The article that ditched paracetamol in osteoarthritis (Part 2)

Early this year, many colleagues were surprised by headlines concerning paracetamol in osteoarthritis. Many started to ask, “Does paracetamol really not work in osteoarthritis?”

The network meta analysis by da Costa et al (Lancet 2016) focused on the comparative efficacy of paracetamol and NSAIDs for osteoarthritis. Similar to the study by Bannuru et al (Ann Int Med 2015), the review again used effect size to compare the impact of interventions on pain and function. However, the population was expanded to include patients with hip osteoarthritis and separated effect size depending on drug dose. It also included only large scale studies (i.e. only trials with at least 100 patients per treatment group) to minimize bias resulting from small samples sizes. The paper included some drugs which had been pulled out of the market (rofecoxib, lumiracoxib) so this will not be discussed in this post.

The effect size of different drugs and drug doses (expressed as daily dose) on improving pain are as follows:

  • Paracetamol 2000mg -0.08 (-0.41, 0.27)
  • Paracetamol 3000mg -0.17 (-0.66, 0.32)
  • Paracetamol 4000mg -0.17 (-0.27, -0.06)
  • Diclofenac 70mg -0.22 (-0.61, 0.16)
  • Diclofenac 100mg -0.34 (-0.58, -0.10)
  • Diclofenac 150mg -0.57 (-0.69, -0.46)
  • Ibuprofen 1200mg -0.30 (-0.84, 0.25)
  • Ibuprofen 2400mg -0.43 (-0.55, -0.31)
  • Naproxen 750mg -0.05 (-0.42, 0.32)
  • Naproxen 1000mg -0.40 (-0.47, -0.33)
  • Celecoxib 100mg -0.16 (-0.29, -0.03)
  • Celecoxib 200mg -0.36 (-0.41, -0.32)
  • Celecoxib 400mg -0.33 (-0.45, -0.21)
  • Etoricoxib 30mg -0.49 (-0.61, -0.37)
  • Etoricoxib 60mg -0.58 (-0.73, -0.43)
  • Etoricoxib 90mg -0.61 (-0.90, -0.33)

Going over the confidence intervals, we see five preparations as having no significant effect in reducing pain – paracetamol 2000mg, paracetamol 3000mg, diclofenac 70mg, ibuprofen 1200mg and naproxen 750mg. Paracetamol 4000mg has a small effect size compared to oral placebo and it fell short of the study’s 0.37 cut off for meaningful change in pain (based on their explanation, the cut off would translate to a 9mm reduction in pain VAS on a 100mm scale). But if we have a hard time verifying this concept, we can still say that paracetamol (at 4000mg) is better than placebo. Going over the remaining NSAIDs and their doses, we see once again that diclofenac (at 150mg) is one of the more effective pain relievers. And, true to my previous theory, etoricoxib (at 60 and 90mg) produced pain relief of similar magnitude as diclofenac.

A dose response (the higher the dose, the better the pain relief) was seen only in diclofenac (100 vs 150mg), celecoxib (100 vs 200mg) and naproxen (750mg vs 1000mg). But given that the effect was not significant at the lower dose of naproxen, it was only in diclofenac and celecoxib where a relevant dose response was observed. The analysis for etoricoxib showed no significant dose response at its three doses – this supports the approved maximum recommended dose of the drug for OA which is 60mg/day.

So based on the above results, what are the appropriate doses of meds to be used in OA for pain relief.

  • Paracetamol 4000mg
  • Diclofenac 100mg/day with the option to increase to 150mg.
  • Ibuprofen 2400mg/day
  • Naproxen 1000mg/day
  • Celecoxib 100mg/day with the option to increase to 200mg (There was no difference in the effect size between the 200 and 400mg preparation)
  • Etoricoxib 60mg

Of these, paracetamol and celecoxib 100mg produced small differences in pain. Diclofenac 100mg, celecoxib 200mg and naproxen produced moderate (~10mm change) improvements in pain. While, diclofenac 150mg and etoricoxib produced large (~20mm change) improvements in pain.

Going over the effect sizes for function, the same drugs at the same doses were observed to produce significant changes. And similarly, the magnitude of improvement in function mirrors what was seen in improvements observed in pain relief.

So did the results of da Costa’s analysis on paracetamol and NSAIDs differ from Bannuru’s review? In one word, NO. Despite differences in the study population (disease target, population size and some trials included), the estimates of effect size were very close for all agents in question. Go back to part 1 of this series and see that differences in the effect size were less than 0.05 and the magnitude of effect size for each drug was similar.

But how does this newer study add to our understanding of OA treatment? Bannuru et al was the first to put a hierarchy on the efficacy of treatments for OA. da Costa et al confirmed this hierarchy and added the appropriate doses for using these agents. It would be interesting to note how guidelines would change (if ever) after these two studies.

Effectiveness of non-steroidal anti-inflammatory drugs for the treatment of pain in knee and hip osteoarthritis: a network meta-analysis.


Author: Sids Manahan MD 🇵🇭

Rheumatology. Internal Medicine. Educating Patients and Colleagues. Curating Rheumatology. Bloggero-Wanabe.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s