doi:10.1111/codi.12958
Original article
Impact of smoking on the surgical outcome of Crohn’s disease: a propensity-score matched National Surgical Quality Improvement Program analysis A. N. Kulaylat*, C. S. Hollenbeak*†, W. Sangster* and D. B. Stewart Sr* *Department of Surgery, The Pennsylvania State University, College of Medicine, Hershey, Pennsylvania, USA and †Department of Public Health Sciences, The Pennsylvania State University, College of Medicine, Hershey, Pennsylvania, USA Received 3 December 2014; accepted 30 January 2015; Accepted Article online 25 March 2015
Abstract Aim Smoking is known to have a deleterious effect on Crohn’s disease (CD). The present study addressed the specific impact of smoking on the outcome of surgery for CD. Method A review of the National Surgical Quality Improvement Program (NSQIP) database (2005–2012) identified 7631 patients with CD who underwent surgical resection. Patients were stratified based on smoking status and were compared with univariate statistical tests. Generalized linear regression and multiple logistic regressions were used to model the impact of smoking on the surgical outcome [length of stay (LOS), mortality, postoperative complications and readmission]. To confirm the validity of the regression models and to evaluate the influence of smoking in comparable patient cohorts, a propensity score match was also performed. Results There were 2047 (26.8%) patients with CD identified as current smokers, and 5584 (74.2%) identified as non- or ex-smokers. Smokers were more likely to have a pulmonary comorbidity, preoperative weight loss and a higher American Society of Anesthesiologists classification. No differences in mortality were observed between smokers and non- or ex-smokers in univariate
Introduction The aetiology of Crohn’s disease (CD) is incompletely understood, but its general pattern follows the development of symptoms in genetically predisposed individuals who are exposed to a variety of environmental factors [1,2]. Tobacco exposure is one of the most well-established causal associations with CD [3–5]. Beyond being Correspondence to: David B. Stewart, MD, FACS, FASCRS, The Pennsylvania State University, College of Medicine, Department of Surgery, 500 University Drive, H137, PO Box 850, Hershey, Pennsylvania 17033-0850, USA. E-mail: [email protected]
analysis. In multivariate analysis, smoking status was not significantly associated with LOS. Morbidity (OR 1.20, P = 0.003), particularly infectious (OR 1.30, P < 0.001) and pulmonary (OR 1.87, P < 0.001) complications, and readmission (OR 1.58, P = 0.004) were significantly associated with smoking status. These findings were validated on propensity-score matching analysis. Conclusion In patients with CD, the detrimental effects of smoking on surgical outcomes are driven by infectious and pulmonary complications, and by an increased likelihood of readmission. Keywords Inflammatory bowel disease, tobacco, NSQIP, readmissions, complications, propensity-score What does this paper add to the literature? Tobacco use is associated with deleterious effects on the activity and severity of Crohn’s disease. This study provides a more focused estimate of the attributable impact of smoking on adverse postoperative results, demonstrating that they are driven by infectious and pulmonary complications.
implicated in the development of CD, tobacco exposure exerts a negative influence on the course of the disease [1,2,6–10]. Exacerbations in disease activity [1,7,8], a greater tendency toward more severe and complicated disease [6,10], reduced efficacy of medical therapy [2,7], more frequent clinical relapses and need for hospitalization [1,7–9] and a possible higher likelihood of requiring surgery [7,9,11] have all been found to be associated with smoking. What is less clear, however, is whether the use of tobacco specifically worsens the outcome for patients who require surgery for CD, as an effect separate from
Colorectal Disease ª 2015 The Association of Coloproctology of Great Britain and Ireland. 17, 891–902
891
A. N. Kulaylat et al.
Smoking and Crohn’s disease: a NSQIP analysis
the influence tobacco exerts on the behaviour of CD. The distinction between preoperative clinical disease activity and the postoperative outcome is important because surgeons may assume an association between smoking and surgery which may or may not be present, and this could influence operative management, such as the decision to construct a stoma. Characterizing the influence of smoking on the postoperative outcome of CD is valuable because nearly 80% of all patients with CD require at least one surgical procedure during their lifetime [1]. The primary aim of this study was to evaluate the impact of smoking on the surgical outcome in patients with CD using propensityscore matching to compare similarly matched cohorts of CD patients. We hypothesized that among patients with CD, current smoking status would be associated with a poorer surgical outcome than in non- or ex-smokers.
Method Data and population
Data were collected from the American College of Surgeons (ACS) National Surgical Quality Improvement Program (NSQIP) participant use data file from 2005 to 2012. The ACS NSQIP collects data from more than 450 institutions across the United States and Canada, and NSQIP is a validated database for measuring surgical outcomes designed to improve hospital surgical quality [12]. Unlike administrative databases based on claims or billing data, NSQIP is focused on the surgical outcome derived from the medical chart. A trained clinical reviewer abstracted over 250 variables, including preoperative risk factors, intra-operative details and clinical end-points such as 30-day postoperative morbidity and mortality. Patients with CD were initially identified using the International Classification of Diseases, Ninth Revision (ICD-9) diagnosis codes: 555, 555.1, 555.2, 555.9 (n = 9172). Patients undergoing surgical resection, identified using the following Current Procedural Terminology (CPT) codes for the principal procedure, were maintained for analysis (n = 7688): small bowel (44120, 44202, 44125, 44130, 45136, 44121, 44203, 44126, 44128), ileocolic (44160, 44205), partial colectomy (44140, 44204, 44143, 44145, 44144, 44141, 44207, 44206, 44146, 44208, 44139, 44147, 44213, 45550), total colectomy (44150, 44155, 44210, 44212, 44211, 44158, 44157, 44151, 44156, 44153, 44152) and proctectomy (45110, 45395, 45123, 45111, 45113, 45112, 45119, 45397, 45114, 45121, 45120, 45135, 45116, 45130). Patients with an American Society of Anesthesiologists (ASA) class of 5, preoperative coma or disseminated cancer were excluded
892
from analysis (n = 20). Patients with missing demographic information or comorbidities were also excluded (n = 37). A final cohort of 7631 patients were included in the study. Covariates and outcome
Data regarding patient characteristics and operative details were abstracted from the NSQIP database. Demographic information included age, sex and race. Body mass index (BMI) was categorized according to the World Health Organization (WHO) classification scheme. The anatomical site of CD was codified based on the ICD-9 designation: intestinal (555), colonic (555.1), ileocolic (i.e. intestinal and colonic) (555.2) and not otherwise specified (NOS) (555.9). Comorbidities and preoperative conditions were derived from the NSQIP database [12], and refer to conditions present within 30 days of the principal surgery. Surgical procedures were further categorized into the following types of resections: small bowel, ileocolic, partial colectomy, total colectomy and proctectomy. Other operative details included whether the surgery was elective or emergency, as well as wound classification. Smoking status was identified in ACS NSQIP by two variables. The primary variable identified patients as current or noncurrent smokers, and did not contain any missing variables. The second variable identified packyear history, which was defined as the number of packs of cigarettes smoked per day multiplied by the number of years the patient has smoked. This variable contained 2964 missing values, and therefore our analysis was based on the former variable. The primary end-points of interest were length of stay (LOS), postoperative complications, readmission and mortality. LOS refers to the total length of hospital stay from the date of admission to the date of discharge. Postoperative complications refer to the presence of any NSQIP-defined postoperative complication [12]. We further defined pulmonary complications as pneumonia, failure to wean from the ventilator in 48 h and unplanned re-intubation, while infectious complications included any surgical site infection (SSI), wound dehiscence, pneumonia, sepsis or septic shock. Both readmission and mortality refer to events that occurred within 30 days of surgery. Statistical analysis
Patient characteristics were stratified by smoking status and were compared using the chi-square test, Student’s t-test and the Mann–Whitney U-test as appropriate. LOS was determined using a generalized linear regres-
Colorectal Disease ª 2015 The Association of Coloproctology of Great Britain and Ireland. 17, 891–902
A. N. Kulaylat et al.
sion model to account for severe skewness of the data. The models assumed a gamma family of distributions and an identity link function. Postoperative complications and 30-day readmission were modelled using logistic regression. Due to differences in the incidence of infectious and pulmonary complications between smokers and non- or ex-smokers as observed by univariate analysis, these two types of complications were additionally modelled using logistic regression. As 30-day readmission only began to be captured in 2011, a subgroup analysis was performed for this outcome. All the multivariate analyses were adjusted for the covariates described above. For LOS and 30-day readmission, the analyses were additionally adjusted for the occurrence of any postoperative complication. Propensity-score matching was used to compare the surgical end-points of interest between non- and exsmokers with CD and a similar group of patients who were current smokers. Both cohorts were matched to ensure that both groups were undergoing similar surgical procedures, as well as possessing otherwise similar characteristics such as age, sex, chronic comorbid conditions, anatomical site of CD, the elective or emergency nature of the operation and wound classification. Propensity-score matching was used to reduce bias caused by imbalanced covariates, and therefore served to confirm the validity of the regression models. CD patients who were non- or ex-smokers were matched 1:1 without replacement to current smokers using a nearestneighbour approach with a calliper restriction. After matching, the distributions of patient characteristics were not found to differ significantly. The primary metric for the propensity scoring was the average effect of treatment on the treated (ATT). The ATT estimates the difference between the outcome for CD patients who were non- or ex-smokers had those patients been current smokers, allowing an estimation of differences in outcome between similarly matched patients who were otherwise comparable with respect to their health status. To assess the sampling variability in the propensity-score model, and to account for variation induced by the matching procedure, a bootstrapping algorithm with 500 replicates was used to compute 95% confidence intervals. All analyses were performed using STATA (version 10/MP; Stata Corp., College Station, Texas, USA) and the PSMATCH2 routines [13]. Significance (P) values < 0.05 were considered statistically significant.
Results A total of 7631 patients with CD were identified from the ACS NSQIP database. Of these, 2047 (26.8%) were
Smoking and Crohn’s disease: a NSQIP analysis
current smokers and 5584 (74.2%) were non- or exsmokers. Patient demographics and characteristics are summarized in Table 1. The average study patient was a 41.8-year-old white woman with a BMI of 24.9 and ASA Class II. Systemic hypertension was the most common comorbidity (16.5%), with a minority of patients experiencing other comorbidities. Notably, 40.3% of study patients were receiving chronic steroid therapy prior to resection, 9.4% of patients had experienced significant preoperative weight loss and 8.5% met sepsis criteria on admission. Ileocolic CD was the most common anatomical site (29.0%) and ileocolectomy was the most frequently performed resection (41.8%). Smokers and non- or ex-smokers were comparable in patient characteristics, but statistically significant differences were observed between smokers and non- or ex-smokers in the following variables (P < 0.001): anatomical site of CD [colonic: smokers (16.1%) vs non- or exsmokers (20.7%)], pulmonary comorbidities [dyspnoea: smokers (5.4%) vs non- or ex-smokers (3.5%)], preoperative weight loss [smokers (12.1%) vs non- or ex-smokers (8.5%)], ASA Class [Class III: smokers (36.0%) vs non- or ex-smokers (29.9%)] and type of surgery performed [ileocolectomy: smokers (45.7%) vs non- or exsmokers (40.4%)]. A summary of the surgical outcome stratified by smoking status is provided in Table 2. In univariate comparisons, the median LOS was longer for smokers than for non- or ex-smokers (7 vs 6 days, P = 0.006). Smokers were also more likely to experience a postoperative complication (30.1% vs 26.5%, P = 0.002). Specifically, a worse outcome was observed in pulmonary and infectious complications among smokers (P < 0.05 for all). Smokers were also more likely to be readmitted (9.5% vs 6.0%, P = 0.001) within 30 days of surgery, with no observed differences in mortality. Smoking status was not significantly associated with LOS when evaluated with a multivariate generalized linear regression model (Table 3). A summary of factors associated with any postoperative complication, infectious and pulmonary complications in particular, and 30-day readmission is presented in Table 4. Current smokers demonstrated a 20% increased odds of developing a postoperative complication following a resection for CD (P = 0.003) with significantly higher odds of infective (OR 1.35, P < 0.001) and pulmonary (OR 1.87, P < 0.001) complications. Finally, in a subgroup analysis of patients for whom 30-day readmission information was available, smoking status was associated with a 58% increase in the odds of readmission (P = 0.004). The results of the propensity-score matching analysis are summarized in Table 5. Comparing smokers with similar non- or ex-smokers, the effect of smoking status
Colorectal Disease ª 2015 The Association of Coloproctology of Great Britain and Ireland. 17, 891–902
893
A. N. Kulaylat et al.
Smoking and Crohn’s disease: a NSQIP analysis
Table 1 Patient demographics and characteristics.
Variable Age (years), mean (SD) < 30 30–45 45–60 > 60 Sex Male Female Race* White African–American Other BMI (kg m 2), mean (SD)† < 18.5 (underweight) 18.5–24.9 (normal weight) 25.0–29.9 (overweight) > 30.0 (obese) Transfer (yes) Crohn’s anatomical site Intestinal Colonic Ileocolic (i.e. intestinal and colonic) NOS Comorbidities or preoperative conditions Chronic steroid use Hypertension Weight loss > 10% in last 6 months Sepsis criteria Wound infection Diabetes Dyspnoea Functionally dependent Bleeding disorder COPD Ascites Previous transfusion Ventilator dependent History of CHF Dialysis Renal failure ASA class I II III IV Operative details Emergent Operation
894
All patients (n = 7631)
Current smoker (n = 2047)
Non- or ex-smoker (n = 5584)
P-value
41.8 (15.3) 26.3
41.2 (13.2) 22.7
42.0 (16.0) 27.7
< 0.001
35.6 24.2 13.9
40.8 27.9 8.6
33.7 22.9 15.8
45.6 54.4
43.6 56.4
46.3 53.6
0.03
90.0 8.8 1.2 24.9 (6.4) 10.4 46.8 25.3 17.5 4.9
89.0 10.1 0.8 24.9 (6.4) 12.0 45.5 25.7 16.9 4.9
90.3 8.4 1.3 25.0 (6.4) 9.8 47.3 25.2 17.7 4.9
0.05
29.0 19.4 24.6 26.9
30.7 16.1 25.1 28.2
28.4 20.7 24.4 26.5
< 0.001
40.3 16.5 9.4 8.5 4.1 4.0 4.0 2.4 2.2 1.7 0.8 1.0 0.2 0.1 0.1 0.1
41.3 16.0 12.1 9.3 4.5 3.9 5.4 2.7 2.1 3.1 0.7 0.6 0.3 0.2 0.1 0.0
39.9 16.7 8.5 8.2 4.0 4.1 3.5 2.3 2.2 1.1 0.8 1.1 0.2 0.1 0.1 0.1
0.25 0.51 < 0.001 0.12 0.35 0.69 < 0.001 0.22 0.89 < 0.001 0.87 0.04 0.34 0.35 0.92 0.73
2.4 64.7 31.5 1.4
1.4 61.2 36.0 1.4
2.8 66.0 29.9 1.4
< 0.001
6.2
6.4
6.2
0.76
0.26
1.00
Colorectal Disease ª 2015 The Association of Coloproctology of Great Britain and Ireland. 17, 891–902
A. N. Kulaylat et al.
Smoking and Crohn’s disease: a NSQIP analysis
Table 1 (Continued).
Variable Enterectomy Ileocolectomy Partial colectomy Total abdominal/proctocolectomy Proctectomy Wound classification Clean Clean-contaminated Contaminated Dirty
All patients (n = 7631)
Current smoker (n = 2047)
Non- or ex-smoker (n = 5584)
14.3 41.8 25.1 14.1 4.7
13.2 45.7 26.2 10.7 4.1
14.7 40.4 24.6 15.3 5.0
< 0.001
0.6 64.2 22.3 13.0
0.6 62.3 23.5 13.5
0.5 64.9 21.8 12.8
0.20
P-value
Values are percentages unless otherwise indicated. *n = 5685. †World Health Organization classification. BMI, body mass index; NOS, not otherwise specified; COPD, chronic obstructive pulmonary disease; CHF, congestive heart failure; ASA, American Society of Anesthesiologists.
Table 2 Outcome following surgical resection in Crohn’s disease patients stratified by smoking status.
Variable LOS (days) Median (IQR) Mean ( SD) Postoperative complications (any)* Cardiovascular Cardiac arrest CVA MI DVT PE Pulmonary Pneumonia Failure to wean off ventilator Unplanned re-intubation Renal Progressive renal failure Acute renal failure UTI SSI Superficial/deep Organ space Wound dehiscence Other Transfusion Sepsis Septic shock
All patients (n = 7631)
Current smoker (n = 2047)
Non- or ex-smoker (n = 5584)
6 (4–10) 8.9 (8.3) 27.5
7 (4–11) 9.2 (8.6) 30.1
6 (4–10) 8.7 (8.2) 26.5
P-value
0.01 0.04 0.002
0.2 0.1 0.2 1.3 0.6
0.2 0.0 0.4 1.0 0.3
0.2 0.1 0.1 1.4 0.6
0.88 0.23 0.04 0.12 0.14
1.7 1.5 1.3
2.6 2.0 2.0
1.4 1.3 1.1
< 0.001 0.04 0.004
0.5 0.3 2.8
0.5 0.3 2.6
0.5 0.3 2.8
0.69 0.86 0.62
10.0 5.9 1.3
11.8 7.0 1.9
9.4 5.5 1.1
0.002 0.01 0.02
6.3 5.7 1.4
5.1 6.8 2.1
6.8 5.3 1.1
0.01 0.01 0.002
Colorectal Disease ª 2015 The Association of Coloproctology of Great Britain and Ireland. 17, 891–902
895
A. N. Kulaylat et al.
Smoking and Crohn’s disease: a NSQIP analysis
Table 2 (Continued). All patients (n = 7631)
Variable Return to the OR 30-day readmission† 30-day mortality
Current smoker (n = 2047)
6.3 6.9 0.5
Non- or ex-smoker (n = 5584)
7.6 9.5 0.4
5.8 6.0 0.5
P-value 0.004 0.001 0.73
Values are percentages unless otherwise indicated. *Patients could have more than 1 complication. †n = 3136 (only collected since 2011). IQR, interquartile range; CVA, cardiovascular accident; MI, myocardial infarction; DVT, deep venous thrombosis; PE, pulmonary embolism; UTI, urinary tract infection; SSI, surgical site infection; OR, operating room.
Table 3 Multivariate generalized linear regression for factors associated with length of stay. 95% CI Variable Smoking status Non- or ex-smoker Current smoker Age (years) < 30 30–45 45–60 > 60 Sex Male Female Race White African–American Other Missing BMI (kg m 2) < 18.5 (underweight) 18.5–24.9 (normal weight) 25.0–29.9 (overweight) > 30.0 (obese) Crohn’s disease anatomical site Intestinal Colonic Ileocolic (i.e. intestinal and colonic) NOS Comorbidities or preoperative conditions Chronic steroid use Hypertension Weight loss > 10% in last 6 months Sepsis criteria Wound infection Diabetes Dyspnoea Functionally dependent Bleeding disorder
896
Marginal effect
Lower
Upper
P-value
Reference 0.22
–0.09
–0.09
0.16
0.15 Reference 0.11 0.26
–0.20
–0.20
0.39
–0.25 –0.22
–0.25 –0.22
0.54 0.28
Reference –0.46
–0.73
–0.73
0.001
Reference 1.98 0.79 0.82
1.32 –0.76 0.49
1.32 –0.76 0.49
< 0.001 0.32 < 0.001
0.15
0.15
0.01
–0.40 –0.44
–0.40 –0.44
0.69 0.79
0.30 0.22 Reference 0.38
–0.08 –0.23
–0.08 –0.23
0.12 0.33
–0.01
–0.01
0.05
0.15 0.11 1.33 2.50 1.12 –0.03 –0.25 2.79 1.53
–0.12 –0.31 0.80 1.84 0.34 –0.74 –0.94 1.57 0.43
–0.12 –0.31 0.80 1.84 0.34 –0.74 –0.94 1.57 0.43
0.28 0.62 < 0.001 < 0.001 0.01 0.93 0.47 < 0.001 0.01
0.65 Reference –0.07 –0.05
Colorectal Disease ª 2015 The Association of Coloproctology of Great Britain and Ireland. 17, 891–902
A. N. Kulaylat et al.
Smoking and Crohn’s disease: a NSQIP analysis
Table 3 (Continued). 95% CI Variable COPD Ascites Previous transfusion Ventilator dependent History of CHF Dialysis Renal failure ASA class I II III IV Operative details Emergent Operation Enterectomy Ileocolectomy Partial colectomy Total abdominal/proctocolectomy Proctectomy Wound classification Clean Clean-contaminated Contaminated Dirty Postoperative complication (yes)
Lower
Upper
P-value
0.31 1.28 1.72 0.43 3.73 –0.40 –0.18
–0.83 –0.51 0.05 –2.90 –1.86 –4.66 –5.24
–0.83 –0.51 0.05 –2.90 –1.86 –4.66 –5.24
0.60 0.16 0.04 0.80 0.19 0.85 0.94
–1.19 Reference 1.65 4.40
–1.94
–1.94
0.002
1.32 2.45
1.32 2.45
< 0.001 < 0.001
0.57
–0.05
–0.05
0.07
0.74 Reference 0.77 1.45 –0.51
0.30
0.30
0.001
0.41 0.93 –1.15
0.41 0.93 –1.15
< 0.001 < 0.001 0.12
–2.11
–2.11
0.65
0.05 0.72 3.94
0.05 0.72 3.94
< 0.001 < 0.001 < 0.001
Marginal effect
–0.40 Reference 0.39 1.19 4.33
BMI, body mass index; NOS, not otherwise specified; COPD, chronic obstructive pulmonary disease; CHF, congestive heart failure; ASA, American Society of Anesthesiologists. R2 = 0.25. R2 is a standalone term here describing goodness-of-fit of the aforementioned model represented in the table.
on postoperative complications was robust (ATT 2.9%, P = 0.04). These differences were largely seen in the influence of smoking status in infectious (ATT 3.7%, P = 0.004) and pulmonary complications (ATT 1.7%, P = 0.004). Additionally, smoking remained independently associated with an elevated increased risk of 30day readmission (ATT 3.5%, P = 0.01).
Discussion The detrimental effects of smoking on postoperative outcome in general surgical populations have been previously evaluated. The literature, however, is limited as it deals chiefly with the impact of tobacco in patients with CD undergoing resection [14–17]. While previous research on patients with CD has emphasized the influence of tobacco exposure on the course of the disease [1,6,8–11,18–21], there has been a limited focus on the effect that smoking may or may not have on the
surgical outcome. The present work demonstrates that current smokers experienced increased 30-day postoperative complications (primarily related to infectious and pulmonary-related postoperative complications) and readmission compared with non- or ex-smokers. Unlike its generalized effects on disease activity among medically treated CD patients, the effect of tobacco on the surgical outcome was not generalized, with no significant differences in mortality, LOS or morbidities such as postoperative cardiovascular complications based on smoking status. The use of propensity-score matching in this analysis allowed the attributable impact of smoking to be assessed among subjects with similar characteristics, anatomical sites of disease and types of resection. The association between the postoperative outcome in smokers with CD is similar in its features to those characterized for other disease processes [14,22–24]. In another NSQIP-based study (2008–2009), Mussallam
Colorectal Disease ª 2015 The Association of Coloproctology of Great Britain and Ireland. 17, 891–902
897
898
< 18.5 (underweight) 18.5–24.9 (normal weight) 25.0–29.9 (overweight) > 30.0 (obese) Crohn’s anatomical site Intestinal Colonic Ileocolic (i.e. intestinal and colonic) NOS Comorbidities or preoperative conditions Chronic steroid use Hypertension Weight loss > 10% in last 6 months Sepsis criteria Wound infection
0.01 0.54 0.32 0.03
Reference 1.31 (1.07–1.61) 0.83 (0.46–1.5) 0.94 (0.83–1.06) 1.23 (1.03–1.47)
1.21 1.05 1.08 1.47 1.63
(1.08–1.34) (0.89–1.22) (0.91–1.3) (1.21–1.79) (1.27–2.09)
0.79 (0.68–0.92) 0.94 (0.79–1.11) Reference 0.99 (0.86–1.15) 0.001 0.57 0.38 < 0.001 < 0.001
0.91
0.002 0.47
0.09 < 0.001
0.81
Reference 1.01 (0.91–1.13)
Reference 1.12 (0.98–1.28) 1.41 (1.22–1.64)
0.56 0.22
0.84 (0.73–0.97) Reference 1.04 (0.91–1.2) 1.12 (0.94–1.34)
0.02
Reference 1.2 (1.07–1.35)
Smoking Status Non- or ex-smoker Current smoker Age (years)
< 30 30–45 45–60 > 60 Sex Male Female Race White African–American Other Missing BMI (kg m 2)
0.003
Odds ratio (95% CI)
Variable
P-value
Postoperative complications (any)
1.19 (1.06–1.35) 1.04 (0.88–1.24) 1.06 (0.86–1.3) 1.3 (1.04–1.62) 1.64 (1.26–2.13)
0.77 (0.65–0.91) 0.94 (0.78–1.14) Reference 1.05 (0.89–1.24)
Reference 1.17 (1.01–1.36) 1.47 (1.25–1.73)
0.9 (0.72–1.12)
Reference 1.19 (0.94–1.5) 0.54 (0.25–1.21) 1.18 (1.03–1.35)
Reference 0.98 (0.87–1.11)
Reference 0.98 (0.84–1.14) 0.97 (0.79–1.19)
0.82 (0.7–0.97)
Reference 1.35 (1.18–1.54)
Odds ratio (95% CI)
Infectious complications
0.004 0.62 0.58 0.02 < 0.001
0.54
0.003 0.54
0.04 < 0.001
0.33
0.15 0.14 0.02
0.76
0.77 0.75
0.02
< 0.001
P-value
1.26 1.35 1.26 2.24 1.22
(0.95–1.67) (0.95–1.91) (0.82–1.93) (1.49–3.35) (0.69–2.14)
0.91 (0.6–1.37) 1.25 (0.8–1.95) Reference 1.09 (0.74–1.62)
Reference 1.05 (0.74–1.5) 1.48 (1.03–2.13)
0.73 (0.42–1.28)
Reference 1.18 (0.71–1.98) 1.12 (0.25–4.93) 0.74 (0.52–1.06)
Reference 0.86 (0.65–1.13)
Reference 1.2 (0.82–1.77) 2.23 (1.47–3.39)
0.82 (0.52–1.27)
Reference 1.87 (1.39–2.51)
Odds ratio (95% CI)
0.10 0.09 0.30 < 0.001 0.50
0.65
0.65 0.33
0.78 0.04
0.27
0.53 0.88 0.10
0.28
0.35 < 0.001
0.37
< 0.001
P-value
Pulmonary complications
0.77 0.77 1.63 0.76 0.52
(0.57–1.02) (0.49–1.22) (1.02–2.59) (0.42–1.36) (0.22–1.25)
0.98 (0.64–1.49) 1.05 (0.65–1.7) Reference 1.21 (0.81–1.79)
Reference 1.21 (0.84–1.75) 1.21 (0.8–1.83)
1.31 (0.83–2.07)
Reference 1.33 (0.84–2.11) 0.95 (0.22–4.15) 0.75 (0.41–1.34)
Reference 1.02 (0.76–1.36)
Reference 1.02 (0.69–1.5) 0.77 (0.45–1.31)
1.05 (0.72–1.55)
Reference 1.58 (1.15–2.16)
Odds ratio (95% CI) e
30-day readmission
Table 4 Multivariate logistic regression for factors affecting postoperative complications, infective and pulmonary complications, and 30-day readmission.
0.07 0.27 0.04 0.36 0.15
0.35
0.91 0.83
0.31 0.36
0.24
0.22 0.94 0.33
0.92
0.92 0.33
0.79
0.004
P-value
Smoking and Crohn’s disease: a NSQIP analysis A. N. Kulaylat et al.
Colorectal Disease ª 2015 The Association of Coloproctology of Great Britain and Ireland. 17, 891–902
Colorectal Disease ª 2015 The Association of Coloproctology of Great Britain and Ireland. 17, 891–902
< 0.001 0.004 0.36 0.66
0.78 (0.52–1.17) Reference 1.43 (1.27–1.61) 1.94 (1.23–3.07) 1.11 (0.88–1.4) 1.04 (0.88–1.23) Reference 1.19 (1.04–1.37) 1.8 (1.51–2.14) 2.19 (1.72–2.8)
< 0.001 0.01 –
0.88
0.93 (0.41–2.13) Reference 1.4 (1.21–1.61) 1.17 (0.97–1.41) –
0.85 (0.69–1.03) Reference 1.11 (0.95–1.3) 1.39 (1.15–1.68) 1.74 (1.33–2.27)
1.02 (0.79–1.33)
0.71 (0.44–1.16) Reference 1.35 (1.18–1.54) 1.28 (0.78–2.1)
1 (0.75–1.33) 1.29 (0.97–1.71) 1.16 (0.82–1.66) 1.3 (0.9–1.87) 1.29 (0.85–1.95) 1.26 (0.68–2.33) 1.49 (0.89–2.51) 0.24 (0.06–1.05) 2.41 (0.62–9.32) 3.82 (0.78–18.84) –
Odds ratio (95% CI)
Infectious complications
< 0.001 0.10 –
0.87
0.17 0.001 < 0.001
0.10
0.87
< 0.001 0.33
0.17
1.00 0.08 0.40 0.16 0.24 0.47 0.13 0.06 0.20 0.10 –
P-value
0.6 (0.08–4.65) Reference 1.64 (1.19–2.26) 1.49 (0.99–2.23) –
0.87 (0.56–1.36) Reference 0.9 (0.63–1.29) 0.78 (0.5–1.23) 0.23 (0.07–0.75)
1.01 (0.62–1.65)
0.003 0.06 –
0.63
0.58 0.29 0.02
0.54
0.96
< 0.001 < 0.001
–
– Reference 1.78 (1.3–2.44) 5.14 (2.68–9.86)
0.23 0.19 0.001 0.03 0.33 0.12 0.34 0.06 0.63 0.67 0.31
P-value
1.35 (0.83–2.2) 0.65 (0.35–1.23) 2.41 (1.44–4.03) 1.9 (1.07–3.38) 1.42 (0.7–2.89) 2.04 (0.84–4.99) 1.52 (0.64–3.59) 3.54 (0.93–13.42) 1.61 (0.23–11.19) 0.64 (0.08–5.12) 3.43 (0.31–37.4)
Odds ratio (95% CI)
Pulmonary complications
(1.11–78.5)
(0.17–2.01)
(0.35–1.68) (0.65–2.86) (0.41–3.09) (0.74–3.5) (0.67–4.74)
1.19 (0.33–4.32) Reference 1.14 (0.81–1.6) 0.81 (0.5–1.31) 4.12 (3.05–5.57)
0.71 (0.42–1.22) Reference 0.59 (0.4–0.88) 1.08 (0.69–1.68) 0.45 (0.11–1.9)
1.65 (0.89–3.05)
1.21 (0.42–3.52) Reference 1.19 (0.86–1.64) 0.7 (0.19–2.59)
0.77 1.37 1.12 1.61 1.78 – 0.59 – – – 9.32
Odds ratio (95% CI) e
30-day readmission
0.44 0.39 < 0.001
0.79
0.01 0.73 0.28
0.22
0.11
0.29 0.59
0.73
0.51 0.41 0.82 0.23 0.25 – 0.40 – – – 0.04
P-value
BMI, body mass index; NOS, not otherwise specified; COPD, chronic obstructive pulmonary disease; CHF, congestive heart failure; ASA, American Society of Anesthesiologists. Area under receiver operating characteristic curve for: postoperative complications (any) (0.66); infectious complications (0.64); pulmonary complications (0.78); readmission (0.74).
1.06 (0.53–2.12) Reference 1.37 (1.2–1.55) 1.25 (1.06–1.48) –
0.24
1.3 (1.01–1.68) 1.1 (0.84–1.44) 1.96 (1.41–2.73) 1.5 (1.07–2.1) 1.03 (0.69–1.54) 1.15 (0.64–2.07) 2.89 (1.72–4.83) 0.81 (0.24–2.7) 1.61 (0.37–7.03) 5.3 (0.57–49.65) 2.27 (0.33–15.43)
Diabetes Dyspnoea Functionally dependent Bleeding disorder COPD Ascites Previous transfusion Ventilator dependent History of CHF Dialysis Renal failure ASA class I II III IV Operative details Emergent Operation Enterectomy Ileocolectomy Partial colectomy Total abdominal/proctocolectomy Proctectomy Wound classification Clean Clean-contaminated Contaminated Dirty Postoperative complication (yes) 0.01 < 0.001 < 0.001
0.05 0.47 < 0.001 0.02 0.87 0.63 < 0.001 0.73 0.53 0.14 0.40
Odds ratio (95% CI)
P-value
Postoperative complications (any)
Variable
Table 4 (Continued).
A. N. Kulaylat et al.
Smoking and Crohn’s disease: a NSQIP analysis
899
A. N. Kulaylat et al.
Smoking and Crohn’s disease: a NSQIP analysis
Table 5 Propensity-score matched primary outcomes for the impact of smoking in Crohn’s disease.
Variable
Smoker (n = 2020)
Non- or ex-smoker (n = 2020)
Length of stay (days) Mortality Postoperative complications (any) Cardiovascular complications Infectious complications Pulmonary complications Readmission*
9.12 0.45 30.1 1.8 22.5 7.7 9.6
9.06 0.50 27.3 2.1 18.9 6.0 6.1
95% CI ATT 0.06 0.05 2.9 0.2 3.7 1.7 3.5
Lower 0.45 0.005 0.001 0.01 0.01 0.01 0.01
Upper
P-value
0.56 0.004 0.06 0.01 0.06 0.06 0.06
0.83 0.82 0.04 0.589 0.004 0.004 0.01
Values are percentages unless otherwise indicated. *Smoker, n = 771; non- or ex-smoker, n = 771.
et al. [14] evaluated the impact of smoking on major surgeries of various kinds. They concluded that smoking increased mortality and exerted its greatest effects on morbidity, particularly for arterial events such as cerebrovascular accident and myocardial infarction and respiratory events (using a classification schema for pulmonary events similar to the present study). This is in contrast to the present study; although the present study also demonstrated increased odds of respiratory events in CD patients, no increased risk of arterial events or mortality was observed. Using data from the Veterans Affairs’ Surgical Quality Improvement Program (2002–2008), Hawn et al. [22] demonstrated a dose-dependent increase in pulmonary complications based on cigarette pack-year exposure, with more than 20 pack years’ exposure leading to a significant increase in smoking-related surgical complications (including SSI and 30-day mortality). We similarly found significant effects on SSI and pulmonary complications in our population without the demonstrated effect on 30-day mortality. The broader inclusion criteria in the aforementioned studies, and the underlying differences in the age and comorbidity of the populations in these studies, were likely to have contributed to the differences in outcome regarding arterial events and 30-day mortality. Two previous studies have described the effect of smoking status in patients undergoing surgery for CD [19,25]. In a single-institution retrospective review, Joyce et al. [19] reported that smoking did not appear to predispose CD patients to postoperative complications, but Joyce et al. found a greater overall mortality and lower quality of life scores among current smokers than non- or ex-smokers. These analyses were limited by a smaller sample size (n = 691) and the authors were unable to adjust for covariates in a multivariate analysis, which may account for the difference in results compared with the present study. In the study of Sharma
900
et al. [25] the ACS NSQIP database (2005–2010) was also used to evaluate the impact of smoking on postoperative outcome after colorectal surgery, and as part of their analysis the authors broadly evaluated the effects of smoking in elective colorectal resection for malignant and benign pathology, which included patients with inflammatory bowel disease. In a subgroup analysis of 3770 patients with CD, they reported that current smokers were at higher risk of developing incisional, infective and other major complications compared with never-smokers. Although we used different end-points for the postoperative outcome, we found similar results for the influence of smoking on infective complications. However, it is likely that the present work underestimates the potential impact of smoking in the CD population as non- or ex-smokers were analysed together. The undesirable effects of smoking on postoperative outcome in the present study are consistent with the currently understood effects of tobacco exposure on the tissue microenvironment and wound healing. Prolonged exposure to tobacco affects inflammatory and reparative cell functions, leading to delayed healing and subsequent complications [15–17]. In colon and rectal surgery, specifically, smoking is an important predictive factor for anastomotic leakage [16]. While this complication was not catalogued in the present study, due to the limitations of the NSQIP database, increased odds of organ space infection were observed. While not an ideal surrogate [26], the development of procedure-targeted NSQIP data has allowed relevant complications (such as anastomotic leakage) to be captured, allowing such variables to be deliberately evaluated in future studies. Improvements in the tissue microenvironment and in inflammatory cellular functions are observed as soon as 4 weeks following smoking cessation [15], reinforcing the importance of provider-driven smoking cessation to mitigate these negative effects on the course of a patient’s disease and postoperative recovery.
Colorectal Disease ª 2015 The Association of Coloproctology of Great Britain and Ireland. 17, 891–902
A. N. Kulaylat et al.
The success of targeted smoking cessation strategies to reduce postoperative complications has been previously demonstrated [27–29]. Interventions to promote smoking cessation utilizing individual counselling initiated at least 4 weeks before noncardiac elective surgery (including general abdominal and orthopaedic) have demonstrated an improvement in the surgical outcome, including reductions in surgical-site complications [27,28]. Success has also been demonstrated in the CD population [29,30]. In one study, 31% of patients achieved complete smoking cessation in an 18-month follow-up period [30]. The present study has several limitations. The determination of smoking status is subject to recall bias as it is historical information provided by the patient. Furthermore, current smokers in NSQIP are defined as anyone who has smoked in the year prior to admission for surgery; therefore the true effect of current smoking status may be diluted by the misclassification of former smokers as current smokers in this cohort. Due to missing data related to measurement of pack-years, and with no other consistent variable to classify ex-smokers, we grouped non- and ex-smokers together for the purposes of our analysis. This consequently biases our findings towards measuring a smaller effect of tobacco on our measured outcomes. Additionally, we classified noncurrent smoking patients with ‘missing’ pack year information as never-smokers, as this approach erred on the side of caution in estimating the effect of tobacco [14,22], while the previously referenced study by Sharma et al. [25] excluded this population. We fitted several different models to estimate dose-dependent risk as a function of increasing pack-year history, but a significant proportion of patients were dropped due to missing pack-year data, and the results were inconsistent. Consequently, we decided against any further analysis or reporting of dose-dependent stratified outcomes. Finally, as a potential weakness related to the available data, we were unable to measure or qualify CD activity or to assess the duration of chronic steroid use or any other CD-related medication. In conclusion, the present study has demonstrated that when matched cohorts of patients with CD are compared on the basis of smoking, infective and pulmonary complications will be more frequently encountered, but with no measurable effects on mortality, LOS or cardiovascular complications. The present study suggests that the deleterious effects of tobacco specifically related to the outcome after surgery for CD are more focused than their general negative effects on the activity of the disease. This information may be helpful for surgeons in terms of decision-making based upon a more precise estimate of the complications CD that patients are more
Smoking and Crohn’s disease: a NSQIP analysis
likely to experience when they are active smokers. Lastly, although this study confirms that tobacco is a risk factor for an adverse postoperative outcome specifically in CD patients, the present study may help surgeons to avoid an exaggerated estimate of the influence of tobacco on the outcome in this population.
Funding source No internal or external financial support was used for this report.
Conflicts of interest All authors report no potential conflicts of interest.
Author contributions Study design: CS Hollenbeak, DB Stewart Sr; Statistical analysis: AN Kulaylat, CS Hollenbeak; manuscript creation and editing: AN Kulaylat, W Sangster; Final approval of manuscript: AN Kulaylat, CS Hollenbeak, W Sangster, DB Stewart Sr.
References 1 Kane SV, Flicker M, Katz-Nelson F. Tobacco use is associated with accelerated clinical recurrence of Crohn’s disease after surgically induced remission. J Clin Gastroenterol 2005; 39: 32–5. 2 Nunes T, Etchevers MJ, Domenech E et al. Smoking does influence disease behaviour and impacts the need for therapy in Crohn’s disease in the biologic era. Aliment Pharmacol Ther 2013; 38: 752–60. 3 Silverstein MD, Lashner BA, Hanauer SB, Kirsner JB. Cigarette smoking in Crohn’s disease. Am J Gastroenterol 1989; 84: 31–3. 4 Tobin MV, Logan RF, Langman MJ, McConnell RB, Gilmore IT. Cigarette smoking and inflammatory bowel disease. Gastroenterology 1987; 93: 316–21. 5 Somerville KW, Logan RF, Edmond M, Langman MJ. Smoking and Crohn’s disease. Br Med J (Clin Res Ed) 1984; 289: 954–6. 6 Aldhous MC, Drummond HE, Anderson N, Smith LA, Arnott ID, Satsangi J. Does cigarette smoking influence the phenotype of Crohn’s disease? Analysis using the Montreal classification. Am J Gastroenterol 2007; 102: 577–88. 7 Cosnes J, Carbonnel F, Beaugerie L, Le Quintrec Y, Gendre JP. Effects of cigarette smoking on the long-term course of Crohn’s disease. Gastroenterology 1996; 110: 424–31. 8 Cosnes J, Carbonnel F, Carrat F, Beaugerie L, Cattan S, Gendre J. Effects of current and former cigarette smoking on the clinical course of Crohn’s disease. Aliment Pharmacol Ther 1999; 13: 1403–11.
Colorectal Disease ª 2015 The Association of Coloproctology of Great Britain and Ireland. 17, 891–902
901
Smoking and Crohn’s disease: a NSQIP analysis
9 Reese GE, Nanidis T, Borysiewicz C, Yamamoto T, Orchard T, Tekkis PP. The effect of smoking after surgery for Crohn’s disease: a meta-analysis of observational studies. Int J Colorectal Dis 2008; 23: 1213–21. 10 Lakatos PL, Czegledi Z, Szamosi T et al. Perianal disease, small bowel disease, smoking, prior steroid or early azathioprine/biological therapy are predictors of disease behavior change in patients with Crohn’s disease. World J Gastroenterol 2009; 15: 3504–10. 11 Sutherland LR, Ramcharan S, Bryant H, Fick G. Effect of cigarette smoking on recurrence of Crohn’s disease. Gastroenterology 1990; 98: 1123–8. 12 Hall BL, Hamilton BH, Richards K, Bilimoria KY, Cohen ME, Ko CY. Does surgical quality improve in the American College of Surgeons National Surgical Quality Improvement Program: an evaluation of all participating hospitals. Ann Surg 2009; 250: 363–76. 13 Leuven E, Sianesi B. PSMATCH2: Stata module to perform full Mahalanobis and propensity score matching, commonsupport graphing, and covariate imbalance testing. 2003. http://ideas.repec.org/c/boc/bocode/s432001. html (accessed October 2013). 14 Musallam KM, Rosendaal FR, Zaatari G et al. Smoking and the risk of mortality and vascular and respiratory events in patients undergoing major surgery. JAMA Surg 2013; 148: 755–62. 15 Sorensen LT. Wound healing and infection in surgery: the pathophysiological impact of smoking, smoking cessation, and nicotine replacement therapy: a systematic review. Ann Surg 2012; 255: 1069–79. 16 Sorensen LT, Jorgensen T, Kirkeby LT, Skovdal J, Vennits B, Wille-Jørgensen P. Smoking and alcohol abuse are major risk factors for anastomotic leakage in colorectal surgery. Br J Surg 1999; 86: 927–31. 17 Sorensen LT, Hemmingsen U, Kallehave F et al. Risk factors for tissue and wound complications in gastrointestinal surgery. Ann Surg 2005; 241: 654–8. 18 Higuchi LM, Khalili H, Chan AT, Richter JM, Bousvaros A, Fuchs CS. A prospective study of cigarette smoking and the risk of inflammatory bowel disease in women. Am J Gastroenterol 2012; 107: 1399–406.
902
A. N. Kulaylat et al.
19 Joyce MR, Hannaway CD, Strong SA, Fazio VW, Kiran RP. Impact of smoking on disease phenotype and postoperative outcomes for Crohn’s disease patients undergoing surgery. Langenbecks Arch Surg 2013; 398: 39–45. 20 Picco MF, Bayless TM. Tobacco consumption and disease duration are associated with fistulizing and stricturing behaviors in the first 8 years of Crohn’s disease. Am J Gastroenterol 2003; 98: 363–8. 21 Sands BE, Arsenault JE, Rosen MJ et al. Risk of early surgery for Crohn’s disease: implications for early treatment strategies. Am J Gastroenterol 2003; 98: 2712–8. 22 Hawn MT, Houston TK, Campagna EJ et al. The attributable risk of smoking on surgical complications. Ann Surg 2011; 254: 914–20. 23 Turan A, Mascha EJ, Roberman D et al. Smoking and perioperative outcomes. Anesthesiology 2011; 114: 837–46. 24 Gajdos C, Hawn MT, Campagna EJ, Henderson WG, Singh JA, Houston T. Adverse effects of smoking on postoperative outcomes in cancer patients. Ann Surg Oncol 2012; 19: 1430–8. 25 Sharma A, Deeb AP, Iannuzzi JC, Rickles AS, Monson JR, Fleming FJ. Tobacco smoking and postoperative outcomes after colorectal surgery. Ann Surg 2013; 258: 296–300. 26 Rickles AS, Iannuzzi JC, Kelly KN et al. Anastomotic leak or organ space surgical site infection: What are we missing in our quality improvement programs? Surgery 2013; 154: 680–7; discussion 687-9. 27 Thomsen T, Tonnesen H, Moller AM. Effect of preoperative smoking cessation interventions on postoperative complications and smoking cessation. Br J Surg 2009; 96: 451–61. 28 Moller AM, Villebro N, Pedersen T, Tonnesen H. Effect of preoperative smoking intervention on postoperative complications: a randomised clinical trial. Lancet 2002; 359: 114–7. 29 Hilsden RJ, Hodgins DC, Timmer A, Sutherland LR. Helping patients with Crohn’s disease quit smoking. Am J Gastroenterol 2000; 95: 352–8. 30 Nunes T, Etchevers MJ, Merino O et al. High smoking cessation rate in Crohn’s disease patients after physician advice–the TABACROHN Study. J Crohns Colitis 2013; 7: 202–7.
Colorectal Disease ª 2015 The Association of Coloproctology of Great Britain and Ireland. 17, 891–902
Original article
Impact of smoking on the surgical outcome of Crohn’s disease: a propensity-score matched National Surgical Quality Improvement Program analysis A. N. Kulaylat*, C. S. Hollenbeak*†, W. Sangster* and D. B. Stewart Sr* *Department of Surgery, The Pennsylvania State University, College of Medicine, Hershey, Pennsylvania, USA and †Department of Public Health Sciences, The Pennsylvania State University, College of Medicine, Hershey, Pennsylvania, USA Received 3 December 2014; accepted 30 January 2015; Accepted Article online 25 March 2015
Abstract Aim Smoking is known to have a deleterious effect on Crohn’s disease (CD). The present study addressed the specific impact of smoking on the outcome of surgery for CD. Method A review of the National Surgical Quality Improvement Program (NSQIP) database (2005–2012) identified 7631 patients with CD who underwent surgical resection. Patients were stratified based on smoking status and were compared with univariate statistical tests. Generalized linear regression and multiple logistic regressions were used to model the impact of smoking on the surgical outcome [length of stay (LOS), mortality, postoperative complications and readmission]. To confirm the validity of the regression models and to evaluate the influence of smoking in comparable patient cohorts, a propensity score match was also performed. Results There were 2047 (26.8%) patients with CD identified as current smokers, and 5584 (74.2%) identified as non- or ex-smokers. Smokers were more likely to have a pulmonary comorbidity, preoperative weight loss and a higher American Society of Anesthesiologists classification. No differences in mortality were observed between smokers and non- or ex-smokers in univariate
Introduction The aetiology of Crohn’s disease (CD) is incompletely understood, but its general pattern follows the development of symptoms in genetically predisposed individuals who are exposed to a variety of environmental factors [1,2]. Tobacco exposure is one of the most well-established causal associations with CD [3–5]. Beyond being Correspondence to: David B. Stewart, MD, FACS, FASCRS, The Pennsylvania State University, College of Medicine, Department of Surgery, 500 University Drive, H137, PO Box 850, Hershey, Pennsylvania 17033-0850, USA. E-mail: [email protected]
analysis. In multivariate analysis, smoking status was not significantly associated with LOS. Morbidity (OR 1.20, P = 0.003), particularly infectious (OR 1.30, P < 0.001) and pulmonary (OR 1.87, P < 0.001) complications, and readmission (OR 1.58, P = 0.004) were significantly associated with smoking status. These findings were validated on propensity-score matching analysis. Conclusion In patients with CD, the detrimental effects of smoking on surgical outcomes are driven by infectious and pulmonary complications, and by an increased likelihood of readmission. Keywords Inflammatory bowel disease, tobacco, NSQIP, readmissions, complications, propensity-score What does this paper add to the literature? Tobacco use is associated with deleterious effects on the activity and severity of Crohn’s disease. This study provides a more focused estimate of the attributable impact of smoking on adverse postoperative results, demonstrating that they are driven by infectious and pulmonary complications.
implicated in the development of CD, tobacco exposure exerts a negative influence on the course of the disease [1,2,6–10]. Exacerbations in disease activity [1,7,8], a greater tendency toward more severe and complicated disease [6,10], reduced efficacy of medical therapy [2,7], more frequent clinical relapses and need for hospitalization [1,7–9] and a possible higher likelihood of requiring surgery [7,9,11] have all been found to be associated with smoking. What is less clear, however, is whether the use of tobacco specifically worsens the outcome for patients who require surgery for CD, as an effect separate from
Colorectal Disease ª 2015 The Association of Coloproctology of Great Britain and Ireland. 17, 891–902
891
A. N. Kulaylat et al.
Smoking and Crohn’s disease: a NSQIP analysis
the influence tobacco exerts on the behaviour of CD. The distinction between preoperative clinical disease activity and the postoperative outcome is important because surgeons may assume an association between smoking and surgery which may or may not be present, and this could influence operative management, such as the decision to construct a stoma. Characterizing the influence of smoking on the postoperative outcome of CD is valuable because nearly 80% of all patients with CD require at least one surgical procedure during their lifetime [1]. The primary aim of this study was to evaluate the impact of smoking on the surgical outcome in patients with CD using propensityscore matching to compare similarly matched cohorts of CD patients. We hypothesized that among patients with CD, current smoking status would be associated with a poorer surgical outcome than in non- or ex-smokers.
Method Data and population
Data were collected from the American College of Surgeons (ACS) National Surgical Quality Improvement Program (NSQIP) participant use data file from 2005 to 2012. The ACS NSQIP collects data from more than 450 institutions across the United States and Canada, and NSQIP is a validated database for measuring surgical outcomes designed to improve hospital surgical quality [12]. Unlike administrative databases based on claims or billing data, NSQIP is focused on the surgical outcome derived from the medical chart. A trained clinical reviewer abstracted over 250 variables, including preoperative risk factors, intra-operative details and clinical end-points such as 30-day postoperative morbidity and mortality. Patients with CD were initially identified using the International Classification of Diseases, Ninth Revision (ICD-9) diagnosis codes: 555, 555.1, 555.2, 555.9 (n = 9172). Patients undergoing surgical resection, identified using the following Current Procedural Terminology (CPT) codes for the principal procedure, were maintained for analysis (n = 7688): small bowel (44120, 44202, 44125, 44130, 45136, 44121, 44203, 44126, 44128), ileocolic (44160, 44205), partial colectomy (44140, 44204, 44143, 44145, 44144, 44141, 44207, 44206, 44146, 44208, 44139, 44147, 44213, 45550), total colectomy (44150, 44155, 44210, 44212, 44211, 44158, 44157, 44151, 44156, 44153, 44152) and proctectomy (45110, 45395, 45123, 45111, 45113, 45112, 45119, 45397, 45114, 45121, 45120, 45135, 45116, 45130). Patients with an American Society of Anesthesiologists (ASA) class of 5, preoperative coma or disseminated cancer were excluded
892
from analysis (n = 20). Patients with missing demographic information or comorbidities were also excluded (n = 37). A final cohort of 7631 patients were included in the study. Covariates and outcome
Data regarding patient characteristics and operative details were abstracted from the NSQIP database. Demographic information included age, sex and race. Body mass index (BMI) was categorized according to the World Health Organization (WHO) classification scheme. The anatomical site of CD was codified based on the ICD-9 designation: intestinal (555), colonic (555.1), ileocolic (i.e. intestinal and colonic) (555.2) and not otherwise specified (NOS) (555.9). Comorbidities and preoperative conditions were derived from the NSQIP database [12], and refer to conditions present within 30 days of the principal surgery. Surgical procedures were further categorized into the following types of resections: small bowel, ileocolic, partial colectomy, total colectomy and proctectomy. Other operative details included whether the surgery was elective or emergency, as well as wound classification. Smoking status was identified in ACS NSQIP by two variables. The primary variable identified patients as current or noncurrent smokers, and did not contain any missing variables. The second variable identified packyear history, which was defined as the number of packs of cigarettes smoked per day multiplied by the number of years the patient has smoked. This variable contained 2964 missing values, and therefore our analysis was based on the former variable. The primary end-points of interest were length of stay (LOS), postoperative complications, readmission and mortality. LOS refers to the total length of hospital stay from the date of admission to the date of discharge. Postoperative complications refer to the presence of any NSQIP-defined postoperative complication [12]. We further defined pulmonary complications as pneumonia, failure to wean from the ventilator in 48 h and unplanned re-intubation, while infectious complications included any surgical site infection (SSI), wound dehiscence, pneumonia, sepsis or septic shock. Both readmission and mortality refer to events that occurred within 30 days of surgery. Statistical analysis
Patient characteristics were stratified by smoking status and were compared using the chi-square test, Student’s t-test and the Mann–Whitney U-test as appropriate. LOS was determined using a generalized linear regres-
Colorectal Disease ª 2015 The Association of Coloproctology of Great Britain and Ireland. 17, 891–902
A. N. Kulaylat et al.
sion model to account for severe skewness of the data. The models assumed a gamma family of distributions and an identity link function. Postoperative complications and 30-day readmission were modelled using logistic regression. Due to differences in the incidence of infectious and pulmonary complications between smokers and non- or ex-smokers as observed by univariate analysis, these two types of complications were additionally modelled using logistic regression. As 30-day readmission only began to be captured in 2011, a subgroup analysis was performed for this outcome. All the multivariate analyses were adjusted for the covariates described above. For LOS and 30-day readmission, the analyses were additionally adjusted for the occurrence of any postoperative complication. Propensity-score matching was used to compare the surgical end-points of interest between non- and exsmokers with CD and a similar group of patients who were current smokers. Both cohorts were matched to ensure that both groups were undergoing similar surgical procedures, as well as possessing otherwise similar characteristics such as age, sex, chronic comorbid conditions, anatomical site of CD, the elective or emergency nature of the operation and wound classification. Propensity-score matching was used to reduce bias caused by imbalanced covariates, and therefore served to confirm the validity of the regression models. CD patients who were non- or ex-smokers were matched 1:1 without replacement to current smokers using a nearestneighbour approach with a calliper restriction. After matching, the distributions of patient characteristics were not found to differ significantly. The primary metric for the propensity scoring was the average effect of treatment on the treated (ATT). The ATT estimates the difference between the outcome for CD patients who were non- or ex-smokers had those patients been current smokers, allowing an estimation of differences in outcome between similarly matched patients who were otherwise comparable with respect to their health status. To assess the sampling variability in the propensity-score model, and to account for variation induced by the matching procedure, a bootstrapping algorithm with 500 replicates was used to compute 95% confidence intervals. All analyses were performed using STATA (version 10/MP; Stata Corp., College Station, Texas, USA) and the PSMATCH2 routines [13]. Significance (P) values < 0.05 were considered statistically significant.
Results A total of 7631 patients with CD were identified from the ACS NSQIP database. Of these, 2047 (26.8%) were
Smoking and Crohn’s disease: a NSQIP analysis
current smokers and 5584 (74.2%) were non- or exsmokers. Patient demographics and characteristics are summarized in Table 1. The average study patient was a 41.8-year-old white woman with a BMI of 24.9 and ASA Class II. Systemic hypertension was the most common comorbidity (16.5%), with a minority of patients experiencing other comorbidities. Notably, 40.3% of study patients were receiving chronic steroid therapy prior to resection, 9.4% of patients had experienced significant preoperative weight loss and 8.5% met sepsis criteria on admission. Ileocolic CD was the most common anatomical site (29.0%) and ileocolectomy was the most frequently performed resection (41.8%). Smokers and non- or ex-smokers were comparable in patient characteristics, but statistically significant differences were observed between smokers and non- or ex-smokers in the following variables (P < 0.001): anatomical site of CD [colonic: smokers (16.1%) vs non- or exsmokers (20.7%)], pulmonary comorbidities [dyspnoea: smokers (5.4%) vs non- or ex-smokers (3.5%)], preoperative weight loss [smokers (12.1%) vs non- or ex-smokers (8.5%)], ASA Class [Class III: smokers (36.0%) vs non- or ex-smokers (29.9%)] and type of surgery performed [ileocolectomy: smokers (45.7%) vs non- or exsmokers (40.4%)]. A summary of the surgical outcome stratified by smoking status is provided in Table 2. In univariate comparisons, the median LOS was longer for smokers than for non- or ex-smokers (7 vs 6 days, P = 0.006). Smokers were also more likely to experience a postoperative complication (30.1% vs 26.5%, P = 0.002). Specifically, a worse outcome was observed in pulmonary and infectious complications among smokers (P < 0.05 for all). Smokers were also more likely to be readmitted (9.5% vs 6.0%, P = 0.001) within 30 days of surgery, with no observed differences in mortality. Smoking status was not significantly associated with LOS when evaluated with a multivariate generalized linear regression model (Table 3). A summary of factors associated with any postoperative complication, infectious and pulmonary complications in particular, and 30-day readmission is presented in Table 4. Current smokers demonstrated a 20% increased odds of developing a postoperative complication following a resection for CD (P = 0.003) with significantly higher odds of infective (OR 1.35, P < 0.001) and pulmonary (OR 1.87, P < 0.001) complications. Finally, in a subgroup analysis of patients for whom 30-day readmission information was available, smoking status was associated with a 58% increase in the odds of readmission (P = 0.004). The results of the propensity-score matching analysis are summarized in Table 5. Comparing smokers with similar non- or ex-smokers, the effect of smoking status
Colorectal Disease ª 2015 The Association of Coloproctology of Great Britain and Ireland. 17, 891–902
893
A. N. Kulaylat et al.
Smoking and Crohn’s disease: a NSQIP analysis
Table 1 Patient demographics and characteristics.
Variable Age (years), mean (SD) < 30 30–45 45–60 > 60 Sex Male Female Race* White African–American Other BMI (kg m 2), mean (SD)† < 18.5 (underweight) 18.5–24.9 (normal weight) 25.0–29.9 (overweight) > 30.0 (obese) Transfer (yes) Crohn’s anatomical site Intestinal Colonic Ileocolic (i.e. intestinal and colonic) NOS Comorbidities or preoperative conditions Chronic steroid use Hypertension Weight loss > 10% in last 6 months Sepsis criteria Wound infection Diabetes Dyspnoea Functionally dependent Bleeding disorder COPD Ascites Previous transfusion Ventilator dependent History of CHF Dialysis Renal failure ASA class I II III IV Operative details Emergent Operation
894
All patients (n = 7631)
Current smoker (n = 2047)
Non- or ex-smoker (n = 5584)
P-value
41.8 (15.3) 26.3
41.2 (13.2) 22.7
42.0 (16.0) 27.7
< 0.001
35.6 24.2 13.9
40.8 27.9 8.6
33.7 22.9 15.8
45.6 54.4
43.6 56.4
46.3 53.6
0.03
90.0 8.8 1.2 24.9 (6.4) 10.4 46.8 25.3 17.5 4.9
89.0 10.1 0.8 24.9 (6.4) 12.0 45.5 25.7 16.9 4.9
90.3 8.4 1.3 25.0 (6.4) 9.8 47.3 25.2 17.7 4.9
0.05
29.0 19.4 24.6 26.9
30.7 16.1 25.1 28.2
28.4 20.7 24.4 26.5
< 0.001
40.3 16.5 9.4 8.5 4.1 4.0 4.0 2.4 2.2 1.7 0.8 1.0 0.2 0.1 0.1 0.1
41.3 16.0 12.1 9.3 4.5 3.9 5.4 2.7 2.1 3.1 0.7 0.6 0.3 0.2 0.1 0.0
39.9 16.7 8.5 8.2 4.0 4.1 3.5 2.3 2.2 1.1 0.8 1.1 0.2 0.1 0.1 0.1
0.25 0.51 < 0.001 0.12 0.35 0.69 < 0.001 0.22 0.89 < 0.001 0.87 0.04 0.34 0.35 0.92 0.73
2.4 64.7 31.5 1.4
1.4 61.2 36.0 1.4
2.8 66.0 29.9 1.4
< 0.001
6.2
6.4
6.2
0.76
0.26
1.00
Colorectal Disease ª 2015 The Association of Coloproctology of Great Britain and Ireland. 17, 891–902
A. N. Kulaylat et al.
Smoking and Crohn’s disease: a NSQIP analysis
Table 1 (Continued).
Variable Enterectomy Ileocolectomy Partial colectomy Total abdominal/proctocolectomy Proctectomy Wound classification Clean Clean-contaminated Contaminated Dirty
All patients (n = 7631)
Current smoker (n = 2047)
Non- or ex-smoker (n = 5584)
14.3 41.8 25.1 14.1 4.7
13.2 45.7 26.2 10.7 4.1
14.7 40.4 24.6 15.3 5.0
< 0.001
0.6 64.2 22.3 13.0
0.6 62.3 23.5 13.5
0.5 64.9 21.8 12.8
0.20
P-value
Values are percentages unless otherwise indicated. *n = 5685. †World Health Organization classification. BMI, body mass index; NOS, not otherwise specified; COPD, chronic obstructive pulmonary disease; CHF, congestive heart failure; ASA, American Society of Anesthesiologists.
Table 2 Outcome following surgical resection in Crohn’s disease patients stratified by smoking status.
Variable LOS (days) Median (IQR) Mean ( SD) Postoperative complications (any)* Cardiovascular Cardiac arrest CVA MI DVT PE Pulmonary Pneumonia Failure to wean off ventilator Unplanned re-intubation Renal Progressive renal failure Acute renal failure UTI SSI Superficial/deep Organ space Wound dehiscence Other Transfusion Sepsis Septic shock
All patients (n = 7631)
Current smoker (n = 2047)
Non- or ex-smoker (n = 5584)
6 (4–10) 8.9 (8.3) 27.5
7 (4–11) 9.2 (8.6) 30.1
6 (4–10) 8.7 (8.2) 26.5
P-value
0.01 0.04 0.002
0.2 0.1 0.2 1.3 0.6
0.2 0.0 0.4 1.0 0.3
0.2 0.1 0.1 1.4 0.6
0.88 0.23 0.04 0.12 0.14
1.7 1.5 1.3
2.6 2.0 2.0
1.4 1.3 1.1
< 0.001 0.04 0.004
0.5 0.3 2.8
0.5 0.3 2.6
0.5 0.3 2.8
0.69 0.86 0.62
10.0 5.9 1.3
11.8 7.0 1.9
9.4 5.5 1.1
0.002 0.01 0.02
6.3 5.7 1.4
5.1 6.8 2.1
6.8 5.3 1.1
0.01 0.01 0.002
Colorectal Disease ª 2015 The Association of Coloproctology of Great Britain and Ireland. 17, 891–902
895
A. N. Kulaylat et al.
Smoking and Crohn’s disease: a NSQIP analysis
Table 2 (Continued). All patients (n = 7631)
Variable Return to the OR 30-day readmission† 30-day mortality
Current smoker (n = 2047)
6.3 6.9 0.5
Non- or ex-smoker (n = 5584)
7.6 9.5 0.4
5.8 6.0 0.5
P-value 0.004 0.001 0.73
Values are percentages unless otherwise indicated. *Patients could have more than 1 complication. †n = 3136 (only collected since 2011). IQR, interquartile range; CVA, cardiovascular accident; MI, myocardial infarction; DVT, deep venous thrombosis; PE, pulmonary embolism; UTI, urinary tract infection; SSI, surgical site infection; OR, operating room.
Table 3 Multivariate generalized linear regression for factors associated with length of stay. 95% CI Variable Smoking status Non- or ex-smoker Current smoker Age (years) < 30 30–45 45–60 > 60 Sex Male Female Race White African–American Other Missing BMI (kg m 2) < 18.5 (underweight) 18.5–24.9 (normal weight) 25.0–29.9 (overweight) > 30.0 (obese) Crohn’s disease anatomical site Intestinal Colonic Ileocolic (i.e. intestinal and colonic) NOS Comorbidities or preoperative conditions Chronic steroid use Hypertension Weight loss > 10% in last 6 months Sepsis criteria Wound infection Diabetes Dyspnoea Functionally dependent Bleeding disorder
896
Marginal effect
Lower
Upper
P-value
Reference 0.22
–0.09
–0.09
0.16
0.15 Reference 0.11 0.26
–0.20
–0.20
0.39
–0.25 –0.22
–0.25 –0.22
0.54 0.28
Reference –0.46
–0.73
–0.73
0.001
Reference 1.98 0.79 0.82
1.32 –0.76 0.49
1.32 –0.76 0.49
< 0.001 0.32 < 0.001
0.15
0.15
0.01
–0.40 –0.44
–0.40 –0.44
0.69 0.79
0.30 0.22 Reference 0.38
–0.08 –0.23
–0.08 –0.23
0.12 0.33
–0.01
–0.01
0.05
0.15 0.11 1.33 2.50 1.12 –0.03 –0.25 2.79 1.53
–0.12 –0.31 0.80 1.84 0.34 –0.74 –0.94 1.57 0.43
–0.12 –0.31 0.80 1.84 0.34 –0.74 –0.94 1.57 0.43
0.28 0.62 < 0.001 < 0.001 0.01 0.93 0.47 < 0.001 0.01
0.65 Reference –0.07 –0.05
Colorectal Disease ª 2015 The Association of Coloproctology of Great Britain and Ireland. 17, 891–902
A. N. Kulaylat et al.
Smoking and Crohn’s disease: a NSQIP analysis
Table 3 (Continued). 95% CI Variable COPD Ascites Previous transfusion Ventilator dependent History of CHF Dialysis Renal failure ASA class I II III IV Operative details Emergent Operation Enterectomy Ileocolectomy Partial colectomy Total abdominal/proctocolectomy Proctectomy Wound classification Clean Clean-contaminated Contaminated Dirty Postoperative complication (yes)
Lower
Upper
P-value
0.31 1.28 1.72 0.43 3.73 –0.40 –0.18
–0.83 –0.51 0.05 –2.90 –1.86 –4.66 –5.24
–0.83 –0.51 0.05 –2.90 –1.86 –4.66 –5.24
0.60 0.16 0.04 0.80 0.19 0.85 0.94
–1.19 Reference 1.65 4.40
–1.94
–1.94
0.002
1.32 2.45
1.32 2.45
< 0.001 < 0.001
0.57
–0.05
–0.05
0.07
0.74 Reference 0.77 1.45 –0.51
0.30
0.30
0.001
0.41 0.93 –1.15
0.41 0.93 –1.15
< 0.001 < 0.001 0.12
–2.11
–2.11
0.65
0.05 0.72 3.94
0.05 0.72 3.94
< 0.001 < 0.001 < 0.001
Marginal effect
–0.40 Reference 0.39 1.19 4.33
BMI, body mass index; NOS, not otherwise specified; COPD, chronic obstructive pulmonary disease; CHF, congestive heart failure; ASA, American Society of Anesthesiologists. R2 = 0.25. R2 is a standalone term here describing goodness-of-fit of the aforementioned model represented in the table.
on postoperative complications was robust (ATT 2.9%, P = 0.04). These differences were largely seen in the influence of smoking status in infectious (ATT 3.7%, P = 0.004) and pulmonary complications (ATT 1.7%, P = 0.004). Additionally, smoking remained independently associated with an elevated increased risk of 30day readmission (ATT 3.5%, P = 0.01).
Discussion The detrimental effects of smoking on postoperative outcome in general surgical populations have been previously evaluated. The literature, however, is limited as it deals chiefly with the impact of tobacco in patients with CD undergoing resection [14–17]. While previous research on patients with CD has emphasized the influence of tobacco exposure on the course of the disease [1,6,8–11,18–21], there has been a limited focus on the effect that smoking may or may not have on the
surgical outcome. The present work demonstrates that current smokers experienced increased 30-day postoperative complications (primarily related to infectious and pulmonary-related postoperative complications) and readmission compared with non- or ex-smokers. Unlike its generalized effects on disease activity among medically treated CD patients, the effect of tobacco on the surgical outcome was not generalized, with no significant differences in mortality, LOS or morbidities such as postoperative cardiovascular complications based on smoking status. The use of propensity-score matching in this analysis allowed the attributable impact of smoking to be assessed among subjects with similar characteristics, anatomical sites of disease and types of resection. The association between the postoperative outcome in smokers with CD is similar in its features to those characterized for other disease processes [14,22–24]. In another NSQIP-based study (2008–2009), Mussallam
Colorectal Disease ª 2015 The Association of Coloproctology of Great Britain and Ireland. 17, 891–902
897
898
< 18.5 (underweight) 18.5–24.9 (normal weight) 25.0–29.9 (overweight) > 30.0 (obese) Crohn’s anatomical site Intestinal Colonic Ileocolic (i.e. intestinal and colonic) NOS Comorbidities or preoperative conditions Chronic steroid use Hypertension Weight loss > 10% in last 6 months Sepsis criteria Wound infection
0.01 0.54 0.32 0.03
Reference 1.31 (1.07–1.61) 0.83 (0.46–1.5) 0.94 (0.83–1.06) 1.23 (1.03–1.47)
1.21 1.05 1.08 1.47 1.63
(1.08–1.34) (0.89–1.22) (0.91–1.3) (1.21–1.79) (1.27–2.09)
0.79 (0.68–0.92) 0.94 (0.79–1.11) Reference 0.99 (0.86–1.15) 0.001 0.57 0.38 < 0.001 < 0.001
0.91
0.002 0.47
0.09 < 0.001
0.81
Reference 1.01 (0.91–1.13)
Reference 1.12 (0.98–1.28) 1.41 (1.22–1.64)
0.56 0.22
0.84 (0.73–0.97) Reference 1.04 (0.91–1.2) 1.12 (0.94–1.34)
0.02
Reference 1.2 (1.07–1.35)
Smoking Status Non- or ex-smoker Current smoker Age (years)
< 30 30–45 45–60 > 60 Sex Male Female Race White African–American Other Missing BMI (kg m 2)
0.003
Odds ratio (95% CI)
Variable
P-value
Postoperative complications (any)
1.19 (1.06–1.35) 1.04 (0.88–1.24) 1.06 (0.86–1.3) 1.3 (1.04–1.62) 1.64 (1.26–2.13)
0.77 (0.65–0.91) 0.94 (0.78–1.14) Reference 1.05 (0.89–1.24)
Reference 1.17 (1.01–1.36) 1.47 (1.25–1.73)
0.9 (0.72–1.12)
Reference 1.19 (0.94–1.5) 0.54 (0.25–1.21) 1.18 (1.03–1.35)
Reference 0.98 (0.87–1.11)
Reference 0.98 (0.84–1.14) 0.97 (0.79–1.19)
0.82 (0.7–0.97)
Reference 1.35 (1.18–1.54)
Odds ratio (95% CI)
Infectious complications
0.004 0.62 0.58 0.02 < 0.001
0.54
0.003 0.54
0.04 < 0.001
0.33
0.15 0.14 0.02
0.76
0.77 0.75
0.02
< 0.001
P-value
1.26 1.35 1.26 2.24 1.22
(0.95–1.67) (0.95–1.91) (0.82–1.93) (1.49–3.35) (0.69–2.14)
0.91 (0.6–1.37) 1.25 (0.8–1.95) Reference 1.09 (0.74–1.62)
Reference 1.05 (0.74–1.5) 1.48 (1.03–2.13)
0.73 (0.42–1.28)
Reference 1.18 (0.71–1.98) 1.12 (0.25–4.93) 0.74 (0.52–1.06)
Reference 0.86 (0.65–1.13)
Reference 1.2 (0.82–1.77) 2.23 (1.47–3.39)
0.82 (0.52–1.27)
Reference 1.87 (1.39–2.51)
Odds ratio (95% CI)
0.10 0.09 0.30 < 0.001 0.50
0.65
0.65 0.33
0.78 0.04
0.27
0.53 0.88 0.10
0.28
0.35 < 0.001
0.37
< 0.001
P-value
Pulmonary complications
0.77 0.77 1.63 0.76 0.52
(0.57–1.02) (0.49–1.22) (1.02–2.59) (0.42–1.36) (0.22–1.25)
0.98 (0.64–1.49) 1.05 (0.65–1.7) Reference 1.21 (0.81–1.79)
Reference 1.21 (0.84–1.75) 1.21 (0.8–1.83)
1.31 (0.83–2.07)
Reference 1.33 (0.84–2.11) 0.95 (0.22–4.15) 0.75 (0.41–1.34)
Reference 1.02 (0.76–1.36)
Reference 1.02 (0.69–1.5) 0.77 (0.45–1.31)
1.05 (0.72–1.55)
Reference 1.58 (1.15–2.16)
Odds ratio (95% CI) e
30-day readmission
Table 4 Multivariate logistic regression for factors affecting postoperative complications, infective and pulmonary complications, and 30-day readmission.
0.07 0.27 0.04 0.36 0.15
0.35
0.91 0.83
0.31 0.36
0.24
0.22 0.94 0.33
0.92
0.92 0.33
0.79
0.004
P-value
Smoking and Crohn’s disease: a NSQIP analysis A. N. Kulaylat et al.
Colorectal Disease ª 2015 The Association of Coloproctology of Great Britain and Ireland. 17, 891–902
Colorectal Disease ª 2015 The Association of Coloproctology of Great Britain and Ireland. 17, 891–902
< 0.001 0.004 0.36 0.66
0.78 (0.52–1.17) Reference 1.43 (1.27–1.61) 1.94 (1.23–3.07) 1.11 (0.88–1.4) 1.04 (0.88–1.23) Reference 1.19 (1.04–1.37) 1.8 (1.51–2.14) 2.19 (1.72–2.8)
< 0.001 0.01 –
0.88
0.93 (0.41–2.13) Reference 1.4 (1.21–1.61) 1.17 (0.97–1.41) –
0.85 (0.69–1.03) Reference 1.11 (0.95–1.3) 1.39 (1.15–1.68) 1.74 (1.33–2.27)
1.02 (0.79–1.33)
0.71 (0.44–1.16) Reference 1.35 (1.18–1.54) 1.28 (0.78–2.1)
1 (0.75–1.33) 1.29 (0.97–1.71) 1.16 (0.82–1.66) 1.3 (0.9–1.87) 1.29 (0.85–1.95) 1.26 (0.68–2.33) 1.49 (0.89–2.51) 0.24 (0.06–1.05) 2.41 (0.62–9.32) 3.82 (0.78–18.84) –
Odds ratio (95% CI)
Infectious complications
< 0.001 0.10 –
0.87
0.17 0.001 < 0.001
0.10
0.87
< 0.001 0.33
0.17
1.00 0.08 0.40 0.16 0.24 0.47 0.13 0.06 0.20 0.10 –
P-value
0.6 (0.08–4.65) Reference 1.64 (1.19–2.26) 1.49 (0.99–2.23) –
0.87 (0.56–1.36) Reference 0.9 (0.63–1.29) 0.78 (0.5–1.23) 0.23 (0.07–0.75)
1.01 (0.62–1.65)
0.003 0.06 –
0.63
0.58 0.29 0.02
0.54
0.96
< 0.001 < 0.001
–
– Reference 1.78 (1.3–2.44) 5.14 (2.68–9.86)
0.23 0.19 0.001 0.03 0.33 0.12 0.34 0.06 0.63 0.67 0.31
P-value
1.35 (0.83–2.2) 0.65 (0.35–1.23) 2.41 (1.44–4.03) 1.9 (1.07–3.38) 1.42 (0.7–2.89) 2.04 (0.84–4.99) 1.52 (0.64–3.59) 3.54 (0.93–13.42) 1.61 (0.23–11.19) 0.64 (0.08–5.12) 3.43 (0.31–37.4)
Odds ratio (95% CI)
Pulmonary complications
(1.11–78.5)
(0.17–2.01)
(0.35–1.68) (0.65–2.86) (0.41–3.09) (0.74–3.5) (0.67–4.74)
1.19 (0.33–4.32) Reference 1.14 (0.81–1.6) 0.81 (0.5–1.31) 4.12 (3.05–5.57)
0.71 (0.42–1.22) Reference 0.59 (0.4–0.88) 1.08 (0.69–1.68) 0.45 (0.11–1.9)
1.65 (0.89–3.05)
1.21 (0.42–3.52) Reference 1.19 (0.86–1.64) 0.7 (0.19–2.59)
0.77 1.37 1.12 1.61 1.78 – 0.59 – – – 9.32
Odds ratio (95% CI) e
30-day readmission
0.44 0.39 < 0.001
0.79
0.01 0.73 0.28
0.22
0.11
0.29 0.59
0.73
0.51 0.41 0.82 0.23 0.25 – 0.40 – – – 0.04
P-value
BMI, body mass index; NOS, not otherwise specified; COPD, chronic obstructive pulmonary disease; CHF, congestive heart failure; ASA, American Society of Anesthesiologists. Area under receiver operating characteristic curve for: postoperative complications (any) (0.66); infectious complications (0.64); pulmonary complications (0.78); readmission (0.74).
1.06 (0.53–2.12) Reference 1.37 (1.2–1.55) 1.25 (1.06–1.48) –
0.24
1.3 (1.01–1.68) 1.1 (0.84–1.44) 1.96 (1.41–2.73) 1.5 (1.07–2.1) 1.03 (0.69–1.54) 1.15 (0.64–2.07) 2.89 (1.72–4.83) 0.81 (0.24–2.7) 1.61 (0.37–7.03) 5.3 (0.57–49.65) 2.27 (0.33–15.43)
Diabetes Dyspnoea Functionally dependent Bleeding disorder COPD Ascites Previous transfusion Ventilator dependent History of CHF Dialysis Renal failure ASA class I II III IV Operative details Emergent Operation Enterectomy Ileocolectomy Partial colectomy Total abdominal/proctocolectomy Proctectomy Wound classification Clean Clean-contaminated Contaminated Dirty Postoperative complication (yes) 0.01 < 0.001 < 0.001
0.05 0.47 < 0.001 0.02 0.87 0.63 < 0.001 0.73 0.53 0.14 0.40
Odds ratio (95% CI)
P-value
Postoperative complications (any)
Variable
Table 4 (Continued).
A. N. Kulaylat et al.
Smoking and Crohn’s disease: a NSQIP analysis
899
A. N. Kulaylat et al.
Smoking and Crohn’s disease: a NSQIP analysis
Table 5 Propensity-score matched primary outcomes for the impact of smoking in Crohn’s disease.
Variable
Smoker (n = 2020)
Non- or ex-smoker (n = 2020)
Length of stay (days) Mortality Postoperative complications (any) Cardiovascular complications Infectious complications Pulmonary complications Readmission*
9.12 0.45 30.1 1.8 22.5 7.7 9.6
9.06 0.50 27.3 2.1 18.9 6.0 6.1
95% CI ATT 0.06 0.05 2.9 0.2 3.7 1.7 3.5
Lower 0.45 0.005 0.001 0.01 0.01 0.01 0.01
Upper
P-value
0.56 0.004 0.06 0.01 0.06 0.06 0.06
0.83 0.82 0.04 0.589 0.004 0.004 0.01
Values are percentages unless otherwise indicated. *Smoker, n = 771; non- or ex-smoker, n = 771.
et al. [14] evaluated the impact of smoking on major surgeries of various kinds. They concluded that smoking increased mortality and exerted its greatest effects on morbidity, particularly for arterial events such as cerebrovascular accident and myocardial infarction and respiratory events (using a classification schema for pulmonary events similar to the present study). This is in contrast to the present study; although the present study also demonstrated increased odds of respiratory events in CD patients, no increased risk of arterial events or mortality was observed. Using data from the Veterans Affairs’ Surgical Quality Improvement Program (2002–2008), Hawn et al. [22] demonstrated a dose-dependent increase in pulmonary complications based on cigarette pack-year exposure, with more than 20 pack years’ exposure leading to a significant increase in smoking-related surgical complications (including SSI and 30-day mortality). We similarly found significant effects on SSI and pulmonary complications in our population without the demonstrated effect on 30-day mortality. The broader inclusion criteria in the aforementioned studies, and the underlying differences in the age and comorbidity of the populations in these studies, were likely to have contributed to the differences in outcome regarding arterial events and 30-day mortality. Two previous studies have described the effect of smoking status in patients undergoing surgery for CD [19,25]. In a single-institution retrospective review, Joyce et al. [19] reported that smoking did not appear to predispose CD patients to postoperative complications, but Joyce et al. found a greater overall mortality and lower quality of life scores among current smokers than non- or ex-smokers. These analyses were limited by a smaller sample size (n = 691) and the authors were unable to adjust for covariates in a multivariate analysis, which may account for the difference in results compared with the present study. In the study of Sharma
900
et al. [25] the ACS NSQIP database (2005–2010) was also used to evaluate the impact of smoking on postoperative outcome after colorectal surgery, and as part of their analysis the authors broadly evaluated the effects of smoking in elective colorectal resection for malignant and benign pathology, which included patients with inflammatory bowel disease. In a subgroup analysis of 3770 patients with CD, they reported that current smokers were at higher risk of developing incisional, infective and other major complications compared with never-smokers. Although we used different end-points for the postoperative outcome, we found similar results for the influence of smoking on infective complications. However, it is likely that the present work underestimates the potential impact of smoking in the CD population as non- or ex-smokers were analysed together. The undesirable effects of smoking on postoperative outcome in the present study are consistent with the currently understood effects of tobacco exposure on the tissue microenvironment and wound healing. Prolonged exposure to tobacco affects inflammatory and reparative cell functions, leading to delayed healing and subsequent complications [15–17]. In colon and rectal surgery, specifically, smoking is an important predictive factor for anastomotic leakage [16]. While this complication was not catalogued in the present study, due to the limitations of the NSQIP database, increased odds of organ space infection were observed. While not an ideal surrogate [26], the development of procedure-targeted NSQIP data has allowed relevant complications (such as anastomotic leakage) to be captured, allowing such variables to be deliberately evaluated in future studies. Improvements in the tissue microenvironment and in inflammatory cellular functions are observed as soon as 4 weeks following smoking cessation [15], reinforcing the importance of provider-driven smoking cessation to mitigate these negative effects on the course of a patient’s disease and postoperative recovery.
Colorectal Disease ª 2015 The Association of Coloproctology of Great Britain and Ireland. 17, 891–902
A. N. Kulaylat et al.
The success of targeted smoking cessation strategies to reduce postoperative complications has been previously demonstrated [27–29]. Interventions to promote smoking cessation utilizing individual counselling initiated at least 4 weeks before noncardiac elective surgery (including general abdominal and orthopaedic) have demonstrated an improvement in the surgical outcome, including reductions in surgical-site complications [27,28]. Success has also been demonstrated in the CD population [29,30]. In one study, 31% of patients achieved complete smoking cessation in an 18-month follow-up period [30]. The present study has several limitations. The determination of smoking status is subject to recall bias as it is historical information provided by the patient. Furthermore, current smokers in NSQIP are defined as anyone who has smoked in the year prior to admission for surgery; therefore the true effect of current smoking status may be diluted by the misclassification of former smokers as current smokers in this cohort. Due to missing data related to measurement of pack-years, and with no other consistent variable to classify ex-smokers, we grouped non- and ex-smokers together for the purposes of our analysis. This consequently biases our findings towards measuring a smaller effect of tobacco on our measured outcomes. Additionally, we classified noncurrent smoking patients with ‘missing’ pack year information as never-smokers, as this approach erred on the side of caution in estimating the effect of tobacco [14,22], while the previously referenced study by Sharma et al. [25] excluded this population. We fitted several different models to estimate dose-dependent risk as a function of increasing pack-year history, but a significant proportion of patients were dropped due to missing pack-year data, and the results were inconsistent. Consequently, we decided against any further analysis or reporting of dose-dependent stratified outcomes. Finally, as a potential weakness related to the available data, we were unable to measure or qualify CD activity or to assess the duration of chronic steroid use or any other CD-related medication. In conclusion, the present study has demonstrated that when matched cohorts of patients with CD are compared on the basis of smoking, infective and pulmonary complications will be more frequently encountered, but with no measurable effects on mortality, LOS or cardiovascular complications. The present study suggests that the deleterious effects of tobacco specifically related to the outcome after surgery for CD are more focused than their general negative effects on the activity of the disease. This information may be helpful for surgeons in terms of decision-making based upon a more precise estimate of the complications CD that patients are more
Smoking and Crohn’s disease: a NSQIP analysis
likely to experience when they are active smokers. Lastly, although this study confirms that tobacco is a risk factor for an adverse postoperative outcome specifically in CD patients, the present study may help surgeons to avoid an exaggerated estimate of the influence of tobacco on the outcome in this population.
Funding source No internal or external financial support was used for this report.
Conflicts of interest All authors report no potential conflicts of interest.
Author contributions Study design: CS Hollenbeak, DB Stewart Sr; Statistical analysis: AN Kulaylat, CS Hollenbeak; manuscript creation and editing: AN Kulaylat, W Sangster; Final approval of manuscript: AN Kulaylat, CS Hollenbeak, W Sangster, DB Stewart Sr.
References 1 Kane SV, Flicker M, Katz-Nelson F. Tobacco use is associated with accelerated clinical recurrence of Crohn’s disease after surgically induced remission. J Clin Gastroenterol 2005; 39: 32–5. 2 Nunes T, Etchevers MJ, Domenech E et al. Smoking does influence disease behaviour and impacts the need for therapy in Crohn’s disease in the biologic era. Aliment Pharmacol Ther 2013; 38: 752–60. 3 Silverstein MD, Lashner BA, Hanauer SB, Kirsner JB. Cigarette smoking in Crohn’s disease. Am J Gastroenterol 1989; 84: 31–3. 4 Tobin MV, Logan RF, Langman MJ, McConnell RB, Gilmore IT. Cigarette smoking and inflammatory bowel disease. Gastroenterology 1987; 93: 316–21. 5 Somerville KW, Logan RF, Edmond M, Langman MJ. Smoking and Crohn’s disease. Br Med J (Clin Res Ed) 1984; 289: 954–6. 6 Aldhous MC, Drummond HE, Anderson N, Smith LA, Arnott ID, Satsangi J. Does cigarette smoking influence the phenotype of Crohn’s disease? Analysis using the Montreal classification. Am J Gastroenterol 2007; 102: 577–88. 7 Cosnes J, Carbonnel F, Beaugerie L, Le Quintrec Y, Gendre JP. Effects of cigarette smoking on the long-term course of Crohn’s disease. Gastroenterology 1996; 110: 424–31. 8 Cosnes J, Carbonnel F, Carrat F, Beaugerie L, Cattan S, Gendre J. Effects of current and former cigarette smoking on the clinical course of Crohn’s disease. Aliment Pharmacol Ther 1999; 13: 1403–11.
Colorectal Disease ª 2015 The Association of Coloproctology of Great Britain and Ireland. 17, 891–902
901
Smoking and Crohn’s disease: a NSQIP analysis
9 Reese GE, Nanidis T, Borysiewicz C, Yamamoto T, Orchard T, Tekkis PP. The effect of smoking after surgery for Crohn’s disease: a meta-analysis of observational studies. Int J Colorectal Dis 2008; 23: 1213–21. 10 Lakatos PL, Czegledi Z, Szamosi T et al. Perianal disease, small bowel disease, smoking, prior steroid or early azathioprine/biological therapy are predictors of disease behavior change in patients with Crohn’s disease. World J Gastroenterol 2009; 15: 3504–10. 11 Sutherland LR, Ramcharan S, Bryant H, Fick G. Effect of cigarette smoking on recurrence of Crohn’s disease. Gastroenterology 1990; 98: 1123–8. 12 Hall BL, Hamilton BH, Richards K, Bilimoria KY, Cohen ME, Ko CY. Does surgical quality improve in the American College of Surgeons National Surgical Quality Improvement Program: an evaluation of all participating hospitals. Ann Surg 2009; 250: 363–76. 13 Leuven E, Sianesi B. PSMATCH2: Stata module to perform full Mahalanobis and propensity score matching, commonsupport graphing, and covariate imbalance testing. 2003. http://ideas.repec.org/c/boc/bocode/s432001. html (accessed October 2013). 14 Musallam KM, Rosendaal FR, Zaatari G et al. Smoking and the risk of mortality and vascular and respiratory events in patients undergoing major surgery. JAMA Surg 2013; 148: 755–62. 15 Sorensen LT. Wound healing and infection in surgery: the pathophysiological impact of smoking, smoking cessation, and nicotine replacement therapy: a systematic review. Ann Surg 2012; 255: 1069–79. 16 Sorensen LT, Jorgensen T, Kirkeby LT, Skovdal J, Vennits B, Wille-Jørgensen P. Smoking and alcohol abuse are major risk factors for anastomotic leakage in colorectal surgery. Br J Surg 1999; 86: 927–31. 17 Sorensen LT, Hemmingsen U, Kallehave F et al. Risk factors for tissue and wound complications in gastrointestinal surgery. Ann Surg 2005; 241: 654–8. 18 Higuchi LM, Khalili H, Chan AT, Richter JM, Bousvaros A, Fuchs CS. A prospective study of cigarette smoking and the risk of inflammatory bowel disease in women. Am J Gastroenterol 2012; 107: 1399–406.
902
A. N. Kulaylat et al.
19 Joyce MR, Hannaway CD, Strong SA, Fazio VW, Kiran RP. Impact of smoking on disease phenotype and postoperative outcomes for Crohn’s disease patients undergoing surgery. Langenbecks Arch Surg 2013; 398: 39–45. 20 Picco MF, Bayless TM. Tobacco consumption and disease duration are associated with fistulizing and stricturing behaviors in the first 8 years of Crohn’s disease. Am J Gastroenterol 2003; 98: 363–8. 21 Sands BE, Arsenault JE, Rosen MJ et al. Risk of early surgery for Crohn’s disease: implications for early treatment strategies. Am J Gastroenterol 2003; 98: 2712–8. 22 Hawn MT, Houston TK, Campagna EJ et al. The attributable risk of smoking on surgical complications. Ann Surg 2011; 254: 914–20. 23 Turan A, Mascha EJ, Roberman D et al. Smoking and perioperative outcomes. Anesthesiology 2011; 114: 837–46. 24 Gajdos C, Hawn MT, Campagna EJ, Henderson WG, Singh JA, Houston T. Adverse effects of smoking on postoperative outcomes in cancer patients. Ann Surg Oncol 2012; 19: 1430–8. 25 Sharma A, Deeb AP, Iannuzzi JC, Rickles AS, Monson JR, Fleming FJ. Tobacco smoking and postoperative outcomes after colorectal surgery. Ann Surg 2013; 258: 296–300. 26 Rickles AS, Iannuzzi JC, Kelly KN et al. Anastomotic leak or organ space surgical site infection: What are we missing in our quality improvement programs? Surgery 2013; 154: 680–7; discussion 687-9. 27 Thomsen T, Tonnesen H, Moller AM. Effect of preoperative smoking cessation interventions on postoperative complications and smoking cessation. Br J Surg 2009; 96: 451–61. 28 Moller AM, Villebro N, Pedersen T, Tonnesen H. Effect of preoperative smoking intervention on postoperative complications: a randomised clinical trial. Lancet 2002; 359: 114–7. 29 Hilsden RJ, Hodgins DC, Timmer A, Sutherland LR. Helping patients with Crohn’s disease quit smoking. Am J Gastroenterol 2000; 95: 352–8. 30 Nunes T, Etchevers MJ, Merino O et al. High smoking cessation rate in Crohn’s disease patients after physician advice–the TABACROHN Study. J Crohns Colitis 2013; 7: 202–7.
Colorectal Disease ª 2015 The Association of Coloproctology of Great Britain and Ireland. 17, 891–902
