On 2016 Apr 04, R Andrew Moore commented:

Do topical nonsteroidal anti-inflammatory drugs for acute musculoskeletal pain work?

It is good to see Peter taking an interest in another of our publications.

The evidence shows that topical NSAIDs can be effective for acute musculosketal pain. That was the result we found in our first review of the topic (BMJ 1998;316:333–8), and it has not changed in subsequent updates. Even 20 years ago we actively examined the issue of study size, and concluded that small studies tended to overestimate treatment effects, as well as testing other issue of quality, and of the particular topical NSAID tested. In the current update we were able to add formulation to the list of topics that might affect study results.

Peter dismisses the results for a number of reasons, we think incorrectly. For example, the only risk of bias measure where there was potentially high risk of bias was small study size, a measure included in our risk of bias assessments rather than ignored, as is so often the case. On the issue of industry funding affecting the results of randomised trials, not only is there no evidence of any such effect, but there is positive evidence that there is no effect. We showed the lack of any effect in analgesic trials a decade ago (Pain 2006;121:207-18), and no effect was found for statins (BMJ 2014;349:g5741). That does not mean that industry has clean hands, of course, and we have pointed out, for example, the biases that might arise from multiple publication (BMJ 1997;315:635-40) or inappropriate imputation methods (Pain. 2012;153:265-8).

The review did include a number of older studies of relatively poorer quality, involving nine NSAIDs (other than diclofenac and ketoprofen) but we make clear in the review that there were insufficient data of adequate quality to draw any conclusions about these. Studies were underpowered for adverse events, which were inconsistently reported, but we have again drawn attention to these limitations, which are common in many clinical trials.

The review is not sui generis, but an ongoing dynamic of updated reports over the years. For example we wrote to 88 pharmaceutical companies for our 2004 update (BMC Family Practice 2004;5:10), with only one providing otherwise unpublished data. Topical NSAIDs (like most drugs in pain) are usually generic, with no requirement to perform clinical testing, or older, when it is much more difficult to obtain CTRs, as we have frequently been able to do previously. Experience suggested that trying to get hold of CTRs from hundreds of companies worldwide, many of which had done no trials on their product, was probably a lost cause. In this update we were able to identify that trial data from almost 6,000 patients was not available; a known unknown rather than an unknown unknown.

Of course there was heterogeneity for all formulations together, because different formulations produced different levels of efficacy (and probably with different doses as well, though that was difficult to assess). But within formulations like Flector plaster or Emulgel the I2 was 0; Figures 5 and 6 in the Cochrane review make the point. There are arguments to have about the use of heterogeneity tests, but this is not the place.

In terms of declarations, the issues here are of time, what it is that journals want, and relevance. Both RB and Menarini have topical NSAID products. RAM has worked with both those companies, and while that did not involve topical products it is a relevant disclosure. Futura Pharma dropped out of the three year period for declarations. None of the others were relevant, and for the most part involved investigator-initiated research using patient-level data to elucidate issues around evidence in analgesic trials, for example the importance of formulation (Pain 2014;155:14-21; Eur J Pain 2015;19:187-92) or how best to conduct multiple dose studies in acute pain (Br J Anaesth. 2016;116:269-76; BMC Anesthesiol 2016;16:9).

There is much that could not be done without constructive involvement with pharmaceutical companies, as they produce the trial data on which we base our evidence. Understanding that evidence thoroughly is what it is all about. But we think there is confusion over declarations on interest and what is required, and are actively working on just that topic.

On 2016 Mar 29, Peter Gøtzsche commented:

Do topical nonsteroidal anti-inflammatory drugs for acute musculoskeletal pain work?

Referring to their Cochrane review, the authors asserted that topical NSAIDs are effective for acute musculoskeletal pain (1). However, there are major problems with the trials they reviewed and also with the Cochrane review itself (2). The authors included 8,781 patients in their review but found that the results were missing from another 5,900 patients. Moreover, the trials were industry funded, of relatively poor quality, and the authors analysed published data, not data from clinical study reports, and did not try to obtain all the missing trials and data from the manufacturers.

There was extreme heterogeneity in their meta-analyses, e.g. I2 was 92% for the diclofenac trials, which were the most common ones, and there was extreme funnel plot asymmetry, with the largest trials showing the smallest effects (the authors didn’t show funnel plots but I constructed one for diclofenac).

The authors cautioned that the large amounts of unpublished data “could influence results in updates of this review” (2). They certainly could and I believe it is plain wrong to perform meta-analyses on the authors’ data. When I most recently reviewed this area for the BMJ in 2010, I concluded that we don't know whether topical NSAIDs are beneficial (3).

One of the authors, Andrew Moore, “reported receiving a grant and personal fees from Reckitt Benckiser and personal fees from Menarini. No other disclosures were reported” (1). However, in 2015, Moore published another systematic review of NSAIDs where his competing interests, in addition to those declared in JAMA, were: “personal fees from Novartis, grants and personal fees from Grunenthal, personal fees from Orion Pharma, personal fees from Futura Pharma, personal fees from Astellas, personal fees from Eli Lilly, personal fees from Pfizer and personal fees from Menarini” (4).

1 DerryS, Wiffen P, Moore A. Topical nonsteroidal anti-inflammatory drugs for acute musculoskeletal pain. JAMA 2016;315:813-4.

2 Derry S, Moore RA, Gaskell H, McIntyreM, Wiffen PJ. Topical NSAIDs for acute musculoskeletal pain in adults. Cochrane Database Syst Rev. 2015;6:CD007402.

3 Gøtzsche PC. NSAIDs. Clin Evid (Online). 2010 Jun 28.

4 Moore RA, Derry S, Wiffen PJ, Straube S. Effects of food on pharmacokinetics of immediate release oral formulations of aspirin, dipyrone, paracetamol and NSAIDs - a systematic review. Br J Clin Pharmacol 2015;80:381-8.

On 2016 Mar 29, Peter Gøtzsche commented:

Do topical nonsteroidal anti-inflammatory drugs for acute musculoskeletal pain work?

Referring to their Cochrane review, the authors asserted that topical NSAIDs are effective for acute musculoskeletal pain (1). However, there are major problems with the trials they reviewed and also with the Cochrane review itself (2). The authors included 8,781 patients in their review but found that the results were missing from another 5,900 patients. Moreover, the trials were industry funded, of relatively poor quality, and the authors analysed published data, not data from clinical study reports, and did not try to obtain all the missing trials and data from the manufacturers.

There was extreme heterogeneity in their meta-analyses, e.g. I2 was 92% for the diclofenac trials, which were the most common ones, and there was extreme funnel plot asymmetry, with the largest trials showing the smallest effects (the authors didn’t show funnel plots but I constructed one for diclofenac).

The authors cautioned that the large amounts of unpublished data “could influence results in updates of this review” (2). They certainly could and I believe it is plain wrong to perform meta-analyses on the authors’ data. When I most recently reviewed this area for the BMJ in 2010, I concluded that we don't know whether topical NSAIDs are beneficial (3).

One of the authors, Andrew Moore, “reported receiving a grant and personal fees from Reckitt Benckiser and personal fees from Menarini. No other disclosures were reported” (1). However, in 2015, Moore published another systematic review of NSAIDs where his competing interests, in addition to those declared in JAMA, were: “personal fees from Novartis, grants and personal fees from Grunenthal, personal fees from Orion Pharma, personal fees from Futura Pharma, personal fees from Astellas, personal fees from Eli Lilly, personal fees from Pfizer and personal fees from Menarini” (4).

1 DerryS, Wiffen P, Moore A. Topical nonsteroidal anti-inflammatory drugs for acute musculoskeletal pain. JAMA 2016;315:813-4.

2 Derry S, Moore RA, Gaskell H, McIntyreM, Wiffen PJ. Topical NSAIDs for acute musculoskeletal pain in adults. Cochrane Database Syst Rev. 2015;6:CD007402.

3 Gøtzsche PC. NSAIDs. Clin Evid (Online). 2010 Jun 28.

4 Moore RA, Derry S, Wiffen PJ, Straube S. Effects of food on pharmacokinetics of immediate release oral formulations of aspirin, dipyrone, paracetamol and NSAIDs - a systematic review. Br J Clin Pharmacol 2015;80:381-8.

On 2016 Mar 29, Peter Gøtzsche commented:

Do topical nonsteroidal anti-inflammatory drugs for acute musculoskeletal pain work?

Referring to their Cochrane review, the authors asserted that topical NSAIDs are effective for acute musculoskeletal pain (1). However, there are major problems with the trials they reviewed and also with the Cochrane review itself (2). The authors included 8,781 patients in their review but found that the results were missing from another 5,900 patients. Moreover, the trials were industry funded, of relatively poor quality, and the authors analysed published data, not data from clinical study reports, and did not try to obtain all the missing trials and data from the manufacturers.

There was extreme heterogeneity in their meta-analyses, e.g. I2 was 92% for the diclofenac trials, which were the most common ones, and there was extreme funnel plot asymmetry, with the largest trials showing the smallest effects (the authors didn’t show funnel plots but I constructed one for diclofenac).

The authors cautioned that the large amounts of unpublished data “could influence results in updates of this review” (2). They certainly could and I believe it is plain wrong to perform meta-analyses on the authors’ data. When I most recently reviewed this area for the BMJ in 2010, I concluded that we don't know whether topical NSAIDs are beneficial (3).

One of the authors, Andrew Moore, “reported receiving a grant and personal fees from Reckitt Benckiser and personal fees from Menarini. No other disclosures were reported” (1). However, in 2015, Moore published another systematic review of NSAIDs where his competing interests, in addition to those declared in JAMA, were: “personal fees from Novartis, grants and personal fees from Grunenthal, personal fees from Orion Pharma, personal fees from Futura Pharma, personal fees from Astellas, personal fees from Eli Lilly, personal fees from Pfizer and personal fees from Menarini” (4).

1 DerryS, Wiffen P, Moore A. Topical nonsteroidal anti-inflammatory drugs for acute musculoskeletal pain. JAMA 2016;315:813-4.

2 Derry S, Moore RA, Gaskell H, McIntyreM, Wiffen PJ. Topical NSAIDs for acute musculoskeletal pain in adults. Cochrane Database Syst Rev. 2015;6:CD007402.

3 Gøtzsche PC. NSAIDs. Clin Evid (Online). 2010 Jun 28.

4 Moore RA, Derry S, Wiffen PJ, Straube S. Effects of food on pharmacokinetics of immediate release oral formulations of aspirin, dipyrone, paracetamol and NSAIDs - a systematic review. Br J Clin Pharmacol 2015;80:381-8.

On 2016 Mar 29, Peter Gøtzsche commented:

Do topical nonsteroidal anti-inflammatory drugs for acute musculoskeletal pain work?

Referring to their Cochrane review, the authors asserted that topical NSAIDs are effective for acute musculoskeletal pain (1). However, there are major problems with the trials they reviewed and also with the Cochrane review itself (2). The authors included 8,781 patients in their review but found that the results were missing from another 5,900 patients. Moreover, the trials were industry funded, of relatively poor quality, and the authors analysed published data, not data from clinical study reports, and did not try to obtain all the missing trials and data from the manufacturers.

There was extreme heterogeneity in their meta-analyses, e.g. I2 was 92% for the diclofenac trials, which were the most common ones, and there was extreme funnel plot asymmetry, with the largest trials showing the smallest effects (the authors didn’t show funnel plots but I constructed one for diclofenac).

The authors cautioned that the large amounts of unpublished data “could influence results in updates of this review” (2). They certainly could and I believe it is plain wrong to perform meta-analyses on the authors’ data. When I most recently reviewed this area for the BMJ in 2010, I concluded that we don't know whether topical NSAIDs are beneficial (3).

One of the authors, Andrew Moore, “reported receiving a grant and personal fees from Reckitt Benckiser and personal fees from Menarini. No other disclosures were reported” (1). However, in 2015, Moore published another systematic review of NSAIDs where his competing interests, in addition to those declared in JAMA, were: “personal fees from Novartis, grants and personal fees from Grunenthal, personal fees from Orion Pharma, personal fees from Futura Pharma, personal fees from Astellas, personal fees from Eli Lilly, personal fees from Pfizer and personal fees from Menarini” (4).

1 DerryS, Wiffen P, Moore A. Topical nonsteroidal anti-inflammatory drugs for acute musculoskeletal pain. JAMA 2016;315:813-4.

2 Derry S, Moore RA, Gaskell H, McIntyreM, Wiffen PJ. Topical NSAIDs for acute musculoskeletal pain in adults. Cochrane Database Syst Rev. 2015;6:CD007402.

3 Gøtzsche PC. NSAIDs. Clin Evid (Online). 2010 Jun 28.

4 Moore RA, Derry S, Wiffen PJ, Straube S. Effects of food on pharmacokinetics of immediate release oral formulations of aspirin, dipyrone, paracetamol and NSAIDs - a systematic review. Br J Clin Pharmacol 2015;80:381-8.

On 2016 Apr 04, R Andrew Moore commented:

Do topical nonsteroidal anti-inflammatory drugs for acute musculoskeletal pain work?

It is good to see Peter taking an interest in another of our publications.

The evidence shows that topical NSAIDs can be effective for acute musculosketal pain. That was the result we found in our first review of the topic (BMJ 1998;316:333–8), and it has not changed in subsequent updates. Even 20 years ago we actively examined the issue of study size, and concluded that small studies tended to overestimate treatment effects, as well as testing other issue of quality, and of the particular topical NSAID tested. In the current update we were able to add formulation to the list of topics that might affect study results.

Peter dismisses the results for a number of reasons, we think incorrectly. For example, the only risk of bias measure where there was potentially high risk of bias was small study size, a measure included in our risk of bias assessments rather than ignored, as is so often the case. On the issue of industry funding affecting the results of randomised trials, not only is there no evidence of any such effect, but there is positive evidence that there is no effect. We showed the lack of any effect in analgesic trials a decade ago (Pain 2006;121:207-18), and no effect was found for statins (BMJ 2014;349:g5741). That does not mean that industry has clean hands, of course, and we have pointed out, for example, the biases that might arise from multiple publication (BMJ 1997;315:635-40) or inappropriate imputation methods (Pain. 2012;153:265-8).

The review did include a number of older studies of relatively poorer quality, involving nine NSAIDs (other than diclofenac and ketoprofen) but we make clear in the review that there were insufficient data of adequate quality to draw any conclusions about these. Studies were underpowered for adverse events, which were inconsistently reported, but we have again drawn attention to these limitations, which are common in many clinical trials.

The review is not sui generis, but an ongoing dynamic of updated reports over the years. For example we wrote to 88 pharmaceutical companies for our 2004 update (BMC Family Practice 2004;5:10), with only one providing otherwise unpublished data. Topical NSAIDs (like most drugs in pain) are usually generic, with no requirement to perform clinical testing, or older, when it is much more difficult to obtain CTRs, as we have frequently been able to do previously. Experience suggested that trying to get hold of CTRs from hundreds of companies worldwide, many of which had done no trials on their product, was probably a lost cause. In this update we were able to identify that trial data from almost 6,000 patients was not available; a known unknown rather than an unknown unknown.

Of course there was heterogeneity for all formulations together, because different formulations produced different levels of efficacy (and probably with different doses as well, though that was difficult to assess). But within formulations like Flector plaster or Emulgel the I2 was 0; Figures 5 and 6 in the Cochrane review make the point. There are arguments to have about the use of heterogeneity tests, but this is not the place.

In terms of declarations, the issues here are of time, what it is that journals want, and relevance. Both RB and Menarini have topical NSAID products. RAM has worked with both those companies, and while that did not involve topical products it is a relevant disclosure. Futura Pharma dropped out of the three year period for declarations. None of the others were relevant, and for the most part involved investigator-initiated research using patient-level data to elucidate issues around evidence in analgesic trials, for example the importance of formulation (Pain 2014;155:14-21; Eur J Pain 2015;19:187-92) or how best to conduct multiple dose studies in acute pain (Br J Anaesth. 2016;116:269-76; BMC Anesthesiol 2016;16:9).

There is much that could not be done without constructive involvement with pharmaceutical companies, as they produce the trial data on which we base our evidence. Understanding that evidence thoroughly is what it is all about. But we think there is confusion over declarations on interest and what is required, and are actively working on just that topic.