Postoperative use of non-steroidal anti-inflammatory drugs in patients with anastomotic leakage requiring reoperation after colorectal resection: cohort study based on prospective data
Cite this as: BMJ 2012;345:e6166
Rapid responses are electronic letters to the editor. They enable our users to debate issues raised in articles published on bmj.com. Although a selection of rapid responses will be included as edited readers' letters in the weekly print issue of the BMJ, their first appearance online means that they are published articles. If you need the url (web address) of an individual response, perhaps for citation purposes, simply click on the response headline and copy the url from the browser window.
Displaying 1-3 out of 3 published
We read with interest the recent paper “Postoperative use of non-steroidal anti-inflammatory drugs in patients with anastomotic leakage requiring reoperation after colorectal resection: cohort study based on prospective data”(1). In this cohort study of a Danish population undergoing elective colorectal cancer resection with primary anastomosis, Klein et al. examined the association between post-operative leak and regular post-operative non-steroidal anti-inflammatory drug (NSAID) use (predominantly ibuprofen or diclofenac) as well as a number of accepted risk factors for anastomotic leakage. In univariate analysis, both ibuprofen and diclofenac were associated with an increased risk of anastomotic leak, however in the final multivariate model only regular post-operative diclofenac use was associated with increased risk (odds ratio 7.16, 95% confidence interval 3.82 to 13.4), alongside intra-operative transfusion, rectal resection, male sex and treatment centre.
This paper highlights an interesting area of concern, particularly in the current era of enhanced recovery following colorectal surgery where routine NSAID use is becoming a standard of post-operative care(2). Although the authors used an accurate, validated and prospectively collected database with strict criteria for post-operative NSAID use, we feel however that the subsequent results should be interpreted with caution. Firstly, the indication and timing of post-operative NSAID use was not evident and did not stratify for patients who received NSAIDs as part of a standard post-operative analgesic pathway. Indeed, those patients receiving NSAIDs as an adjunct to standard analgesia later in their post-operative course may have been developing clinical signs and symptoms of anastomotic leak prior to their initiation. Secondly, the definition of an anastomotic leak as one which required repeat surgical intervention may not reflect daily clinical practice, where a significant number of patients with anastomotic leak may require medical or radiological intervention only, and up to 40% of patients with confirmed anastomotic leak may remain asymptomatic(3, 4). Furthermore, the authors did not report on several other factors associated with both post-operative morbidity and anastomotic leak following colorectal resection and primary anastomosis. The use of primary defunctioning stomas, epidural rates, uptake of enhanced recovery pathways and medical comorbidities may have differed significantly between groups and subsequently confounded the results. Certainly, the results of several meta-analyses of enhanced recovery after colorectal surgery suggest that avoidance of opioid analgesia and use of other analgesic modalities may in fact reduce risk of post-operative complications(5).
Recent attention has turned to the potential use of NSAIDs in the treatment of colorectal cancer. The body of evidence predominantly arising from epidemiological studies and survival data from cardiovascular disease prevention trials have shown aspirin and non-aspirin NSAIDs increase survival and reduce risk of metastatic disease following potentially curative surgery(6-8), even when NSAIDS are commenced following diagnosis(9). Similarly, pre-operative NSAID use may also result in increased tumour infiltration of activated lymphocytes and changes in tumour gene expression associated with decreased proliferation and increased susceptibility to oxidative stress and immune surveillance(10, 11). Furthermore, NSAID use in the post-operative phase has previously been shown to restore cell-mediated immune competence following major surgery(12).
Given the potential benefits of NSAIDs in the treatment of colorectal cancer, further trials are required. Certainly, the argument for NSAID use in the adjuvant treatment of colorectal cancer is so compelling that clinical trials of aspirin following surgery for Dukes B/C colorectal cancer are currently recruiting(13). Since both traditional NSAIDs and specific cyclooxygenase-2 inhibitors are associated with significant gastrointestinal and cardiovascular side effects, it is clear that a need remains to identify the patient population most likely to benefit. Given the weight of evidence supporting the use of NSAIDs not only as an effective component of enhanced recovery pathways but also in improving oncological outcomes, further trials of NSAIDs as adjuvant treatment are required and clinicians should not be discouraged from entering patients in to such trials.
1. Klein M, Gogenur I, Rosenberg J. Postoperative use of non-steroidal anti-inflammatory drugs in patients with anastomotic leakage requiring reoperation after colorectal resection: cohort study based on prospective data. BMJ. 2012;345:e6166.
2. Kehlet H, Wilmore DW. Evidence-based surgical care and the evolution of fast-track surgery. Annals of Surgery. 2008;248(2):189-98.
3. Ogilvie JW, Dietz DW, Stocchi L. Anastomotic leak after restorative proctosigmoidectomy for cancer: what are the chances of a permanent ostomy? International Journal of Colorectal Disease. 2012;27(10):1259-66.
4. Maggiori L, Bretagnol F, Lefevre JH, Ferron M, Vicaut E, Panis Y. Conservative management is associated with a decreased risk of definitive stoma after anastomotic leakage complicating sphincter-saving resection for rectal cancer. Colorectal Disease. 2011;13(6):632-7.
5. Varadhan KK, Neal KR, Dejong CHC, Fearon KCH, Ljungqvist O, Lobo DN. The enhanced recovery after surgery (ERAS) pathway for patients undergoing major elective open colorectal surgery: A meta-analysis of randomized controlled trials. Clinical Nutrition. 2010;29(4):434-40.
6. Coghill AE, Newcomb PA, Campbell PT, Burnett-Hartman AN, Adams SV, Poole EM, et al. Prediagnostic non-steroidal anti-inflammatory drug use and survival after diagnosis of colorectal cancer. Gut. 2011;60(4):491-8.
7. Rothwell PM, Wilson M, Elwin CE, Norrving B, Algra A, Warlow CP, et al. Long-term effect of aspirin on colorectal cancer incidence and mortality: 20-year follow-up of five randomised trials. Lancet. 2010;376(9754):1741-50.
8. Rothwell PM, Wilson M, Price JF, Belch JF, Meade TW, Mehta Z. Effect of daily aspirin on risk of cancer metastasis: a study of incident cancers during randomised controlled trials. Lancet. 2012;379(9826):1591-601.
9. Chan AT, Ogino S, Fuchs CS. Aspirin use and survival after diagnosis of colorectal cancer. JAMA. 2009;302(6):649-58.
10. Lönnroth C, Andersson M, Arvidsson A, Nordgren S, Brevinge H, Lagerstedt K, et al. Preoperative treatment with a non-steroidal anti-inflammatory drug (NSAID) increases tumor tissue infiltration of seemingly activated immune cells in colorectal cancer. Cancer immunity: a journal of the Academy of Cancer Immunology. 2008;8:5
11. Auman JT, Church R, Lee SY, Watson MA, Fleshman JW, McLeod HL. Celecoxib pre-treatment in human colorectal adenocarcinoma patients is associated with gene expression alterations suggestive of diminished cellular proliferation. Eur J Cancer. 2008;44(12):1754-60.
12. Gogos CA, Maroulis J, Zoumbos NC, Salsa B, Kalfarentzos F. Effect of Parenteral Indomethacin on T-Lymphocyte Subpopulations and Cytokine Production in Patients under Major Surgical Operations. Res Exp Med. 1995;195(2):85-92.
13. Ali R, Toh HC, Chia WK. The utility of Aspirin in Dukes C and High Risk Dukes B Colorectal cancer--the ASCOLT study: study protocol for a randomized controlled trial. Trials. 2011;12:261.
Competing interests: None declared
University Department of Surgery, Glasgow Royal Infirmary, Castle Street, Glasgow G4 0SF
Click to like:
The question whether postoperative NSAID use is associated with anastomotic leakage requiring reoperation after colorectal resection addressed by Klein et al is highly interesting and the authors have compiled high quality data to address this issue(1). We do, however, have some concerns regarding the analysis of these data and the conclusions drawn.
An important issue to carefully consider is the potential for residual confounding by indication and the problem of reverse causation. This issue is to some extent recognised by the authors but the problem cannot be substantially reduced by the fact that there are treatment guidelines for NSAID use. Only 32% of the study population receive NSAID treatment as defined by the authors. Even if this exposure is prescribed exclusively according to predefined guidelines (which is unlikely), the fact that only one third of the patients receive treatment means that some sort of selection mechanisms must be in play. These selection mechanisms and individual adaptation may very likely correlate with factors that can influence the risk for anastomotic leakage. The problem of reverse causation from a medication exposure used to treat symptoms related to the outcome is well known and remains a threat to the conclusions of the study(2). A possible way forward would be to study a graded exposure, such as 2-4 days versus 4-6 days of NSAID use, on late (after ≥ 7 days post surgery) anastomotic leakage. A dose-response pattern could strengthen conclusions regarding a potential causal relation.
The finding that the risk appears to be substantially higher with diclofenac than ibuprofen is intriguing. A potential problem here is if there is a reduction in the incidence of anastomotic leakage over time (e.g. caused by an overall improvement in surgical technique), and if in parallel it becomes more common over time to use ibuprofen instead of diclofenac (e.g. resulting from safety concerns relating to diclofenac). Such a period effect could induce a spurious non-causal association between diclofenac and the risk for anastomotic leakage. A plot of incidence of anastomotic leakage and the proportion of diclofenac among NSAID use over time would be informative and the problem could perhaps partly be addressed by including year as a variable in the model.
A crucial methodological issue is the inclusion of product terms in the regression model. It is commendable to look for potential effect-measure modification (statistical interaction/product terms) in the regression analysis but there are a number of issues with the selected approach. The authors use a very conservative detection level (P < 0.05) for identification of potential effect-measure modifiers and from the data presented (table 4) at least sex could also be an important effect-measure modifier. In the selection of potential effect-measure modifiers these P-values should only be seen as indicative and their interpretation must be strongly influenced by pre-specified subject knowledge based on plausible biological mechanisms. An alternative and more conventional approach would be to identify potential candidate variables mainly based on subject knowledge, apply a less conservative p-value level (e.g. P < 0.1) for initial screening, and finally look at the influence of added product terms to the model on the model fit by using a likelihood ratio test. If a factor is strongly suspected to be an important effect-measure modifier based on clinical reasoning, a rational approach would be to include it in the model regardless of the P-value for the product term parameter estimate.
The main problem is, however, that models with one or more product term(s) including the exposure (in this case NSAID use), while highly relevant, will mean that the parameter estimate for the exposure is no longer interpretable as an odds ratio independent from other risk factors(3). The odds ratio will instead in this case vary across levels of the variable for intraoperative transfusion. This means that the odds ratios presented from multi-variable models in this study are hard to interpret in a meaningful way.
We would also caution against the stated conclusion that “use of the COX-1-selective NSAID ibuprofen did not increase risk” based on the odds ratio estimate 1.5 (0.8 to 2.9). Apart from the caveats discussed above this estimate indicates a 50% increase in risk and a “non-significant result” does not prove an absence of increase in risk.
The question addressed by the authors is highly interesting but a modified and expanded analytical approach would be helpful to increase the understanding of a potential causal relation between postoperative NSAID use the risk for anastomotic leakage after colorectal resection.
Medical Products Agency, Uppsala, Sweden
Disclaimer: The opinions expressed by the authors may not represent the official positions of the Medical Products Agency.
1. Klein M, Gogenur I, Rosenberg J. Postoperative use of non-steroidal anti-inflammatory drugs in patients with anastomotic leakage requiring reoperation after colorectal resection: cohort study based on prospective data. BMJ 2012;345:e6166.
2. Poulsen AH, Christensen S, McLaughlin JK, Thomsen RW, Sorensen HT, Olsen JH, et al. Proton pump inhibitors and risk of gastric cancer: a population-based cohort study. Br J Cancer 2009;100(9):1503-7.
3. Rothman KJ, Greenland S, Lash TL, editors. Modern epidemiology. Philadelphia, 2008.
Competing interests: None declared
Medical Products Agency, P.O. Box 26, SE-751 03 Uppsala, Sweden
Click to like:
We read with interest the recent article by Klein et al highlighting the negative impact of diclofenac on anastomotic leakage after colorectal resection with primary anastomosis across several Danish centres . The authors are to be congratulated on the rigorous approach to data collection, analysis, and insightful reflection on the potential limitations of the study design while providing interesting data relevant to a large proportion of colorectal surgical practice.
In particular we focus on the issue of cofounding association as particularly relevant and suggest that it has an important role in considering the study results implicating diclofenac. While the decision to treat with NSAID is made preoperatively, the vast majority (>75%) receive ibuprofen instead of diclofenac. Is this arbitrary or indeed a centre specific decision? This has relevance if the proportion of the patients receiving diclofenac (226/2756 [8%]) are from one or only a few centres and perhaps the link with diclofenac and leak more likely to reflect local surgical practices (laparoscopic learning curve, proportion of rectal cancers, bowel preparation protocols, use of covering defunctioning stomas etc) and indeed patients factors (BMI, nutritional factors etc.) peculiar to that centre(s).
The authors report an increased use of blood transfusion (a recognised risk factor for anastomotic leak) in the cohort given ibuprofen compared to diclofenac and controls and conclude that this “strengthens the results of increased risk of anastomotic leakage among patients treated with diclofenac”. Does the data really support this statement as if we look at the operative variables we find that despite a statistically significant greater blood loss in the diclofenac group the use of intraoperative blood transfusion is more in the ibuprofen and control groups ? Again are the criteria for blood transfusion uniform across the centres or is there variation potentially creating relevant cofounding factor(s)? Indeed the authors go onto describe the negative impact of blood transfusion on immune function and as a surrogate marker for extent of surgery and “...perhaps insufficient anastomotic perfusion caused by anaemia or hypotension”. We propose this to be more relevant in the diclofenac group as they have greater blood loss and potential for perfusion insufficiency .
The authors have incorporated both colon and rectal cancer surgery as one group when considering anastomotic leakage and could present a challenge to the results as the reported factors for leakage vary in importance from right sided colonic resections to low rectal tumours and even hand vs. stapled anastomosis. It would be of interest to note the results when excluded for the rectal cancers thus providing more ‘homogenised’ data relevant to colonic and/or rectal cancer surgery. Furthermore we are not provided with information regarding the use of neoadjuvant chemo/radiotherapies, corticosteroids, the protocols for the use of defunctioning ‘protective’ stomas, or the use of low molecular weight heparin postoperatively (which may help in management of “microthomboses or micro emboli” and the “possible important mechanism behind the adverse effect of [cyclo-oxygenase -2 (COX-2) selective NSAIDS)”. We acknowledge the recent laboratory work suggesting a putative link between cyclo-oxygenase – 2 selective NSAIDs and await further in vivo studies that may indeed add weight to the authors’ conclusions. However data from a recent clinical study in 795 patients indicated that non-selective NSAIDs and not COX-2 inhibitors are associated with anastomotic leakage .
The univariate analysis of the risk of anastomotic leak by centre does highlight statistical significant variation. Taking this and all of the above into consideration in a study population using diclofenac of only 8%, the results of the study must be interpreted with significant weight given to the effect of cofounding by indication. Finally we view the recruitment of patients with informed consent into a randomised controlled trial in light of these finding to present a considerable challenge (n=2100 α=0.05, β=0.20).
 Klein M, Gögenur I, Rosenberg J. Postoperative use of non-steroidal anti-inflammatory drugs in patients with anastomotic leakage requiring reoperation after colorectal resection: cohort study based on prospective data. BMJ. 2012 Sep 26;345.
 Iancu C, Mocan LC, Todea-Iancu D et al. Host-related predictive factors for anastomotic leakage following large bowel resections for colorectal cancer. J Gastrointestin Liver Dis. 2008 Sep;17(3):299-303.
 Gorissen KJ, Benning D, Berghmans T et al. Risk of anastomotic leakage with non-steroidal anti-inflammatory drugs in colorectal surgery. Br J Surg. 2012 May;99(5):721-7.
Competing interests: None declared
Northampton General Hospital, Cliftonville, Northampton NN5 5NQ
Click to like: