- Email: [email protected]
Predictors of 30-day readmissions after gastrectomy for malignancy
j o u r n a l o f s u r g i c a l r e s e a r c h a p r i l 2 0 1 8 ( 2 2 4 ) 1 7 6 e1 8 4
Available online at www.sciencedirect.com
ScienceDirect journal homepage: www.JournalofSurgicalResearch.com
Predictors of 30-d readmissions after gastrectomy for malignancy John B. Ammori, MD,a,* Suparna Navale, MS, MPH,b Nicholas Schiltz, PhD,b and Siran M. Koroukian, PhDb a
Department of Surgery, University Hospitals Seidman Cancer Center and Case Comprehensive Cancer Center, Cleveland, Ohio b Department of Epidemiology and Biostatistics, School of Medicine, Case Western Reserve University, Cleveland, Ohio
article info
abstract
Article history:
Background: The objective of this study is to identify risk factors associated with read-
Received 27 June 2017
mission after gastrectomy to potentially identify potential areas for targeted improve-
Received in revised form
ments. Hospital readmission after surgery is a topic of interest in health-care policy among
3 November 2017
hospitals, payers, and providers. Readmissions are associated with increased costs,
Accepted 5 December 2017
morbidity, and mortality. Readmission rates have been proposed as a quality metric for
Available online xxx
hospitals and quality indicator of individual surgeon’s performance. In addition, the Centers for Medicare and Medicaid Services has reduced payments to hospitals with
Keywords:
excessive readmissions for certain diagnoses.
Readmission
Materials and methods: All gastrectomy procedures for malignancy in patients aged 18 y
Gastrectomy
from 2005 to 2011 were queried from the California State Inpatient Database. Patients who
Gastric cancer
died during index admission were excluded. Descriptive statistics were examined between all baseline variables and readmission status. Logistic regression models were adjusted for age, race, sex, and insurance status. Results: A total of 6985 patients underwent gastrectomy for malignancy; 16.5% of the patients were readmitted after postoperative discharge. Readmission rate did not change significantly over time. Multivariable analysis demonstrated that the occurrence of any postoperative complications, postoperative length of stay greater than 10 d, discharge to skilled nursing facility or home health care, combined resection with distal pancreatectomy and/or splenectomy, and patient comorbidities like diabetes mellitus and renal failure were independently associated with readmissions. Conclusions: The findings suggest that focusing on quality improvement efforts by targeting reduction of postoperative complications may reduce readmission rates. ª 2017 Elsevier Inc. All rights reserved.
Introduction Reducing hospital readmissions has become a focus in the realm of health-care policy and inpatient care quality improvement. In Medicare patient population, approximately
16% of surgical patients and 21% of medical patients had an unplanned readmission in 2004, accounting for $17 billion in excess health-care costs.1,2 On average, the cost per readmission for any cause is $11,200.3 The Affordable Care Act requires the establishment of a Hospital Readmission
* Corresponding author. Department of Surgery, University Hospitals Seidman Cancer Center and Case Comprehensive Cancer Center, 11100 Euclid Avenue, LKS 7001, Cleveland, OH 44106. Tel.: þ1 216 844 1777; fax: þ1 216 286 3294. E-mail address: [email protected] (J.B. Ammori). 0022-4804/$ e see front matter ª 2017 Elsevier Inc. All rights reserved. https://doi.org/10.1016/j.jss.2017.12.004
ammori et al readmission after gastrectomy
Reduction Program. The purpose of the program is to improve quality and lower costs while attempting to ensure that patients are fully prepared and safe for continued care outside the hospital setting at the time of discharge. The program is designed to penalize hospitals with excessive 30d readmissions by withholding 1% of Medicare reimbursement starting from the fiscal year 2013 and increasing the penalty to 3% for the fiscal year 2015. The American Cancer Society estimates over 26,000 new stomach cancer diagnoses and more than 10,000 deaths in the Unites States in 2016.4 Gastrectomy provides the only chance for cure. Gastrectomy for malignancy is a complex surgical procedure, and readmission rates after this procedure have been estimated to be up to 20%.5-13 Patient and hospital-level factors have been found to be associated with readmissions. The goal of this study is to identify factors associated with readmissions after gastrectomy for malignancy, thus informing future interventions to improve care, reduce excess readmissions, and curtail health-care costs.
Materials and methods The present study was deemed exempt from approval by our hospital’s Institutional Review Board of Case Western Reserve University.
177
as any unplanned (or unscheduled) admission to a hospital within 30 d of discharge from the index admission. Postoperative complications were identified using specific definitions (Appendix). In addition, “any complication” was defined as one or more of these postoperative complications.
Independent variables The independent variables used in the statistical models are as follows: (1) patient demographics such as age, sex, race, and insurance; (2) discharge status; (3) hospital teaching status; and (4) patient comorbidities. The variables were defined as those recorded by the HCUP with a few category modifications to facilitate model specification. Age was grouped as 18-49, 5059, 51-60, 60-69, 70-79, and 80 y. Race was categorized as white, black, Hispanic, and other. Insurance status was identified as Medicare, Medicaid, private/health maintenance organization, uninsured/self-pay, and other. Discharge status was classified as routine, skilled nursing facility, home health care, and other. Comorbidities were identified and classified using the Comorbidity Software provided by the Agency for Healthcare Research and Quality.14 LOS was grouped into three categories based on the median LOS of 10 d, as 1-10, 1120, and 21 d. In addition, blood transfusion was identified based on ICD-9 procedure codes 99.00, 99.02, 99.03, and 99.04.
Statistical analysis Data source This study represents a retrospective cross-sectional secondary data analysis of the California State Inpatient Database (CA SID) file. State Inpatient Databases (SIDs) are state-specific files that contain all inpatient care records of respective participating states; they were developed as a part of the Healthcare Cost and Utilization Project (HCUP). The SIDs encompass 97% of all US hospital discharges. Over 100 clinical and nonclinical data elements are available, including demographics, hospital characteristics, primary and secondary diagnoses, inpatient procedures, insurance type, discharge status, length of stay (LOS), and total charges, among others. The CA SID is one of the few SIDs that also allows for tracking hospital readmissions, given that it includes a synthetic patient identifier and a variable that reflects the lapse of time between two hospital admissions.
Study population Cases that were not assigned a Clinical Classification Software code for neoplasm (Clinical Classification Software 11-47; n ¼ 29,531) or did not have a diagnosis of malignant neoplasm of the stomach (ICD-9 diagnosis code 151; n ¼ 35,430) were excluded. In addition, admissions for individuals under the age of 18 y or those who died during the index admission were excluded from analysis, leaving the study population at 6985 individuals.
Outcomes of interest The outcomes of interest were 30-d readmissions and postoperative complications. Thirty-day readmission was defined
All analyses were performed using SAS System for Windows, version 9.3 (SAS Institute Inc, Cary, NC). We calculated the distribution of data and descriptive statistics for all baseline variables included in the analyses for all gastrectomy patients and those who were readmitted. Median LOS for readmitted and not readmitted patients was also calculated for each baseline variable. Annual trend in gastrectomy procedures and readmission rates were also examined. Univariate comparisons were made using Pearson’s chi-squared test for categorical variables. We calculated the complication rates and comorbidity rates by using readmission and discharge status, as well as the comorbidity rates of patients with any complication by using readmission status. We compared the discharge status after index admission based on index LOS and the complication rates based on index LOS and surgery type. Logistic regression models, adjusted for age, race, sex, and insurance, were fit to examine the relationship between readmission and a combination of factors (LOS, discharge status, any complications, blood transfusion, and comorbidities). Finally, we identified the causes for readmission based on the time to readmission, categorized as 1-7, 8-14, and 15 d.
Results A total of 6985 patients who underwent gastrectomy for malignancy are included in this analysis. Of these patients, 1152 (16.5%) required readmission. Median postoperative LOS for all patients was 10 d. The median LOS after readmission was 5 d. Also, 1972 patients (28.2%) underwent total gastrectomy, 2511 patients (35.95%) suffered a postoperative complication,
178
j o u r n a l o f s u r g i c a l r e s e a r c h a p r i l 2 0 1 8 ( 2 2 4 ) 1 7 6 e1 8 4
and 4083 patients (58.5%) were discharged routinely to home, while 1791 (25.6%) and 1039 (14.9%) patients were discharged with home health care or to a skilled nursing facility, respectively. Readmissions were to the index hospital in 81.1% of patients and to a nonindex hospital in 18.9% of patients. The association of readmissions with numerous factors was examined, including patient demographics, type of gastrectomy, LOS from the index admission, discharge status, and postoperative complications (Table 1). Following are the notable comparisons. Total gastrectomy was associated with a higher readmission rate compared to partial gastrectomy (18.4% versus 15.7%, P ¼ 0.007). Combined organ resection with the addition of distal pancreatectomy and/or splenectomy to gastrectomy was significantly associated with readmission (24.8% versus 16%, P < 0.001). Women were less likely to be readmitted compared to men (15.8% versus 17.1%, P ¼ 0.03). Using the median LOS of 10 d as a cutoff, those with an index LOS of 10 d were readmitted at a rate of 12.9% compared to a 25.2% readmission rate for those with an index readmission LOS of more than 20 d, P < 0.001. Discharge status after surgical recovery was significantly associated with readmission, with a readmission rate of 24.6% in patients discharged to a skilled nursing facility, 18.9% in those discharged with home health care, and 13.4% in those discharged home, P < 0.001. Transfusion of blood in the perioperative period was associated with increased readmission rate (18.6% versus 15.4%, P < 0.001). Patients who suffered a postoperative complication had a higher rate of readmission, 21% versus 14%, P < 0.001. Several individual comorbidities were also associated with readmissions, including chronic lung disease, diabetes mellitus, hypertension, fluid and electrolyte imbalance, renal failure, and weight loss. There was a trend toward a reduced rate of readmissions over time from 17.6% in 2005 down to 13% in 2011, but this did not reach statistical significance, P ¼ 0.074 (data not shown). Rates of readmission were not different between teaching and nonteaching hospitals (data not shown). A multivariable logistic regression was performed to identify correlates of readmission after gastrectomy. Numerous factors were included in the regression model, including any comorbidity with a prevalence of >10%. Independent predictors of readmission, including initial LOS >10 d, discharge to skilled nursing facility or home health care, the occurrence of any postoperative complications, combined organ resection, and patient comorbidities of diabetes mellitus and renal failure, were also included in the model. Independent predictors were significantly associated with readmission based on our model that included initial LOS >10 d, discharge to skilled nursing facility or home health care, and the occurrence of any postoperative complications. Table 2 shows the results of the model. We analyzed the time elapsed from discharge to readmission and reasons for readmission. Over half of the readmissions occurred in the first 9 d after discharge (Fig.). The primary cause for readmission was complication of surgery (Table 3). The most common complications associated with readmission were postoperative infection which occurred in 40% of those readmitted for a complication, digestive system complication in 28.39%, and hemorrhage in 4.19%. Other relatively frequent causes of readmission included
septicemia, intestinal obstruction, and fluid and electrolyte abnormities. Fluid and electrolyte abnormalities were more commonly seen as a reason for readmission in those patients readmitted at 2 wk after discharge. In addition, we examined some additional associations of interest based on the analyses described previously. Extent of gastrectomy (partial versus total) was associated with a significantly higher rate of complications, with a rate of 41.58% after total gastrectomy compared to 33.73% after partial gastrectomy, P < 0.001. However, extent of gastrectomy by itself was not independently associated with readmission in the multivariate model. Combined resection with the addition of distal pancreatectomy and/or splenectomy was not only independently associated with readmission but was also associated with higher rate of complication than gastrectomy alone (52.7% versus 35%, P < 0.001). Complications were associated with LOS. Compared to LOS of 10 d for the index admission, those with a LOS of >20 d suffered significantly higher rates of complications (22.3% versus 70.1%, respectively, P < 0.001). LOS for the index admission was also associated with discharge status. Patients with an LOS 10 d were routinely discharged 72.0% of the time compared to 30% routine discharge for those with an LOS >20 d, P < 0.001. Only 8.15% of patients with LOS 10 d for the index admission were discharged to skilled nursing facility compared to 33.3% of those with LOS >20 d, P < 0.001. Despite these associations in univariable analysis, the independent predictors of readmission were initial LOS >10 d, discharge to skilled nursing facility or home health care, and the occurrence of any postoperative complications.
Discussion There is significant attention directed at reducing hospital readmissions in the United States as an effort to reduce waste and inefficiency in health care. Postoperative readmission rate has been proposed as a quality metric for hospitals and quality indicator of individual surgeon’s performance.15-17 This is the fourth year of the Hospital Readmission Reduction Program, which was created under the Affordable Care Act. In this year, 2592 hospitals will receive reduced Medicare payments ranging from 0.01% to 3% for excessive readmissions.18 These penalties are projected to cost these hospitals $420 million from October 1, 2015, through September 30, 2016.19 In an effort to not only improve quality and efficiency but also to avoid financial penalties, it is important to understand the variables that are associated with readmissions. This understanding may allow for targeted quality improvement efforts of modifiable factors. Given the paucity of data regarding readmissions after gastric malignancy surgery, this study contributes to the literature by identifying factors associated with our outcome of interest. Using the HCUP data from California, we documented a readmission rate of 16.5% after gastrectomy for gastric malignancy, with over half of the readmissions occurring within 10 d of discharge. The readmission rate in this study slightly exceeds the range of 7.5%-15.4% reported in previous studies examining readmissions for the same procedure.5,10,13,20 A study of nearly 9000 patients using data linked from the Office
179
ammori et al readmission after gastrectomy
Table 1 e Patient characteristics. All gastrectomy patients Variable Total
Table 1 e (continued )
N (%)
N (%)
6985
1152
Total gastrectomy
1972 (28.23)
363 (18.41)
Partial gastrectomy
5013 (71.77)
789 (15.74)
69
70
Age <50 y
P valuey
Variable Alcohol abuse
<0.01
Surgery type
Age (median), y
All gastrectomy patients
Readmitted
N (%)
P valuey <0.01
203 (2.91)
50 (24.63)
2206 (31.58)
374 (16.95)
0.48
410 (5.87)
101 (24.63)
<0.01
1042 (14.92)
203 (19.48)
<0.01
249 (3.56)
62 (24.90)
<0.01
Diabetes mellitusuncomplicated
1389 (19.89)
265 (19.08)
<0.01
219 (3.14)
48 (21.92)
0.03
Deficiency anemia Congestive heart failure Chronic lung disease
0.07
N (%)
Readmitted
Coagulopathy
748 (10.71)
119 (15.91)
50-59 y
1113 (15.93)
163 (14.65)
60-69 y
1780 (25.48)
275 (15.45)
Diabetes mellituscomplicated
70-79 y
2081 (29.79)
366 (17.59)
Hypertension
3737 (53.50)
662 (17.71)
<0.01
80þ y
1263 (18.08)
229 (18.13)
Fluid and electrolyte imbalance
1548 (22.16)
329 (21.25)
<0.01
472 (6.76)
70 (14.83)
0.31
Pulmonary circulation disorders
95 (1.36)
24 (25.26)
0.02
Renal failure
384 (5.50)
116 (30.21)
<0.01
Valvular disease
278 (3.98)
61 (21.94)
0.01
Weight loss
965 (13.82)
185 (19.17)
0.02
<0.01
Race White Black
2796 (40.03)
495 (17.70)
403 (5.77)
68 (16.87)
Hispanic
1654 (23.68)
295 (17.84)
Other
1760 (25.20)
248 (14.09)
372 (5.33)
46 (12.37)
Missing Sex
0.03
Male
4220 (60.42)
721 (17.09)
Female
2691 (38.53)
426 (15.83)
Missing
74 (1.06)
Index admission LOS (median), d
11.5
1-10 d
3977 (56.94)
137 (3.44)
11-20 d
2097 (30.02)
196 (9.35)
910 (13.03)
178 (19.56)
<0.01
Index admission LOS
<0.01
Discharge status after index admission Routine
4083 (58.45)
545 (13.35)
Skilled nursing facility
1039 (14.87)
256 (24.64)
Home health care
1791 (25.64)
339 (18.93)
72 (1.03)
12 (16.67)
Other
* Cells with frequency <11 are suppressed due to requirements in the HCUP data use agreement. y P value of readmitted versus not readmitted participants.
*
10
21þ d
Obesity
<0.01
Combined organ resection (pancreas or spleen) Yes
391 (5.6)
No
6594 (94.4)
97 (24.8) 1055 (16) <0.01
Received blood transfusion Yes
2406 (34.45)
448 (18.62)
No
4579 (65.55)
704 (15.37)
Yes
2511 (35.95)
527 (20.99)
No
4474 (64.05)
625 (13.97)
<0.01
Complications
Comorbidities (continued)
of Statewide Health Planning and Development database and the California Cancer Registry showed a readmission rate of 15.42% at 30 d postoperatively, which increased to 28.8% at 90 d after surgery. Interestingly, this study also showed a worsened overall 5-y survival rate associated with postoperative readmissions.20 A single-institution study from Massachusetts General Hospital reported a 30-d readmission rate of 14.6% in 418 studied patients, with an increased rate of readmissions after total gastrectomy compared to that after partial gastrectomy, 23% versus 13%, respectively.5 Data from the American College of Surgeons National Surgical Quality Improvement Program (NSQIP) showed a 13.5% 30d unplanned readmission rate in 1400 patients who underwent esophagectomy or gastrectomy. These rates of readmission are similar to those of the present study but higher than those reported from Asia. For example, a study from Korea showed a 7.5% readmission rate after radical subtotal gastrectomy in 530 gastric cancer patients.10 Similarly, a study of 376 gastric cancer patients in China demonstrated a 30d readmission rate of 7.2%.13 The differences in readmissions between Asian and North American studies may be related to patient factors and cultural differences. For instance, in the Chinese study, the median body mass index was 22, the median LOS was 13 d, 82% of patients were discharged home without services, and no patients were discharged to rehabilitation facilities.13
180
j o u r n a l o f s u r g i c a l r e s e a r c h a p r i l 2 0 1 8 ( 2 2 4 ) 1 7 6 e1 8 4
Table 2 e Multivariate logistic regression model of predictors for readmission after gastrectomy. Variable
Odds ratio (95% CI)
P value
Age <50
Reference
50-59
0.84 (0.64-1.09)
0.18
60-69
0.83 (0.64-1.07)
0.15
70-79
0.90 (0.68-1.20)
0.49
80þ
0.84 (0.61-1.14)
0.26
Race White
Reference
Black
0.92 (0.69-1.22)
0.54
Hispanic
1.08 (0.91-1.28)
0.37
Other
0.83 (0.70-0.98)
0.03
Missing
0.85 (0.60-1.21)
0.36
Sex Male
Reference
Female
0.95 (0.83-1.09)
0.49
Missing
0.57 (0.22-1.53)
0.27
Medicare
0.97 (0.80-1.19)
0.80
Medicaid
0.98 (0.77-1.24)
0.85
Insurance
Private insurance
Reference
Self-pay
0.92 (0.46-1.82)
0.80
Other
1.76 (1.11-2.79)
0.02
Index admission LOS 1-10 d
Reference
11-20 d
1.36 (1.16-1.59)
<0.01
21þ d
1.61 (1.31-1.98)
<0.01
Discharge status Routine
Reference
Skilled nursing facility
1.62 (1.34-1.97)
<0.01
Home health care
1.29 (1.10-1.51)
<0.01
Other
0.97 (0.51-1.83)
0.91
1.48 (1.16-1.90)
<0.01
1.23 (1.07-1.42)
<0.01
Deficiency anemia
0.90 (0.78-1.04)
0.14
Chronic lung disease
1.08 (0.91-1.29)
0.37
Diabetes mellitus-uncomplicated
1.19 (1.01-1.40)
0.03
Hypertension
1.06 (0.91-1.22)
0.45
Fluid and electrolyte imbalance
1.15 (0.98-1.34)
0.09
Metastatic cancer
1.00 (0.88-1.15)
0.95
Renal failure
1.97 (1.55-2.51)
<0.01
Weight loss
0.89 (0.73-1.07)
0.22
Organ resection Distal pancreas resection or splenectomy Complications Yes Comorbidities
CI ¼ confidence interval.
The present study shows that increased rates of readmission after gastrectomy for malignancy are associated with postoperative LOS, discharge status, the occurrence of postoperative complications, combined resection with distal pancreatectomy and/or splenectomy, and patient comorbidities such as either diabetes mellitus or renal failure. In determining the factors associated with readmission, different combinations of the factors were added to the multivariate model separately. The results of each model showed that the addition of any factor to the model did not change the strength or degree of association of all factors currently in the model; therefore, indicating that each of these factors were independently associated with readmission. Gastrectomy for malignancy is a high-risk procedure associated with complication rates ranging from 20% to 45%.69,11,12,21-23 The occurrence of postoperative complications after gastric cancer resection was associated with readmissions in the study from Massachusetts General Hospital, but not in the studies from Korea and China.5,10,13 In two separate studies using the American College of Surgeons NSQIP database to assess readmissions after major gastrointestinal resections and general surgery procedures (both intra-abdominal and extra-abdominal surgeries), the occurrence of postoperative complications was the major risk factor for readmissions.24,25 Other factors that were associated with readmissions after gastric cancer surgery were preoperative malnutrition, LOS, extent of surgery, and preexisting cardiovascular disease.5,10,13 These variables were not consistent among the published studies. Factors noted in reports using administrative datasets examining major cancer surgery and major gastrointestinal resections included discharge to a facility other than home, longer LOS, higher stage of disease at diagnosis, longer travel distance, prolonged operative time of more than 4 h, and chronic steroid use.25,26 Overall, the factors associated with readmission appear to be multifactorial and differ depending on the data which are used. The occurrence of postoperative complications and discharge to destination other than home appear to be the most consistent factors demonstrated in the larger series. A novel finding of the present study was the association of diabetes mellitus and renal failure with readmissions after gastrectomy for malignancy. A report examining 1442 general surgery patients using NSQIP which included gastrectomy patients showed no association of diabetes mellitus or chronic renal failure with readmissions.24 A single-institution study from China showed that diabetes mellitus was not associated with readmissions.13 However, both diabetes and renal failure have been reported to increase readmissions after coronary artery bypass surgery. Both diabetes and renal failure are wellknown risk factors for postoperative complications, the occurrence of which is a strong predictor of readmissions. An important limitation of previous studies lies in the fact that the data originate from single institutions; therefore, their findings may not be generalizable to other patient populations, especially given that they may reflect the local culture and hospital/surgeon practice patterns, as well as the study population’s socioeconomic and comorbidity case mix.5
ammori et al readmission after gastrectomy
181
Fig e Time to readmission after hospital discharge after gastrectomy. The bars in the graph represent the percent of total patients admitted on that specific postdischarge day. The numbers on the graph represent cumulative totals at 5d intervals. (Color version of figure is available online.)
To address some of these weaknesses, we use data from the CA SID for the years 2005-2011, which include 100% of hospital discharges occurring in the state. In addition to yielding a large number of patients undergoing gastrectomy for malignancy, we exploited a unique feature of the CA SID, which makes it possible to identify all the discharges for a given patient during the study period, including 30d readmissions.
There are some inherent limitations to the present study. With the use of administrative data, we cannot account for factors that are not adequately captured in billing codes. For example, data were extracted for the diagnosis of “malignant neoplasm of stomach”, and it is unclear how many of these were adenocarcinoma diagnoses versus other diagnoses such as gastrointestinal stromal tumor. In addition, more than 35,000 gastrectomies for benign etiologies were excluded
Table 3 e Reason for readmission for all patients by time from discharge to readmission. 30-d readmission
Reason for readmission
N (%)
Complications of surgical procedures or medical care
310 (26.91)
Septicemia
72 (6.25)
Fluid and electrolyte disorders
66 (5.73)
Intestinal obstruction without hernia
52 (4.51)
Pneumonia
40 (3.47)
Cancer of stomach13
39 (3.39)
Gastrointestinal hemorrhage
36 (3.13)
Complications of surgical procedures or medical care
160 (31.75)
Septicemia
29 (5.75)
Intestinal obstruction without hernia
22 (4.37)
Time to readmission 1-7 d
8-14 d
15þ d
Complications of surgical procedures or medical care
86 (28.38)
Intestinal obstruction without hernia
16 (5.28)
Septicemia
15 (4.95)
Complications of surgical procedures or medical care
64 (18.55)
Fluid and electrolyte disorders
35 (10.14)
Septicemia
28 (8.12)
Intestinal obstruction without hernia
14 (4.06)
182
j o u r n a l o f s u r g i c a l r e s e a r c h a p r i l 2 0 1 8 ( 2 2 4 ) 1 7 6 e1 8 4
from the study. While the authors did not delve into the reasons for these gastrectomies, given the relatively low incidence of gastric cancer in the United States and the large numbers of US patients undergoing weight loss surgery, the large number of gastrectomies performed for reasons other than malignant diagnosis are presumed to be related to coding of bariatric procedures such as sleeve gastrectomy. Eighteen thousand of these patients had a diagnosis of obesity, but this diagnosis is known to be under-reported in administrative data.27 Furthermore, the extent of partial gastrectomy and specifics related to anastomoses were not available in the data. In addition, details about readmissions due to a complication of surgery are lumped together as it was not possible to parse the specific details of the complications. Thus, the effect of the complication severity or grade on rate of readmission could not be examined because of the limitations of the administrative data. Although complications that occurred during the index admission are noted in the database, we were unable to account for complications that occurred after discharge. The factors that could potentially be managed in outpatient settings, such as dehydration or pain control, are therefore modifiable and could not be examined in detail. In addition, this is a patient level analysis and may not fully account for hospital and provider characteristics related to readmissions. For example, higher hospital volume has been associated with decreased readmission rates and improved perioperative mortality; therefore, the current data may not reflect the outcomes of high-volume tertiary institutions.17 Social factors leading to readmissions could not be adequately captured. For instance, patients from skilled nursing facilities may be more likely to be sent to the emergency department and admitted for psychosocial issues, which could have been addressed in the outpatient setting. These may result in readmission for social reasons, even if not medically indicated. Data from all Medicare beneficiaries discharged to a skilled nursing facility from 2000 to 2006 showed a readmission rate of 23.5%.28 A readmission rate of 28.6% was reported from 2007 to 2009 in over 400,000 patients from five states, with readmission being the strongest predictor of death in the coming years.29 The Centers for Medicare and Medicaid Services will adopt a skilled nursing facility 30-d all-cause readmission measure as the sole measure in the skilled nursing facility valueebased purchasing program.30 Despite the limitations, the data in the present study are statewide reflection of real-world practice and most likely reflect the true rate of readmissions. This study could not examine which readmissions were potentially preventable. Although not specifically addressed in this analysis, there are strategies that may help reduce readmissions. One overarching strategy is to have patients evaluated in an outpatient setting rather than the emergency department. Readmission to the hospital is the path of least resistance for patients presenting to the emergency department, particularly during offhours. Some practical approaches to help reduce readmissions include a phone call follow-up to identify any postdischarge problems that can be addressed as an outpatient. For example, dehydration can be treated with intravenous fluids delivered in an infusion center. Also, wound problems can be examined as an outpatient rather
than through a visit to the emergency department. Furthermore, another potentially useful approach includes patient education regarding issues that would require a phone call to the physician’s office. Finally, because patients discharged to skilled nursing facilities have a higher readmission rate, education directed to skilled nursing facilities may reduce preventable readmissions.
Conclusion In conclusion, based on a statewide administrative dataset, there is a 16.49% rate of readmission after gastrectomy for gastric malignancy. Readmissions are associated with postoperative complications, LOS, discharge to skilled nursing facility or home health care, combined resection with distal pancreatectomy and/or splenectomy, and patient comorbidities such as diabetes mellitus and renal failure. Some of these factors may not be modifiable, such as preoperative comorbidity. In addition, because of the somewhat urgent nature of surgery, there may not be a prolonged time to optimize these comorbidities. Increased LOS and discharge status may be related to preoperative functional status, postoperative complications or debility, or social factors. The rate of readmissions has not significantly improved over time. Many postoperative readmissions are appropriate and necessary for safe patient care. Although current health-care policy strives to reduce excessive readmissions, it is important to note that readmissions are not all preventable. Future efforts should focus on best practices to reduce postoperative complications. Reducing readmissions requires a multifaceted approach including patient-centered discharge planning, medication reconciliation, using a nurse transition advocate who meets regularly with the patient in hospital to coordinate discharge planning among team members, and early postdischarge telephone follow-up to identify potential problems.27 These strategies will not only reduce patient suffering but also reduce readmission rates and overall health expenditures.
Acknowledgment Funding: This study was supported by an award (award number: NIH K12CA076917) from the National Cancer Institute. Authors’ contributions: J.B.A. made substantial contributions to the conception and design of the work, analysis and interpretation of data for the work, drafting the manuscript, and final approval of the version to be published. S.N. contributed to acquisition and analysis of data for the work, was involved in drafting the work, and gave final approval of the version to be published. S.M.K. and N.S. contributed to the conception and design of the work, were involved in analysis and interpretation of data for the work, revised the work critically for important intellectual content, and gave final approval of the version to be published.
ammori et al readmission after gastrectomy
Disclosure There are no conflicts of interest to be disclosed by any of the authors. Although this has no bearing on the present work, S.M.K. would like to note, for full transparency, that her spouse has ownership interests in American Renal Associates.
references
1. Jencks SF, Williams MV, Coleman EA. Rehospitalizations among patients in the Medicare fee-for-service program. N Engl J Med. 2009;360:1418e1428. 2. [cited 2016 January 12]. Available at: http://www.modernhe althcare.com/article/20150808/MAGAZINE/308089963. 3. [cited 2016 January 13]. Available at: http://www. beckershospitalreview.com/quality/6-stats-on-the-cost-ofreadmission-for-cms-tracked-conditions.html. 4. [ March 14, 2016]. Available at: http://www.cancer.org/cancer/ stomachcancer/detailedguide/stomach-cancer-key-statistics. 5. Ahmad R, Schmidt BH, Rattner DW, Mullen JT. Factors influencing readmission after curative gastrectomy for gastric cancer. J Am Coll Surg. 2014;218:1215e1222. 6. Bonenkamp JJ, Songun I, Hermans J, et al. Randomised comparison of morbidity after D1 and D2 dissection for gastric cancer in 996 Dutch patients. Lancet. 1995;345:745e748. 7. Cuschieri A, Fayers P, Fielding J, et al. Postoperative morbidity and mortality after D1 and D2 resections for gastric cancer: preliminary results of the MRC randomised controlled surgical trial. The Surgical Cooperative Group. Lancet. 1996;347:995e999. 8. Gil-Rendo A, Hernandez-Lizoain JL, Martinez-Regueira F, et al. Risk factors related to operative morbidity in patients undergoing gastrectomy for gastric cancer. Clin Transl Oncol. 2006;8:354e361. 9. Grossmann EM, Longo WE, Virgo KS, et al. Morbidity and mortality of gastrectomy for cancer in Department of Veterans Affairs Medical Centers. Surgery. 2002;131:484e490. 10. Kim MC, Kim KH, Jung GJ. A 5 year analysis of readmissions after radical subtotal gastrectomy for early gastric cancer. Ann Surg Oncol. 2012;19:2459e2464. 11. Papenfuss WA, Kukar M, Oxenberg J, et al. Morbidity and mortality associated with gastrectomy for gastric cancer. Ann Surg Oncol. 2014;21:3008e3014. 12. Selby LV, Vertosick EA, Sjoberg DD, et al. Morbidity after total gastrectomy: analysis of 238 patients. J Am Coll Surg. 2015;220:863e871 e2. 13. Zhuang CL, Wang SL, Huang DD, et al. Risk factors for hospital readmission after radical gastrectomy for gastric cancer: a prospective study. PLoS One. 2015;10:e0125572. 14. Elixhauser A, Steiner C, Harris DR, Coffey RM. Comorbidity measures for use with administrative data. Med Care. 1998;36:8e27.
183
15. Birkmeyer JD, Finks JF, O’Reilly A, et al. Surgical skill and complication rates after bariatric surgery. N Engl J Med. 2013;369:1434e1442. 16. Keller DS, Chien HL, Hashemi L, Senagore AJ, Delaney CP. The HARM score: a novel, easy measure to evaluate quality and outcomes in colorectal surgery. Ann Surg. 2014;259:1119e1125. 17. Tsai TC, Joynt KE, Orav EJ, Gawande AA, Jha AK. Variation in surgical-readmission rates and quality of hospital care. N Engl J Med. 2013;369:1134e1142. 18. [cited 2016 January 14]. Available at: http://www.beckersho spitalreview.com/quality/the-38-hospitals-facing-highestmedicare-penalties-for-readmissions.html. 19. [cited 2016 January 15]. Available at: http://www.beckersho spitalreview.com/quality/2-5k-hospitals-penalized-by-cmsfor-high-readmissions-10-things-to-know.html. 20. Merchant SJ, Ituarte PH, Choi A, et al. Hospital readmission following surgery for gastric cancer: frequency, timing, etiologies, and survival. J Gastrointest Surg. 2015;19:1769e1781. 21. Bartlett EK, Roses RE, Kelz RR, Drebin JA, Fraker DL, Karakousis GC. Morbidity and mortality after total gastrectomy for gastric malignancy using the American College of Surgeons National Surgical Quality Improvement Program database. Surgery. 2014;156:298e304. 22. Hartgrink HH, van de Velde CJ, Putter H, et al. Extended lymph node dissection for gastric cancer: who may benefit? Final results of the randomized Dutch gastric cancer group trial. J Clin Oncol. 2004;22:2069e2077. 23. Sano T, Sasako M, Yamamoto S, et al. Gastric cancer surgery: morbidity and mortality results from a prospective randomized controlled trial comparing D2 and extended para-aortic lymphadenectomyeJapan Clinical Oncology Group study 9501. J Clin Oncol. 2004;22:2767e2773. 24. Kassin MT, Owen RM, Perez SD, et al. Risk factors for 30-day hospital readmission among general surgery patients. J Am Coll Surg. 2012;215:322e330. 25. Kelly KN, Iannuzzi JC, Rickles AS, Monson JR, Fleming FJ. Risk factors associated with 30-day postoperative readmissions in major gastrointestinal resections. J Gastrointest Surg. 2014;18:35e43. discussion 43-44. 26. Stitzenberg KB, Chang Y, Smith AB, Nielsen ME. Exploring the burden of inpatient readmissions after major cancer surgery. J Clin Oncol. 2015;33:455e464. 27. Martin BJ, Chen G, Graham M, Quan H. Coding of obesity in administrative hospital discharge abstract data: accuracy and impact for future research studies. BMC Health Serv Res. 2014;14:70. 28. Mor V, Intrator O, Feng Z, Grabowski DC. The revolving door of rehospitalization from skilled nursing facilities. Health Aff (Millwood). 2010;29:57e64. 29. Hakkarainen TW, Arbabi S, Willis MM, Davidson GH, Flum DR. Outcomes of patients discharged to skilled nursing facilities after acute care hospitalizations. Ann Surg. 2016;263:280e285. 30. [ March 20, 2016]. Available at: http://www.fha.org/show Document.aspx?f¼SNFMCARE_-FY2016-FR-Summary.pdf.
184
j o u r n a l o f s u r g i c a l r e s e a r c h a p r i l 2 0 1 8 ( 2 2 4 ) 1 7 6 e1 8 4
Appendix. ICD-9-CM complication codes.
Complication
ICD-9-CM code (secondary diagnosis codes)
Digestive system
997.49
Infection
998.5
Wound infection
998.3
Persistent postoperative fistula
998.6
Complications peculiar to certain specified procedures
996
Complications affecting specified body system not elsewhere classified
997
Other complications of procedures not elsewhere classified
998
Complications of medical care not elsewhere classified
999
Acute posthemorrhagic anemia
285.1
Pulmonary embolism and deep venous thrombosis Pulmonary insufficiency after surgery
415.1, 451.1, 451.2, 451.8, 451.9, 453.2, 453.8, 453.9 518.5
Available online at www.sciencedirect.com
ScienceDirect journal homepage: www.JournalofSurgicalResearch.com
Predictors of 30-d readmissions after gastrectomy for malignancy John B. Ammori, MD,a,* Suparna Navale, MS, MPH,b Nicholas Schiltz, PhD,b and Siran M. Koroukian, PhDb a
Department of Surgery, University Hospitals Seidman Cancer Center and Case Comprehensive Cancer Center, Cleveland, Ohio b Department of Epidemiology and Biostatistics, School of Medicine, Case Western Reserve University, Cleveland, Ohio
article info
abstract
Article history:
Background: The objective of this study is to identify risk factors associated with read-
Received 27 June 2017
mission after gastrectomy to potentially identify potential areas for targeted improve-
Received in revised form
ments. Hospital readmission after surgery is a topic of interest in health-care policy among
3 November 2017
hospitals, payers, and providers. Readmissions are associated with increased costs,
Accepted 5 December 2017
morbidity, and mortality. Readmission rates have been proposed as a quality metric for
Available online xxx
hospitals and quality indicator of individual surgeon’s performance. In addition, the Centers for Medicare and Medicaid Services has reduced payments to hospitals with
Keywords:
excessive readmissions for certain diagnoses.
Readmission
Materials and methods: All gastrectomy procedures for malignancy in patients aged 18 y
Gastrectomy
from 2005 to 2011 were queried from the California State Inpatient Database. Patients who
Gastric cancer
died during index admission were excluded. Descriptive statistics were examined between all baseline variables and readmission status. Logistic regression models were adjusted for age, race, sex, and insurance status. Results: A total of 6985 patients underwent gastrectomy for malignancy; 16.5% of the patients were readmitted after postoperative discharge. Readmission rate did not change significantly over time. Multivariable analysis demonstrated that the occurrence of any postoperative complications, postoperative length of stay greater than 10 d, discharge to skilled nursing facility or home health care, combined resection with distal pancreatectomy and/or splenectomy, and patient comorbidities like diabetes mellitus and renal failure were independently associated with readmissions. Conclusions: The findings suggest that focusing on quality improvement efforts by targeting reduction of postoperative complications may reduce readmission rates. ª 2017 Elsevier Inc. All rights reserved.
Introduction Reducing hospital readmissions has become a focus in the realm of health-care policy and inpatient care quality improvement. In Medicare patient population, approximately
16% of surgical patients and 21% of medical patients had an unplanned readmission in 2004, accounting for $17 billion in excess health-care costs.1,2 On average, the cost per readmission for any cause is $11,200.3 The Affordable Care Act requires the establishment of a Hospital Readmission
* Corresponding author. Department of Surgery, University Hospitals Seidman Cancer Center and Case Comprehensive Cancer Center, 11100 Euclid Avenue, LKS 7001, Cleveland, OH 44106. Tel.: þ1 216 844 1777; fax: þ1 216 286 3294. E-mail address: [email protected] (J.B. Ammori). 0022-4804/$ e see front matter ª 2017 Elsevier Inc. All rights reserved. https://doi.org/10.1016/j.jss.2017.12.004
ammori et al readmission after gastrectomy
Reduction Program. The purpose of the program is to improve quality and lower costs while attempting to ensure that patients are fully prepared and safe for continued care outside the hospital setting at the time of discharge. The program is designed to penalize hospitals with excessive 30d readmissions by withholding 1% of Medicare reimbursement starting from the fiscal year 2013 and increasing the penalty to 3% for the fiscal year 2015. The American Cancer Society estimates over 26,000 new stomach cancer diagnoses and more than 10,000 deaths in the Unites States in 2016.4 Gastrectomy provides the only chance for cure. Gastrectomy for malignancy is a complex surgical procedure, and readmission rates after this procedure have been estimated to be up to 20%.5-13 Patient and hospital-level factors have been found to be associated with readmissions. The goal of this study is to identify factors associated with readmissions after gastrectomy for malignancy, thus informing future interventions to improve care, reduce excess readmissions, and curtail health-care costs.
Materials and methods The present study was deemed exempt from approval by our hospital’s Institutional Review Board of Case Western Reserve University.
177
as any unplanned (or unscheduled) admission to a hospital within 30 d of discharge from the index admission. Postoperative complications were identified using specific definitions (Appendix). In addition, “any complication” was defined as one or more of these postoperative complications.
Independent variables The independent variables used in the statistical models are as follows: (1) patient demographics such as age, sex, race, and insurance; (2) discharge status; (3) hospital teaching status; and (4) patient comorbidities. The variables were defined as those recorded by the HCUP with a few category modifications to facilitate model specification. Age was grouped as 18-49, 5059, 51-60, 60-69, 70-79, and 80 y. Race was categorized as white, black, Hispanic, and other. Insurance status was identified as Medicare, Medicaid, private/health maintenance organization, uninsured/self-pay, and other. Discharge status was classified as routine, skilled nursing facility, home health care, and other. Comorbidities were identified and classified using the Comorbidity Software provided by the Agency for Healthcare Research and Quality.14 LOS was grouped into three categories based on the median LOS of 10 d, as 1-10, 1120, and 21 d. In addition, blood transfusion was identified based on ICD-9 procedure codes 99.00, 99.02, 99.03, and 99.04.
Statistical analysis Data source This study represents a retrospective cross-sectional secondary data analysis of the California State Inpatient Database (CA SID) file. State Inpatient Databases (SIDs) are state-specific files that contain all inpatient care records of respective participating states; they were developed as a part of the Healthcare Cost and Utilization Project (HCUP). The SIDs encompass 97% of all US hospital discharges. Over 100 clinical and nonclinical data elements are available, including demographics, hospital characteristics, primary and secondary diagnoses, inpatient procedures, insurance type, discharge status, length of stay (LOS), and total charges, among others. The CA SID is one of the few SIDs that also allows for tracking hospital readmissions, given that it includes a synthetic patient identifier and a variable that reflects the lapse of time between two hospital admissions.
Study population Cases that were not assigned a Clinical Classification Software code for neoplasm (Clinical Classification Software 11-47; n ¼ 29,531) or did not have a diagnosis of malignant neoplasm of the stomach (ICD-9 diagnosis code 151; n ¼ 35,430) were excluded. In addition, admissions for individuals under the age of 18 y or those who died during the index admission were excluded from analysis, leaving the study population at 6985 individuals.
Outcomes of interest The outcomes of interest were 30-d readmissions and postoperative complications. Thirty-day readmission was defined
All analyses were performed using SAS System for Windows, version 9.3 (SAS Institute Inc, Cary, NC). We calculated the distribution of data and descriptive statistics for all baseline variables included in the analyses for all gastrectomy patients and those who were readmitted. Median LOS for readmitted and not readmitted patients was also calculated for each baseline variable. Annual trend in gastrectomy procedures and readmission rates were also examined. Univariate comparisons were made using Pearson’s chi-squared test for categorical variables. We calculated the complication rates and comorbidity rates by using readmission and discharge status, as well as the comorbidity rates of patients with any complication by using readmission status. We compared the discharge status after index admission based on index LOS and the complication rates based on index LOS and surgery type. Logistic regression models, adjusted for age, race, sex, and insurance, were fit to examine the relationship between readmission and a combination of factors (LOS, discharge status, any complications, blood transfusion, and comorbidities). Finally, we identified the causes for readmission based on the time to readmission, categorized as 1-7, 8-14, and 15 d.
Results A total of 6985 patients who underwent gastrectomy for malignancy are included in this analysis. Of these patients, 1152 (16.5%) required readmission. Median postoperative LOS for all patients was 10 d. The median LOS after readmission was 5 d. Also, 1972 patients (28.2%) underwent total gastrectomy, 2511 patients (35.95%) suffered a postoperative complication,
178
j o u r n a l o f s u r g i c a l r e s e a r c h a p r i l 2 0 1 8 ( 2 2 4 ) 1 7 6 e1 8 4
and 4083 patients (58.5%) were discharged routinely to home, while 1791 (25.6%) and 1039 (14.9%) patients were discharged with home health care or to a skilled nursing facility, respectively. Readmissions were to the index hospital in 81.1% of patients and to a nonindex hospital in 18.9% of patients. The association of readmissions with numerous factors was examined, including patient demographics, type of gastrectomy, LOS from the index admission, discharge status, and postoperative complications (Table 1). Following are the notable comparisons. Total gastrectomy was associated with a higher readmission rate compared to partial gastrectomy (18.4% versus 15.7%, P ¼ 0.007). Combined organ resection with the addition of distal pancreatectomy and/or splenectomy to gastrectomy was significantly associated with readmission (24.8% versus 16%, P < 0.001). Women were less likely to be readmitted compared to men (15.8% versus 17.1%, P ¼ 0.03). Using the median LOS of 10 d as a cutoff, those with an index LOS of 10 d were readmitted at a rate of 12.9% compared to a 25.2% readmission rate for those with an index readmission LOS of more than 20 d, P < 0.001. Discharge status after surgical recovery was significantly associated with readmission, with a readmission rate of 24.6% in patients discharged to a skilled nursing facility, 18.9% in those discharged with home health care, and 13.4% in those discharged home, P < 0.001. Transfusion of blood in the perioperative period was associated with increased readmission rate (18.6% versus 15.4%, P < 0.001). Patients who suffered a postoperative complication had a higher rate of readmission, 21% versus 14%, P < 0.001. Several individual comorbidities were also associated with readmissions, including chronic lung disease, diabetes mellitus, hypertension, fluid and electrolyte imbalance, renal failure, and weight loss. There was a trend toward a reduced rate of readmissions over time from 17.6% in 2005 down to 13% in 2011, but this did not reach statistical significance, P ¼ 0.074 (data not shown). Rates of readmission were not different between teaching and nonteaching hospitals (data not shown). A multivariable logistic regression was performed to identify correlates of readmission after gastrectomy. Numerous factors were included in the regression model, including any comorbidity with a prevalence of >10%. Independent predictors of readmission, including initial LOS >10 d, discharge to skilled nursing facility or home health care, the occurrence of any postoperative complications, combined organ resection, and patient comorbidities of diabetes mellitus and renal failure, were also included in the model. Independent predictors were significantly associated with readmission based on our model that included initial LOS >10 d, discharge to skilled nursing facility or home health care, and the occurrence of any postoperative complications. Table 2 shows the results of the model. We analyzed the time elapsed from discharge to readmission and reasons for readmission. Over half of the readmissions occurred in the first 9 d after discharge (Fig.). The primary cause for readmission was complication of surgery (Table 3). The most common complications associated with readmission were postoperative infection which occurred in 40% of those readmitted for a complication, digestive system complication in 28.39%, and hemorrhage in 4.19%. Other relatively frequent causes of readmission included
septicemia, intestinal obstruction, and fluid and electrolyte abnormities. Fluid and electrolyte abnormalities were more commonly seen as a reason for readmission in those patients readmitted at 2 wk after discharge. In addition, we examined some additional associations of interest based on the analyses described previously. Extent of gastrectomy (partial versus total) was associated with a significantly higher rate of complications, with a rate of 41.58% after total gastrectomy compared to 33.73% after partial gastrectomy, P < 0.001. However, extent of gastrectomy by itself was not independently associated with readmission in the multivariate model. Combined resection with the addition of distal pancreatectomy and/or splenectomy was not only independently associated with readmission but was also associated with higher rate of complication than gastrectomy alone (52.7% versus 35%, P < 0.001). Complications were associated with LOS. Compared to LOS of 10 d for the index admission, those with a LOS of >20 d suffered significantly higher rates of complications (22.3% versus 70.1%, respectively, P < 0.001). LOS for the index admission was also associated with discharge status. Patients with an LOS 10 d were routinely discharged 72.0% of the time compared to 30% routine discharge for those with an LOS >20 d, P < 0.001. Only 8.15% of patients with LOS 10 d for the index admission were discharged to skilled nursing facility compared to 33.3% of those with LOS >20 d, P < 0.001. Despite these associations in univariable analysis, the independent predictors of readmission were initial LOS >10 d, discharge to skilled nursing facility or home health care, and the occurrence of any postoperative complications.
Discussion There is significant attention directed at reducing hospital readmissions in the United States as an effort to reduce waste and inefficiency in health care. Postoperative readmission rate has been proposed as a quality metric for hospitals and quality indicator of individual surgeon’s performance.15-17 This is the fourth year of the Hospital Readmission Reduction Program, which was created under the Affordable Care Act. In this year, 2592 hospitals will receive reduced Medicare payments ranging from 0.01% to 3% for excessive readmissions.18 These penalties are projected to cost these hospitals $420 million from October 1, 2015, through September 30, 2016.19 In an effort to not only improve quality and efficiency but also to avoid financial penalties, it is important to understand the variables that are associated with readmissions. This understanding may allow for targeted quality improvement efforts of modifiable factors. Given the paucity of data regarding readmissions after gastric malignancy surgery, this study contributes to the literature by identifying factors associated with our outcome of interest. Using the HCUP data from California, we documented a readmission rate of 16.5% after gastrectomy for gastric malignancy, with over half of the readmissions occurring within 10 d of discharge. The readmission rate in this study slightly exceeds the range of 7.5%-15.4% reported in previous studies examining readmissions for the same procedure.5,10,13,20 A study of nearly 9000 patients using data linked from the Office
179
ammori et al readmission after gastrectomy
Table 1 e Patient characteristics. All gastrectomy patients Variable Total
Table 1 e (continued )
N (%)
N (%)
6985
1152
Total gastrectomy
1972 (28.23)
363 (18.41)
Partial gastrectomy
5013 (71.77)
789 (15.74)
69
70
Age <50 y
P valuey
Variable Alcohol abuse
<0.01
Surgery type
Age (median), y
All gastrectomy patients
Readmitted
N (%)
P valuey <0.01
203 (2.91)
50 (24.63)
2206 (31.58)
374 (16.95)
0.48
410 (5.87)
101 (24.63)
<0.01
1042 (14.92)
203 (19.48)
<0.01
249 (3.56)
62 (24.90)
<0.01
Diabetes mellitusuncomplicated
1389 (19.89)
265 (19.08)
<0.01
219 (3.14)
48 (21.92)
0.03
Deficiency anemia Congestive heart failure Chronic lung disease
0.07
N (%)
Readmitted
Coagulopathy
748 (10.71)
119 (15.91)
50-59 y
1113 (15.93)
163 (14.65)
60-69 y
1780 (25.48)
275 (15.45)
Diabetes mellituscomplicated
70-79 y
2081 (29.79)
366 (17.59)
Hypertension
3737 (53.50)
662 (17.71)
<0.01
80þ y
1263 (18.08)
229 (18.13)
Fluid and electrolyte imbalance
1548 (22.16)
329 (21.25)
<0.01
472 (6.76)
70 (14.83)
0.31
Pulmonary circulation disorders
95 (1.36)
24 (25.26)
0.02
Renal failure
384 (5.50)
116 (30.21)
<0.01
Valvular disease
278 (3.98)
61 (21.94)
0.01
Weight loss
965 (13.82)
185 (19.17)
0.02
<0.01
Race White Black
2796 (40.03)
495 (17.70)
403 (5.77)
68 (16.87)
Hispanic
1654 (23.68)
295 (17.84)
Other
1760 (25.20)
248 (14.09)
372 (5.33)
46 (12.37)
Missing Sex
0.03
Male
4220 (60.42)
721 (17.09)
Female
2691 (38.53)
426 (15.83)
Missing
74 (1.06)
Index admission LOS (median), d
11.5
1-10 d
3977 (56.94)
137 (3.44)
11-20 d
2097 (30.02)
196 (9.35)
910 (13.03)
178 (19.56)
<0.01
Index admission LOS
<0.01
Discharge status after index admission Routine
4083 (58.45)
545 (13.35)
Skilled nursing facility
1039 (14.87)
256 (24.64)
Home health care
1791 (25.64)
339 (18.93)
72 (1.03)
12 (16.67)
Other
* Cells with frequency <11 are suppressed due to requirements in the HCUP data use agreement. y P value of readmitted versus not readmitted participants.
*
10
21þ d
Obesity
<0.01
Combined organ resection (pancreas or spleen) Yes
391 (5.6)
No
6594 (94.4)
97 (24.8) 1055 (16) <0.01
Received blood transfusion Yes
2406 (34.45)
448 (18.62)
No
4579 (65.55)
704 (15.37)
Yes
2511 (35.95)
527 (20.99)
No
4474 (64.05)
625 (13.97)
<0.01
Complications
Comorbidities (continued)
of Statewide Health Planning and Development database and the California Cancer Registry showed a readmission rate of 15.42% at 30 d postoperatively, which increased to 28.8% at 90 d after surgery. Interestingly, this study also showed a worsened overall 5-y survival rate associated with postoperative readmissions.20 A single-institution study from Massachusetts General Hospital reported a 30-d readmission rate of 14.6% in 418 studied patients, with an increased rate of readmissions after total gastrectomy compared to that after partial gastrectomy, 23% versus 13%, respectively.5 Data from the American College of Surgeons National Surgical Quality Improvement Program (NSQIP) showed a 13.5% 30d unplanned readmission rate in 1400 patients who underwent esophagectomy or gastrectomy. These rates of readmission are similar to those of the present study but higher than those reported from Asia. For example, a study from Korea showed a 7.5% readmission rate after radical subtotal gastrectomy in 530 gastric cancer patients.10 Similarly, a study of 376 gastric cancer patients in China demonstrated a 30d readmission rate of 7.2%.13 The differences in readmissions between Asian and North American studies may be related to patient factors and cultural differences. For instance, in the Chinese study, the median body mass index was 22, the median LOS was 13 d, 82% of patients were discharged home without services, and no patients were discharged to rehabilitation facilities.13
180
j o u r n a l o f s u r g i c a l r e s e a r c h a p r i l 2 0 1 8 ( 2 2 4 ) 1 7 6 e1 8 4
Table 2 e Multivariate logistic regression model of predictors for readmission after gastrectomy. Variable
Odds ratio (95% CI)
P value
Age <50
Reference
50-59
0.84 (0.64-1.09)
0.18
60-69
0.83 (0.64-1.07)
0.15
70-79
0.90 (0.68-1.20)
0.49
80þ
0.84 (0.61-1.14)
0.26
Race White
Reference
Black
0.92 (0.69-1.22)
0.54
Hispanic
1.08 (0.91-1.28)
0.37
Other
0.83 (0.70-0.98)
0.03
Missing
0.85 (0.60-1.21)
0.36
Sex Male
Reference
Female
0.95 (0.83-1.09)
0.49
Missing
0.57 (0.22-1.53)
0.27
Medicare
0.97 (0.80-1.19)
0.80
Medicaid
0.98 (0.77-1.24)
0.85
Insurance
Private insurance
Reference
Self-pay
0.92 (0.46-1.82)
0.80
Other
1.76 (1.11-2.79)
0.02
Index admission LOS 1-10 d
Reference
11-20 d
1.36 (1.16-1.59)
<0.01
21þ d
1.61 (1.31-1.98)
<0.01
Discharge status Routine
Reference
Skilled nursing facility
1.62 (1.34-1.97)
<0.01
Home health care
1.29 (1.10-1.51)
<0.01
Other
0.97 (0.51-1.83)
0.91
1.48 (1.16-1.90)
<0.01
1.23 (1.07-1.42)
<0.01
Deficiency anemia
0.90 (0.78-1.04)
0.14
Chronic lung disease
1.08 (0.91-1.29)
0.37
Diabetes mellitus-uncomplicated
1.19 (1.01-1.40)
0.03
Hypertension
1.06 (0.91-1.22)
0.45
Fluid and electrolyte imbalance
1.15 (0.98-1.34)
0.09
Metastatic cancer
1.00 (0.88-1.15)
0.95
Renal failure
1.97 (1.55-2.51)
<0.01
Weight loss
0.89 (0.73-1.07)
0.22
Organ resection Distal pancreas resection or splenectomy Complications Yes Comorbidities
CI ¼ confidence interval.
The present study shows that increased rates of readmission after gastrectomy for malignancy are associated with postoperative LOS, discharge status, the occurrence of postoperative complications, combined resection with distal pancreatectomy and/or splenectomy, and patient comorbidities such as either diabetes mellitus or renal failure. In determining the factors associated with readmission, different combinations of the factors were added to the multivariate model separately. The results of each model showed that the addition of any factor to the model did not change the strength or degree of association of all factors currently in the model; therefore, indicating that each of these factors were independently associated with readmission. Gastrectomy for malignancy is a high-risk procedure associated with complication rates ranging from 20% to 45%.69,11,12,21-23 The occurrence of postoperative complications after gastric cancer resection was associated with readmissions in the study from Massachusetts General Hospital, but not in the studies from Korea and China.5,10,13 In two separate studies using the American College of Surgeons NSQIP database to assess readmissions after major gastrointestinal resections and general surgery procedures (both intra-abdominal and extra-abdominal surgeries), the occurrence of postoperative complications was the major risk factor for readmissions.24,25 Other factors that were associated with readmissions after gastric cancer surgery were preoperative malnutrition, LOS, extent of surgery, and preexisting cardiovascular disease.5,10,13 These variables were not consistent among the published studies. Factors noted in reports using administrative datasets examining major cancer surgery and major gastrointestinal resections included discharge to a facility other than home, longer LOS, higher stage of disease at diagnosis, longer travel distance, prolonged operative time of more than 4 h, and chronic steroid use.25,26 Overall, the factors associated with readmission appear to be multifactorial and differ depending on the data which are used. The occurrence of postoperative complications and discharge to destination other than home appear to be the most consistent factors demonstrated in the larger series. A novel finding of the present study was the association of diabetes mellitus and renal failure with readmissions after gastrectomy for malignancy. A report examining 1442 general surgery patients using NSQIP which included gastrectomy patients showed no association of diabetes mellitus or chronic renal failure with readmissions.24 A single-institution study from China showed that diabetes mellitus was not associated with readmissions.13 However, both diabetes and renal failure have been reported to increase readmissions after coronary artery bypass surgery. Both diabetes and renal failure are wellknown risk factors for postoperative complications, the occurrence of which is a strong predictor of readmissions. An important limitation of previous studies lies in the fact that the data originate from single institutions; therefore, their findings may not be generalizable to other patient populations, especially given that they may reflect the local culture and hospital/surgeon practice patterns, as well as the study population’s socioeconomic and comorbidity case mix.5
ammori et al readmission after gastrectomy
181
Fig e Time to readmission after hospital discharge after gastrectomy. The bars in the graph represent the percent of total patients admitted on that specific postdischarge day. The numbers on the graph represent cumulative totals at 5d intervals. (Color version of figure is available online.)
To address some of these weaknesses, we use data from the CA SID for the years 2005-2011, which include 100% of hospital discharges occurring in the state. In addition to yielding a large number of patients undergoing gastrectomy for malignancy, we exploited a unique feature of the CA SID, which makes it possible to identify all the discharges for a given patient during the study period, including 30d readmissions.
There are some inherent limitations to the present study. With the use of administrative data, we cannot account for factors that are not adequately captured in billing codes. For example, data were extracted for the diagnosis of “malignant neoplasm of stomach”, and it is unclear how many of these were adenocarcinoma diagnoses versus other diagnoses such as gastrointestinal stromal tumor. In addition, more than 35,000 gastrectomies for benign etiologies were excluded
Table 3 e Reason for readmission for all patients by time from discharge to readmission. 30-d readmission
Reason for readmission
N (%)
Complications of surgical procedures or medical care
310 (26.91)
Septicemia
72 (6.25)
Fluid and electrolyte disorders
66 (5.73)
Intestinal obstruction without hernia
52 (4.51)
Pneumonia
40 (3.47)
Cancer of stomach13
39 (3.39)
Gastrointestinal hemorrhage
36 (3.13)
Complications of surgical procedures or medical care
160 (31.75)
Septicemia
29 (5.75)
Intestinal obstruction without hernia
22 (4.37)
Time to readmission 1-7 d
8-14 d
15þ d
Complications of surgical procedures or medical care
86 (28.38)
Intestinal obstruction without hernia
16 (5.28)
Septicemia
15 (4.95)
Complications of surgical procedures or medical care
64 (18.55)
Fluid and electrolyte disorders
35 (10.14)
Septicemia
28 (8.12)
Intestinal obstruction without hernia
14 (4.06)
182
j o u r n a l o f s u r g i c a l r e s e a r c h a p r i l 2 0 1 8 ( 2 2 4 ) 1 7 6 e1 8 4
from the study. While the authors did not delve into the reasons for these gastrectomies, given the relatively low incidence of gastric cancer in the United States and the large numbers of US patients undergoing weight loss surgery, the large number of gastrectomies performed for reasons other than malignant diagnosis are presumed to be related to coding of bariatric procedures such as sleeve gastrectomy. Eighteen thousand of these patients had a diagnosis of obesity, but this diagnosis is known to be under-reported in administrative data.27 Furthermore, the extent of partial gastrectomy and specifics related to anastomoses were not available in the data. In addition, details about readmissions due to a complication of surgery are lumped together as it was not possible to parse the specific details of the complications. Thus, the effect of the complication severity or grade on rate of readmission could not be examined because of the limitations of the administrative data. Although complications that occurred during the index admission are noted in the database, we were unable to account for complications that occurred after discharge. The factors that could potentially be managed in outpatient settings, such as dehydration or pain control, are therefore modifiable and could not be examined in detail. In addition, this is a patient level analysis and may not fully account for hospital and provider characteristics related to readmissions. For example, higher hospital volume has been associated with decreased readmission rates and improved perioperative mortality; therefore, the current data may not reflect the outcomes of high-volume tertiary institutions.17 Social factors leading to readmissions could not be adequately captured. For instance, patients from skilled nursing facilities may be more likely to be sent to the emergency department and admitted for psychosocial issues, which could have been addressed in the outpatient setting. These may result in readmission for social reasons, even if not medically indicated. Data from all Medicare beneficiaries discharged to a skilled nursing facility from 2000 to 2006 showed a readmission rate of 23.5%.28 A readmission rate of 28.6% was reported from 2007 to 2009 in over 400,000 patients from five states, with readmission being the strongest predictor of death in the coming years.29 The Centers for Medicare and Medicaid Services will adopt a skilled nursing facility 30-d all-cause readmission measure as the sole measure in the skilled nursing facility valueebased purchasing program.30 Despite the limitations, the data in the present study are statewide reflection of real-world practice and most likely reflect the true rate of readmissions. This study could not examine which readmissions were potentially preventable. Although not specifically addressed in this analysis, there are strategies that may help reduce readmissions. One overarching strategy is to have patients evaluated in an outpatient setting rather than the emergency department. Readmission to the hospital is the path of least resistance for patients presenting to the emergency department, particularly during offhours. Some practical approaches to help reduce readmissions include a phone call follow-up to identify any postdischarge problems that can be addressed as an outpatient. For example, dehydration can be treated with intravenous fluids delivered in an infusion center. Also, wound problems can be examined as an outpatient rather
than through a visit to the emergency department. Furthermore, another potentially useful approach includes patient education regarding issues that would require a phone call to the physician’s office. Finally, because patients discharged to skilled nursing facilities have a higher readmission rate, education directed to skilled nursing facilities may reduce preventable readmissions.
Conclusion In conclusion, based on a statewide administrative dataset, there is a 16.49% rate of readmission after gastrectomy for gastric malignancy. Readmissions are associated with postoperative complications, LOS, discharge to skilled nursing facility or home health care, combined resection with distal pancreatectomy and/or splenectomy, and patient comorbidities such as diabetes mellitus and renal failure. Some of these factors may not be modifiable, such as preoperative comorbidity. In addition, because of the somewhat urgent nature of surgery, there may not be a prolonged time to optimize these comorbidities. Increased LOS and discharge status may be related to preoperative functional status, postoperative complications or debility, or social factors. The rate of readmissions has not significantly improved over time. Many postoperative readmissions are appropriate and necessary for safe patient care. Although current health-care policy strives to reduce excessive readmissions, it is important to note that readmissions are not all preventable. Future efforts should focus on best practices to reduce postoperative complications. Reducing readmissions requires a multifaceted approach including patient-centered discharge planning, medication reconciliation, using a nurse transition advocate who meets regularly with the patient in hospital to coordinate discharge planning among team members, and early postdischarge telephone follow-up to identify potential problems.27 These strategies will not only reduce patient suffering but also reduce readmission rates and overall health expenditures.
Acknowledgment Funding: This study was supported by an award (award number: NIH K12CA076917) from the National Cancer Institute. Authors’ contributions: J.B.A. made substantial contributions to the conception and design of the work, analysis and interpretation of data for the work, drafting the manuscript, and final approval of the version to be published. S.N. contributed to acquisition and analysis of data for the work, was involved in drafting the work, and gave final approval of the version to be published. S.M.K. and N.S. contributed to the conception and design of the work, were involved in analysis and interpretation of data for the work, revised the work critically for important intellectual content, and gave final approval of the version to be published.
ammori et al readmission after gastrectomy
Disclosure There are no conflicts of interest to be disclosed by any of the authors. Although this has no bearing on the present work, S.M.K. would like to note, for full transparency, that her spouse has ownership interests in American Renal Associates.
references
1. Jencks SF, Williams MV, Coleman EA. Rehospitalizations among patients in the Medicare fee-for-service program. N Engl J Med. 2009;360:1418e1428. 2. [cited 2016 January 12]. Available at: http://www.modernhe althcare.com/article/20150808/MAGAZINE/308089963. 3. [cited 2016 January 13]. Available at: http://www. beckershospitalreview.com/quality/6-stats-on-the-cost-ofreadmission-for-cms-tracked-conditions.html. 4. [ March 14, 2016]. Available at: http://www.cancer.org/cancer/ stomachcancer/detailedguide/stomach-cancer-key-statistics. 5. Ahmad R, Schmidt BH, Rattner DW, Mullen JT. Factors influencing readmission after curative gastrectomy for gastric cancer. J Am Coll Surg. 2014;218:1215e1222. 6. Bonenkamp JJ, Songun I, Hermans J, et al. Randomised comparison of morbidity after D1 and D2 dissection for gastric cancer in 996 Dutch patients. Lancet. 1995;345:745e748. 7. Cuschieri A, Fayers P, Fielding J, et al. Postoperative morbidity and mortality after D1 and D2 resections for gastric cancer: preliminary results of the MRC randomised controlled surgical trial. The Surgical Cooperative Group. Lancet. 1996;347:995e999. 8. Gil-Rendo A, Hernandez-Lizoain JL, Martinez-Regueira F, et al. Risk factors related to operative morbidity in patients undergoing gastrectomy for gastric cancer. Clin Transl Oncol. 2006;8:354e361. 9. Grossmann EM, Longo WE, Virgo KS, et al. Morbidity and mortality of gastrectomy for cancer in Department of Veterans Affairs Medical Centers. Surgery. 2002;131:484e490. 10. Kim MC, Kim KH, Jung GJ. A 5 year analysis of readmissions after radical subtotal gastrectomy for early gastric cancer. Ann Surg Oncol. 2012;19:2459e2464. 11. Papenfuss WA, Kukar M, Oxenberg J, et al. Morbidity and mortality associated with gastrectomy for gastric cancer. Ann Surg Oncol. 2014;21:3008e3014. 12. Selby LV, Vertosick EA, Sjoberg DD, et al. Morbidity after total gastrectomy: analysis of 238 patients. J Am Coll Surg. 2015;220:863e871 e2. 13. Zhuang CL, Wang SL, Huang DD, et al. Risk factors for hospital readmission after radical gastrectomy for gastric cancer: a prospective study. PLoS One. 2015;10:e0125572. 14. Elixhauser A, Steiner C, Harris DR, Coffey RM. Comorbidity measures for use with administrative data. Med Care. 1998;36:8e27.
183
15. Birkmeyer JD, Finks JF, O’Reilly A, et al. Surgical skill and complication rates after bariatric surgery. N Engl J Med. 2013;369:1434e1442. 16. Keller DS, Chien HL, Hashemi L, Senagore AJ, Delaney CP. The HARM score: a novel, easy measure to evaluate quality and outcomes in colorectal surgery. Ann Surg. 2014;259:1119e1125. 17. Tsai TC, Joynt KE, Orav EJ, Gawande AA, Jha AK. Variation in surgical-readmission rates and quality of hospital care. N Engl J Med. 2013;369:1134e1142. 18. [cited 2016 January 14]. Available at: http://www.beckersho spitalreview.com/quality/the-38-hospitals-facing-highestmedicare-penalties-for-readmissions.html. 19. [cited 2016 January 15]. Available at: http://www.beckersho spitalreview.com/quality/2-5k-hospitals-penalized-by-cmsfor-high-readmissions-10-things-to-know.html. 20. Merchant SJ, Ituarte PH, Choi A, et al. Hospital readmission following surgery for gastric cancer: frequency, timing, etiologies, and survival. J Gastrointest Surg. 2015;19:1769e1781. 21. Bartlett EK, Roses RE, Kelz RR, Drebin JA, Fraker DL, Karakousis GC. Morbidity and mortality after total gastrectomy for gastric malignancy using the American College of Surgeons National Surgical Quality Improvement Program database. Surgery. 2014;156:298e304. 22. Hartgrink HH, van de Velde CJ, Putter H, et al. Extended lymph node dissection for gastric cancer: who may benefit? Final results of the randomized Dutch gastric cancer group trial. J Clin Oncol. 2004;22:2069e2077. 23. Sano T, Sasako M, Yamamoto S, et al. Gastric cancer surgery: morbidity and mortality results from a prospective randomized controlled trial comparing D2 and extended para-aortic lymphadenectomyeJapan Clinical Oncology Group study 9501. J Clin Oncol. 2004;22:2767e2773. 24. Kassin MT, Owen RM, Perez SD, et al. Risk factors for 30-day hospital readmission among general surgery patients. J Am Coll Surg. 2012;215:322e330. 25. Kelly KN, Iannuzzi JC, Rickles AS, Monson JR, Fleming FJ. Risk factors associated with 30-day postoperative readmissions in major gastrointestinal resections. J Gastrointest Surg. 2014;18:35e43. discussion 43-44. 26. Stitzenberg KB, Chang Y, Smith AB, Nielsen ME. Exploring the burden of inpatient readmissions after major cancer surgery. J Clin Oncol. 2015;33:455e464. 27. Martin BJ, Chen G, Graham M, Quan H. Coding of obesity in administrative hospital discharge abstract data: accuracy and impact for future research studies. BMC Health Serv Res. 2014;14:70. 28. Mor V, Intrator O, Feng Z, Grabowski DC. The revolving door of rehospitalization from skilled nursing facilities. Health Aff (Millwood). 2010;29:57e64. 29. Hakkarainen TW, Arbabi S, Willis MM, Davidson GH, Flum DR. Outcomes of patients discharged to skilled nursing facilities after acute care hospitalizations. Ann Surg. 2016;263:280e285. 30. [ March 20, 2016]. Available at: http://www.fha.org/show Document.aspx?f¼SNFMCARE_-FY2016-FR-Summary.pdf.
184
j o u r n a l o f s u r g i c a l r e s e a r c h a p r i l 2 0 1 8 ( 2 2 4 ) 1 7 6 e1 8 4
Appendix. ICD-9-CM complication codes.
Complication
ICD-9-CM code (secondary diagnosis codes)
Digestive system
997.49
Infection
998.5
Wound infection
998.3
Persistent postoperative fistula
998.6
Complications peculiar to certain specified procedures
996
Complications affecting specified body system not elsewhere classified
997
Other complications of procedures not elsewhere classified
998
Complications of medical care not elsewhere classified
999
Acute posthemorrhagic anemia
285.1
Pulmonary embolism and deep venous thrombosis Pulmonary insufficiency after surgery
415.1, 451.1, 451.2, 451.8, 451.9, 453.2, 453.8, 453.9 518.5