.13]	0.08	0.89 [0.76–1.04]	0.14
Urban academic medical center	1.09 [1.07–1.11]	<0.01	0.88 [0.83–0.92]	<0.01	1.09 [1.04–1.15]	<0.01	0.94 [0.84–1.06]	0.30

OR = odds ratio; CI = confidence interval.

Risk factors for re-admission for infection included: LOS >30 days, non-elective operation, or initial hospitalization at a large or urban academic hospital. On the other hand, non-elective surgery or initial hospitalization at a large hospital was actually protective against re-admission to a different hospital for a post-operative infection. Risk factors for re-admission for infection to a different hospital included: LOS >30 days, age ≥65 years, and CCI >1.

Discussion

This is the first study to examine post-operative infection re-admission patterns and risk factors across different hospitals nationally. Overall re-admission rates for the top five operations ranged from 2.9% to 6.2%, slightly less than reported previously [12,13,19 –21]. Previously unreported, one in five non-elective 30-day re-admissions for post-operative infections occur at a hospital different from index admission and are missed by current data collection efforts. The incidence rises to one in four for the three most common operations performed. Knee arthroplasty, the most common operation performed in this dataset, also has the highest different hospital admission incidence at nearly 27%. In fact, spinal and orthopedic re-admissions were far more likely to occur at different hospital than intra-abdominal procedures.

CMS and the Hospital Quality Alliance publicly report 30-day re-admission rates for hip and knee procedures under the Hospital Quality Initiative [22]. The HRRP tracks re-admissions after knee arthroplasty as a quality indicator and penalizes hospitals with excess re-admission with reduced payment. Thus, this has significant implications for institution-specific quality metrics and national benchmarking. High rates of re-admission after knee arthroplasty are effectively hidden by occurring at different hospitals and are not tracked by current benchmarking. Therefore, re-admission rates used to determine hospital payment are inaccurate and thus benchmarking and policy based on readmission quality metrics are also inaccurate.

Infection is the most common cause of post-operative non-elective 30-day re-admission, with non-elective procedures with higher likelihood of bacterial contamination, such as appendectomy, having the highest rates. The operations at highest risk for different hospital post-operative infection re-admission, however, were also the operations at the highest risk for different hospital re-admission (spinal and orthopedic procedures), not the procedures at highest risk for post-operative infection re-admission overall, such as intra-abdominal procedures. This suggests that a substantial proportion of post-operative infections are missed currently and that estimates of the post-operative infection rate based on same hospital re-admission for infection is not a reliable measure. Given the importance of post-operative infection as a national outcomes benchmark to track and compare hospital quality, such as in the American College of Surgeons National Surgical Quality Improvement Program (ACS-NSQIP), the inaccuracy of post-operative infection rates that only account for same hospital re-admission is troublesome [23].

Risk factors differed for re-admission overall compared with re-admission to a different hospital. The person experiencing re-admission was more likely to be an older, poorer, sicker, publicly insured patient undergoing a non-elective procedure necessitating a longer hospitalization at a large or urban academic hospital. Indeed, these patient and hospital characteristics have been shown previously to increase re-admission [10, 24 –33]. While several patient and LOS characteristics continued to be risk factors, non-elective operation was actually protective against re-admission to a different hospital. Although evidence-based understanding is lacking, it is possible that urgent procedures of acute disease processes are more complex, and patients seek continuity of care. Indeed, fragmentation of care, particularly after complex hospitalizations, worsens outcomes and reduces quality [34]. The risk profile for hospital characteristics also differed, with index hospitalization at a large or urban academic hospital predicting against re-admission to a different hospital. Again, there is a paucity of data, but it is possible that these hospitals are taking care of more complex cases where continuity is deemed important to the patient, that a viable alternative hospital does not exist nearby, or patients are less able to access a different institution when a complication arises.

Predictors of re-admission for post-operative infection also differed between hospitalization at index and different hospitals. The patient most likely to be re-admitted for a post-operative infection was someone who underwent a non-elective procedure necessitating a prolonged LOS at either a large or urban academic hospital. The most likely patient to experience a post-operative infectious re-admission to a different hospital, however, was an elderly patient with comorbidities having a long LOS after an elective surgical procedure at a hospital that did not quality as either large or urban academic. In other words, infections in patients with risk factors for greater complications and cost of care, such as elderly patients with comorbidities and previous long hospitalizations, are more likely to be missed when using only same-hospital re-admission data [10,35]. As state previously, care fragmentation in these cases has a detrimental effect on quality and outcomes [34].

While patient and peri-operative characteristics have been well correlated to increased risk of infection, hospital characteristics have been less well studied [10]. In addition, given the different risk factors between post-operative infectious re-admission and post-operative infectious re-admission to a different hospital, interventions designed to prevent re-admission that are based on only same-hospital data will not necessarily be appropriate for preventing those patients who are at higher risk for different hospital re-admission.

Strengths of this study include that it utilizes national records that track re-admissions across different hospitals. For single-state databases that track re-admissions across hospitals, the ability to distinguish elective from non-elective re-admission is often not possible [10, 36]. The NRD, however, is able to distinguish elective from non-elective index admission and re-admission, improving accuracy of results. For studies using national datasets such as the ACS-NSQIP, only same-hospital re-admission is captured [11,12,14]. The NRD offers for the first time a more complete picture of re-admission patterns and risk factors after common operations.

As with any administrative database, conclusions drawn from the NRD have limitations. First, the NRD relies on hospital reporting and as such is subject to related errors. Second, as a sample of discharge records, the NRD does not record physiologic parameters that are important for adjusting risk to accurately compare outcomes. Other databases, such ACS-NSQIP, contain physical and biochemical characteristics of patients but lack the ability to track re-admissions across different hospitals, missing out on a significant and unique cohort as demonstrated by this study. The ideal database would be able to track physiologic variables in patients with an identifier that follows the patient across hospitals. Development of a national database that includes patient- and hospital-specific risk factors as well as hospitalizations across different hospitals should be of interest to clinicians, hospitals, and policymakers. Third, while patients underwent a surgical procedure, the primary admitting service is unknown. Whether being admitted to a medical service with a consultation from a surgical service or being admitted to a surgical service leads to different outcomes is unknown. Thus, admitting service could be a missed confounder.

Fourth, because NRD does not track geographic information, no analysis of geographic variance of patterns and risk factors based on state or city population size could be performed. Nevertheless, the sample is representative of the entire country including all regions and population distributions from urban to rural settings. Finally, unique to the NRD, re-admissions are not tracked across years or states. Of note, HCUP does track re-admissions of Medicare patients across state lines. Including interstate re-admissions resulted in a 1.9%–3.8% increase in 30-day re-admission rates [17]. In this study, Medicare patients were the largest group, accounting for 43.5% of the sample, so the effect of excluding interstate re-admissions is also likely minimal.

Future areas of research include further study of risk factors including clinical data as well as more demographic variables such as race and socioeconomic status, which have been associated previously with re-admission [37]. To date, all studies, including this one, only provide a partial understanding of why re-admissions occur. A more comprehensive understanding can allow more effective interventions to improve care quality. For example, if the data contained within ACS-NSQIP could be associated with a unique patient identifier that tracked patients across hospital re-admissions, findings of any study would be far more robust than findings based on NRD or ACS-NSQIP alone. As predictors of re-admission to either index or different hospital are further refined, efforts at prevention through hospital protocols and patient education should be implemented and studied.

While the ideal national dataset is lacking, it follows from statewide collaborations reducing postoperative complications that national standardization of data collection that tracks patients in hospitalizations across the country through time and subsequent refinement of benchmarks will allow for wider local implementation and lead to reductions in re-admissions as well as improvement in overall outcomes and quality [38]. Finally, the possible effect of the unique subgroup experiencing different hospital non-elective re-admission on quality benchmarking and costs requires further study and suggests policy reformation is needed.

Conclusions

A substantial proportion of post-operative re-admissions are missed by same-hospital re-admission data. All-cause and infection-related post-operative re-admissions to a different hospital are affected by unique patient and institution-specific factors. Re-admission reduction programs, quality metrics, and policy developed using only same hospital re-admission data should be updated to incorporate different hospital re-admission.

Footnotes

Author Disclosure Statement

No competing financial interests exist.

References

Benbassat

, Taragin

. Hospital readmissions as a measure of quality of health care: Advantages and limitations. Arch Intern Med, 2000; 160:1074–1081.

Vest

, Gamm

, Oxford

, et al. Determinants of preventable readmissions in the United States: A systematic review. Implement Sci, 2010; 5:88.

Hines

, Barrett

, Jiang

, et al. Conditions with the largest number of adult hospital readmissions by payer, 2011: Statistical Brief #172. Healthcare Cost and Utilization Project (HCUP) Statistical Briefs. Rockville, MD: 2006.

Muthuvel

, Tevis

, Liepert

, et al. A composite index for predicting readmission following emergency general surgery. J Trauma Acute Care Surg, 2014; 76:1467–1472.

Lawson

, Hall

, Louie

, et al. Association between occurrence of a postoperative complication and readmission: Implications for quality improvement and cost savings. Ann Surg, 2013; 258:10–18.

Kwaan

, Vogler

, Sun

, et al. Readmission after colorectal surgery is related to preoperative clinical conditions and major complications. Dis Colon Rectum, 2013; 56:1087–1092.

Manilich

, Vogel

, Kiran

, et al. Key factors associated with postoperative complications in patients undergoing colorectal surgery. Dis Colon Rectum, 2013; 56:64–71.

Marks

, Loehrer

, McCarthy

. Hospital readmissions: Measuring for improvement, accountability, and patients. Issue Brief (Commonw Fund), 2013; 24:1–8.

Hospital Readmissions Reduction Program: Overview: QualityNet; 2016. www.qualitynet.org/dcs/ContentServer?c=Page&pagename=QnetPublic%2FPage%2FQnetTier2&cid=1228772412458. (Last accessed October 6, 2017).

10.

Havens

, Olufajo

, Cooper

, et al. Defining rates and risk factors for readmissions following emergency general surgery. JAMA Surg, 2016; 151:330–336.

11.

Glebova

, Bronsert

, Hammermeister

, et al. Drivers of readmissions in vascular surgery patients. J Vasc Surg, 2016; 64:185-194 e3.

12.

Moghadamyeghaneh

, Hwang

, Hanna

, et al. Unplanned readmission after appendectomy. Am J Surg, 2016; 212:493–500.

13.

Awolaran

, Gana

, Samuel

, Oalkhinan

Readmissions after laparoscopic cholecystectomy in a UK District General Hospital. Surg Endosc, 2016.

14.

Kurtz

, Lau

, Ong

, et al. Hospital, patient, and clinical factors influence 30- and 90-day readmission after primary total hip arthroplasty. J Arthroplasty, 2016; 31:2130–2138.

15.

Park

, Kim

, Werner

. Inpatient hospital charge variability of U.S. hospitals. J Gen Intern Med, 2015; 30:1627–1632.

16.

Cost-to-charge ratio files: 2013 Nationwide Readmissions Database (NRD) User Guide: Rockville, MD: Agency for Healthcare Research and Quality, 2015. https://www.hcup-us.ahrq.gov/db/state/CCR-NRD-UserGuide_2010-2014.pdf. (Last accessed October 6, 2017).

17.

Faul

, Sasser

, Lairet

, et al. Trauma center staffing, infrastructure, and patient characteristics that influence trauma center need. West J Emerg Med, 2015; 16:98–106.

18.

Clark

, Osler

, Hahn

. ICDPIC: Stata module to provide methods for translating International Classification of Diseases (Ninth Revision) diagnosis codes into standard injury categories and/or scores: Boston College, 2010. www:ideas.repec.org/c/boc/bocode/s457028.html. (Last accessed September 27, 2017).

19.

Rana

, Bhullar

, Subhas

, et al. Thirty-day readmissions after inpatient laparoscopic cholecystectomy: Factors and outcomes. Am J Surg, 2016; 211:626–630.

20.

Varacallo

, Herzog

, Toossi

, Johanson

. Ten-year trends and independent risk factors for unplanned readmission following elective total joint arthroplasty at a large urban academic hospital. J Arthroplasty, 2017; 32:1739–1746.

21.

Welsh

, Graham

, Karmarkar

, et al. Effects of postacute settings on readmission rates and reasons for readmission following total knee arthroplasty. J Am Med Dir Assoc, 2017; 18:367.

22.

Hospital Quality Initative Outcome Meausures. Baltimore, MD: Centers for Medicare & Medicaid Services, 2015. www.cms.gov/Medicare/Quality-Initiatives-Patient-Assessment-Instruments/HospitalQualityInits/OutcomeMeasures.html. (Last accessed September 27, 2017).

23.

About ACS NSQIP: American College of Surgeons, 2017. www.facs.org/quality-programs/acs-nsqip/about. (Last accessed September 27, 2017).

24.

Shafi

, Aboutanos

, Agarwal

Jr. , et al. Emergency general surgery: Definition and estimated burden of disease. J Trauma Acute Care Surg, 2013; 74:1092–1097.

25.

Schwartz

, Hui

, Schneider

, et al. Worse outcomes among uninsured general surgery patients: Does the need for an emergency operation explain these disparities?. Surgery, 2014; 156:345–351.

26.

Tsai

, Joynt

, Orav

, et al. Variation in surgical-readmission rates and quality of hospital care. N Engl J Med, 2013; 369:1134–1142.

27.

Tsai

, Orav

, Joynt

. Disparities in surgical 30-day readmission rates for Medicare beneficiaries by race and site of care. Ann Surg, 2014; 259:1086–1090.

28.

Merkow

, Ju

, Chung

, et al. Underlying reasons associated with hospital readmission following surgery in the United States. JAMA, 2015; 313:483–495.

29.

Havens

, Peetz

, Do

, et al. The excess morbidity and mortality of emergency general surgery. J Trauma Acute Care Surg, 2015; 78:306–311.

30.

Glance

, Kellermann

, Osler

, et al. Hospital readmission after noncardiac surgery: The role of major complications. JAMA Surg, 2014; 149:439–445.

31.

Hendren

, Morris

, Zhang

, Dimick

. Early discharge and hospital readmission after colectomy for cancer. Dis Colon Rectum, 2011; 54:1362–1367.

32.

Dawes

, Sacks

, Russell

, et al. Preventable readmissions to surgical services: Lessons learned and targets for improvement. J Am Coll Surg, 2014; 219:382–389.

33.

Schneider

, Hyder

, Brooke

, et al. Patient readmission and mortality after colorectal surgery for colon cancer: Impact of length of stay relative to other clinical factors. J Am Coll Surg, 2012; 214:390–398.

34.

Tsai

, Orav

, Jha

. Care fragmentation in the postdischarge period: Surgical readmissions, distance of travel, and postoperative mortality. JAMA Surg, 2015; 150:59–64.

35.

Gonzalez

, Shih

, Dimick

, Ghaferi

. Using same-hospital readmission rates to estimate all-hospital readmission rates. J Am Coll Surg, 2014; 219:656–663.

36.

Metcalfe

, Olufajo

, Zogg

, et al. Unplanned 30-day readmissions in orthopaedic trauma. Injury, 2016; 47:1794–1797.

37.

Rathore

, Foody

, Wang

, et al. Race, quality of care, and outcomes of elderly patients hospitalized with heart failure. JAMA, 2003; 289:2517–2524.

38.

Campbell

Jr ., Englesbe

, Kubus

, et al. Accelerating the pace of surgical quality improvement: The power of hospital collaboration. Arch Surg, 2010; 145:985–991.