NPC Archive Item: Gastroprotection – no protective advantage for coxibs over traditional NSAIDs among elderly patients on a PPI

NOTE – This is an archive post from the NPC and has not been updated since first publication. Therefore, some hyperlinks may no longer be working.
MeReC Rapid Review NPC Logo

Abraham NS, Hartman C, Castillo D, et al. Effectiveness of national provider prescription of PPI gastroprotection among elderly NSAID users. Am J Gastroenterol 2008;103:323-32

Limited observational data question using a coxib plus PPI in preference to using a non-selective NSAID plus PPI to reduce the risk of serious upper GI bleeds in older people with osteoarthritis.  If NSAID use is essential in someone with OA, first choice for most people will be ibuprofen 1200 mg/day or naproxen 1000 mg/day, plus a PPI with a low acquisition cost. This approach minimises both cardiovascular and GI risks of NSAIDs.

Oral NSAIDs, including coxibs, should be used only when other, safer treatments are ineffective or not tolerated. When their use is essential in OA, NICE guidance to use a traditional NSAID or a Cox-2 inhibitor, with co-prescription of a PPI should be followed. All NSAIDs/coxibs should be prescribed at the lowest effective dose for the shortest possible period of time. When choosing an individual agent, prescribing should be based on the safety profiles of individual NSAIDs/coxibs and on individual patient risk factors. Some traditional NSAIDs (e.g. diclofenac 150mg/day) carry the same increased risk of thrombotic events as coxibs. Low-dose ibuprofen (<1200mg/day), and naproxen 1000mg/day do not appear to share these cardiovascular (CV) side effects.  These are appropriate first choice NSAIDs for most people as they also have a low absolute risk increase of gastro-intestinal (GI) side effects which may be very low when combined with a PPI. (See Rahme E et al and Lanas A et al.)

What is the background to this?
All non-steroidal anti-inflammatory drugs (NSAIDs), including Cox-2 inhibitors (coxibs), carry the risk of side effects, which can be serious and life-threatening. Although the risks may vary between individual NSAIDs, important side effects include gastrointestinal (GI) complications (e.g. perforation, ulcer, bleeding) and cardiovascular (CV) events (e.g. stroke, myocardial infarction). For more information on the GI and CV safety of traditional NSAIDs and coxibs, see MeReC Extra No 30, November 2007 and the related blog.

This retrospective cohort study, carried out in the US, provides new data on the risk of serious upper GI events (e.g. ulcers and bleeds) with traditional NSAIDs and coxibs in elderly patients, and the reductions in these risks with the co-prescribing of proton pump inhibitors (PPIs).

What does this study claim?
After adjusting for numerous confounding variables and risk factors, the risk (hazard ratio [HR]) of serious upper GI events (UGIEs) in the cohort of nearly 500,000 largely white, male patients (mean age 74 years) was 1.8 (95% CI 1.6 to 2.0) during exposure to non-selective NSAIDs alone and 1.8 (95% CI 1.5 to 2.0) during exposure to coxibs alone, compared with time not exposed to these drugs. That is, the risk of UGIEs was not statistically significantly different in the coxib group from that in the NSAID group. Other important predictors of GI risk were antiplatelet drugs (HR 1.6, 95% CI 1.4 to 1.9), selective serotonin reuptake inhibitors (HR 1.4, 95% CI 1.2 to 1.6) and serotonin-noradrenaline reuptake inhibitors (HR 1.8 95% CI 1.3 to 2.4). Low-dose aspirin and anticoagulants also increased the risk of serious upper GI events, but not to a statistically significant degree. Risk increased with age, with the greatest risk being seen in patients 85 years of age and over (HR 1.9, 95% CI 1.5 to 2.3).

Co-prescription of a PPI was observed to reduce the risk of UGIEs, to 1.1 (95% CI 0.7 to 4.6) with an NSAID plus a PPI and 1.1 (95% CI 0.6 to 5.2) with a coxib plus a PPI. That is, the risk of UGIEs was not statistically significantly different in the coxib + PPI group from in the NSAID + PPI group.

How does this relate to other studies?
The results of this cohort study suggest that, as a class, traditional NSAIDs and coxibs carry the same degree of GI risk; increasing the risk of a serious GI event by 80% in an elderly population. However, this is observational data, which is subject to various confounding factors and bias, as patients were not randomised to their treatment regimens. A recent systematic review of eight randomised controlled trials (n=73,449) found that, compared with traditional NSAIDS, coxibs as a class produced significantly fewer clinically important ulcer complications (relative risk [RR] 0.39, 95%CI 0.31 to 0.50). These results are reasonably homogeneous with different coxibs in these RCTs and it seems pretty clear that coxibs do offer some level of gastroprotection. However, they do increase cardiovascular risk compared with ibuoprofen 1200mg per day or less, or naproxen 1000mg per day (see MeReC Extra 30).

NICE has recently published its guidance on the management of osteoarthritis (OA) (see blog). When offering treatment with an oral NSAID/Cox-2 inhibitor for OA, NICE recommends that the first choice should be either a traditional NSAID or a Cox-2 inhibitor (other than etoricoxib 60 mg). In either case, they recommend that these should be co-prescribed with a PPI in all patients. The MeReC Extra pointed out that the benefits from gastroprotection largely depend on the individual patient’s baseline risk of GI complications, and that “there is, as yet, no good evidence that adding a PPI to a coxib is a more beneficial, equivalent or worse option than adding a PPI to a traditional NSAID.” This study suggests that adding a PPI to a coxib is only of similar benefit to adding a PPI to a traditional NSAID.

So what?
This new data is observational data, and when the data is soft we should speak softly. A major limitation of this study is the overlapping 95% confidence intervals – the risk of UGIEs in both groups co-prescribed a PPI overlap with the confidence intervals in both groups not co-prescribed a PPI.  In other words, although co-prescription led to an observed reduction in UGIE risk this was not statistically significant in this study.  That doesn’t necessarily mean that there is not a real difference, but we can’t be sure from this study that the observed difference was not just a chance finding.  This paper does not provide evidence beyond reasonable doubt that a coxib + PPI is less risky, more risky, or has the same risks as a coxib alone, a non-selective NSAID alone or a non-selective NSAID +PPI.

So, although the observed protective GI benefit of a coxib plus a PPI was similar to an NSAID plus a PPI, this finding should be regarded as provisional and replicated by additional observational analyses, and ideally in a large randomised controlled trial. We cannot currently be certain whether a coxib plus a PPI offers any reduction in UGIEs compared with a traditional NSAID plus PPI. However, there is good evidence that choosing a coxib and some traditional NSAIDs instead of ibuoprofen 200mg per day or less, or naproxen 1000mg per day will increase the cardiovascular event rate.

You can find more information about osteoarthritis and its management, and the risks and benefits of NSAIDs and analgesics, on the Musculoskeletal Pain section of NPC

Study details
Patients – This US cohort study identified veterans aged 65 years or older from national pharmacy records who were prescribed an NSAID, a coxib, or aspirin (>325mg/day) from January 2000 to December 2002. Prescription data was linked to inpatient, outpatient, and death files.

Intervention and comparators – each person-day of follow-up was assessed for exposure to NSAID alone, NSAID plus PPI, coxib, or coxib plus PPI.

Outcomes – Serious upper GI events were defined using a published, validated algorithm. The risk of serious upper GI events was assessed using Cox proportional hazards models, while adjusting for demographics, GI risk factors, comorbidity, prescription channeling, geographic location, and multiple time-dependent pharmacological covariates, including aspirin, steroids, anticoagulants, antiplatelets, statins, and selective serotonin reuptake inhibitors.

Results – The cohort consisted of 481,980 patients (98% male, 85% white, mean age 74 years). Co-prescription of a PPI with either an NSAID or a coxib occurred in about 20% of the cohort. 2,753 serious upper GI events occurred in 220,662 person-years of follow-up. When adjusted for prescription channeling, confounders, and effect modification-associated PPI, risk of serious upper GI event was 1.8 (95% CI 1.6–2.0) on NSAID alone, 1.8 (95% CI 1.5–2.0) on coxib alone, 1.1 (95% CI 0.7–4.6) on NSAID plus PPI, and 1.1 (0.6–5.2) on coxib plus PPI.

Feedback: Please comment on this blog in the NPC discussion rooms, or using our feedback form