COX-2 inhibitors measured on what the patient can do, not on a number they report

Design

Systematic review and meta-analysis of randomised controlled trials with GRADE certainty assessment

Population

5,424 adults across 38 trials of systemic perioperative selective COX-2 inhibitors versus placebo, opioids or usual care

Primary outcome

Acute postoperative pain-related interference with daily living on a validated multidimensional instrument within 1 month

Effect

Brief Pain Inventory MD -1.1 (95% CI -1.4 to -0.8, moderate certainty); chronic pain incidence OR 0.44 (0.21-0.93, low certainty); QoR-9 MD 0.84 (0.33-1.35)

Perioperative COX-2 inhibitors are given routinely, and the evidence behind them is mostly pain scores at fixed timepoints — a measure that tells you little about whether the patient could sit up, sleep, or walk to the bathroom. This review took the harder outcome: multidimensional instruments assessing pain-related interference with daily living, within one month of surgery.

MEDLINE, Embase and Cochrane CENTRAL were searched; 9,071 citations screened; 38 randomised trials and 5,424 adults met criteria, comparing systemic perioperative selective COX-2 inhibitors with placebo, opioids or usual care. Two reviewers did selection, extraction, risk of bias and certainty assessment independently.

The Brief Pain Inventory mean difference was -1.1 (95% CI -1.4 to -0.8, seven trials, moderate certainty) on a 0-10 scale whose minimal important difference is 1.0 — so the point estimate sits just past the threshold and the whole interval is in the right direction. Chronic pain incidence fell (OR 0.44, 95% CI 0.21-0.93, five trials, low certainty), though chronic pain intensity did not change significantly (MD -1.0, 95% CI -2.7 to 0.7, one trial, very low certainty). Quality of Recovery-9 improved by 0.84 points (95% CI 0.33-1.35, low certainty). Intraoperative blood loss was 22 ml lower (95% CI -40 to -4).

Safety was the other question, and nothing appeared: no difference in acute renal failure, gastrointestinal bleeding, impaired bone healing, myocardial infarction, stroke or death.

• The effect is on interference with function, which is the outcome patients notice
• A mean difference of 1.1 against a threshold of 1.0 is a real but modest benefit
• No signal on renal failure, GI bleeding or bone healing across 38 trials
• The chronic pain finding rests on five trials at low certainty — promising, not settled
• Blood loss 22 ml lower is statistically real and clinically irrelevant; do not quote it as a benefit

Why it matters

It moves the evidence for a routine drug from pain scores to whether the patient could function.

Don't overread it

The absence of a safety signal across 38 trials does not establish safety in the patients trials exclude — established renal impairment, active ulcer disease, high cardiovascular risk.

The statistics, in plain English

The minimal important difference is the smallest change a patient can actually notice, and here it was set at 1.0 on the Brief Pain Inventory. The pooled difference of -1.1 with an interval from -1.4 to -0.8 means the average benefit clears that threshold but the lower end of the range does not — some patients gain meaningfully, others not. 'Moderate certainty' for that result and 'low' or 'very low' for the chronic pain results tells you which numbers to lean on.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.