Revision of the current guideline module is warranted due to clinical questions concerning the comparative effectiveness of selective versus non-selective NSAIDs in pain management and opioid use.

Summary of Findings

Population: Adult patients undergoing a surgical procedure

Intervention: Postoperative non-selective NSAIDs

Comparator: Postoperative COX-2 inhibitors

_{^1. Risk of Bias: serious. Inadequate concealment of allocation during randomization process, resulting in potential for selection bias, due to differences in interventions and not enough power; Inconsistency: serious. Point estimates vary widely; Imprecision: serious. due to differences between studies in intervention and comparator;}

_{^2. Risk of Bias: serious. Inadequate concealment of allocation during randomization process, resulting in potential for selection bias, due to differences in interventions, not enough power; Imprecision: serious. due to differences between studies in intervention and comparator;}

_{^3. Risk of Bias: serious. Inadequate concealment of allocation during randomization process, resulting in potential for selection bias, due to differences in interventions and not enough power; Inconsistency: serious. The direction of the effect is not consistent between the included studies; Imprecision: serious. due to differences between studies in intervention and comparison;}

_{^4. Risk of Bias: serious. Inadequate concealment of allocation during randomization process, resulting in potential for selection bias, due to differences in interventions, not enough power; Inconsistency: serious. The direction of the effect is not consistent between the included studies; Imprecision: serious. due to differences between studies in interventions and comparisons;}

_{^5. Risk of Bias: serious. Inadequate concealment of allocation during randomization process, resulting in potential for selection bias, due to differences in interventions and not enough power; Inconsistency: serious. The direction of the effect is not consistent between the included studies; Imprecision: serious. due to differences between studies in interventions and comparisons;}

_{^6. Risk of Bias: serious. Inadequate concealment of allocation during randomization process, resulting in potential for selection bias, due to differences in interventions and not enough power; Imprecision: very serious. Due to overlap of the lower and upper limit of the 95% confidence interval with the minimally clinically important difference;}

_{^7. Risk of Bias: serious. Inadequate concealment of allocation during randomization process, resulting in potential for selection bias, due to differences in interventions and not enough power; Imprecision: very serious. Wide confidence intervals, due to differences between studies in interventions and comparators.}

Description of studies

A total of three randomized controlled trials were included in the analysis of the literature. Important study characteristics and results are summarized in table 2. The assessment of the risk of bias is summarized in the risk of bias tables (under the tab ‘Evidence tabellen’).

Rawal (2013) performed a multicenter, randomized, double-blind, placebo- and active-comparator controlled, multiple-dose study in patients undergoing unilateral total knee replacement surgery. The objective of the study was to evaluate the tolerability and efficacy of etoricoxib. Patients were eligible for trial participation if they were men or women, at least eighteen years of age, considered to be in good general health, and scheduled to have a unilateral total knee replacement completed in 3 hours or less. Women of childbearing potential were required to have a β-hCG consistent with a non-gravid state prior to randomization and were required to use adequate birth control methods or remain abstinent during the study. Postoperatively, patients needed to tolerate clear liquids and have a pain intensity of at least five in the recovery room. Exclusion criteria were: allergy or intolerance to study medications; recent history of chronic drug abuse/dependence, received an investigational agent/device (within four weeks), morbid obesity, or certain cardiovascular and gastrointestinal conditions.

White (2011) performed a single-center, randomized, double-blind study in patients who underwent ambulatory surgery. The objective of the study was to evaluate the effect of postoperative administration of either a nonselective NSAID (ibuprofen) or the cyclooxygenase-2 selective inhibitor (celecoxib) on the severity of pain, the need for rescue analgesics and clinically relevant patient outcomes. Patients scheduled for superficial (noncavitary) surgical procedures were enrolled in the study. Exclusion criteria were: difficulty understanding English, allergy or contraindication to taking NSAIDs, chronic use of NSAIDs, opioid analgesic medication within a 12-hour period before the operation, pregnancy or breast-feeding, history of alcohol or drug abuse, bleeding disorder, clinically significant neurologic, cardiovascular, renal, hepatic or gastrointestinal diseases.

Wattchow (2009) performed a multicenter, randomized, double-blind, controlled trial in patients undergoing elective major abdominal surgery. The objective of the study was to determine whether prostaglandin inhibition by cyclooxygenase inhibitors (celecoxib or diclofenac) could enhance gastrointestinal recovery and reduce post-operative ileus in humans. Patients were eligible for trial participation if they were to undergo elective surgery that involved substantial handling of the intestines (laparotomy for colorectal procedures and small bowel resections). Exclusion criteria were: calculated creatinine clearance less than 30 mL/min, allergy to NSAIDs, recent gastrointestinal ulcers, asthma or therapeutic anticoagulation.

Table 2. Characteristics of included studies

_{*For further details, see the risk of bias table in the appendix}

Results

Postoperative pain (critical)

Three studies reported the outcome measure postoperative pain at 24 hours and 48 hours (Rawal, 2013; Wattchow, 2009; White, 2011). Since the interventions and controls varied across studies, results were not pooled.

At 24 hours

Rawal (2013) reported the average pain intensity at rest on day one (change from baseline). Only a figure without exact scores was reported in the study. The patients who received ibuprofen (n=224) had a pain intensity score (change from baseline) of approximately -3.2, the patients who received etoricoxib 90 mg (n=224) had a pain intensity score (change from baseline) of approximately -3.1, and the patients who received etoricoxib 120 mg (n=230) had a pain intensity score (change from baseline) of approximately -3.0. The change in average pain intensity from baseline was the largest for the group who received ibuprofen, with a difference of 0.1 and 0.2 compared to etoricoxib 90 mg and etoricoxib 120 mg. This difference was not clinically relevant.

Wattchow (2009) reported median post-operative pain scores with 25^th and 75^th percentiles at day 1. The patients who received diclofenac (n=69) had a median pain score of 2 (1, 4), and the patients who received celecoxib (n=74) had a median pain score of 3 (2, 5). The median difference was -1, in favour of the patients who received diclofenac. This difference was clinically relevant.

White (2011) reported mean postoperative pain scores with standard deviations at 24 hours. The patients who received ibuprofen (n=60) had a mean pain score of 5 (SD ± 3), and the patients who received celecoxib (n=60) had a mean pain score of 5 (SD ± 3). There was no difference in pain scores between the groups (95% CI -1.07 to 1.07).

At 48 hours

Rawal (2013) reported average pain intensity at rest at day 2 (change from baseline). Only a figure but no exact numbers were reported in the study. The patients who received ibuprofen (n=224) had a pain intensity score (change from baseline) of approximately -3.9, the patients who received etoricoxib 90 mg (n=224) had a pain intensity score (change from baseline) of approximately -4.0, and the patients who received etoricoxib 120 mg (n=230) had a pain intensity score (change from baseline) of approximately -4.0. The change in average pain intensity from baseline was the largest for the group who received etoricoxib (both 90 mg and 120 mg), with a difference of 0.1 compared to ibuprofen. This was not clinically relevant.

Wattchow (2009) reported median post-operative pain scores with 25^th and 75^th percentiles at day 2. The patients who received diclofenac (n=69) had a median pain score of 2 (1, 3), and the patients who received celecoxib (n=74) had a median pain score of 2 (1, 3). There was no difference in pain scores between the groups.

White (2011) reported mean postoperative pain scores with standard deviations at 48 hours. The patients who received ibuprofen (n=60) had a mean pain score of 4 (SD ± 3), and the patients who received celecoxib (n=60) had a mean pain score of 4 (SD ± 3). There was no difference in pain scores between the groups (95% CI -1.07 to 1.07).

Postoperative opioid use (important)

Three studies reported the outcome measure postoperative opioid use 24 hours and 48 hours (Rawal, 2013; Wattchow, 2009; White, 2011). Since the interventions and controls varied across studies, results were not pooled.

In 24 hours

Rawal (2013) reported postoperative morphine consumption (morphine equivalent) at day 1. Only a figure but no exact numbers were reported in the study. The patients who received ibuprofen (n=224) had a morphine equivalent of approximately 35 mg, the patients who received etoricoxib 90 mg (n=224) had a morphine equivalent of approximately 45 mg, and the patients who received etoricoxib 120 mg (n=230) had a morphine equivalent of approximately 45 mg. The difference in morphine consumption between the patients who received ibuprofen and the patients who received etoricoxib 90 mg and 120 mg was -10 mg, in favour of the patients who received ibuprofen. This difference was clinically relevant.

Wattchow (2009) reported median parenteral narcotics with 25^th and 75^th percentiles at day 1. The patients who received diclofenac (n=69) received a median of 17 mg of narcotics (0, 67), and the patients who received celecoxib (n=74) received a median of 28 mg of narcotics (0, 57). The median difference was -11 mg, in favour of the patients who received diclofenac. This difference was clinically relevant.

White (2011) reported rescue opioid analgesic medication in mean number of pills at 24 hours. The patients who received ibuprofen (n=60) received 2 rescue opioid pills (SD ± 2), and the patients who received celecoxib (n=60) received 1 rescue opioid pill (SD ± 1). The mean difference was 1.00 (95% CI 0.43 to 1.57), in favour of the patients who received celecoxib. As the dose is not reported, the clinical relevance cannot be assessed.

In 48 hours

Rawal (2013) reported postoperative morphine consumption at day 2. Only a figure but no exact numbers were reported in the study. The patients who received ibuprofen (n=224) had a morphine equivalent of approximately 7 mg, the patients who received etoricoxib 90 mg (n=224) had a morphine equivalent of approximately 8.5 mg, and the patients who received etoricoxib 120 mg (n=230) had a morphine equivalent of approximately 8.5 mg. The difference in morphine consumption between the patients who received ibuprofen and the patients who received etoricoxib 90 mg and 120 mg was -1.5 mg, in favour of the patients who received ibuprofen. This difference was not clinically relevant.

Wattchow (2009) reported median parenteral narcotics with 25^th and 75^th percentiles at day 2. The patients who received diclofenac (n=69) received a median of 6 mg of narcotics (0, 36), and the patients who received celecoxib (n=74) received a median of 15.5 mg of narcotics (0, 39). The median difference was -9.5 mg, in favour of the patients who received diclofenac. This difference was not clinically relevant.

White (2011) reported rescue opioid analgesic medication in mean number of pills at 48 hours. The patients who received ibuprofen (n=60) received 1 rescue opioid pills (SD ± 2), and the patients who received celecoxib (n=60) received 1 rescue opioid pill (SD ± 1). There was no difference in rescue opioid analgesic medication between the groups (95% CI -0.57 to 0.57).

(Major) Complications: gastrointestinal adverse events and cardiovascular adverse events (important)

Three studies reported the outcome measure (major) complications: gastrointestinal adverse events and cardiovascular adverse events (Rawal, 2013; Wattchow, 2009; White, 2011). Since the interventions and controls varied across studies, results were not pooled.

Bleeding

Rawal (2013) reported bleeding-related adverse events. Of the 224 patients who received ibuprofen, 14 patients (6.3%) had a bleeding-related adverse event, of the 224 patients who received etoricoxib 90 mg, 16 patients (7.2%) had a bleeding-related adverse event, and of the 230 patients who received etoricoxib 120 mg, 11 patients (4.8%) had a bleeding-related adverse event.

The risk ratio, comparing ibuprofen with etoricoxib 90 mg, was 0.88 (95% CI 0.44 to 1.75), in favour of the patients who received ibuprofen. This difference was clinically relevant.

The risk ratio, comparing ibuprofen with etoricoxib 120 mg, was 1.31 (95% CI 0.61 to 2.82), in favour of the patients who received etoricoxib 120 mg. This difference was clinically relevant.

White (2011) reported bleeding. None of the 60 patients who received ibuprofen and none of the 60 patients who received celecoxib suffered from bleeding.

Wound-related adverse events

Rawal (2013) reported wound-related adverse events. Of the 224 patients who received ibuprofen, 2 patients (0.9%) had a wound-related adverse event, of the 224 patients who received etoricoxib 90 mg, 1 patient (0.5%) had a wound-related adverse event, and of the 230 patients who received etoricoxib 120 mg, 2 patients (0.9%) had a wound-related adverse event.

The risk ratio, comparing ibuprofen with etoricoxib 90 mg, was 2.00 (95% CI 0.18 to 21.90), in favor of the patients who received etoricoxib 90 mg. This difference was clinically relevant.

The risk ratio, comparing ibuprofen with etoricoxib 120 mg, was 1.03 (95% CI 0.15 to 7.23), in favor of the patients who received etoricoxib 120 mg. This difference was not clinically relevant.

Myocardial infarction

Rawal (2013) reported myocardial infarction. None of the 224 patients who received ibuprofen had myocardial infarction, one of the 224 patients (0.5%) who received etoricoxib 90 mg had myocardial infarction, and one of the 230 patients (0.4%) who received etoricoxib 120 mg had myocardial infarction.

The risk ratio, comparing ibuprofen with etoricoxib 90 mg, was 0.33 (95% CI 0.01 to 8.14), in favor of the patients who received ibuprofen. This difference was clinically relevant.

The risk ratio, comparing ibuprofen with etoricoxib 120 mg, was 0.34 (95% CI 0.01 to 8.36), in favor of the patients who received ibuprofen. This difference was clinically relevant.

Wattchow (2009) reported myocardial infarction. None of the 69 patients who received diclofenac and none of the 74 patients who received celecoxib had myocardial infarction.

Rawal (2013) reported cardiac arrest. None of the 224 patients who received ibuprofen, none of the 224 patients who received etoricoxib 90 mg, and none of the patients who received etoricoxib 120 mg had cardiac arrest.

Cardiovascular death

None of the studies reported the outcome measure cardiovascular death.

A systematic review of the literature was performed to answer the following question(s):

What is the effect of postoperative use of non-selective NSAIDs compared to COX-2 inhibitors on outcomes: postoperative pain, postoperative opioid use and (major) complications?

Table 1. PICO

Relevant outcome measures

The guideline panel considered postoperative pain at 24 hours and 48 hours as a critical outcome measure for decision making; and postoperative opioid use in 24 hours and 48 hours, and (major) complications as important outcome measures for decision making.

A priori, the guideline panel defined the outcome measures as follows:

• Postoperative pain: measured at rest on a validated pain scale (VAS or NRS) at 24 hours and 48 hours.
• Postoperative opioid use: dose consumed in 24 hours and 48 hours.
• (Major) complications: gastrointestinal adverse effects (bleeding or ulcers), cardiovascular events (myocardial infarct or cardiovascular death).

For pain, the working group defined one point as a minimal clinically (patient) important difference on a 10-point scale and 10 mm on a 100 mm scale. Regarding postoperative opioid use, a difference of 10 mg MME between groups was considered clinically relevant. For complications, a difference of 10% was considered clinically relevant (RR <0.91 or >1.10; RD 0.10).

Search and select (Methods)

A systematic literature search was performed by a medical information specialist using the following bibliographic databases: Embase.com and Ovid/Medline. Both databases were searched from 2005 to 27 May 2025 for systematic reviews and RCTs. Systematic searches were completed using a combination of controlled vocabulary/subject headings (e.g., Emtree-terms, MeSH) wherever they were available and natural language keywords. The overall search strategy was derived from two primary search concepts: (1) postoperative; (2) cyclooxygenase 2 inhibitor. Duplicates were removed using EndNote software. After deduplication, a total of 2,235 records were imported for title/abstract screening. Titles and abstracts were screened using the ASReview software. The settings ELAS u4 (TF-IDF and SVM) were used. White (2011) and Rawal (2013) were used as prior knowledge for inclusion. No articles were pre-labelled for exclusion. The first 10% of hits were screened by the working group and the guideline methodologist. The remaining articles were subsequently screened by the guideline methodologist, using the following stopping rule: stop after 200 subsequent exclusions.

Based on the titles and abstracts, three articles were selected in total: two were provided as prior knowledge, and one additional study was identified through ASReview screening. In total 2232 studies were excluded. After screening the full-text articles, a total of three studies were included, and no  studies were excluded at this stage.

1. Chan FK, Hung LC, Suen BY, Wu JC, Lee KC, Leung VK, et al. Celecoxib versus diclofenac and omeprazole in reducing the risk of recurrent ulcer bleeding in patients with arthritis. N Engl J Med 2002; 347: 2104-2110.
2. Milne, T. G. E., Jaung, R., O'Grady, G., & Bissett, I. P. (2018). Nonsteroidal anti-inflammatory drugs reduce the time to recovery of gut function after elective colorectal surgery: a systematic review and meta-analysis. Colorectal disease : the official journal of the Association of Coloproctology of Great Britain and Ireland, 20(8), O190–O198. https://doi.org/10.1111/codi.14268.
3. Rawal N, Viscusi E, Peloso PM, Minkowitz HS, Chen L, Shah S, Mehta A, Chitkara DK, Curtis SP, Papanicolaou DA. Evaluation of etoricoxib in patients undergoing total knee replacement surgery in a double-blind, randomized controlled trial. BMC Musculoskelet Disord. 2013 Oct 24;14:300. doi: 10.1186/1471-2474-14-300. PMID:24156640; PMCID: PMC3840772.
4. Rijksinstituut voor Volksgezondheid en Milieu (RIVM), Risico’s van pijnstillers in het oppervlaktewater Keuzes voor gebruik op basis van milieueffecten, 2023.
5. Wattchow DA, De Fontgalland D, Bampton PA, Leach PL, McLaughlin K, Costa M. Clinical trial: the impact of cyclooxygenase inhibitors on gastrointestinal recovery after major surgery - a randomized double blind controlled trial of celecoxib or diclofenac vs. placebo. Aliment Pharmacol Ther. 2009 Nov 15;30(10):987-98. doi:10.1111/j.1365-2036.2009.04126.x. Epub 2009 Aug 20. PMID: 19694636.
6. White PF, Tang J, Wender RH, Zhao M, Time M, Zaentz A, Yumul R, Sloninsky A, Naruse R, Kariger R, Webb T, Fermelia DE, Tsushima GK. The effects of oral ibuprofen and celecoxib in preventing pain, improving recovery outcomes and patient satisfaction after ambulatory surgery. Anesth Analg. 2011 Feb;112(2):323-9. doi:10.1213/ANE.0b013e3182025a8a. Epub 2010 Dec 14. PMID: 21156974.