- Email: [email protected]
Statin medications are associated with decreased risk of sepsis and anastomotic leaks after rectal resections
Accepted Manuscript Statin Medications Are Associated with Decreased Risk of Sepsis and Anastomotic Leaks after Rectal Resections David Disbrow, MD, Corie L. Seelbach, DO, Jeremy Albright, PhD, Jane Ferraro, MPP, RN, Juan Wu, ScD, Jon M. Hain, MD, Beth-Ann Shanker, MD, Robert K. Cleary, MD PII:
S0002-9610(17)31471-X
DOI:
10.1016/j.amjsurg.2018.01.024
Reference:
AJS 12732
To appear in:
The American Journal of Surgery
Received Date: 14 October 2017 Revised Date:
30 November 2017
Accepted Date: 4 January 2018
Please cite this article as: Disbrow D, Seelbach CL, Albright J, Ferraro J, Wu J, Hain JM, Shanker B-A, Cleary RK, Statin Medications Are Associated with Decreased Risk of Sepsis and Anastomotic Leaks after Rectal Resections, The American Journal of Surgery (2018), doi: 10.1016/j.amjsurg.2018.01.024. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
Title: Statin Medications Are Associated with Decreased Risk of Sepsis and Anastomotic Leaks after Rectal Resections Authors: David Disbrow MDa , Corie L. Seelbach DOb , Jeremy Albright PhDc , Jane Ferraro
RI PT
MPP, RNd , Juan Wu ScDe , Jon M. Hain MDf , Beth-Ann Shanker MDg , Robert K Cleary MDh
a
Colon and Rectal Surgery Resident, Department of Surgery, St. Joseph Mercy Hospital Ann
SC
Arbor, Ann Arbor, MI, [email protected] b
Department of Surgery, William Beaumont Hospital Troy, Troy, MI, [email protected]
c
[email protected] d
M AN U
Senior Data Scientist, Methods Consultants, 25 East Cross St., Ypsilanti, MI.
Program Manager of Colorectal Research, St. Joseph Mercy Hospital Ann Arbor, Ann Arbor,
MI, [email protected] e
Outcomes Research Scientist, Academic Research Department, St. Joseph Mercy Hospital
f
TE D
Ann Arbor, Ann Arbor, MI, [email protected] Department of Surgery, William Beaumont Hospital Troy, Troy, MI, [email protected]
g
Department of Surgery, St Joseph Mercy Hospital Ann Arbor, Ann Arbor, MI,
EP
[email protected] h
Program Director Colon and Rectal Surgery, Director of Research, St. Joseph Mercy Hospital
AC C
Ann Arbor, Ann Arbor, MI, [email protected]
Corresponding Author: Robert K. Cleary, MD 5333 McAuley Dr.
Reichert Health Center, Suite 2115 Ann Arbor, MI 48106
ACCEPTED MANUSCRIPT
tel: 734-358-9730 fax: 734-712-8151
RI PT
Email: [email protected]
Disclaimers: None to Report
M AN U
public, commercial, or not-for-profit sectors.
SC
Funding Sources: This research did not receive any specific grant from funding agencies in the
This research was presented by Dr. Drisbow at the annual American Society of Colon and
AC C
EP
TE D
Rectal Surgeons Meeting, Seattle, WA, June 10-14, 2017.
RESEARCH HIGHLIGHTS ●
Statins are associated with decreased risk of sepsis after colorectal surgery
●
Statins are associated with fewer anastomotic leaks in rectal resection group
ACCEPTED MANUSCRIPT
Statins are not associated with surgical site infections or mortality
AC C
EP
TE D
M AN U
SC
RI PT
●
ABSTRACT
Background: This study was designed to determine the effect of statins on colorectal postoperative complications related to sepsis. Previous studies have reported conflicting results.
ACCEPTED MANUSCRIPT
Methods: This is a retrospective propensity score analysis of postoperative outcomes from a large regional database of patients who underwent elective colorectal resection from June 2012-
RI PT
July 2015.
Results: 7285 patients met inclusion criteria: 34.5% received statins. Propensity score matching revealed that patients taking statins had reduced risk of sepsis (3.75% vs 5.32%, p=0.03).
SC
Subgroup analysis revealed that this difference was driven by patients undergoing rectal
resections. Among the rectal resection group, anastomotic leaks were more common in the non-
M AN U
statins group (4.1% vs. 1.3%, p=0.01). There was no significant difference between those taking statins and those not on statin medications with respect to composite SSI or 30-day mortality.
Conclusions: Statin medications are associated with decreased risk of sepsis after colorectal
TE D
surgery and anastomotic leaks after rectal resection. Future studies should focus on medication type, dosage, and duration to confirm these results and identify patient populations that would
EP
benefit most from statin therapy.
AC C
Keywords: Statin Medication, Colorectal, Colon and Rectal Surgery, Anastomotic Leak, Sepsis
ACCEPTED MANUSCRIPT
INTRODUCTION: Infectious complications occur in 9.4-44% of patients after colorectal surgery and are a source of significant morbidity and mortality.1-6 The most severe complications are sepsis and
RI PT
anastomotic leak. Sepsis occurs in 5.4% of patients undergoing elective colorectal surgery while anastomotic leaks occur in 1-24% depending on the location of the anastomosis and other
SC
clinical and patient risk factors.7-10
Three-hydroxy-3-methylglutaryl coenzyme A (HMG-CoA) reductase inhibitors, or statins, have
M AN U
numerous lipid-independent pleiotropic effects that include decreased inflammation, improved endothelial function, antithrombotic properties, reduction of cytokines, and increased angiogenesis.11-14 Clinical studies have found statins to be beneficial for wound healing, the prevention of sepsis, and the mitigation of the effects of sepsis.14-17 The impact of statin use in the peri-operative colorectal patient has not yet been determined, with some studies showing a
TE D
decrease in septic complications and others showing no difference.7,17-20
The purpose of this study is to determine if colorectal surgery patients on statins have
EP
fewer complications related to sepsis and anastomotic leaks than those patients not on statins using a protocol-driven regional database. This study was also done to
AC C
determine if complications are different depending on surgery location – colectomy vs rectal resection. Confounding was addressed through propensity score matching on demographics and pre-surgery comorbidities.
MATERIALS & METHODS: Data Source This observational retrospective study of the Michigan Surgical Quality Collaborative (MSQC) database was composed of an analysis of all patients who underwent elective colorectal
ACCEPTED MANUSCRIPT
resection from June 2012 through July 2015. The MSQC is a consortium of 64 hospitals representing diverse practice settings throughout the state of Michigan and includes a large protocol-driven regional database devoted to quality, improving patient outcomes, and cost-
RI PT
effectiveness. Specially trained data abstractors prospectively collect patient demographics, characteristics, intraoperative processes of care, and 30-day postoperative outcomes and
complications. A standard sampling algorithm is applied to minimize selection bias. Regular
SC
data audits ensure registry data validity. Data collection for MSQC is institutional review board exempt at participating hospitals. The quality and validity of MSQC methodology have been
M AN U
demonstrated in several reports.21-22
Study Data
Patients were identified for study inclusion by Current Procedural Terminology codes for standard colon and rectal operations (44140, 44145, 44146, 44147, 44150, 44155, 44158,
TE D
44160, 44204, 44205, 44207, 44208, 44210, 44211, 44212, 45110, 45112, 45119, 45395, 45397, 45400, 45402, 45540, 45550). Only those codes that include an anastomosis were used for analysis of the anastomotic leak outcome. Exclusion criteria were age under 18 years,
EP
current pregnancy, ASA class 5 and 6, emergent cases, and the presence of preoperative open wounds with or without infection. The statin variable in the MSQC database is reported as yes
AC C
or no. There is no information available with regard to statin medication choice, dose, and duration of use.
Outcomes of interest were: 1) anastomotic leak; 2) any SSI including superficial SSI, deep SSI, and organ/space SSI; 3) sepsis; and 4) death within 30 days of surgery. The MSQC definition for anastomotic leak is a clinically diagnosed leak at the site of the intestinal anastomosis requiring one or more of the following interventions: antibiotic treatment, percutaneous drainage, reoperation with new anastomosis, reoperation with proximal diversion, or reoperation with end
ACCEPTED MANUSCRIPT
stoma. The definitions for SSIs were based on Centers for Disease Control and Prevention definitions. Finally, the definition for sepsis was based on the Surviving Sepsis definition.23 All outcome measures were abstracted in clinical (not administrative) sources to include
RI PT
intraoperative records, operative reports, nursing notes, progress notes, discharge summaries, physician office notes, or 30-day follow-up from the patient. Secondary outcomes considered were acute renal failure, myocardial infarction, deep venous thrombosis, unplanned intubation,
SC
pneumonia, urinary tract infection, 30-day readmission, 30-day reoperations, and discharge to a
M AN U
rehabilitation or skilled nursing facility.
Patients on statin therapy were initially compared with patients not on statin therapy based on demographic factors (age, sex, race, and BMI), general health factors (tobacco use, alcohol abuse, ASA class, number of comorbidities), the presence of comorbidities (coronary artery disease, hypertension, congestive heart failure, peripheral vascular disease (claudication or rest
TE D
pain or gangrene, history of revascularization or amputation for peripheral vascular disease), renal failure requiring dialysis, chronic obstructive pulmonary disease, weight loss more than 10 pounds over the past 6 months, bleeding disorder, beta blocker use, hypoalbuminemia,
EP
obstructive sleep apnea, diabetes, and cancer diagnosis). Categorical variables were compared by using χ2, and continuous variables (e.g. age) were compared by using an independent
AC C
samples t test. This was repeated for both the primary and secondary outcomes.
Statistical Analysis
Because this was not a randomized trial and significant differences between statin users and non-users were expected, it was necessary to control for potential confounders. The large number of potential pre-treatment confounders relative to the number of events, especially for the subgroup analysis, meant that logistic regressions may overfit the data. We therefore opted
ACCEPTED MANUSCRIPT
for matching statin and non-statin patients on the basis of propensity scores, which are the probability of taking statins given the set of covariates. The matching was done by finding the nearest neighbor that fell within .1 standard deviations of the propensity score. This approach
RI PT
dropped cases from both the statin and non-statin groups for whom a similar counterpart was unavailable in the other treatment group. The variables used in the matching are those listed in Table 1 and the tables in the appendix. The accuracy of the propensity score matching was
SC
assessed by comparing the distribution of the confounders between groups after the matching. Pre-surgery comorbidity counts still displayed some unbalance in the matched samples
M AN U
(p=0.01). We therefore additionally controlled for this variable in a logistic regression in the matched samples as a "doubly robust" approach. Additional graphical assessments were also conducted and are presented in the Appendix. The propensity score matching was performed separately for the entire colorectal sample, the rectal resection subsample, and the colectomy subsamples. The post-matching confounder distributions are presented in the body of the
relegated to the Appendix.
TE D
paper for the whole sample. For space considerations, similar tables for the two subgroups are
EP
After creating matched subsamples with similar demographics and pre-surgery comorbidities, the distribution of the outcomes in each treatment group are presented along with χ2 tests to
AC C
assess statistical significance (p < 0.05). All analyses were performed using R version 3.3.2. The propensity score matching was performed using the matchit package.24
RESULTS All Cases
From June 2012 to July 2015, a total of 10,273 patients undergoing elective colorectal surgery were identified in the MSQC database. After applying exclusion criteria, 7,285 patients were available for analysis with 2515 (34.5%) on statin therapy and 4770 (65.5%) not on statin
ACCEPTED MANUSCRIPT
therapy. Statistical differences in patient characteristics both before and after propensity score matching are shown in Table 1. As expected, the patients in the statin therapy group were significantly older (68.2 vs 59.2, p < 0.001) prior to the matching and had a significantly higher
RI PT
average number of comorbidities (2.44 vs 1.14, p < 0.001). There was also a significant difference between groups with respect to indication for surgery and type of surgery performed (colectomy vs. rectal resection). In the statin group, 28% had rectal resections compared to 33%
SC
of those not on statins. Following propensity score matching, the only variable that remained significantly different between the two treatment groups was the mean number of preoperative
M AN U
comorbidities (2.07 vs 1.95, p = 0.01). Although the p-value is less than .05, this difference is much smaller compared to the pre-matched sample.
Table 2 shows the comparisons of the primary and secondary outcomes by statin group along with p-values both before and after the propensity score matching. Before adjusting for
TE D
potential confounders, there are no statistically significant differences in any of the four primary outcomes. The statin group had a small but statistically significant higher percentage of unplanned intubation (2.07% versus 1.40%, p = 0.04) as well as a greater percentage of
EP
patients not discharged to home (26.6% versus 22.2%, p < 0.001.). There was no statistically
AC C
significant difference in 30-day mortality between groups (0.7% versus 1.1%, p = .172)
The propensity score matching reduced the sample to 1786 comparable cases in each treatment group. After the matching, the primary outcome of sepsis becomes significant, with the non-statin group having higher rates of sepsis (5.32 vs 3.75, p = 0.03). For the secondary outcomes, unplanned intubation loses its significance and discharge destination sits right at the .05 threshold. At the same time, 30-day readmissions become significantly more likely for the non-statin group relative to the statin group (11.25% vs 9.07%, p=0.04). The difference in 30day mortality remained non-significant in the matched sample (1.3% vs 0.7%, p = .065).
ACCEPTED MANUSCRIPT
Subsample Analysis
RI PT
Table 3 shows a similar analysis for colectomy patients. In the unadjusted analysis, there were 3196 in the no-statins group and 1810 in the statins group. Unplanned intubations were
significantly higher in the statins group (2.43% vs 1.50%, p = 0.03). Discharge destination was
SC
also significantly different, with the statin group more likely to be discharged to a rehab or skilled nursing facility (22.21 versus 16.65%, p < 0.001.) These differences disappeared following the
M AN U
matched analysis, which was based on a sample of 1322 in each group. One secondary outcome became significant following propensity score matching. Those on statins were less likely to be readmitted within 30 days (7.41% vs 10.29%, p = 0.01). No other outcomes differed significantly by treatment.
TE D
Table 4 presents the treatment group comparisons for the subset of patients undergoing rectal resections. There were 1,574 cases in the data set not on statins and 705 that were taking statins. The unadjusted comparisons show a significant difference in the rate of anastomotic
EP
leaks, with statin takers having a lower rate of occurrence (1.31% vs 3.41%, p = 0.01). No other
AC C
outcomes were significantly different.
The matching reduced the sample to 485 cases in each treatment group. The significant difference in anastomotic leaks remained and was slightly more pronounced, with outcomes favoring the statin patients (1.29% vs 4.13%, p = 0.01). In addition, sepsis now emerges as statistically significant, again favoring the statin takers (2.89% vs 6.60%, p = 0.01). The remaining comparisons are not statistically significant.
DISCUSSION
ACCEPTED MANUSCRIPT
This large regional database analysis of patients undergoing elective colon and rectal surgery revealed that patients on statins have significantly less postoperative sepsis than those not on
RI PT
statins according to an analysis of propensity score matched samples. Subgroup analysis of those having rectal resections supported the postoperative sepsis outcome advantage for the statin group. In addition, patients on statins undergoing rectal resections have significantly fewer
SC
anastomotic leaks as confirmed by both statistical analyses. These advantages for statin takers
M AN U
were not observed for colectomies.
An American College of Surgeons-National Surgical Quality Improvement Program (ACSNSQIP) study of 7777 non-cardiac surgery patients, 19.7% of whom had colorectal resections, concluded that statins were protective against major non-cardiac complications (OR 0.62, 95% CI [.49,.92], P < .001), respiratory complications (OR 0.63, 95% CI [0.50,0.79], p=.017), venous
TE D
thromboembolism (OR 0.41, 95% CI [.18,.98], p=.044), and infectious complications defined as organ space surgical site infections, sepsis, and systemic inflammatory response syndrome (OR 0.65, 95% CI [.45,.94], p=.023).25 This study differs from ours in that their data source was
EP
composed of single institution data merged with the ACS-NSQIP dataset, and their study population was not limited to colorectal surgery patients. Our study is similar to this ACS-NSQIP
AC C
study in that it supported the protective effect of statins on postoperative sepsis. In contrast, our study did not reveal a protective effect for statins against surgical site infections, perhaps because our study included all 3 surgical site infection classes and not just organ space infections. Statin use was also associated with fewer 30-day readmissions in our study, though there was no significant difference between groups with respect to venous thromboembolism.
Khan, et. al., reported a retrospective cohort analysis of 577 patients who underwent curative resections for colon cancer, 21.7% of whom were on statins.16 Though the statin cohort was
ACCEPTED MANUSCRIPT
significantly older (74.7 vs 69.2) and had more comorbid conditions, their study also demonstrated no significant difference in mortality rates between those on and those not on statins (7.2% vs 6.4%, p=0.77), and a difference in sepsis favoring the statin group, though it
RI PT
was not statistically significant (13.3% vs 20.7%, p=0.22). In contrast to our study, this ACSNSQIP study also showed significantly fewer wound infections (8.8% vs 14.6%, p=0.04) in the
SC
statin group when compared to those not on statins.16
Statins are competitive inhibitors of HMG-CoA reductase, the rate-limiting enzyme of the cholesterol biosynthetic pathway that results in anti-inflammatory effects on vascular endothelial
M AN U
cells.14 The anti-inflammatory benefits of statins are related to decreased plasma concentrations of TNF-alpha and IL-6 in postoperative colon and rectal surgery patients on statin medications.14-20 Statins also have antiplatelet and antithrombotic actions, promote angiogenesis, protect against oxidative stress, and have direct antimicrobial effects. Animal
TE D
models investigating the strength of intestinal anastomoses have shown that statin groups have higher anastomotic bursting pressures and hydroxyproline content.12 The clinical benefit of these actions has been suggested in retrospective studies with respect to improving wound
EP
healing and decreasing infectious complications.7,17-20,25,26 Our study also reveals a potential
AC C
benefit for anastomotic leak rates on rectal resections that requires further study.
A Danish study of 2755 colorectal surgery patients,19% of whom took statins, revealed that there was no significant difference in anastomotic leaks when comparing those on statins to those not on statins (OR 1.31, 95% CI [.84,2.05], p=0.23).7 This finding was the same for high and low dose statin therapy. This is in contrast to our study that showed significantly fewer anastomotic leaks for rectal resections. These authors identified rectal resection as a risk factor for anastomotic leak but did not perform subgroup analysis of colectomy and rectal resection for statin users. In addition, their patient population was limited to those with colorectal cancer and
ACCEPTED MANUSCRIPT
included only those anastomotic leaks that required operative intervention. Our study included colorectal surgery patients with benign diagnoses and anastomotic leaks that were treated nonoperatively. Singh, et al. revealed fewer anastomotic leaks in patients on statins undergoing
RI PT
colectomies.18 This finding is in contrast to our study that showed no difference in anastomotic leaks for colectomies. These authors did not include rectal resections in their analysis. The reason for the different finding in their study compared to ours is not obvious, though their patient population and study design were different in that it was a prospective analysis of
M AN U
SC
elective colectomies in a hospital database within an enhanced recovery program.18
This study has several strengths and limitations. Our analysis was limited to available variables in the database, which does not allow evaluation of all potential confounders. We utilized subsamples matched on propensity scores – the probability of being on a statin as a function of pre-surgery characteristics - to best adjust for the known differences between the two treatment
TE D
groups. We nonetheless cannot exclude the influence of the unmeasured or residual confounding to our results. Specifically, outcomes that were originally not statistically significant between statin and non-statin groups became significant after matching, which indicates that we
EP
have captured and controlled for most, if not all, negative confounders that have obscured the associations in the original unmatched study population. We further found that ASA Class and
AC C
preoperative comorbidities count that broadly reflect pre-surgery health are two principal negative confounders - both are more prevalent in the statin group. Information regarding types, dosages and duration of use for statins received are not available in the MSQC database. Therefore, future studies with more detailed assessment of statin usage are warranted to determine the optimal benefits. Privacy restrictions also meant that no zip code information was available that could be used to incorporate socioeconomic variables, and the data did not contain an indicator for proper pre-operative use of antibiotics. A small number of cases for the secondary outcomes limits our study power to detect a statistically significant difference. Lastly,
ACCEPTED MANUSCRIPT
the difference between anastomotic leak and deep organ space surgical site infection definitions can be subtle. It is possible some leaks may have been assigned to an organ space SSI, thereby underestimating the true anastomotic leak rate. However, such non-differential
RI PT
misclassification of the anastomotic leak outcome with respect to statin therapy should lead to an underestimation of the true association. The strength of this study is the source of the data a protocol-driven, risk adjusted, regularly validated large database with data collected by trained
M AN U
chart review, letters, and telephone calls).
SC
registered nurses using clinical confirmation (chart ascertainment and 30-day follow-up with
Our study and several others raise the question of whether or not statin therapy should be considered perioperatively and started preoperatively to protect against postoperative septic complications. However, the value of preoperative statin intervention has not yet been confirmed in studies that administer preoperative statins for this purpose. A meta-analysis of
TE D
randomized controlled trials of patients undergoing cardiac and noncardiac surgery who had statins prescribed one to 3 weeks prior to surgery showed insufficient evidence to support starting statins preoperatively when compared to placebo.26 Another prospective randomized
EP
controlled trial of 132 colorectal surgery patients who received statins 3-7 days preoperatively and postoperatively showed decreased levels of inflammatory markers IL-6, IL-8, and TNF-
AC C
alpha, but no significant difference in postoperative complications.20 If there is any benefit to preoperative statin therapy, how long a patient must be on statin therapy before surgery to benefit from the pleiotropic effects is not clear and it may be that months are required rather than days or weeks. For colorectal cancer patients, prolonged preoperative statin therapy may not be practical except in selected patients with rectal cancer undergoing neoadjuvant therapy. Future studies may serve to confirm or refute these possibilities with attention to statin dosage and duration, and operative types and indications.
ACCEPTED MANUSCRIPT
CONCLUSION This large regional database analysis showed that patients on statin medications are associated with a significant reduction in the rate of sepsis after colorectal surgery and in anastomotic leaks
RI PT
after rectal resection. Further studies with attention to medication dose and duration are needed to confirm or refute these results and identify patient populations that would benefit most from
References:
M AN U
SC
statin therapy.
1. Smith RL, Bohl JK, McElearney ST, et al. Wound infection after elective colorectal resection. Ann Surg. 2004;239:599–605.
TE D
2. Paun BC, Cassie S, MacLean AR, Dixon E, Buie WD. Postoperative complications following surgery for rectal cancer. Ann Surg. 2010;251:807–818.
EP
3. Aimaq R, Akopian G, Kaufman HS. Surgical site infection rates in laparoscopic versus open
AC C
colorectal surgery. Am Surg. 2011;77:1290–1294.
4. Bishawi M, Fakhoury M, Denoya PI, Stein S, Bergamaschi R. Surgical site infection rates: open versus hand-assisted colorectal resections. Tech Coloproctol. 2014;18:381–386.
5. Bot J, Piessen G, Robb WB, Roger V, Mariette C. Advanced tumor stage is an independent risk factor of postoperative infectious complications after colorectal surgery: arguments from a case-matched series. Dis Colon Rectum. 2013;56(5):568–576.
ACCEPTED MANUSCRIPT
6. Olivier F, Paquette B, Orry D, et al. Diagnostic accuracy of inflammatory markers as early predictors of infection after elective colorectal surgery: results from the IMACORS study.
RI PT
Ann Surg. 2016;263(5):961–966.
7. Bisgard AS, Noack MW, Klein M, et al. Perioperative statin therapy is not associated with
SC
reduced risk of anastomotic leakage after colorectal resection. Dis Colon Rectum.
M AN U
2013;56:980--986.
8. Paulasir S, Kaoutzanis C, Welch K, et al. Nonsteroidal anti-inflammatory drugs: do they increase the risk of anastomotic leaks following colorectal operations? Dis Colon Rectum. 2015;58(9):870--877.
TE D
9. Goligher JC, Graham NG, De Dombal FT. Anastomotic dehiscence after anterior resection of rectum and sigmoid. Br J Surg. 1970;57:109–118.
EP
10. Ogilvie JW Jr, Dietz DW, Stocchi L. Anastomotic leak after restorative proctosigmoidectomy for cancer: what are the chances of a permanent ostomy? Int J Colorectal Dis.
AC C
2012;27:1259–1266.
11. Derici J, Yaman I, Kara C, Kamer E, Diniz G, Ortac R. Simvastatin improves incisional wound healing in a rat model: an experimental study. Wounds. 2012;24(7):195--200
12. Karadeniz Cakmak G, Irkorucu O, Ucan BH, et al. Simvastatin improves wound strength after intestinal anastomosis in the rat. J Gastrointest Surg. 2009;13:1707--1716.
ACCEPTED MANUSCRIPT
13. Liao JK, Laufs U. Pleiotropic effects of statins. Annu Rev Pharmacol Toxicol. 2005;45:89– 118.
RI PT
14. Brookes ZL, McGown CC, Reilly CS. Statins for all: the new premed? Br J Anaesth. 2009;103:99–107
SC
15. Fitzmaurice GJ, McWilliams B, Nolke L, Redmond JM, McGuiness JG, O’Donnell ME. Do statins have a role in the promotion of postoperative wound healing in cardiac surgical
M AN U
patients? Ann Thorac Surg. 2014;98(2):756-764
16. Khan A, Yeung D, Wyatt B, et al. Effects of statins on postoperative sepsis, systemic inflammatory response syndrome and mortality after colorectal surgery. Crit Care
TE D
2009;13(Suppl 4):47
17. Santos J. Can statins improve outcomes in colorectal surgery? Journal of Coloproctology.
EP
2012;32(3):334-338.
18. Singh PP, Srinivasa S, Bambarawana S, et al. Perioperative use of statins in elective
AC C
colectomy. Dis Colon Rectum. 2012;55(2):205--210.
19. Battersby CLF, Green P, Vyapury V, Rooney P, Hunt J. Statins may modify colorectal anastomotic leak risk in high risk patients. Gut. 2015;64(Suppl 1):A162--163
ACCEPTED MANUSCRIPT
20. Sing PP, Lemanu DP, Soop M, MD, Bissett IP, Harrison, Hill AG. Perioperative simvastatin therapy in major colorectal surgery: a prospective, double-blind randomized controlled trial.
RI PT
J Am Coll Surg. 2016;223:308-320.
21. Khuri SF, Daley J, Henderson W, et al. The Department of Veterans Affairs’ NSQIP: the first national, validated, outcome-based, risk-adjusted, and peer-controlled program for the
M AN U
Improvement Program. Ann Surg. 1998;228:491–507.
SC
measurement and enhancement of the quality of surgical care. National VA Surgical Quality
22. Campbell DA Jr, Kubus JJ, Henke PK, Hutton M, Englesbe MJ. The Michigan Surgical Quality Collaborative: a legacy of Shukri Khuri. Am J Surg. 2009;198(5 suppl):S49–S55.
23. Dellinger RP, Levy MM, Rhodes A, et al. Surviving sepsis campaign: international guidelines
2013;39:165–228.
TE D
for management of severe sepsis and septic shock, 2012. Intensive Care Med.
EP
24. Ho DE, Imai K, King G, Stuart EA. MatchIt: Nonparametric Preprocessing for Parametric Causal Inference. Journal of Statistical Software, 2011;Vol. 42, No. 8, pp. 1-28. URL
AC C
http://www.jstatsoft.org/v42/i08/.
25. Iannuzzi JC, Rickles AS, Kelly NK, et al. Perioperative pleiotropic statin effects in general surgery. Surgery. 2014;155:398-407.
26. Li H, Lin YL, Diao SL, Ma BX, Liu XL. Does short preoperative statin therapy prevent infectious complications in adults undergoing cardiac or non-cardiacsurgery? A metaanalysis of 5 randomized placebo-controlled trials. Saudi Med J. 2016;37(5):492-497
ACCEPTED MANUSCRIPT
Appendix
EP
TE D
M AN U
SC
RI PT
Propensity score matching attempts to balance the treatment groups across covariates such that the groups differ only by a random component. The variables used in our propensity score model included the pre-surgery variables as well as demographic information. A tabular summary of post-matching covariate distributions was presented in the body for the entire sample. This appendix provides figures that visualize the pre and post matching cases in terms of the distribution of propensity scores and covariate balance. These figures are presented for the entire sample, the rectal resection subsample, and the colectomy subsample. It also provides tabular comparisons for the two subgroups.
AC C
Figure A1. Propensity score distribution of matched and unmatched cases – All patients.
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
Figure A2. Propensity score histograms before and after matching – All patients.
Figure A3. Standardized mean covariate differences before and after matching – All patients.
ACCEPTED MANUSCRIPT
Table A1. Pre and Post Matching Comparisons of Covariates – Rectal Resection Cases Only
29 (1.84%) 841 (53.43%) 670 (42.57%) 34 (2.16%)
2 (0.28%) 214 (30.35%) 445 (63.12%) 44 (6.24%)
SC
RI PT
Unadjusted Statins - Yes (n = 705) 20 (2.84%) 42 (5.96%) 260 (36.88%) 2 (0.28%) 130 (18.44%) 75 (10.64%) 6 (0.85%) 556 (78.87%) 308 (43.69%) 33 (4.68%) 46 (6.52%) 8 (1.13%) 26 (3.69%) 27 (3.83%)
AC C
1 2 3 4
TE D
M AN U
Category
EP
Variable Preop Cancer, n (%) Preop Albumin, n (%) Preop Diabetes, n (%) Preop Ascites/Cirrhosis, n (%) Preop Sleep Apnea, n (%) Preop COPD, n (%) Preop CHF, n (%) Preop Hypertension, n (%) Preop Beta Blocker, n (%) Preop Peripheral Vascular Disease, n (%) Preop DVT, n (%) Preop Dialysis, n (%) Preop Weight Loss, n (%) Preop Bleeding Disorder, n (%) ASA Class ID, n (%)
Statins - No (n = 1,574) 59 (3.75%) 67 (4.26%) 174 (11.05%) 11 (0.7%) 140 (8.89%) 107 (6.8%) 5 (0.32%) 626 (39.77%) 271 (17.22%) 26 (1.65%) 76 (4.83%) 2 (0.13%) 53 (3.37%) 28 (1.78%)
p-value 0.329 0.098 < 0.001 0.36 < 0.001 0.002 0.17 < 0.001 < 0.001 < 0.001 0.118 0.003 0.793 0.005 < 0.001
Statins - No (n = 485) 16 (3.3%) 25 (5.15%) 113 (23.3%) 2 (0.41%) 72 (14.85%) 47 (9.69%) 3 (0.62%) 361 (74.43%) 175 (36.08%) 16 (3.3%) 31 (6.39%) 1 (0.21%) 10 (2.06%) 16 (3.3%)
PS Matched Statins - Yes (n = 485) 13 (2.68%) 25 (5.15%) 143 (29.48%) 1 (0.21%) 82 (16.91%) 56 (11.55%) 2 (0.41%) 360 (74.23%) 183 (37.73%) 20 (4.12%) 31 (6.39%) 2 (0.41%) 13 (2.68%) 15 (3.09%)
3 (0.62%) 180 (37.11%) 285 (58.76%) 17 (3.51%)
2 (0.41%) 178 (36.7%) 285 (58.76%) 20 (4.12%)
p-value 0.706 > 0.999 0.035 > 0.999 0.429 0.404 > 0.999 > 0.999 0.641 0.61 > 0.999 > 0.999 0.673 > 0.999 0.929
ACCEPTED MANUSCRIPT
Table A1 (continued)
26 (1.65%) 488 (31%) 692 (43.96%) 309 (19.63%) 59 (3.75%)
3 (0.43%) 221 (31.35%) 326 (46.24%) 141 (20%) 14 (1.99%)
1.101 (1.284)
2.487 (1.531)
< 0.001
66.635 (10.492) 29.773 (6.313)
< 0.001 < 0.001
EP
Preop Comorbidities Count, Mean (SD)
617 (87.52%) 51 (7.23%) 8 (1.13%) 29 (4.11%) 111 (15.74%) 26 (3.69%)
TE D
Smoker, n (%) Alcohol Consumption, n (%) Reason for Surgery, n (%) Crohn's Diverticulitis Neoplasm Other Ulcerative Colitis
1335 (84.82%) 156 (9.91%) 19 (1.21%) 64 (4.07%) 389 (24.71%) 55 (3.49%)
M AN U
White Af Am Other Unknown
AC C
Age, Mean (SD) 57.601 (13.655) BMI, Mean (SD) 28.498 (6.734) Note. p-values from chi-square tests and independent samples t-tests.
Statins - No (n = 485) 220 (45.36%)
PS Matched Statins - Yes (n = 485) 222 (45.77%)
441 (90.93%) 24 (4.95%) 4 (0.82%) 16 (3.3%) 84 (17.32%) 19 (3.92%)
422 (87.01%) 34 (7.01%) 6 (1.24%) 23 (4.74%) 89 (18.35%) 17 (3.51%)
0 (0.0%) 184 (37.94%) 204 (42.06%) 90 (18.56%) 7 (1.44%)
0 (0.0%) 169 (34.85%) 208 (42.89%) 102 (21.03%) 6 (1.24%)
1.957 (1.389) 65.833 (11.225) 29.438 (6.502)
2.216 (1.474)
0.005
65.115 (9.913) 29.707 (6.354)
0.292 0.515
RI PT
Category
Unadjusted Statins - Yes (n = 705) 331 (46.95%)
p-value 0.605 0.233
SC
Variable Sex = Male, n (%) Race, n (%)
Statins - No (n = 1,574) 719 (45.68%)
< 0.001 0.914 0.026
p-value 0.949 0.284
0.737 0.865 0.682
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
Figure A4. Propensity score distribution of matched and unmatched cases – Rectal Resections.
Figure A5. Propensity score histograms before and after matching – Rectal Resections.
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
Figure A6. Standardized mean covariate differences before and after matching – Rectal Resections.
ACCEPTED MANUSCRIPT
Table A2. Pre and Post Matching Comparisons of Covariates – Rectal Resection Cases Only
68 (2.13%) 1753 (54.85%) 1292 (40.43%) 83 (2.6%)
9 (0.5%) 556 (30.72%) 1121 (61.93%) 124 (6.85%)
TE D EP
1 2 3 4
Statins - No (n = 1,322) 21 (1.59%) 48 (3.63%) 309 (23.37%) 2 (0.15%) 198 (14.98%) 139 (10.51%) 5 (0.38%) 959 (72.54%) 478 (36.16%) 29 (2.19%) 76 (5.75%) 5 (0.38%) 33 (2.5%) 38 (2.87%)
PS Matched Statins - Yes (n = 1,322) 22 (1.66%) 44 (3.33%) 358 (27.08%) 3 (0.23%) 203 (15.36%) 136 (10.29%) 5 (0.38%) 955 (72.24%) 505 (38.2%) 40 (3.03%) 85 (6.43%) 7 (0.53%) 33 (2.5%) 47 (3.56%)
9 (0.68%) 523 (39.56%) 729 (55.14%) 61 (4.61%)
9 (0.68%) 504 (38.12%) 742 (56.13%) 67 (5.07%)
RI PT
Unadjusted Statins - Yes (n = 1,810) 32 (1.77%) 56 (3.09%) 599 (33.09%) 5 (0.28%) 305 (16.85%) 208 (11.49%) 11 (0.61%) 1410 (77.9%) 842 (46.52%) 88 (4.86%) 129 (7.13%) 8 (0.44%) 40 (2.21%) 82 (4.53%)
p-value 0.065 0.105 < 0.001 0.11 < 0.001 < 0.001 0.05 < 0.001 < 0.001 < 0.001 < 0.001 0.494 0.036 < 0.001
SC
M AN U
Category
AC C
Variable Preop Cancer, n (%) Preop Albumin, n (%) Preop Diabetes, n (%) Preop Ascites/Cirrhosis, n (%) Preop Sleep Apnea, n (%) Preop COPD, n (%) Preop CHF, n (%) Preop Hypertension, n (%) Preop Beta Blocker, n (%) Preop Peripheral Vascular Disease, n (%) Preop DVT, n (%) Preop Dialysis, n (%) Preop Weight Loss, n (%) Preop Bleeding Disorder, n (%) ASA Class ID, n (%)
Statins - No (n = 3,196) 84 (2.63%) 129 (4.04%) 388 (12.14%) 21 (0.66%) 315 (9.86%) 264 (8.26%) 7 (0.22%) 1341 (41.96%) 593 (18.55%) 32 (1%) 140 (4.38%) 9 (0.28%) 105 (3.29%) 69 (2.16%)
< 0.001
p-value > 0.999 0.75 0.032 > 0.999 0.828 0.899 > 0.999 0.896 0.295 0.223 0.515 0.772 > 0.999 0.378 0.862
ACCEPTED MANUSCRIPT
Table A2 (continued)
Smoker, n (%) Alcohol Consumption, n (%) Reason for Surgery, n (%)
Preop Comorbidities Count, Mean (SD)
130 (4.07%) 784 (24.53%) 1476 (46.18%) 24 (0.75%) 782 (24.47%)
4 (0.22%) 340 (18.78%) 1098 (60.66%) 3 (0.17%) 365 (20.17%)
1.166 (1.327)
2.418 (1.547)
59.993 (14.697) 68.801 (10.443) Age, Mean (SD) BMI, Mean (SD) 28.575 (7.895) 29.907 (6.402) Note. p-values from chi-square tests and independent samples t-tests.
AC C
Statins - No (n = 1,322) 594 (44.93%)
PS Matched Statins - Yes (n = 1,322) 614 (46.44%)
1119 (84.64%) 145 (10.97%) 10 (0.76%) 48 (3.63%) 229 (17.32%) 38 (2.87%)
1111 (84.04%) 150 (11.35%) 9 (0.68%) 52 (3.93%) 244 (18.46%) 36 (2.72%)
3 (0.23%) 272 (20.57%) 767 (58.02%) 1 (0.08%) 279 (21.1%)
4 (0.3%) 264 (19.97%) 780 (59%) 2 (0.15%) 272 (20.57%)
1.924 (1.389) 68.096 (12.306) 29.639 (9.608)
2.042 (1.446) 67.45 (10.381) 29.789 (6.43)
RI PT
1524 (84.2%) 205 (11.33%) 12 (0.66%) 69 (3.81%) 317 (17.51%) 57 (3.15%)
EP
Crohn's Diverticulitis Neoplasm Ulcerative Colitis Other
2696 (84.36%) 368 (11.51%) 23 (0.72%) 109 (3.41%) 707 (22.12%) 94 (2.94%)
p-value < 0.001
0.893
SC
White Af Am Other Unknown
M AN U
Category
Unadjusted Statins - Yes (n = 1,810) 899 (49.67%)
TE D
Variable Sex = Male, n (%) Race, n (%)
Statins - No (n = 3,196) 1362 (42.62%)
< 0.001 0.743 < 0.001
< 0.001 < 0.001 < 0.001
p-value 0.458 0.955
0.477 0.906 0.939
0.032 0.145 0.637
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
Figure A7. Propensity score distribution of matched and unmatched cases – Colectomies.
Figure A8. Propensity score histograms before and after matching – Colectomies.
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
Figure A9. Standardized mean covariate differences before and after matching – Colectomies.
ACCEPTED MANUSCRIPT
Table 1. Pre-surgery Patient Characteristics Colectomies and Rectal Resections Combined Unadjusted
PS-Matched Statins -Yes
Statins - No
Statins - Yes
(n = 4,770)
(n = 2,515)
p-value
(n = 1,786)
(n = 1,786)
p-value
Age, Mean (SD)
59.20 (14.40)
68.19 (10.5)
< 0.001
67.47(12.08)
67.06(10.51)
0.28
Sex = Male, n (%)
2081 (43.63%)
1230 (48.91%)
790 (44.23%)
808 (45.24%)
CATEGORY
< 0.001 0.67
White
4031 (84.51%)
2141 (85.13%)
African American Other
524 (10.99%)
256 (10.18%)
42 (0.88%)
20 (0.8%)
Unknown
173 (3.63%)
98 (3.9%)
28.55 (7.53)
29.87 (6.38)
BMI, Mean (SD) ASA Class ID, n (%)
1517 (84.94%)
1531 (85.72%)
185 (10.36%)
174 (9.74%)
`
15 (0.84%)
SC
Race, n (%)
RI PT
Statins - No VARIABLE
< 0.001
68 (3.81%)
66 (3.7%)
29.76 (8.93)
29.64 (6.25)
< 0.001
97 (2.03%)
11 (0.44%)
M AN U
1
0.57 0.93
0.62 0.90
12 (0.67%)
11 (0.62%)
2
2594 (54.38%)
770 (30.62%)
680 (38.07%)
667 (37.35%)
3
1962 (41.13%)
1566 (62.27%)
1020 (57.11%)
1026 (57.45%)
4
168 (6.68%)
74 (4.14%)
82 (4.59%)
1.14 (1.31)
2.44 (1.54)
< 0.001
1.95 (1.40)
2.07 (1.46)
0.01
Cancer, n (%)
143 (3.0%)
52 (2.07%)
0.02
39 (2.18%)
37 (2.07%)
0.91
Albumin, n (%)
196 (4.11%)
98 (3.9%)
0.71
76 (4.26%)
72 (4.03%)
0.80
Diabetes, n (%)
562 (11.78%)
859 (34.16%)
< 0.001
433 (24.24%)
471 (26.37%)
0.15
Ascites/Cirrhosis, n (%)
32 (0.67%)
7 (0.28%)
0.04
4 (0.22%)
4 (0.22%)
> 0.999
455 (9.54%)
435 (17.3%)
< 0.001
259 (14.5%)
279 (15.62%)
0.37
371 (7.78%)
283 (11.25%)
< 0.001
186 (10.41%)
186 (10.41%)
> 0.999
12 (0.25%)
17 (0.68%)
0.01
9 (0.5%)
11 (0.62%)
0.82
1967 (41.24%)
1966 (78.17%)
< 0.001
1309 (73.29%)
1298 (72.68%)
0.71
864 (18.11%)
TE D
117 (2.45%)
Comorbidities Count, Mean(SD)
Sleep Apnea, n (%) COPD, n (%) CHF, n (%) Hypertension, n (%) Peripheral Vascular Disease DVT, n (%) Renal Failure on Dialysis, n (%)
< 0.001
632 (35.39%)
665 (37.23%)
0.27
121 (4.81%)
< 0.001
44 (2.46%)
61 (3.42%)
0.11
216 (4.53%)
175 (6.96%)
< 0.001
110 (6.16%)
122 (6.83%)
0.46
11 (0.23%)
16 (0.64%)
0.01
7 (0.39%)
9 (0.5%)
0.80
158 (3.31%)
66 (2.62%)
0.12
42 (2.35%)
48 (2.69%)
0.59
Bleeding Disorder, n (%)
97 (2.03%)
109 (4.33%)
< 0.001
52 (2.91%)
60 (3.36%)
0.50
Alcohol Consumption, n (%)
149 (3.12%)
83 (3.3%)
0.74
56 (3.14%)
52 (2.91%)
0.77
Crohn's
156 (3.27%)
7 (0.28%)
0 (0%)
4 (0.22%)
Diverticulitis
1272 (26.67%)
561 (22.31%)
458 (25.64%)
433 (24.24%)
Neoplasm
2168 (45.45%)
1424 (56.62%)
964 (53.98%)
961 (53.81%)
Ulcerative Colitis Other
83 (1.74%)
17 (0.68%)
8 (0.45%)
9 (0.5%)
1091 (22.87%)
506 (20.12%)
356 (19.93%)
379 (21.22%)
AC C
Weight Loss, n (%)
1150 (45.73%)
58 (1.22%)
EP
Beta Blocker, n (%)
Reason for Surgery, n (%)
Colectomy/Rectum, n (%)
< 0.001
< 0.001
0.24
0.74
ACCEPTED MANUSCRIPT
Table 1. Pre-surgery Patient Characteristics Colectomies and Rectal Resections Combined 3196 (67%)
1810 (72%)
1297 (72.62%)
1307 (73.18%)
Rectum
1574 (33%)
705 (28%)
489 (27.38%)
479 (26.82%)
AC C
EP
TE D
M AN U
SC
RI PT
Colectomy
ACCEPTED MANUSCRIPT
Table 2
Outcomes – Colectomies and Rectal Resections Combined
p-value
49 (2.05%) 125 (4.97%) 95 (3.78%) 27 (1.07%)
0.603 > 0.999 0.317 0.172
48 (2.76%) 81 (4.54%) 95 (5.32%) 24 (1.34%)
35 (2.01%) 80 (4.48%) 67 (3.75%) 12 (0.67%)
0.181 > 0.999 0.03 0.065
51 (2.03%) 29 (1.15%) 31 (1.23%) 52 (2.07%) 37 (1.47%) 62 (2.47%) 258 (10.26%) 165 (6.56%) 669 (26.6%)
0.079 0.072 0.301 0.043 0.787 0.817 0.955 0.118 < 0.001
33 (1.85%) 24 (1.34%) 18 (1.01%) 41 (2.3%) 37 (2.07%) 52 (2.91%) 201 (11.25%) 135 (7.56%) 452 (25.31%)
31 (1.74%) 12 (0.67%) 22 (1.23%) 27 (1.51%) 25 (1.4%) 43 (2.41%) 162 (9.07%) 118 (6.61%) 401 (22.45%)
0.9 0.065 0.633 0.111 0.159 0.405 0.035 0.297 0.05
RI PT
p-value
AC C
EP
TE D
M AN U
Variable Primary Outcomes, n (%) Anastomotic Leak 101 (2.28%) Surgical Site Infection 237 (4.97%) Sepsis 205 (4.3%) 30-Day Mortality 35 (0.73%) Secondary Outcome, n (%) Acute Renal Failure 69 (1.45%) Myocardial Infarction 34 (0.71%) Deep Venous Thrombosis 45 (0.94%) Unplanned Intubation 67 (1.4%) Pneumonia 65 (1.36%) Urinary Tract Infection 112 (2.35%) 30-Day Readmission 489 (10.25%) 30-Day Reoperation 360 (7.55%) Discharge to Rehab/SNF 1061 (22.24%) Note. p-values from chi-square tests.
PS Matched Statins - No Statins - Yes (n = 1,786) (n = 1,786)
SC
Unadjusted Statins - No Statins - Yes (n = 4,770) (n = 2,515)
ACCEPTED MANUSCRIPT
Table 3 Outcomes – Colectomies Only
p-value
41 (2.31%) 75 (4.14%) 69 (3.81%) 23 (1.27%)
0.261 > 0.999 > 0.999 0.093
29 (2.23%) 63 (4.77%) 59 (4.46%) 20 (1.51%)
30 (2.31%) 53 (4.01%) 51 (3.86%) 16 (1.21%)
> 0.999 0.393 0.495 0.615
33 (1.82%) 22 (1.22%) 19 (1.05%) 44 (2.43%) 28 (1.55%) 35 (1.93%) 164 (9.06%) 115 (6.35%) 402 (22.21%)
0.265 0.258 0.347 0.025 0.649 > 0.999 0.897 0.544 < 0.001
29 (2.19%) 20 (1.51%) 14 (1.06%) 33 (2.5%) 29 (2.19%) 35 (2.65%) 136 (10.29%) 93 (7.03%) 297 (22.47%)
19 (1.44%) 15 (1.13%) 10 (0.76%) 26 (1.97%) 17 (1.29%) 23 (1.74%) 98 (7.41%) 82 (6.2%) 265 (20.05%)
0.19 0.496 0.538 0.43 0.102 0.144 0.011 0.434 0.141
RI PT
p-value
AC C
EP
TE D
M AN U
Variable Primary Outcomes, n (%) Anastomotic Leak 56 (1.8%) Surgical Site Infection 133 (4.16%) Sepsis 121 (3.79%) 30-Day Mortality 24 (0.75%) Secondary Outcomes, n (%) Acute Renal Failure 44 (1.38%) Myocardial Infarction 27 (0.84%) Deep Venous Thrombosis 24 (0.75%) Unplanned Intubation 48 (1.5%) Pneumonia 43 (1.35%) Urinary Tract Infection 62 (1.94%) 30-Day Readmission 293 (9.17%) 30-Day Reoperation 218 (6.82%) Discharge to Rehab/SNF 532 (16.65%) Note. p-values from chi-square tests.
PS Matched Statins - No Statins - Yes (n = 1,322) (n = 1,322)
SC
Unadjusted Statins - No Statins - Yes (n = 3,196) (n = 1,810)
ACCEPTED MANUSCRIPT
Table 4 Outcomes – Rectal Resections Only
Unadjusted Statins - No Statins - Yes (n = 1,574) (n = 705)
p-value
p-value
0.014 0.737 0.111 0.937
19 (4.13%) 32 (6.6%) 32 (6.6%) 7 (1.44%)
6 (1.29%) 31 (6.39%) 14 (2.89%) 1 (0.21%)
0.014 > 0.999 0.01 0.076
18 (2.55%) 7 (0.99%) 12 (1.7%) 8 (1.13%) 9 (1.28%) 27 (3.83%) 94 (13.33%) 50 (7.09%) 267 (37.87%)
0.162 0.208 0.624 > 0.999 0.972 0.501 0.7 0.126 0.054
8 (1.65%) 4 (0.82%) 6 (1.24%) 9 (1.86%) 13 (2.68%) 23 (4.74%) 71 (14.64%) 46 (9.48%) 158 (32.58%)
12 (2.47%) 4 (0.82%) 4 (0.82%) 3 (0.62%) 5 (1.03%) 14 (2.89%) 64 (13.2%) 32 (6.6%) 146 (30.1%)
0.498 > 0.999 0.751 0.146 0.096 0.18 0.578 0.125 0.446
SC
RI PT
8 (1.31%) 50 (7.09%) 26 (3.69%) 4 (0.57%)
AC C
EP
TE D
M AN U
Variable Primary Outcomes, n (%) Anastomotic Leak 45 (3.41%) Surgical Site Infection 104 (6.61%) Sepsis 84 (5.34%) 30-Day Mortality 11 (0.7%) Secondary Outcomes, n (%) Acute Renal Failure 25 (1.59%) Myocardial Infarction 7 (0.44%) Deep Venous Thrombosis 21 (1.33%) Unplanned Intubation 19 (1.21%) Pneumonia 22 (1.4%) Urinary Tract Infection 50 (3.18%) 30-Day Readmission 196 (12.45%) 30-Day Reoperation 142 (9.02%) Discharge to Rehab/SNF 529 (33.61%) Note. p-values from chi-square tests.
PS Matched Statins - No Statins - Yes (n = 485) (n = 485)
ACCEPTED MANUSCRIPT
RESEARCH HIGHLIGHTS Statins are associated with decreased risk of sepsis after colorectal surgery
•
Statins are associated with fewer anastomotic leaks in rectal resection group
•
Statins are not associated with surgical site infections or mortality
AC C
EP
TE D
M AN U
SC
RI PT
•
ACCEPTED MANUSCRIPT
Conflicts of Interest
AC C
EP
TE D
M AN U
SC
RI PT
The authors have no conflicts of interest related to this work.
S0002-9610(17)31471-X
DOI:
10.1016/j.amjsurg.2018.01.024
Reference:
AJS 12732
To appear in:
The American Journal of Surgery
Received Date: 14 October 2017 Revised Date:
30 November 2017
Accepted Date: 4 January 2018
Please cite this article as: Disbrow D, Seelbach CL, Albright J, Ferraro J, Wu J, Hain JM, Shanker B-A, Cleary RK, Statin Medications Are Associated with Decreased Risk of Sepsis and Anastomotic Leaks after Rectal Resections, The American Journal of Surgery (2018), doi: 10.1016/j.amjsurg.2018.01.024. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
Title: Statin Medications Are Associated with Decreased Risk of Sepsis and Anastomotic Leaks after Rectal Resections Authors: David Disbrow MDa , Corie L. Seelbach DOb , Jeremy Albright PhDc , Jane Ferraro
RI PT
MPP, RNd , Juan Wu ScDe , Jon M. Hain MDf , Beth-Ann Shanker MDg , Robert K Cleary MDh
a
Colon and Rectal Surgery Resident, Department of Surgery, St. Joseph Mercy Hospital Ann
SC
Arbor, Ann Arbor, MI, [email protected] b
Department of Surgery, William Beaumont Hospital Troy, Troy, MI, [email protected]
c
[email protected] d
M AN U
Senior Data Scientist, Methods Consultants, 25 East Cross St., Ypsilanti, MI.
Program Manager of Colorectal Research, St. Joseph Mercy Hospital Ann Arbor, Ann Arbor,
MI, [email protected] e
Outcomes Research Scientist, Academic Research Department, St. Joseph Mercy Hospital
f
TE D
Ann Arbor, Ann Arbor, MI, [email protected] Department of Surgery, William Beaumont Hospital Troy, Troy, MI, [email protected]
g
Department of Surgery, St Joseph Mercy Hospital Ann Arbor, Ann Arbor, MI,
EP
[email protected] h
Program Director Colon and Rectal Surgery, Director of Research, St. Joseph Mercy Hospital
AC C
Ann Arbor, Ann Arbor, MI, [email protected]
Corresponding Author: Robert K. Cleary, MD 5333 McAuley Dr.
Reichert Health Center, Suite 2115 Ann Arbor, MI 48106
ACCEPTED MANUSCRIPT
tel: 734-358-9730 fax: 734-712-8151
RI PT
Email: [email protected]
Disclaimers: None to Report
M AN U
public, commercial, or not-for-profit sectors.
SC
Funding Sources: This research did not receive any specific grant from funding agencies in the
This research was presented by Dr. Drisbow at the annual American Society of Colon and
AC C
EP
TE D
Rectal Surgeons Meeting, Seattle, WA, June 10-14, 2017.
RESEARCH HIGHLIGHTS ●
Statins are associated with decreased risk of sepsis after colorectal surgery
●
Statins are associated with fewer anastomotic leaks in rectal resection group
ACCEPTED MANUSCRIPT
Statins are not associated with surgical site infections or mortality
AC C
EP
TE D
M AN U
SC
RI PT
●
ABSTRACT
Background: This study was designed to determine the effect of statins on colorectal postoperative complications related to sepsis. Previous studies have reported conflicting results.
ACCEPTED MANUSCRIPT
Methods: This is a retrospective propensity score analysis of postoperative outcomes from a large regional database of patients who underwent elective colorectal resection from June 2012-
RI PT
July 2015.
Results: 7285 patients met inclusion criteria: 34.5% received statins. Propensity score matching revealed that patients taking statins had reduced risk of sepsis (3.75% vs 5.32%, p=0.03).
SC
Subgroup analysis revealed that this difference was driven by patients undergoing rectal
resections. Among the rectal resection group, anastomotic leaks were more common in the non-
M AN U
statins group (4.1% vs. 1.3%, p=0.01). There was no significant difference between those taking statins and those not on statin medications with respect to composite SSI or 30-day mortality.
Conclusions: Statin medications are associated with decreased risk of sepsis after colorectal
TE D
surgery and anastomotic leaks after rectal resection. Future studies should focus on medication type, dosage, and duration to confirm these results and identify patient populations that would
EP
benefit most from statin therapy.
AC C
Keywords: Statin Medication, Colorectal, Colon and Rectal Surgery, Anastomotic Leak, Sepsis
ACCEPTED MANUSCRIPT
INTRODUCTION: Infectious complications occur in 9.4-44% of patients after colorectal surgery and are a source of significant morbidity and mortality.1-6 The most severe complications are sepsis and
RI PT
anastomotic leak. Sepsis occurs in 5.4% of patients undergoing elective colorectal surgery while anastomotic leaks occur in 1-24% depending on the location of the anastomosis and other
SC
clinical and patient risk factors.7-10
Three-hydroxy-3-methylglutaryl coenzyme A (HMG-CoA) reductase inhibitors, or statins, have
M AN U
numerous lipid-independent pleiotropic effects that include decreased inflammation, improved endothelial function, antithrombotic properties, reduction of cytokines, and increased angiogenesis.11-14 Clinical studies have found statins to be beneficial for wound healing, the prevention of sepsis, and the mitigation of the effects of sepsis.14-17 The impact of statin use in the peri-operative colorectal patient has not yet been determined, with some studies showing a
TE D
decrease in septic complications and others showing no difference.7,17-20
The purpose of this study is to determine if colorectal surgery patients on statins have
EP
fewer complications related to sepsis and anastomotic leaks than those patients not on statins using a protocol-driven regional database. This study was also done to
AC C
determine if complications are different depending on surgery location – colectomy vs rectal resection. Confounding was addressed through propensity score matching on demographics and pre-surgery comorbidities.
MATERIALS & METHODS: Data Source This observational retrospective study of the Michigan Surgical Quality Collaborative (MSQC) database was composed of an analysis of all patients who underwent elective colorectal
ACCEPTED MANUSCRIPT
resection from June 2012 through July 2015. The MSQC is a consortium of 64 hospitals representing diverse practice settings throughout the state of Michigan and includes a large protocol-driven regional database devoted to quality, improving patient outcomes, and cost-
RI PT
effectiveness. Specially trained data abstractors prospectively collect patient demographics, characteristics, intraoperative processes of care, and 30-day postoperative outcomes and
complications. A standard sampling algorithm is applied to minimize selection bias. Regular
SC
data audits ensure registry data validity. Data collection for MSQC is institutional review board exempt at participating hospitals. The quality and validity of MSQC methodology have been
M AN U
demonstrated in several reports.21-22
Study Data
Patients were identified for study inclusion by Current Procedural Terminology codes for standard colon and rectal operations (44140, 44145, 44146, 44147, 44150, 44155, 44158,
TE D
44160, 44204, 44205, 44207, 44208, 44210, 44211, 44212, 45110, 45112, 45119, 45395, 45397, 45400, 45402, 45540, 45550). Only those codes that include an anastomosis were used for analysis of the anastomotic leak outcome. Exclusion criteria were age under 18 years,
EP
current pregnancy, ASA class 5 and 6, emergent cases, and the presence of preoperative open wounds with or without infection. The statin variable in the MSQC database is reported as yes
AC C
or no. There is no information available with regard to statin medication choice, dose, and duration of use.
Outcomes of interest were: 1) anastomotic leak; 2) any SSI including superficial SSI, deep SSI, and organ/space SSI; 3) sepsis; and 4) death within 30 days of surgery. The MSQC definition for anastomotic leak is a clinically diagnosed leak at the site of the intestinal anastomosis requiring one or more of the following interventions: antibiotic treatment, percutaneous drainage, reoperation with new anastomosis, reoperation with proximal diversion, or reoperation with end
ACCEPTED MANUSCRIPT
stoma. The definitions for SSIs were based on Centers for Disease Control and Prevention definitions. Finally, the definition for sepsis was based on the Surviving Sepsis definition.23 All outcome measures were abstracted in clinical (not administrative) sources to include
RI PT
intraoperative records, operative reports, nursing notes, progress notes, discharge summaries, physician office notes, or 30-day follow-up from the patient. Secondary outcomes considered were acute renal failure, myocardial infarction, deep venous thrombosis, unplanned intubation,
SC
pneumonia, urinary tract infection, 30-day readmission, 30-day reoperations, and discharge to a
M AN U
rehabilitation or skilled nursing facility.
Patients on statin therapy were initially compared with patients not on statin therapy based on demographic factors (age, sex, race, and BMI), general health factors (tobacco use, alcohol abuse, ASA class, number of comorbidities), the presence of comorbidities (coronary artery disease, hypertension, congestive heart failure, peripheral vascular disease (claudication or rest
TE D
pain or gangrene, history of revascularization or amputation for peripheral vascular disease), renal failure requiring dialysis, chronic obstructive pulmonary disease, weight loss more than 10 pounds over the past 6 months, bleeding disorder, beta blocker use, hypoalbuminemia,
EP
obstructive sleep apnea, diabetes, and cancer diagnosis). Categorical variables were compared by using χ2, and continuous variables (e.g. age) were compared by using an independent
AC C
samples t test. This was repeated for both the primary and secondary outcomes.
Statistical Analysis
Because this was not a randomized trial and significant differences between statin users and non-users were expected, it was necessary to control for potential confounders. The large number of potential pre-treatment confounders relative to the number of events, especially for the subgroup analysis, meant that logistic regressions may overfit the data. We therefore opted
ACCEPTED MANUSCRIPT
for matching statin and non-statin patients on the basis of propensity scores, which are the probability of taking statins given the set of covariates. The matching was done by finding the nearest neighbor that fell within .1 standard deviations of the propensity score. This approach
RI PT
dropped cases from both the statin and non-statin groups for whom a similar counterpart was unavailable in the other treatment group. The variables used in the matching are those listed in Table 1 and the tables in the appendix. The accuracy of the propensity score matching was
SC
assessed by comparing the distribution of the confounders between groups after the matching. Pre-surgery comorbidity counts still displayed some unbalance in the matched samples
M AN U
(p=0.01). We therefore additionally controlled for this variable in a logistic regression in the matched samples as a "doubly robust" approach. Additional graphical assessments were also conducted and are presented in the Appendix. The propensity score matching was performed separately for the entire colorectal sample, the rectal resection subsample, and the colectomy subsamples. The post-matching confounder distributions are presented in the body of the
relegated to the Appendix.
TE D
paper for the whole sample. For space considerations, similar tables for the two subgroups are
EP
After creating matched subsamples with similar demographics and pre-surgery comorbidities, the distribution of the outcomes in each treatment group are presented along with χ2 tests to
AC C
assess statistical significance (p < 0.05). All analyses were performed using R version 3.3.2. The propensity score matching was performed using the matchit package.24
RESULTS All Cases
From June 2012 to July 2015, a total of 10,273 patients undergoing elective colorectal surgery were identified in the MSQC database. After applying exclusion criteria, 7,285 patients were available for analysis with 2515 (34.5%) on statin therapy and 4770 (65.5%) not on statin
ACCEPTED MANUSCRIPT
therapy. Statistical differences in patient characteristics both before and after propensity score matching are shown in Table 1. As expected, the patients in the statin therapy group were significantly older (68.2 vs 59.2, p < 0.001) prior to the matching and had a significantly higher
RI PT
average number of comorbidities (2.44 vs 1.14, p < 0.001). There was also a significant difference between groups with respect to indication for surgery and type of surgery performed (colectomy vs. rectal resection). In the statin group, 28% had rectal resections compared to 33%
SC
of those not on statins. Following propensity score matching, the only variable that remained significantly different between the two treatment groups was the mean number of preoperative
M AN U
comorbidities (2.07 vs 1.95, p = 0.01). Although the p-value is less than .05, this difference is much smaller compared to the pre-matched sample.
Table 2 shows the comparisons of the primary and secondary outcomes by statin group along with p-values both before and after the propensity score matching. Before adjusting for
TE D
potential confounders, there are no statistically significant differences in any of the four primary outcomes. The statin group had a small but statistically significant higher percentage of unplanned intubation (2.07% versus 1.40%, p = 0.04) as well as a greater percentage of
EP
patients not discharged to home (26.6% versus 22.2%, p < 0.001.). There was no statistically
AC C
significant difference in 30-day mortality between groups (0.7% versus 1.1%, p = .172)
The propensity score matching reduced the sample to 1786 comparable cases in each treatment group. After the matching, the primary outcome of sepsis becomes significant, with the non-statin group having higher rates of sepsis (5.32 vs 3.75, p = 0.03). For the secondary outcomes, unplanned intubation loses its significance and discharge destination sits right at the .05 threshold. At the same time, 30-day readmissions become significantly more likely for the non-statin group relative to the statin group (11.25% vs 9.07%, p=0.04). The difference in 30day mortality remained non-significant in the matched sample (1.3% vs 0.7%, p = .065).
ACCEPTED MANUSCRIPT
Subsample Analysis
RI PT
Table 3 shows a similar analysis for colectomy patients. In the unadjusted analysis, there were 3196 in the no-statins group and 1810 in the statins group. Unplanned intubations were
significantly higher in the statins group (2.43% vs 1.50%, p = 0.03). Discharge destination was
SC
also significantly different, with the statin group more likely to be discharged to a rehab or skilled nursing facility (22.21 versus 16.65%, p < 0.001.) These differences disappeared following the
M AN U
matched analysis, which was based on a sample of 1322 in each group. One secondary outcome became significant following propensity score matching. Those on statins were less likely to be readmitted within 30 days (7.41% vs 10.29%, p = 0.01). No other outcomes differed significantly by treatment.
TE D
Table 4 presents the treatment group comparisons for the subset of patients undergoing rectal resections. There were 1,574 cases in the data set not on statins and 705 that were taking statins. The unadjusted comparisons show a significant difference in the rate of anastomotic
EP
leaks, with statin takers having a lower rate of occurrence (1.31% vs 3.41%, p = 0.01). No other
AC C
outcomes were significantly different.
The matching reduced the sample to 485 cases in each treatment group. The significant difference in anastomotic leaks remained and was slightly more pronounced, with outcomes favoring the statin patients (1.29% vs 4.13%, p = 0.01). In addition, sepsis now emerges as statistically significant, again favoring the statin takers (2.89% vs 6.60%, p = 0.01). The remaining comparisons are not statistically significant.
DISCUSSION
ACCEPTED MANUSCRIPT
This large regional database analysis of patients undergoing elective colon and rectal surgery revealed that patients on statins have significantly less postoperative sepsis than those not on
RI PT
statins according to an analysis of propensity score matched samples. Subgroup analysis of those having rectal resections supported the postoperative sepsis outcome advantage for the statin group. In addition, patients on statins undergoing rectal resections have significantly fewer
SC
anastomotic leaks as confirmed by both statistical analyses. These advantages for statin takers
M AN U
were not observed for colectomies.
An American College of Surgeons-National Surgical Quality Improvement Program (ACSNSQIP) study of 7777 non-cardiac surgery patients, 19.7% of whom had colorectal resections, concluded that statins were protective against major non-cardiac complications (OR 0.62, 95% CI [.49,.92], P < .001), respiratory complications (OR 0.63, 95% CI [0.50,0.79], p=.017), venous
TE D
thromboembolism (OR 0.41, 95% CI [.18,.98], p=.044), and infectious complications defined as organ space surgical site infections, sepsis, and systemic inflammatory response syndrome (OR 0.65, 95% CI [.45,.94], p=.023).25 This study differs from ours in that their data source was
EP
composed of single institution data merged with the ACS-NSQIP dataset, and their study population was not limited to colorectal surgery patients. Our study is similar to this ACS-NSQIP
AC C
study in that it supported the protective effect of statins on postoperative sepsis. In contrast, our study did not reveal a protective effect for statins against surgical site infections, perhaps because our study included all 3 surgical site infection classes and not just organ space infections. Statin use was also associated with fewer 30-day readmissions in our study, though there was no significant difference between groups with respect to venous thromboembolism.
Khan, et. al., reported a retrospective cohort analysis of 577 patients who underwent curative resections for colon cancer, 21.7% of whom were on statins.16 Though the statin cohort was
ACCEPTED MANUSCRIPT
significantly older (74.7 vs 69.2) and had more comorbid conditions, their study also demonstrated no significant difference in mortality rates between those on and those not on statins (7.2% vs 6.4%, p=0.77), and a difference in sepsis favoring the statin group, though it
RI PT
was not statistically significant (13.3% vs 20.7%, p=0.22). In contrast to our study, this ACSNSQIP study also showed significantly fewer wound infections (8.8% vs 14.6%, p=0.04) in the
SC
statin group when compared to those not on statins.16
Statins are competitive inhibitors of HMG-CoA reductase, the rate-limiting enzyme of the cholesterol biosynthetic pathway that results in anti-inflammatory effects on vascular endothelial
M AN U
cells.14 The anti-inflammatory benefits of statins are related to decreased plasma concentrations of TNF-alpha and IL-6 in postoperative colon and rectal surgery patients on statin medications.14-20 Statins also have antiplatelet and antithrombotic actions, promote angiogenesis, protect against oxidative stress, and have direct antimicrobial effects. Animal
TE D
models investigating the strength of intestinal anastomoses have shown that statin groups have higher anastomotic bursting pressures and hydroxyproline content.12 The clinical benefit of these actions has been suggested in retrospective studies with respect to improving wound
EP
healing and decreasing infectious complications.7,17-20,25,26 Our study also reveals a potential
AC C
benefit for anastomotic leak rates on rectal resections that requires further study.
A Danish study of 2755 colorectal surgery patients,19% of whom took statins, revealed that there was no significant difference in anastomotic leaks when comparing those on statins to those not on statins (OR 1.31, 95% CI [.84,2.05], p=0.23).7 This finding was the same for high and low dose statin therapy. This is in contrast to our study that showed significantly fewer anastomotic leaks for rectal resections. These authors identified rectal resection as a risk factor for anastomotic leak but did not perform subgroup analysis of colectomy and rectal resection for statin users. In addition, their patient population was limited to those with colorectal cancer and
ACCEPTED MANUSCRIPT
included only those anastomotic leaks that required operative intervention. Our study included colorectal surgery patients with benign diagnoses and anastomotic leaks that were treated nonoperatively. Singh, et al. revealed fewer anastomotic leaks in patients on statins undergoing
RI PT
colectomies.18 This finding is in contrast to our study that showed no difference in anastomotic leaks for colectomies. These authors did not include rectal resections in their analysis. The reason for the different finding in their study compared to ours is not obvious, though their patient population and study design were different in that it was a prospective analysis of
M AN U
SC
elective colectomies in a hospital database within an enhanced recovery program.18
This study has several strengths and limitations. Our analysis was limited to available variables in the database, which does not allow evaluation of all potential confounders. We utilized subsamples matched on propensity scores – the probability of being on a statin as a function of pre-surgery characteristics - to best adjust for the known differences between the two treatment
TE D
groups. We nonetheless cannot exclude the influence of the unmeasured or residual confounding to our results. Specifically, outcomes that were originally not statistically significant between statin and non-statin groups became significant after matching, which indicates that we
EP
have captured and controlled for most, if not all, negative confounders that have obscured the associations in the original unmatched study population. We further found that ASA Class and
AC C
preoperative comorbidities count that broadly reflect pre-surgery health are two principal negative confounders - both are more prevalent in the statin group. Information regarding types, dosages and duration of use for statins received are not available in the MSQC database. Therefore, future studies with more detailed assessment of statin usage are warranted to determine the optimal benefits. Privacy restrictions also meant that no zip code information was available that could be used to incorporate socioeconomic variables, and the data did not contain an indicator for proper pre-operative use of antibiotics. A small number of cases for the secondary outcomes limits our study power to detect a statistically significant difference. Lastly,
ACCEPTED MANUSCRIPT
the difference between anastomotic leak and deep organ space surgical site infection definitions can be subtle. It is possible some leaks may have been assigned to an organ space SSI, thereby underestimating the true anastomotic leak rate. However, such non-differential
RI PT
misclassification of the anastomotic leak outcome with respect to statin therapy should lead to an underestimation of the true association. The strength of this study is the source of the data a protocol-driven, risk adjusted, regularly validated large database with data collected by trained
M AN U
chart review, letters, and telephone calls).
SC
registered nurses using clinical confirmation (chart ascertainment and 30-day follow-up with
Our study and several others raise the question of whether or not statin therapy should be considered perioperatively and started preoperatively to protect against postoperative septic complications. However, the value of preoperative statin intervention has not yet been confirmed in studies that administer preoperative statins for this purpose. A meta-analysis of
TE D
randomized controlled trials of patients undergoing cardiac and noncardiac surgery who had statins prescribed one to 3 weeks prior to surgery showed insufficient evidence to support starting statins preoperatively when compared to placebo.26 Another prospective randomized
EP
controlled trial of 132 colorectal surgery patients who received statins 3-7 days preoperatively and postoperatively showed decreased levels of inflammatory markers IL-6, IL-8, and TNF-
AC C
alpha, but no significant difference in postoperative complications.20 If there is any benefit to preoperative statin therapy, how long a patient must be on statin therapy before surgery to benefit from the pleiotropic effects is not clear and it may be that months are required rather than days or weeks. For colorectal cancer patients, prolonged preoperative statin therapy may not be practical except in selected patients with rectal cancer undergoing neoadjuvant therapy. Future studies may serve to confirm or refute these possibilities with attention to statin dosage and duration, and operative types and indications.
ACCEPTED MANUSCRIPT
CONCLUSION This large regional database analysis showed that patients on statin medications are associated with a significant reduction in the rate of sepsis after colorectal surgery and in anastomotic leaks
RI PT
after rectal resection. Further studies with attention to medication dose and duration are needed to confirm or refute these results and identify patient populations that would benefit most from
References:
M AN U
SC
statin therapy.
1. Smith RL, Bohl JK, McElearney ST, et al. Wound infection after elective colorectal resection. Ann Surg. 2004;239:599–605.
TE D
2. Paun BC, Cassie S, MacLean AR, Dixon E, Buie WD. Postoperative complications following surgery for rectal cancer. Ann Surg. 2010;251:807–818.
EP
3. Aimaq R, Akopian G, Kaufman HS. Surgical site infection rates in laparoscopic versus open
AC C
colorectal surgery. Am Surg. 2011;77:1290–1294.
4. Bishawi M, Fakhoury M, Denoya PI, Stein S, Bergamaschi R. Surgical site infection rates: open versus hand-assisted colorectal resections. Tech Coloproctol. 2014;18:381–386.
5. Bot J, Piessen G, Robb WB, Roger V, Mariette C. Advanced tumor stage is an independent risk factor of postoperative infectious complications after colorectal surgery: arguments from a case-matched series. Dis Colon Rectum. 2013;56(5):568–576.
ACCEPTED MANUSCRIPT
6. Olivier F, Paquette B, Orry D, et al. Diagnostic accuracy of inflammatory markers as early predictors of infection after elective colorectal surgery: results from the IMACORS study.
RI PT
Ann Surg. 2016;263(5):961–966.
7. Bisgard AS, Noack MW, Klein M, et al. Perioperative statin therapy is not associated with
SC
reduced risk of anastomotic leakage after colorectal resection. Dis Colon Rectum.
M AN U
2013;56:980--986.
8. Paulasir S, Kaoutzanis C, Welch K, et al. Nonsteroidal anti-inflammatory drugs: do they increase the risk of anastomotic leaks following colorectal operations? Dis Colon Rectum. 2015;58(9):870--877.
TE D
9. Goligher JC, Graham NG, De Dombal FT. Anastomotic dehiscence after anterior resection of rectum and sigmoid. Br J Surg. 1970;57:109–118.
EP
10. Ogilvie JW Jr, Dietz DW, Stocchi L. Anastomotic leak after restorative proctosigmoidectomy for cancer: what are the chances of a permanent ostomy? Int J Colorectal Dis.
AC C
2012;27:1259–1266.
11. Derici J, Yaman I, Kara C, Kamer E, Diniz G, Ortac R. Simvastatin improves incisional wound healing in a rat model: an experimental study. Wounds. 2012;24(7):195--200
12. Karadeniz Cakmak G, Irkorucu O, Ucan BH, et al. Simvastatin improves wound strength after intestinal anastomosis in the rat. J Gastrointest Surg. 2009;13:1707--1716.
ACCEPTED MANUSCRIPT
13. Liao JK, Laufs U. Pleiotropic effects of statins. Annu Rev Pharmacol Toxicol. 2005;45:89– 118.
RI PT
14. Brookes ZL, McGown CC, Reilly CS. Statins for all: the new premed? Br J Anaesth. 2009;103:99–107
SC
15. Fitzmaurice GJ, McWilliams B, Nolke L, Redmond JM, McGuiness JG, O’Donnell ME. Do statins have a role in the promotion of postoperative wound healing in cardiac surgical
M AN U
patients? Ann Thorac Surg. 2014;98(2):756-764
16. Khan A, Yeung D, Wyatt B, et al. Effects of statins on postoperative sepsis, systemic inflammatory response syndrome and mortality after colorectal surgery. Crit Care
TE D
2009;13(Suppl 4):47
17. Santos J. Can statins improve outcomes in colorectal surgery? Journal of Coloproctology.
EP
2012;32(3):334-338.
18. Singh PP, Srinivasa S, Bambarawana S, et al. Perioperative use of statins in elective
AC C
colectomy. Dis Colon Rectum. 2012;55(2):205--210.
19. Battersby CLF, Green P, Vyapury V, Rooney P, Hunt J. Statins may modify colorectal anastomotic leak risk in high risk patients. Gut. 2015;64(Suppl 1):A162--163
ACCEPTED MANUSCRIPT
20. Sing PP, Lemanu DP, Soop M, MD, Bissett IP, Harrison, Hill AG. Perioperative simvastatin therapy in major colorectal surgery: a prospective, double-blind randomized controlled trial.
RI PT
J Am Coll Surg. 2016;223:308-320.
21. Khuri SF, Daley J, Henderson W, et al. The Department of Veterans Affairs’ NSQIP: the first national, validated, outcome-based, risk-adjusted, and peer-controlled program for the
M AN U
Improvement Program. Ann Surg. 1998;228:491–507.
SC
measurement and enhancement of the quality of surgical care. National VA Surgical Quality
22. Campbell DA Jr, Kubus JJ, Henke PK, Hutton M, Englesbe MJ. The Michigan Surgical Quality Collaborative: a legacy of Shukri Khuri. Am J Surg. 2009;198(5 suppl):S49–S55.
23. Dellinger RP, Levy MM, Rhodes A, et al. Surviving sepsis campaign: international guidelines
2013;39:165–228.
TE D
for management of severe sepsis and septic shock, 2012. Intensive Care Med.
EP
24. Ho DE, Imai K, King G, Stuart EA. MatchIt: Nonparametric Preprocessing for Parametric Causal Inference. Journal of Statistical Software, 2011;Vol. 42, No. 8, pp. 1-28. URL
AC C
http://www.jstatsoft.org/v42/i08/.
25. Iannuzzi JC, Rickles AS, Kelly NK, et al. Perioperative pleiotropic statin effects in general surgery. Surgery. 2014;155:398-407.
26. Li H, Lin YL, Diao SL, Ma BX, Liu XL. Does short preoperative statin therapy prevent infectious complications in adults undergoing cardiac or non-cardiacsurgery? A metaanalysis of 5 randomized placebo-controlled trials. Saudi Med J. 2016;37(5):492-497
ACCEPTED MANUSCRIPT
Appendix
EP
TE D
M AN U
SC
RI PT
Propensity score matching attempts to balance the treatment groups across covariates such that the groups differ only by a random component. The variables used in our propensity score model included the pre-surgery variables as well as demographic information. A tabular summary of post-matching covariate distributions was presented in the body for the entire sample. This appendix provides figures that visualize the pre and post matching cases in terms of the distribution of propensity scores and covariate balance. These figures are presented for the entire sample, the rectal resection subsample, and the colectomy subsample. It also provides tabular comparisons for the two subgroups.
AC C
Figure A1. Propensity score distribution of matched and unmatched cases – All patients.
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
Figure A2. Propensity score histograms before and after matching – All patients.
Figure A3. Standardized mean covariate differences before and after matching – All patients.
ACCEPTED MANUSCRIPT
Table A1. Pre and Post Matching Comparisons of Covariates – Rectal Resection Cases Only
29 (1.84%) 841 (53.43%) 670 (42.57%) 34 (2.16%)
2 (0.28%) 214 (30.35%) 445 (63.12%) 44 (6.24%)
SC
RI PT
Unadjusted Statins - Yes (n = 705) 20 (2.84%) 42 (5.96%) 260 (36.88%) 2 (0.28%) 130 (18.44%) 75 (10.64%) 6 (0.85%) 556 (78.87%) 308 (43.69%) 33 (4.68%) 46 (6.52%) 8 (1.13%) 26 (3.69%) 27 (3.83%)
AC C
1 2 3 4
TE D
M AN U
Category
EP
Variable Preop Cancer, n (%) Preop Albumin, n (%) Preop Diabetes, n (%) Preop Ascites/Cirrhosis, n (%) Preop Sleep Apnea, n (%) Preop COPD, n (%) Preop CHF, n (%) Preop Hypertension, n (%) Preop Beta Blocker, n (%) Preop Peripheral Vascular Disease, n (%) Preop DVT, n (%) Preop Dialysis, n (%) Preop Weight Loss, n (%) Preop Bleeding Disorder, n (%) ASA Class ID, n (%)
Statins - No (n = 1,574) 59 (3.75%) 67 (4.26%) 174 (11.05%) 11 (0.7%) 140 (8.89%) 107 (6.8%) 5 (0.32%) 626 (39.77%) 271 (17.22%) 26 (1.65%) 76 (4.83%) 2 (0.13%) 53 (3.37%) 28 (1.78%)
p-value 0.329 0.098 < 0.001 0.36 < 0.001 0.002 0.17 < 0.001 < 0.001 < 0.001 0.118 0.003 0.793 0.005 < 0.001
Statins - No (n = 485) 16 (3.3%) 25 (5.15%) 113 (23.3%) 2 (0.41%) 72 (14.85%) 47 (9.69%) 3 (0.62%) 361 (74.43%) 175 (36.08%) 16 (3.3%) 31 (6.39%) 1 (0.21%) 10 (2.06%) 16 (3.3%)
PS Matched Statins - Yes (n = 485) 13 (2.68%) 25 (5.15%) 143 (29.48%) 1 (0.21%) 82 (16.91%) 56 (11.55%) 2 (0.41%) 360 (74.23%) 183 (37.73%) 20 (4.12%) 31 (6.39%) 2 (0.41%) 13 (2.68%) 15 (3.09%)
3 (0.62%) 180 (37.11%) 285 (58.76%) 17 (3.51%)
2 (0.41%) 178 (36.7%) 285 (58.76%) 20 (4.12%)
p-value 0.706 > 0.999 0.035 > 0.999 0.429 0.404 > 0.999 > 0.999 0.641 0.61 > 0.999 > 0.999 0.673 > 0.999 0.929
ACCEPTED MANUSCRIPT
Table A1 (continued)
26 (1.65%) 488 (31%) 692 (43.96%) 309 (19.63%) 59 (3.75%)
3 (0.43%) 221 (31.35%) 326 (46.24%) 141 (20%) 14 (1.99%)
1.101 (1.284)
2.487 (1.531)
< 0.001
66.635 (10.492) 29.773 (6.313)
< 0.001 < 0.001
EP
Preop Comorbidities Count, Mean (SD)
617 (87.52%) 51 (7.23%) 8 (1.13%) 29 (4.11%) 111 (15.74%) 26 (3.69%)
TE D
Smoker, n (%) Alcohol Consumption, n (%) Reason for Surgery, n (%) Crohn's Diverticulitis Neoplasm Other Ulcerative Colitis
1335 (84.82%) 156 (9.91%) 19 (1.21%) 64 (4.07%) 389 (24.71%) 55 (3.49%)
M AN U
White Af Am Other Unknown
AC C
Age, Mean (SD) 57.601 (13.655) BMI, Mean (SD) 28.498 (6.734) Note. p-values from chi-square tests and independent samples t-tests.
Statins - No (n = 485) 220 (45.36%)
PS Matched Statins - Yes (n = 485) 222 (45.77%)
441 (90.93%) 24 (4.95%) 4 (0.82%) 16 (3.3%) 84 (17.32%) 19 (3.92%)
422 (87.01%) 34 (7.01%) 6 (1.24%) 23 (4.74%) 89 (18.35%) 17 (3.51%)
0 (0.0%) 184 (37.94%) 204 (42.06%) 90 (18.56%) 7 (1.44%)
0 (0.0%) 169 (34.85%) 208 (42.89%) 102 (21.03%) 6 (1.24%)
1.957 (1.389) 65.833 (11.225) 29.438 (6.502)
2.216 (1.474)
0.005
65.115 (9.913) 29.707 (6.354)
0.292 0.515
RI PT
Category
Unadjusted Statins - Yes (n = 705) 331 (46.95%)
p-value 0.605 0.233
SC
Variable Sex = Male, n (%) Race, n (%)
Statins - No (n = 1,574) 719 (45.68%)
< 0.001 0.914 0.026
p-value 0.949 0.284
0.737 0.865 0.682
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
Figure A4. Propensity score distribution of matched and unmatched cases – Rectal Resections.
Figure A5. Propensity score histograms before and after matching – Rectal Resections.
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
Figure A6. Standardized mean covariate differences before and after matching – Rectal Resections.
ACCEPTED MANUSCRIPT
Table A2. Pre and Post Matching Comparisons of Covariates – Rectal Resection Cases Only
68 (2.13%) 1753 (54.85%) 1292 (40.43%) 83 (2.6%)
9 (0.5%) 556 (30.72%) 1121 (61.93%) 124 (6.85%)
TE D EP
1 2 3 4
Statins - No (n = 1,322) 21 (1.59%) 48 (3.63%) 309 (23.37%) 2 (0.15%) 198 (14.98%) 139 (10.51%) 5 (0.38%) 959 (72.54%) 478 (36.16%) 29 (2.19%) 76 (5.75%) 5 (0.38%) 33 (2.5%) 38 (2.87%)
PS Matched Statins - Yes (n = 1,322) 22 (1.66%) 44 (3.33%) 358 (27.08%) 3 (0.23%) 203 (15.36%) 136 (10.29%) 5 (0.38%) 955 (72.24%) 505 (38.2%) 40 (3.03%) 85 (6.43%) 7 (0.53%) 33 (2.5%) 47 (3.56%)
9 (0.68%) 523 (39.56%) 729 (55.14%) 61 (4.61%)
9 (0.68%) 504 (38.12%) 742 (56.13%) 67 (5.07%)
RI PT
Unadjusted Statins - Yes (n = 1,810) 32 (1.77%) 56 (3.09%) 599 (33.09%) 5 (0.28%) 305 (16.85%) 208 (11.49%) 11 (0.61%) 1410 (77.9%) 842 (46.52%) 88 (4.86%) 129 (7.13%) 8 (0.44%) 40 (2.21%) 82 (4.53%)
p-value 0.065 0.105 < 0.001 0.11 < 0.001 < 0.001 0.05 < 0.001 < 0.001 < 0.001 < 0.001 0.494 0.036 < 0.001
SC
M AN U
Category
AC C
Variable Preop Cancer, n (%) Preop Albumin, n (%) Preop Diabetes, n (%) Preop Ascites/Cirrhosis, n (%) Preop Sleep Apnea, n (%) Preop COPD, n (%) Preop CHF, n (%) Preop Hypertension, n (%) Preop Beta Blocker, n (%) Preop Peripheral Vascular Disease, n (%) Preop DVT, n (%) Preop Dialysis, n (%) Preop Weight Loss, n (%) Preop Bleeding Disorder, n (%) ASA Class ID, n (%)
Statins - No (n = 3,196) 84 (2.63%) 129 (4.04%) 388 (12.14%) 21 (0.66%) 315 (9.86%) 264 (8.26%) 7 (0.22%) 1341 (41.96%) 593 (18.55%) 32 (1%) 140 (4.38%) 9 (0.28%) 105 (3.29%) 69 (2.16%)
< 0.001
p-value > 0.999 0.75 0.032 > 0.999 0.828 0.899 > 0.999 0.896 0.295 0.223 0.515 0.772 > 0.999 0.378 0.862
ACCEPTED MANUSCRIPT
Table A2 (continued)
Smoker, n (%) Alcohol Consumption, n (%) Reason for Surgery, n (%)
Preop Comorbidities Count, Mean (SD)
130 (4.07%) 784 (24.53%) 1476 (46.18%) 24 (0.75%) 782 (24.47%)
4 (0.22%) 340 (18.78%) 1098 (60.66%) 3 (0.17%) 365 (20.17%)
1.166 (1.327)
2.418 (1.547)
59.993 (14.697) 68.801 (10.443) Age, Mean (SD) BMI, Mean (SD) 28.575 (7.895) 29.907 (6.402) Note. p-values from chi-square tests and independent samples t-tests.
AC C
Statins - No (n = 1,322) 594 (44.93%)
PS Matched Statins - Yes (n = 1,322) 614 (46.44%)
1119 (84.64%) 145 (10.97%) 10 (0.76%) 48 (3.63%) 229 (17.32%) 38 (2.87%)
1111 (84.04%) 150 (11.35%) 9 (0.68%) 52 (3.93%) 244 (18.46%) 36 (2.72%)
3 (0.23%) 272 (20.57%) 767 (58.02%) 1 (0.08%) 279 (21.1%)
4 (0.3%) 264 (19.97%) 780 (59%) 2 (0.15%) 272 (20.57%)
1.924 (1.389) 68.096 (12.306) 29.639 (9.608)
2.042 (1.446) 67.45 (10.381) 29.789 (6.43)
RI PT
1524 (84.2%) 205 (11.33%) 12 (0.66%) 69 (3.81%) 317 (17.51%) 57 (3.15%)
EP
Crohn's Diverticulitis Neoplasm Ulcerative Colitis Other
2696 (84.36%) 368 (11.51%) 23 (0.72%) 109 (3.41%) 707 (22.12%) 94 (2.94%)
p-value < 0.001
0.893
SC
White Af Am Other Unknown
M AN U
Category
Unadjusted Statins - Yes (n = 1,810) 899 (49.67%)
TE D
Variable Sex = Male, n (%) Race, n (%)
Statins - No (n = 3,196) 1362 (42.62%)
< 0.001 0.743 < 0.001
< 0.001 < 0.001 < 0.001
p-value 0.458 0.955
0.477 0.906 0.939
0.032 0.145 0.637
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
Figure A7. Propensity score distribution of matched and unmatched cases – Colectomies.
Figure A8. Propensity score histograms before and after matching – Colectomies.
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
Figure A9. Standardized mean covariate differences before and after matching – Colectomies.
ACCEPTED MANUSCRIPT
Table 1. Pre-surgery Patient Characteristics Colectomies and Rectal Resections Combined Unadjusted
PS-Matched Statins -Yes
Statins - No
Statins - Yes
(n = 4,770)
(n = 2,515)
p-value
(n = 1,786)
(n = 1,786)
p-value
Age, Mean (SD)
59.20 (14.40)
68.19 (10.5)
< 0.001
67.47(12.08)
67.06(10.51)
0.28
Sex = Male, n (%)
2081 (43.63%)
1230 (48.91%)
790 (44.23%)
808 (45.24%)
CATEGORY
< 0.001 0.67
White
4031 (84.51%)
2141 (85.13%)
African American Other
524 (10.99%)
256 (10.18%)
42 (0.88%)
20 (0.8%)
Unknown
173 (3.63%)
98 (3.9%)
28.55 (7.53)
29.87 (6.38)
BMI, Mean (SD) ASA Class ID, n (%)
1517 (84.94%)
1531 (85.72%)
185 (10.36%)
174 (9.74%)
`
15 (0.84%)
SC
Race, n (%)
RI PT
Statins - No VARIABLE
< 0.001
68 (3.81%)
66 (3.7%)
29.76 (8.93)
29.64 (6.25)
< 0.001
97 (2.03%)
11 (0.44%)
M AN U
1
0.57 0.93
0.62 0.90
12 (0.67%)
11 (0.62%)
2
2594 (54.38%)
770 (30.62%)
680 (38.07%)
667 (37.35%)
3
1962 (41.13%)
1566 (62.27%)
1020 (57.11%)
1026 (57.45%)
4
168 (6.68%)
74 (4.14%)
82 (4.59%)
1.14 (1.31)
2.44 (1.54)
< 0.001
1.95 (1.40)
2.07 (1.46)
0.01
Cancer, n (%)
143 (3.0%)
52 (2.07%)
0.02
39 (2.18%)
37 (2.07%)
0.91
Albumin, n (%)
196 (4.11%)
98 (3.9%)
0.71
76 (4.26%)
72 (4.03%)
0.80
Diabetes, n (%)
562 (11.78%)
859 (34.16%)
< 0.001
433 (24.24%)
471 (26.37%)
0.15
Ascites/Cirrhosis, n (%)
32 (0.67%)
7 (0.28%)
0.04
4 (0.22%)
4 (0.22%)
> 0.999
455 (9.54%)
435 (17.3%)
< 0.001
259 (14.5%)
279 (15.62%)
0.37
371 (7.78%)
283 (11.25%)
< 0.001
186 (10.41%)
186 (10.41%)
> 0.999
12 (0.25%)
17 (0.68%)
0.01
9 (0.5%)
11 (0.62%)
0.82
1967 (41.24%)
1966 (78.17%)
< 0.001
1309 (73.29%)
1298 (72.68%)
0.71
864 (18.11%)
TE D
117 (2.45%)
Comorbidities Count, Mean(SD)
Sleep Apnea, n (%) COPD, n (%) CHF, n (%) Hypertension, n (%) Peripheral Vascular Disease DVT, n (%) Renal Failure on Dialysis, n (%)
< 0.001
632 (35.39%)
665 (37.23%)
0.27
121 (4.81%)
< 0.001
44 (2.46%)
61 (3.42%)
0.11
216 (4.53%)
175 (6.96%)
< 0.001
110 (6.16%)
122 (6.83%)
0.46
11 (0.23%)
16 (0.64%)
0.01
7 (0.39%)
9 (0.5%)
0.80
158 (3.31%)
66 (2.62%)
0.12
42 (2.35%)
48 (2.69%)
0.59
Bleeding Disorder, n (%)
97 (2.03%)
109 (4.33%)
< 0.001
52 (2.91%)
60 (3.36%)
0.50
Alcohol Consumption, n (%)
149 (3.12%)
83 (3.3%)
0.74
56 (3.14%)
52 (2.91%)
0.77
Crohn's
156 (3.27%)
7 (0.28%)
0 (0%)
4 (0.22%)
Diverticulitis
1272 (26.67%)
561 (22.31%)
458 (25.64%)
433 (24.24%)
Neoplasm
2168 (45.45%)
1424 (56.62%)
964 (53.98%)
961 (53.81%)
Ulcerative Colitis Other
83 (1.74%)
17 (0.68%)
8 (0.45%)
9 (0.5%)
1091 (22.87%)
506 (20.12%)
356 (19.93%)
379 (21.22%)
AC C
Weight Loss, n (%)
1150 (45.73%)
58 (1.22%)
EP
Beta Blocker, n (%)
Reason for Surgery, n (%)
Colectomy/Rectum, n (%)
< 0.001
< 0.001
0.24
0.74
ACCEPTED MANUSCRIPT
Table 1. Pre-surgery Patient Characteristics Colectomies and Rectal Resections Combined 3196 (67%)
1810 (72%)
1297 (72.62%)
1307 (73.18%)
Rectum
1574 (33%)
705 (28%)
489 (27.38%)
479 (26.82%)
AC C
EP
TE D
M AN U
SC
RI PT
Colectomy
ACCEPTED MANUSCRIPT
Table 2
Outcomes – Colectomies and Rectal Resections Combined
p-value
49 (2.05%) 125 (4.97%) 95 (3.78%) 27 (1.07%)
0.603 > 0.999 0.317 0.172
48 (2.76%) 81 (4.54%) 95 (5.32%) 24 (1.34%)
35 (2.01%) 80 (4.48%) 67 (3.75%) 12 (0.67%)
0.181 > 0.999 0.03 0.065
51 (2.03%) 29 (1.15%) 31 (1.23%) 52 (2.07%) 37 (1.47%) 62 (2.47%) 258 (10.26%) 165 (6.56%) 669 (26.6%)
0.079 0.072 0.301 0.043 0.787 0.817 0.955 0.118 < 0.001
33 (1.85%) 24 (1.34%) 18 (1.01%) 41 (2.3%) 37 (2.07%) 52 (2.91%) 201 (11.25%) 135 (7.56%) 452 (25.31%)
31 (1.74%) 12 (0.67%) 22 (1.23%) 27 (1.51%) 25 (1.4%) 43 (2.41%) 162 (9.07%) 118 (6.61%) 401 (22.45%)
0.9 0.065 0.633 0.111 0.159 0.405 0.035 0.297 0.05
RI PT
p-value
AC C
EP
TE D
M AN U
Variable Primary Outcomes, n (%) Anastomotic Leak 101 (2.28%) Surgical Site Infection 237 (4.97%) Sepsis 205 (4.3%) 30-Day Mortality 35 (0.73%) Secondary Outcome, n (%) Acute Renal Failure 69 (1.45%) Myocardial Infarction 34 (0.71%) Deep Venous Thrombosis 45 (0.94%) Unplanned Intubation 67 (1.4%) Pneumonia 65 (1.36%) Urinary Tract Infection 112 (2.35%) 30-Day Readmission 489 (10.25%) 30-Day Reoperation 360 (7.55%) Discharge to Rehab/SNF 1061 (22.24%) Note. p-values from chi-square tests.
PS Matched Statins - No Statins - Yes (n = 1,786) (n = 1,786)
SC
Unadjusted Statins - No Statins - Yes (n = 4,770) (n = 2,515)
ACCEPTED MANUSCRIPT
Table 3 Outcomes – Colectomies Only
p-value
41 (2.31%) 75 (4.14%) 69 (3.81%) 23 (1.27%)
0.261 > 0.999 > 0.999 0.093
29 (2.23%) 63 (4.77%) 59 (4.46%) 20 (1.51%)
30 (2.31%) 53 (4.01%) 51 (3.86%) 16 (1.21%)
> 0.999 0.393 0.495 0.615
33 (1.82%) 22 (1.22%) 19 (1.05%) 44 (2.43%) 28 (1.55%) 35 (1.93%) 164 (9.06%) 115 (6.35%) 402 (22.21%)
0.265 0.258 0.347 0.025 0.649 > 0.999 0.897 0.544 < 0.001
29 (2.19%) 20 (1.51%) 14 (1.06%) 33 (2.5%) 29 (2.19%) 35 (2.65%) 136 (10.29%) 93 (7.03%) 297 (22.47%)
19 (1.44%) 15 (1.13%) 10 (0.76%) 26 (1.97%) 17 (1.29%) 23 (1.74%) 98 (7.41%) 82 (6.2%) 265 (20.05%)
0.19 0.496 0.538 0.43 0.102 0.144 0.011 0.434 0.141
RI PT
p-value
AC C
EP
TE D
M AN U
Variable Primary Outcomes, n (%) Anastomotic Leak 56 (1.8%) Surgical Site Infection 133 (4.16%) Sepsis 121 (3.79%) 30-Day Mortality 24 (0.75%) Secondary Outcomes, n (%) Acute Renal Failure 44 (1.38%) Myocardial Infarction 27 (0.84%) Deep Venous Thrombosis 24 (0.75%) Unplanned Intubation 48 (1.5%) Pneumonia 43 (1.35%) Urinary Tract Infection 62 (1.94%) 30-Day Readmission 293 (9.17%) 30-Day Reoperation 218 (6.82%) Discharge to Rehab/SNF 532 (16.65%) Note. p-values from chi-square tests.
PS Matched Statins - No Statins - Yes (n = 1,322) (n = 1,322)
SC
Unadjusted Statins - No Statins - Yes (n = 3,196) (n = 1,810)
ACCEPTED MANUSCRIPT
Table 4 Outcomes – Rectal Resections Only
Unadjusted Statins - No Statins - Yes (n = 1,574) (n = 705)
p-value
p-value
0.014 0.737 0.111 0.937
19 (4.13%) 32 (6.6%) 32 (6.6%) 7 (1.44%)
6 (1.29%) 31 (6.39%) 14 (2.89%) 1 (0.21%)
0.014 > 0.999 0.01 0.076
18 (2.55%) 7 (0.99%) 12 (1.7%) 8 (1.13%) 9 (1.28%) 27 (3.83%) 94 (13.33%) 50 (7.09%) 267 (37.87%)
0.162 0.208 0.624 > 0.999 0.972 0.501 0.7 0.126 0.054
8 (1.65%) 4 (0.82%) 6 (1.24%) 9 (1.86%) 13 (2.68%) 23 (4.74%) 71 (14.64%) 46 (9.48%) 158 (32.58%)
12 (2.47%) 4 (0.82%) 4 (0.82%) 3 (0.62%) 5 (1.03%) 14 (2.89%) 64 (13.2%) 32 (6.6%) 146 (30.1%)
0.498 > 0.999 0.751 0.146 0.096 0.18 0.578 0.125 0.446
SC
RI PT
8 (1.31%) 50 (7.09%) 26 (3.69%) 4 (0.57%)
AC C
EP
TE D
M AN U
Variable Primary Outcomes, n (%) Anastomotic Leak 45 (3.41%) Surgical Site Infection 104 (6.61%) Sepsis 84 (5.34%) 30-Day Mortality 11 (0.7%) Secondary Outcomes, n (%) Acute Renal Failure 25 (1.59%) Myocardial Infarction 7 (0.44%) Deep Venous Thrombosis 21 (1.33%) Unplanned Intubation 19 (1.21%) Pneumonia 22 (1.4%) Urinary Tract Infection 50 (3.18%) 30-Day Readmission 196 (12.45%) 30-Day Reoperation 142 (9.02%) Discharge to Rehab/SNF 529 (33.61%) Note. p-values from chi-square tests.
PS Matched Statins - No Statins - Yes (n = 485) (n = 485)
ACCEPTED MANUSCRIPT
RESEARCH HIGHLIGHTS Statins are associated with decreased risk of sepsis after colorectal surgery
•
Statins are associated with fewer anastomotic leaks in rectal resection group
•
Statins are not associated with surgical site infections or mortality
AC C
EP
TE D
M AN U
SC
RI PT
•
ACCEPTED MANUSCRIPT
Conflicts of Interest
AC C
EP
TE D
M AN U
SC
RI PT
The authors have no conflicts of interest related to this work.