ww.sciencedirect.com

j o u rn a l o f s u r g i c a l r e s e a r c h x x x ( 2 0 1 4 ) 1e8

Available online at w

ScienceDirect

journal homepage: www.JournalofSurgicalResearch.com

Early discharge and readmission after colorectalresection

Rebecca L. Hoffman, MD,a,* Edmund K. Bartlett, MD,a Clifford Ko, MD,b

Najjia Mahmoud, MD,a Giorgos C. Karakousis, MD,a

and Rachel R. Kelz, MD, MSCEa

aDepartment of Surgery, Perelman School of Medicine at the University of Pennsylvania, Philadelphia, PennsylvaniabDepartment of Surgery, David Geffen School of Medicine at the University of California, Los Angeles, California

a r t i c l e i n f o

Article history:

Received 4 January 2014

Received in revised form

31 January 2014

Accepted 11 February 2014

Available online xxx

Keywords:

Early discharge

Readmission

Colon

Colorectal

Surgery

Laparoscopic

ACS NSQIP

* Corresponding author. Department of SurgeStreet, Philadelphia, PA 19104. Tel.: þ1 267 2

E-mail address: rgaugler@gmail.com (R.L0022-4804/$ e see front matter ª 2014 Elsevhttp://dx.doi.org/10.1016/j.jss.2014.02.006

a b s t r a c t

Background: Emphasis on the provision of high quality, cost-effective healthcare has meant

increasing efforts at reducing postoperative length of stay while reducing 30-d readmission

rates. The aim of this study was to identify factors associated with early discharge (ED) and

to evaluate the effectof ED on readmission after colorectal resection.

Materials and methods: We identified all inpatients aged �18 y who underwent a colorectal

resection in the American College of Surgeons National Surgical Quality Improvement

Program Participant Use File, 2011. ED was defined as a length of stay �25th percentile by

procedure (rectal resection, open colectomy, and laparoscopic colectomy). Multivariate

logistic regression was used to identify factors significantly associated with ED and read-

mission. A subset analysis was performed by procedure type.

Results: Of 28,532 patients, 2171 (7%) underwent rectal resection, 14,976 (52%) underwent

open colectomy, and 11,385 (40%) underwent laparoscopic colectomy with an ED on or

before postoperative days 5, 5, and 3, respectively. The overall cohort included patients

with a mean age of 61 y. A total of 52% were women and 37% were colorectal cancer pa-

tients. Age >65 y, recent steroid use, simultaneous ostomy creation, nonelective surgery,

need for reoperation, and a postoperative occurrence before discharge were significantly

associated with a reduced likelihood of ED. The overall rate of readmission was 12%. Pa-

tients who were discharged early were significantly less likely to be readmitted (odds ratio,

0.77; 95% confidence interval, 0.70e0.84).

Conclusions: In the appropriate patient population, ED after colorectal surgery may be

implemented without any adverse effect on readmission rates.

ª 2014 Elsevier Inc. All rights reserved.

1. Introduction suggested that early discharge (ED) was associated with

A national emphasis on the provision of high quality, cost-

effective health care has meant increasing efforts at

reducing postoperative length of stay (LOS) while simulta-

neously reducing 30-d readmission rates. Early research

ry, Perelman School of M75 3290; fax: þ1 215 662 7. Hoffman).ier Inc. All rights reserved

increased rates of readmission. Therefore, concerns exist

among surgeons that the inability to monitor patients’ clinical

progress and detect complications would result in higher

readmission rates and occurrences diagnosed after discharge

[1e3].

edicine at the University of Pennsylvania, 4 Maloney, 3400 Spruce983.

.

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 x x x ( 2 0 1 4 ) 1e82

Colorectal surgery provides a primemodel for investigating

the effects of ED on patient outcomes and readmission

because of the frequency of procedures performed and the

significant rates of postoperative occurrences (POs). Accord-

ingly, recent data have been published regarding the rela-

tionship among LOS, ED, and readmissions after colorectal

surgery. Single-institution experiences have demonstrated

the benefits of early recovery pathways after colorectal sur-

gery with little to no effect on readmission rates [4e7]. At the

population level, however, LOS has decreased after colon

surgery, whereas readmission rates have increased over the

last two decades [8].

The recent addition of 30-d readmission information to the

American College of Surgeons National Surgical Quality

Improvement Program (ACS NSQIP) [9] provides a new op-

portunity to evaluate ED and readmission rates. In this study,

we sought to identify factors associated with ED and evaluate

the effect of ED on readmission after colorectal resection.

Fig. 1 e Selection of the study cohort.

2. Methods

We performed a retrospective study of prospectively collected

data from the ACS NSQIP Participant Use File (PUF) data set

from 2011. The ACS NSQIP PUF contains data on 442,149 cases

collected from 315 academic and community-based hospitals

located around the United States. A trained Surgical Clinical

Reviewer captures data on 252 variables, including preopera-

tive risk factors, intraoperative variables, and 30-

d postoperative morbidity and mortality outcomes. Patients

aged �18 y undergoing major surgical procedures (both

inpatient and outpatient) are included using an 8-d cycle

sampling procedure. All variables collected in the ACS NSQIP

are predefined in the NSQIP PUF 2011 user guide [10].

We identified all inpatients aged �18 y who underwent

colorectal resection in the 2011 ACS NSQIP PUF. Colorectal

resectionwas defined using Common Procedural Terminology

codes for both open and laparoscopic procedures, including

44140e44147, 44150, 44151, 44155e44158, 44160, 44204e44208,

44210e44212, 45110e45114, 45116, 45119e45123, 45126, 45135,

45136, 45395, and 45397. Patients missing information on LOS

and thosewith a LOS recorded as�0 dwere excluded from the

analysis. In addition, patients who were listed as still in the

hospital and those who died during the initial inpatient hos-

pitalization were excluded (Fig. 1).

Patient demographic characteristics including age, sex,

and race were abstracted from the NSQIP database, as was

information regarding preoperative comorbidities and

whether the procedure was performed electively or emer-

gently. Heart disease was assigned for all patients with a

recorded history of congestive heart failure, myocardial

infarction, percutaneous coronary intervention or stenting,

and/or angina. A history of colorectal cancer was defined

using a postoperative International Classification of Diseases,

Ninth Edition (ICD-9) diagnosis code of 153, 153.0, 154, 154.0, or

197.7. The simultaneous creation of an ostomy was deter-

mined using the Common Procedural Technology codes

44141, 44143, 44144, 44146, 44188, 44206, 44208, 44320, 44322,

44340, 44345, 44346, 45110, 45395, 44150, 44151, 44156, 44157,

44158, 44211, and 44187.

To examine the effects of patient complications on the

likelihood of ED and that of readmission, patients were clas-

sified by the presence or absence of any PO. Information

regarding POs, including cardiac, respiratory, infectious, renal

and neurologic events, and an unplanned return to the oper-

ating room was noted. More specifically, POs categorized as

wound complication (superficial skin infection, deep surgical

site infection, or fascial dehiscence), organ space infection,

sepsis (sepsis or septic shock), renal (progressive renal failure

or acute renal failure requiring dialysis), venous thrombo-

embolism (deep vein thrombosis or pulmonary embolism),

respiratory (pneumonia, intubation for greater than 48 h, or

reintubation), bleeding (hemorrhage requiring transfusion of

at least 4 U of blood), neurologic (stroke or coma), urinary tract

infection, or cardiac (myocardial infarction or arrest requiring

resuscitation) were abstracted directly from the ACS NSQIP

PUF, and patients were classified by occurrence status: no

occurrence, any occurrence before hospital discharge, and

any occurrence after hospital discharge.

The primary outcome variable was procedure-specific ED.

The secondary outcome of interest was 30-d readmission. LOS

was defined as the number of days from the index operation to

hospital discharge (to home or any facility). Because it is

commonly acknowledged that differences in LOS exist by

procedure, procedure-specific EDwas defined as an LOS�25th

percentile for rectal resection, open colectomy, and laparo-

scopic colectomy, respectively [11,12]. ED was reported as a

binary outcome. Readmission refers to an admission to any

hospital within 30 d of the principal surgical procedure, as

defined within the NSQIP PUF.

Descriptive statistics were performed. Patient and proce-

dure characteristics were examined by ED status using the

Student t and chi-square tests, as appropriate. Multivariate

logistic regression was used to identify factors significantly

associated with ED. The association between ED and

Table 1e Patient characteristics for the overall cohort andby discharge status.

Patientcharacteristics

Overallcohort,n (%)

Earlydischarge

Notdischarged

early

Pvalue

N 28,532 8641 (30) 19,891 (70)

Age (y, mean � SD) 61.1 � 15.6 58.4 � 14.8 62.3 � 15.8 <0.01

Sex

Female 14,912 (52) 4537 (53) 10,375 (52)

Male 13,557 (48) 4090 (47) 9467 (48) 0.64

j o u rn a l o f s u r g i c a l r e s e a r c h x x x ( 2 0 1 4 ) 1e8 3

readmission was investigated using univariate and multivar-

iate logistic regression to adjust for potential confounders.

Independent logistic regression models were developed to

examine the association between ED and readmission for

each of the procedure groups with adjustment for potential

confounders. All statistical analyses were performed using

STATA, version 12.1 (STATA Corp, College Station, TX). This

study was reviewed by the Institutional Review Board of the

University of Pennsylvania and deemed exempt from

continuing review (IRB#818956).

Race <0.01

Caucasian 22,459 (79) 7066 (82) 15,393 (77)

Black 2525 (9) 633 (7) 1892 (10)

Asian 723 (3) 243 (3) 480 (2)

Other 2825 (10) 699 (8) 2126 (11)

Ethnicity-hispanic 1396 (5) 439 (5) 957 (5) <0.01

Comorbidities

Smoke 5139 (18) 1512 (18) 3627 (18) 0.14

Diabetes 4187 (15) 1045 (12) 3142 (16) <0.01

Hypertension 13,958 (49) 3756 (43) 10,202 (51) <0.01

COPD 1595 (6) 294 (3) 1301 (7) <0.01

Ascites 320 (1) 38 (0.4) 282 (1) <0.01

BMI � 30 8621 (31) 2546 (30) 6075 (31) <0.01

Heart disease 1329 (5) 274 (3) 1055 (5) <0.01

Acute renal

failure

153 (0.5) 9 (0.1) 144 (0.7) <0.01

Dialysis 241 (0.8) 36 (0.4) 205 (1) <0.01

Cerebrovascular

disease

654 (2) 126 (1) 528 (3) <0.01

Steroid use 2147 (8) 508 (6) 1639 (8) <0.01

Bleeding

disorder

1357 (5) 215 (2) 1142 (6) <0.01

Colorectal

cancer

10,624 (37) 3243 (38) 7381 (37) 0.50

Intraoperative factors

Ostomy 6983 (24) 1345 (16) 5638 (28) <0.01

Emergent

surgery

3976 (14) 625 (7) 3351 (17) <0.01

Complications

Return to OR 1501 (5) 195 (2) 1306 (7) <0.01

No complication 19,875 (74) 7357 (88) 12,518 (68) <0.01

PO before

discharge

4755 (18) 273 (3) 4482 (24) <0.01

PO after

discharge

2253 (8) 778 (9) 1475 (8) <0.01

Readmission 3325 (12) 839 (10) 2486 (13) <0.01

BMI ¼ body mass index; SD ¼ standard deviation.

3. Results

Of 31,267 colorectal procedures included in the data set, 28,532

(91%) patients were included in the study. Patient character-

istics can be viewed in Table 1. In the overall cohort, the me-

dian LOS was 6 d (range, 1e130). A total of 7% (n ¼ 2171) of

patients underwent rectal resection, 52% (n ¼ 14,976) under-

went an open colectomy, and 40% (n ¼ 11,385) underwent a

laparoscopic colectomy. Within these three groups, the me-

dian surgical LOSs were 7 d for rectal resections and open

colectomies, and 4 d for laparoscopic colectomies. Procedure-

specific ED, defined as an LOS �25th percentile, corresponded

to a day of discharge on or before postoperative day 5 for pa-

tients undergoing a rectal resection, day 5 for patients un-

dergoing an open colon resection, and day 3 for patients

undergoing a laparoscopic colectomy (see Fig. 2).

Of all patients, 30% (n ¼ 8641) were discharged early. The

overall readmission rate was 12% (n ¼ 3325) with 17% (n ¼ 346)

of rectal resection patients, 15% (n ¼ 2027) of open colectomy

patients, and 10% (n ¼ 1048) of patients undergoing a laparo-

scopic colon resection requiring a readmission. There were

8280 (29%) patients who experienced a PO. Complete infor-

mation regarding the timing of the PO was available for 7008

patients, and 4755 (68%) experienced the event before

discharge. Despite having a PO diagnosed before discharge,

26% of the patients were discharged early.

In a multivariate analysis of the overall cohort done to

identify factors associated with ED, ages 65e79 y (odds ratio

[OR], 0.75; 95% confidence interval [CI], 0.64e0.87) and >80 y

(OR, 0.50; 95% CI, 0.41e0.60), chronic obstructive pulmonary

disease (COPD; OR, 0.60; 95% CI, 0.49e0.75), history of cere-

brovascular accident (OR, 0.63; 95% CI, 0.42e0.94), recent ste-

roid use (OR, 0.80; 95% CI, 0.68e0.95), nonelective surgery (OR,

0.58; 95% CI, 0.50e0.68), simultaneous ostomy creation (OR,

0.65; 95% CI, 0.58e0.72), need for reoperation (OR, 0.57; 95% CI,

0.44e0.73), and a PO before discharge (OR, 0.13; 95% CI,

0.11e0.16) were significantly associated with a reduced like-

lihood of ED. Simultaneous ostomy creation and a PO before

dischargewere the only covariates that remained significantly

associated with a reduced likelihood of ED in the threemodels

developed for the subset analysis by procedure type. In the

two independentmodels developed for open and laparoscopic

colectomy, age>80 y, a history of COPD, a procedure classified

as emergent, and the need for reoperation were also signifi-

cantly associated with a reduced likelihood of ED. Recent

steroid use was significantly associated with a reduced like-

lihood of ED in only the laparoscopic colectomy model (see

models in Appendix A).

Of the patients in the overall cohort who were discharged

early, 10% were readmitted versus 13% of patients who were

not discharged early (P< 0.01). The proportion of patients that

were readmitted after rectal resection differed significantly by

ED status (14% ED versus 17% not ED; P ¼ 0.02). The same was

true for those undergoing open colon resection (12% ED versus

14% not ED; P � 0.01) and among patients undergoing lapa-

roscopic colon resection (7% ED versus 10% not ED; P� 0.01; see

Table 2 for unadjusted ORs).

In the multivariate model after adjustment for potential

confounders, ED patients remained significantly less likely to

be readmitted after colorectal resection than patients dis-

charged later in the postoperative course (OR, 0.80; 95% CI,

0.70e0.93). In the subset analysis, after adjustment for

Fig. 2 e Median LOS by procedure type. Procedure-specific

ED was defined as an LOS £25th percentile. (For

interpretation of the references to color in this figure, the

reader is referred to the web version of this article).

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 x x x ( 2 0 1 4 ) 1e84

potential confounders, ED remained significantly associated

with a reduced likelihood of readmission in the laparoscopic

colectomy cohort (OR, 0.55; 95% CI, 0.41e0.73) but did not

reach significance for patients undergoing open colectomies

(OR, 0.88; 95% CI; 0.73e1.07) or rectal resections (OR, 0.82; 95%

CI, 0.52e1.30). The simultaneous creation of an ostomy was

significantly associated with a reduced likelihood of read-

mission in patients undergoing rectal resection (OR, 0.60; 95%

CI, 0.39e0.93) and significantly associated with an increased

likelihood of readmission for patients undergoing a colon

resection (laparoscopic: OR, 1.56; 95% CI, 1.08e2.25 and open:

OR, 1.24; 95% CI, 1.04e1.48). The need for emergent surgery

was not associatedwith readmission in the overall cohort (OR,

0.87; 95% CI, 0.72e1.06) or after any specific procedure. The

most highly significant predictors of readmission in the

overall model were POs diagnosed both before (OR, 1.53; 95%

CI, 1.29e1.81) and after discharge (OR, 14.51; 95% CI;

12.41e16.97). A PO diagnosed after discharge was associated

with readmission regardless of procedure type, whereas a PO

diagnosed before discharge increased readmission risk after

all procedures except laparoscopic colectomy (Table 2).

In the overall cohort, factors associated with both a

decreased likelihood of ED and an increased likelihood of

readmission include age >65 y, a history of COPD, recent

steroid use, the simultaneous creation of an ostomy, and a PO

diagnosed before discharge.

4. Discussion

Numerous single-institution reports have commented on the

relationship between ED and readmission using colorectal

surgery as the prototypical example [4e7]. Here, we report the

results of our multi-institutional study of colorectal surgery

patients demonstrating that patient and procedure factors

influenced the likelihood of successful ED, and that ED after

colorectal resection was not associated with an increased risk

of readmission before and after adjustment for potential

confounders. Moreover, we found that EDwas associatedwith

a reduced likelihood of readmission in the overall cohort and

among patients undergoing laparoscopic colon resection with

no difference in the likelihood of readmission after ED among

those undergoing rectal surgery and open colon resection.

Furthermore, we found that a history of significant comor-

bidities, such as recent steroid use and the need for an

emergent operation, and operative factors, such as the

simultaneous creation of an ostomy and a PO diagnosed

before discharge, negatively influenced the likelihood of ED.

In surgery, the national emphasis on quality improvement

and cost containment in healthcare has translated into the

need to facilitate expeditious discharge without affecting the

rate of readmission. The addition of 30-d readmission to the

2011 ACS NSQIP database afforded us the opportunity to test

this balancing act on a national scale. We defined ED using a

procedure-specific LOS �25th percentile. This definition was

conceived after Collins et al. [13], in an analysis of risk factors

for a prolonged LOS using the Veterans Administration Sur-

gical Quality Improvement Program, using an LOS of �75th

percentile to represent “prolonged.” Hendren et al. [14], in an

analysis of ED and readmission in Medicare colon cancer

patients, used a cutoff of 5 d for ED, based on the single-

institution studies of enhanced recovery pathways. Our defi-

nition of early, meaning patients discharged on or before days

5, 5, and 3 for rectal resection, open colectomy, and laparo-

scopic colectomy, respectively, proved to be consistent with

previous studies, which used ED protocols, where early was

defined by an LOS of 2e5 d [1,15,16].

Operative and postoperative factors have been shown to

have the most influence on LOS after major surgery [13]. The

odds of an ED in this study were significantly reduced when

patients’ surgery was classified as emergent, or if patients

experienced a PO. Preoperative patient factors also influence

the likelihood of ED after surgery. Although there was no

difference in ED by sex, the chance of being discharged early

dropped dramatically with increasing age such that patients

>65 y were 30% less likely and those 80 y and older were half

as likely to be discharged early. However, in a study of elderly

patients undergoing open colon resection, DiFronzo et al. [17]

demonstrated the safe implementation of an early recovery

pathway when discharge occurred on postoperative day 3.

The relative reluctance to discharge elderly patients and pa-

tients with a history of serious comorbidities, such as COPD,

steroid use, and new ostomies, may be reflective of appro-

priate patient selection for ED, but may also be reflective of

fears of readmission in an era when readmission is a quality

indicator.

Our findings support the wealth of single-institution

studies demonstrating that ED and readmission are compat-

ible. Readmission rates in these studies, when patients are

discharged early, vary from 11%e16% [14,18] and are compa-

rable, if not improved, to traditional recovery LOS. The rate of

readmission for patients discharged early (10%) in our large

sample was significantly lower than patients discharged after

a longer LOS (13%). Even when controlling for preoperative

comorbidities, the risk of readmission after ED was 20% less

likely than it was for patients discharged after a longer LOS. To

our knowledge, this is only the second study to examine this

relationship between ED and readmission in a large multi-

institutional data set. In a 2011 study by Hendren et al. [14],

an analysis of 477,000Medicare patients across 6 y determined

that readmission rates were not increased with ED. Our study

builds on that work by expanding the study population to

Table 2 e Univariate and multivariate odds ratios of factors associated with readmission after colorectal resection in theoverall cohort and by procedure group.

Variable Overall cohort, OR(95% CI) n ¼ 25,971

Rectal resection, OR(95% CI) n ¼ 1984

Open colon resection,OR (95% CI) n ¼ 13,721

Laparoscopic colonresection, OR (95% CI)

n ¼ 10,266

Univariate models

Discharge >25th

percentile

REF REF REF REF

Early discharge 0.77 (0.70e0.84) 0.84 (0.65e1.07) 0.82 (0.73e0.91) 0.64 (0.55e0.75)

Multivariate models

Discharge >25th

percentile

REF REF REF REF

ED 0.80 (0.69e0.93) 0.82 (0.52e1.30) 0.88 (0.73e1.07) 0.55 (0.41e0.73)

Age category (y)

18e39 REF REF REF REF

40e49 0.78 (0.61e1.00) 0.75 (0.37e1.52) 0.91 (0.64e1.30) 0.70 (0.45e1.09)

50e64 0.74 (0.60e0.92) 0.54 (0.29e1.01) 0.98 (0.72e1.33) 0.60 (0.41e0.89)

65e79 0.68 (0.54e0.86) 0.76 (0.39e1.49) 0.85 (0.62e1.17) 0.57 (0.37e0.86)

�80 0.70 (0.53e0.92) 0.60 (0.25e1.49) 0.88 (0.61e1.28) 0.53 (0.32e0.89)

Race

White REF REF REF REF

Black 1.24 (1.01e1.53) 1.04 (0.48e2.24) 1.12 (0.87e1.46) 1.62 (1.11e2.36)

Asian 0.87 (0.55e1.36) 1.12 (0.40e3.14) 0.92 (0.47e1.80) 0.71 (0.31e1.60)

Other 0.97 (0.80e1.89) 1.05 (0.57e1.92) 0.91 (0.70e1.18) 1.11 (0.77e1.62)

Comorbidities

Diabetes 1.20 (1.00e1.42) 0.97 (0.51e1.84) 1.12 (0.89e1.40) 1.37 (1.00e1.89)

Hypertension 1.00 (0.86e1.16) 1.17 (0.72e1.91) 1.04 (0.86e1.26) 0.91 (0.70e1.20)

COPD 1.35 (1.06e1.72) 2.28 (0.88e5.92) 1.33 (1.00e1.79) 1.17 (0.71e1.95)

Ascites 1.33 (0.79e2.26) d 1.39 (0.81e2.39) 0.67 (0.07e6.18)

BMI � 30 0.96 (0.68e1.36) 0.62 (0.25e1.52) 1.21 (0.77e1.90) 0.85 (0.42e1.75)

Heart disease 1.44 (1.16e1.78) 0.30 (0.11e0.83) 1.50 (1.15e1.96) 1.72 (1.15e2.57)

Acute renal failure 0.87 (0.41e1.86) d 0.89 (0.40e1.98) 0.88 (0.09e8.38)

Dialysis 1.65 (0.96e2.86) 22.75 (1.57e329.39) 1.39 (0.74e2.58) 1.91 (0.47e7.71)

Cerebrovascular

disease

1.18 (0.78e1.79) 11.22 (2.28e55.33) 0.90 (0.53e1.52) 1.54 (0.69e3.44)

Steroid use 1.50 (1.22e1.84) 1.03 (0.41e2.58) 1.39 (1.07e1.81) 1.94 (1.34e2.80)

Bleeding disorder 1.15 (0.87e1.52) 0.90 (0.20e4.02) 1.06 (0.77e1.46) 1.66 (0.92e3.00)

Intraoperative factors

Ostomy 1.30 (1.13e1.50) 0.60 (0.39e0.93) 1.24 (1.04e1.48) 1.56 (1.08e2.25)

Emergent surgery 0.87 (0.72e1.06) 0.95 (0.18e4.92) 0.87 (0.71e1.07) 0.66 (0.33e1.30)

Complications

Return to OR 0.26 (0.21e0.32) 0.22 (0.11e0.44) 0.34 (0.26e0.44) 0.14 (0.10e0.21)

No complication REF REF REF REF

Postoperative

occurrence before

discharge

1.53 (1.29e1.81) 2.46 (1.44e4.20) 1.54 (1.25e1.90) 0.97 (0.66e1.44)

Postoperative

occurrence after

discharge

14.51 (12.41e16.97) 23.89 (14.32e39.84) 14.48 (11.75e17.83) 12.17 (9.12e16.23)

REF ¼ reference.

Bold indicates statistical significance. For all covariates listed in italics, the reference group was the cohort of patients without the factor.

j o u rn a l o f s u r g i c a l r e s e a r c h x x x ( 2 0 1 4 ) 1e8 5

include younger patients and through the use of ACS NSQIP

data that are not subject to the limitations inherent to

administrative claims [19,20]. In fact, just over half (56%) of

patients included in our study cohort were younger than 65 y.

This may explain our ability to identify a reduced risk of

readmission among the ED cohort undergoing laparoscopic

colon resection when compared with that shown by Hendren

et al.[14]

Enhanced recovery pathways have been developed to

facilitate ED after colorectal surgery. Improvements in patient

satisfaction, decreased postoperative morbidity and mortal-

ity, and decreased costs are among the benefits of enhanced

recovery initiatives in procedures ranging from esoph-

agectomy to pulmonary lobectomy, and to intestinal surgery

[21e23]. Yet despite the endorsement of well-defined, evi-

dence-based protocols [24,25], there is still relative reluctance

with which early recovery pathways are used. In a national

survey of general and colorectal surgeons, less than one third

incorporated formal clinical care pathways into practice [26].

To our knowledge, this is the first study that uses the ACS

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 x x x ( 2 0 1 4 ) 1e86

NSQIP sample to identify factors associated with ED and to

provide evidence of a reduced risk of readmission after ED.

The fact that our findings using amulti-institutional sample of

colorectal patients agree with the reports of numerous single-

institution trials, and provide a complementary population to

Medicare ED and readmission data, may provide a significant

impetus for the potential adoption of standardized post-

operative recovery pathways to facilitate the ED of appropri-

ately selected patients.

Our study has several important limitations. First, this

study is subject to selection bias, in that the patientswhowere

discharged early were clearly a different group of patients

than those of the traditional LOS cohort. This selection bias

does, however, allow us to generalize about the group of pa-

tients who were, in a multi-institutional sample, deemed safe

by their physician for discharge at an earlier postoperative

date. Furthermore, the selection bias has no influence on the

readmission status of the patients, and therefore should not

influence the significance of our results. Additionally, this

study suffers the limitations characteristic of any study using

large registry data. We are unable to determine social factors

that may influence the likelihood of discharge, adjust for

intraoperative technical difficulty, surgeon practice prefer-

ence, and hospital case mix and practice patterns. Finally, we

do not have information regarding the use of an enhanced

recovery or fast track protocols. Therefore, we cannot vouch

for the safety of these pathways per se, but can only use our

definition of ED as the closest possible surrogate for the indi-

vidual level data that we lack.

5. Conclusions

One third of patients in the ACS NSQIP PUF undergoing major

colorectal surgery are now discharged in less than 5 d. In the

appropriate patient population, ED after colorectal surgery

may be implemented without any adverse effect on read-

mission rates. The information learned from this study can be

used to inform surgeons about the key relationships among

patient characteristics, ED, and readmission. In doing so, it

may facilitate an increased interest in the adoption of

enhanced recovery pathways.

Acknowledgment

Author contributions: R.L.H. and R.R.K. contributed to the

conception and design. R.L.H., E.K.B., C.K., N.M., G.C.K., and

R.R.K. did the analysis and interpretation. R.L.H. and R.R.K.

collected data. R.L.H. and R.R.K. wrote the article. E.K.B.,

C.K., N.M., G.C.K., and R.R.K. did the critical appraisal and

revision.

Disclosure

The authors reported no proprietary or commercial interest in

any product mentioned or concept discussed in this article.

r e f e r e n c e s

[1] Basse L, Thorbol JE, Lossl K, Kehlet H. Colonic surgery withaccelerated rehabilitation or conventional care. Dis ColonRectum 2004;47:271. discussion 277e8.

[2] McAleese P, Odling-Smee W. The effect of complications onlength of stay. Ann Surg 1994;220:740.

[3] Nygren J, Hausel J, Kehlet H, et al. A comparison in fiveEuropean centres of case mix, clinical management andoutcomes following either conventional or fast-trackperioperative care in colorectal surgery. Clin Nutr 2005;24:455.

[4] Fierens J, Wolthuis AM, Penninckx F, D’Hoore A. Enhancedrecovery after surgery (ERAS) protocol: prospective study ofoutcome in colorectal surgery. Acta Chir Belg 2012;112:355.

[5] Gouvas N, Gogos-Pappas G, Tsimogiannis K, Tsimoyiannis E,Dervenis C, Xynos E. Implementation of fast-track protocolsin open and laparoscopic sphincter-preserving rectal cancersurgery: a multicenter, comparative, prospective, non-randomized study. Dig Surg 2012;29:301.

[6] Lawrence JK, Keller DS, Samia H, et al. Discharge within 24 to72 hours of colorectal surgery is associated with lowreadmission rates when using Enhanced Recovery Pathways.J Am Coll Surg 2013;216:390.

[7] Basse L, Hjort Jakobsen D, Billesbolle P, Werner M, Kehlet H.A clinical pathway to accelerate recovery after colonicresection. Ann Surg 2000;232:51.

[8] Schneider EB, Hyder O, Brooke BS, et al. Patient readmissionand mortality after colorectal surgery for colon cancer:impact of length of stay relative to other clinical factors. J AmColl Surg 2012;214:390. discussion 8e9.

[9] American College of Surgeons National Surgical QualityImprovement Program. http://site.acsnsqip.org/ [Accessed8.01.2013].

[10] Program ACoSNSQI. User guide for the 2011 participant usedata file; 2012.

[11] Lacy AM, Garcia-Valdecasas JC, Delgado S, et al.Laparoscopy-assisted colectomy versus open colectomy fortreatment of non-metastatic colon cancer: a randomisedtrial. Lancet 2002;359:2224.

[12] Lezoche E, Feliciotti F, Paganini AM, et al. Laparoscopic vsopen hemicolectomy for colon cancer. Surg Endosc 2002;16:596.

[13] Collins TC, Daley J, Henderson WH, Khuri SF. Risk factors forprolonged length of stay after major elective surgery. AnnSurg 1999;230:251.

[14] Hendren S, Morris AM, Zhang W, Dimick J. Early dischargeand hospital readmission after colectomy for cancer. DisColon Rectum 2011;54:1362.

[15] Adamina M, Kehlet H, Tomlinson GA, Senagore AJ,Delaney CP. Enhanced recovery pathways optimize healthoutcomes and resource utilization: a meta-analysis ofrandomized controlled trials in colorectal surgery. Surgery2011;149:830.

[16] King PM, Blazeby JM, Ewings P, et al. Randomized clinicaltrial comparing laparoscopic and open surgery for colorectalcancer within an enhanced recovery programme. Br J Surg2006;93:300.

[17] DiFronzo LA, Yamin N, Patel K, O’Connell TX. Benefits ofearly feeding and early hospital discharge in elderly patientsundergoing open colon resection. J Am Coll Surg 2003;197:747.

[18] Andersen J, Hjort-Jakobsen D, Christiansen PS, Kehlet H.Readmission rates after a planned hospital stay of 2 versus 3days in fast-track colonic surgery. Br J Surg 2007;94:890.

[19] Levine MN, Julian JA. Registries that show efficacy: good, butnot good enough. J Clin Oncol 2008;26:5316.

j o u rn a l o f s u r g i c a l r e s e a r c h x x x ( 2 0 1 4 ) 1e8 7

[20] Nathan H, Pawlik TM. Limitations of claims and registrydata in surgical oncology research. Ann Surg Oncol 2008;15:415.

[21] Schwarzbach MH, Ronellenfitsch U, Wang Q, et al. Effects ofa clinical pathway for video-assisted thoracoscopic surgery(VATS) on quality and cost of care. Langenbecks Arch Surg2010;395:333.

[22] Cerfolio RJ, Bryant AS, Bass CS, Alexander JR, Bartolucci AA.Fast tracking after Ivor Lewis esophagogastrectomy. Chest2004;126:1187.

[23] O’Brien DP. Predictors and outcome of readmission afterlaparoscopic intestinal surgery. World J Surg 2007;31:2138.

[24] Gustafsson UO, Scott MJ, Schwenk W, et al. Guidelines forperioperative care in elective colonic surgery: EnhancedRecovery After Surgery (ERAS�)Society recommendations.Clin Nutr 2012;31:783.

[25] Nygren J, Thacker J, Carli F, et al. Guidelines for perioperativecare in elective rectal/pelvic surgery: Enhanced RecoveryAfter Surgery (ERAS�) Society recommendations. Clin Nutr2012;31:801.

[26] Delaney CP, Senagore AJ, Gerkin TM, et al. Association ofsurgical care practices with length of stay and use of clinicalprotocols after elective bowel resection: results of a nationalsurvey. Am J Surg 2010;199:299. discussion.

Appendix A e Multivariable odds ratios of factors associated with ED after colorectal resection in the overall cohort and byprocedure type.

Variable Overall cohort, OR(95% CI) n ¼ 25,971

Rectal resection, OR(95% CI) n ¼ 1984

Open colon resection, OR(95% CI) n ¼ 13,721

Laparoscopic colonresection, OR (95% CI)

n ¼ 10,266

Age categories (y)

18e39 REF REF REF REF

40e49 1.04 (0.88e1.23) 1.60 (0.97e2.63) 0.85 (0.67e1.08) 1.20 (0.91e1.58)

50e64 1.00 (0.87e1.16) 1.26 (0.82e1.94) 0.78 (0.63e0.97) 1.26 (0.99e1.60)

65e79 0.75 (0.64e0.87) 0.82 (0.50e1.34) 0.61 (0.48e0.76) 0.90 (0.70e1.17)

�80 0.50 (0.41e0.60) 0.67 (0.35e1.28) 0.40 (0.30e0.52) 0.50 (0.35e0.71)

Race

White REF REF REF REF

Black 0.72 (0.62e0.84) 0.25 (0.13e0.50) 0.79 (0.65e0.97) 0.64 (0.49e0.84)

Asian 0.91 (0.71e1.17) 0.78 (0.40e1.51) 1.11 (0.74e1.69) 0.84 (0.58e1.22)

Other 0.70 (0.62e0.80) 0.65 (0.43e0.98) 0.61 (0.50e0.73) 0.80 (0.65e0.98)

Comorbidities

Diabetes 0.91 (0.80e1.03) 0.96 (0.61e1.52) 1.03 (0.86e1.22) 0.76 (0.61e0.93)

Hypertension 0.95 (0.86e1.04) 1.26 (0.91e1.76) 0.89 (0.78e1.01) 0.97 (0.84e1.13)

COPD 0.60 (0.49e0.75) 0.49 (0.21e1.11) 0.56 (0.42e0.74) 0.63 (0.43e0.91)

Ascites 0.75 (0.46e1.24) d 0.74 (0.43e1.28) 0.40 (0.09e1.84)

BMI � 30 0.96 (0.75e1.23) 0.98 (0.49e1.96) 0.98 (0.72e1.35) 1.28 (0.77e2.14)

Heart disease 0.90 (0.76e1.07) 1.04 (0.57e1.91) 0.74 (0.59e0.94) 1.11 (0.84e1.46)

Acute renal

failure

0.85 (0.32e2.28) d 0.59 (0.17e2.02) 4.29 (0.53e34.85)

Dialysis 0.57 (0.31e1.07) 1.15 (0.08e16.2) 0.51 (0.24e1.09) 0.50 (0.14e1.84)

Cerebrovascular

disease

0.63 (0.42e0.94) 0.20 (0.03e1.76) 0.61 (0.37e1.01) 0.80 (0.37e1.71)

Steroid use 0.80 (0.68e0.95) 1.33 (0.68e2.59) 0.94 (0.75e1.19) 0.62 (0.46e0.84)

Bleeding

disorder

0.84 (0.65e1.08) 1.21 (0.44e3.32) 0.85 (0.62e1.17) 0.69 (0.42e1.13)

Intraoperative factors

Ostomy 0.65 (0.59e0.73) 0.54 (0.41e0.72) 0.50 (0.44e0.58) 0.45 (0.33e0.61)

Emergent

surgery

0.58 (0.50e0.68) 0.34 (0.06e1.84) 0.50 (0.42e0.59) 0.51 (0.32e0.81)

Complications

Return to OR 0.57 (0.44e0.73) 0.82 (0.39e1.69) 0.52 (0.37e0.73) 0.58 (0.36e0.92)

No

complication

REF REF REF REF

PO before

discharge

0.14 (0.11e0.16) 0.11 (0.06e0.20) 0.13 (0.10e0.16) 0.08 (0.04e0.13)

PO after

discharge

0.99 (0.85e1.14) 0.69 (0.45e1.06) 1.03 (0.85e1.25) 0.81 (0.61e1.06)

BMI ¼ body mass index; REF ¼ reference.

Bold indicates statistical significance. For all covariates listed in italics, the reference group was the cohort of patients without the factor.

Appendix

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 x x x ( 2 0 1 4 ) 1e88