Knee pain/NSAIDs

[Jordan Feigenbaum]

There are MANY pooled studies / meta-analyses as well as high quality individual trials; and there is enough disagreement among them that I cannot responsibly conclude that the risk with naproxen is no different than placebo. I'm not saying I can responsibly conclude the opposite either -- but the fact is that many studies show it's got increased risk; and since increased risk appears to be a class effect of NSAIDs it's more likely that unmeasured differences in naproxen use or in study design account for all the heterogeneity in study findings.

That's a fair shake at it, Paul. I disagree mainly based that when pooled, the small increased risks seen in relatively smaller studies do not persist and due to the multifactorial nature of the adverse events we're discussing I cannot conclude that there's a significant risk. When newer/better data comes out that supplants the analysis below, I may change this view.

To claim with confidence that naproxen has no risk seems to in the very least overreach the current position of the medical community.

Unfortunately, I think there may be a gap in knowledge base on the topic here. When looking at the "current position of the medical community", I think one could argue that meta analyses published in high-impact journals and position statements tend to be ideal. As such, here is what I consider to be high quality evidence in support of my stance. You'll notice the very large sample size and the 280 studies it includes (including all of those mentioned in this thread).

Major cardiovascular events (a composite of nonfatal MI, nonfatal stroke, or vascular death) were significantly increased compared with placebo for high-dose diclofenac (adjusted rate ratio [ARR] 1.41, 95% CI 1.12-1.78) and for the coxibs (ARR 1.37, 95% CI 1.14-1.66), largely due to increases in major coronary heart disease events. A similar, but statistically nonsignificant, increase in risk was observed with high-dose ibuprofen (ARR 1.44, 95% CI 0.89-2.33). For these drugs, the risk of major coronary events was significantly increased. The use of high-dose naproxen did not result in a significant increase in major cardiovascular events (ARR 0.93, 95% CI 0.69-1.27) or coronary events. The data regarding the individual nonselective NSAIDs are somewhat limited, contributing to wide confidence intervals, especially for ibuprofen, which reflect imprecision in the estimates.

Vascular death was significantly increased by diclofenac and by the coxibs (RR 1.65, 99% CI 0.95-2.85, and 1.58, 99% CI 1.00-2.49). The similarly increased risk suggested for ibuprofen did not achieve statistical significance (RR 1.90, 99% CI 0.56-6.41). Risk of vascular death was not significantly increased by use of naproxen (RR 1.08, 99% CI 0.48-2.47).

Coxib and traditional NSAID Trialists' (CNT) Collaboration, Bhala N, Emberson J, et al. Vascular and upper gastrointestinal effects of non-steroidal anti-inflammatory drugs: meta-analyses of individual participant data from randomised trials. Lancet 2013

Is it bad to regularly use Naproxen to treat golfers/tennis elbow?

Bad how?

[Subsistence]

Unfortunately, I think there may be a gap in knowledge base on the topic here. When looking at the "current position of the medical community", I think one could argue that meta analyses published in high-impact journals and position statements tend to be ideal. As such, here is what I consider to be high quality evidence in support of my stance. You'll notice the very large sample size and the 280 studies it includes (including all of those mentioned in this thread)

The authors of that metaanalysis even state that their confidence intervals were rather wide which effects the precision of the results. And the FDA just shot down a bid to remove the cardiovascular warning from naproxen. They may be being overly cautious, and the risk may very well be no different than placebo. But at this point I'm more comfortable telling my patients that the risk for naproxen falls somewhere between the best option we have and zero. I think reporting the results of those studies (no risk found!) while neglecting to mention the side cautions and caveats could be a case of overextension.

It is also likely worth noting that most of the high risk patients I see are on 81mg aspirin as a blood thinner and there is some (minor) data suggesting a risk of interference by naproxen (and other NSAIDs) on aspirin's ability to thin the blood over time.

[Paul1]

Jordan,
I think the following we could agree on:

1 - Avoiding NSAIDs, whichever ones, are about 20th on anyone's priority list for preventing heart disease, and it only really applies to people who take them chronically.

2 - There is a class effect of NSAIDs, and the observational data make it impossible to truly know the magnitude of the effect let alone parse trends within the members of the class. You're right that in SOME pooled data naproxen appears safer, but this is contradicted by data of comparable quality.

If the truth lies anywhere, it is most likely that either 1) all NSAIDs were safe all along, or 2) there is a moderate class effect, every drug in the class has this effect, and there are small differences between drugs that have more to do with which subpopulations take them than anything pharmacologic or biochemical.

[Jordan Feigenbaum]

The authors of that metaanalysis even state that their confidence intervals were rather wide which effects the precision of the results. And the FDA just shot down a bid to remove the cardiovascular warning from naproxen. They may be being overly cautious, and the risk may very well be no different than placebo. But at this point I'm more comfortable telling my patients that the risk for naproxen falls somewhere between the best option we have and zero. I think reporting the results of those studies (no risk found!) while neglecting to mention the side cautions and caveats could be a case of overextension.

Who said failing to mention side effects to patients? I do not think the current body of evidence on Naproxen warrants a big discussion of cardiovascular risk. You do and that's okay too. I think that is a case of "overextension" of the evidence, but it is okay that we disagree. Also consider that the FDA removing black box warnings is extremely unlikely to happen-just look at at the fact they haven't removed the suicide risk for SSRI's in adolescents/kids yet. It's probably never going to happen for reasons outside the scope of this topic.

It is also likely worth noting that most of the high risk patients I see are on 81mg aspirin as a blood thinner and there is some (minor) data suggesting a risk of interference by naproxen (and other NSAIDs) on aspirin's ability to thin the blood over time.

From what I've read, it's not so much interference- but rather competing for the COX-1 receptor by the NSAID so the ASA can't irreversibly inhibit it. Additionally, the TARGET trial suggests that this does not cause a significant difference in cardiovascular events when using naproxen- though there is a difference when using ibuprofen.

Jordan,
I think the following we could agree on:

1 - Avoiding NSAIDs, whichever ones, are about 20th on anyone's priority list for preventing heart disease, and it only really applies to people who take them chronically.

2 - There is a class effect of NSAIDs, and the observational data make it impossible to truly know the magnitude of the effect let alone parse trends within the members of the class. You're right that in SOME pooled data naproxen appears safer, but this is contradicted by data of comparable quality.

If the truth lies anywhere, it is most likely that either 1) all NSAIDs were safe all along, or 2) there is a moderate class effect, every drug in the class has this effect, and there are small differences between drugs that have more to do with which subpopulations take them than anything pharmacologic or biochemical.

I would agree with this with the caveat that I have not seen data of comparable quality/size yet and with both #1 and the likely "subpopulation effect" in mind, I suppose I am less interested. Good discussion, gents.

[Paul1]

There are a lot of pooled studies. The Lancet one is the biggest, but there are a handful with 40-50 included studies.

Bigger is not always better with meta-analyses, because you smooth over a ton of variables that differentiate the constituent studies. The fact though is that all these studies have the same basic design, which is secondary data analysis, and they all fall in the same evidence class. On our guideline our methodologist has taught us that grading protocol is to downgrade evidence quality when you have studies of comparable design that contradict each other. It creates uncertainty.

The subpopulation effect hasn't been studied in anything I've seen. But you can imagine that prescription-only celecoxib patients are a different pool than OTC naproxen patients. Patients who purchase OTC drugs are probably less likely to take every single dose all the time as compared with patients who have a pill bottle specifying the dose and frequency. Also, different specialties have different prescribing patterns, and there may be different (but unmeasured) populations of patients taking different NSAIDs.

At the same time the reason I hate this issue is that I see upper GI bleeding from NSAIDs probably every single week I'm on the general medicine service -- I had 3 or 4 patients with this last week alone and I was only on service for 4 days. We can do a lot more about managing baseline CAD risk than about managing GI bleed risk, which is a lot less predictable.

[vanslix]

From my perspective, I'd be more worried about renal damage than CV events with naproxen. Plus, it's easily monitored.

[Jordan Feigenbaum]

There are a lot of pooled studies. The Lancet one is the biggest, but there are a handful with 40-50 included studies.

Bigger is not always better with meta-analyses, because you smooth over a ton of variables that differentiate the constituent studies. The fact though is that all these studies have the same basic design, which is secondary data analysis, and they all fall in the same evidence class. On our guideline our methodologist has taught us that grading protocol is to downgrade evidence quality when you have studies of comparable design that contradict each other. It creates uncertainty.

I can see your point here and thanks for sharing. It obviously gets murky when multiple pooled analyses exist using many of the same studies, but different statistical methods and come up with slightly different findings- creating uncertainty, as it were. You also make a good point about the big data smoothing over variables that may be important. On the other hand, removing the effect of the smoothed-over variables may be more representative of the gen pop at large, but you're exactly right in being cautious about both big and small data.

In the end, I suppose my bias is that my concern is very low at baseline for cardiovascular events 2/2 NSAIDs- particularly naproxen- and when the Lancet's monster meta analysis doesn't suggest a significant risk I get even less concerned. I, like you, worry more about GI bleeds and renal issues than cardiac stuff from NSAIDs.

[Paul1]

Yup. Cardiology is the world of incremental risk math. And even if you just took celebrex for example, how do you balance the cardiac "risks" of that drug versus the risks from reduced activity due to arthritis pain?

[Hyperfluxe]

Bad how?

Well I've read that NSAIDs prolong recovery and fuck with MPS?

[Jordan Feigenbaum]

Well I've read that NSAIDs prolong recovery and fuck with MPS?

They do not. In fact, there's data suggesting they improve recovery, MPS, anabolism, etc.