Paracetamol versus nonsteroidal anti‐inflammatory drugs for rheumatoid arthritis

Abstract

Background

Nonsteroidal anti‐inflammatory drugs (NSAIDs) are usually preferred for simple analgesics such as paracetamol for rheumatoid arthritis. It is not clear, however, whether the trade‐offs between benefits and harms of NSAIDs are preferable to those of paracetamol (paracetamol is also called acetaminophen).

Objectives

To compare the benefits and harms of paracetamol with NSAIDs in patients with rheumatoid arthritis.

Search methods

PubMed and EMBASE databases were searched up until August 2007. Reference lists of identified articles were also searched.

Selection criteria

Randomised double‐blind studies comparing paracetamol with an NSAID.

Data collection and analysis

Decisions on inclusion of trials and data extraction were performed by the two authors independently.

Main results

Four cross‐over studies, published between 1968 and 1982, involving 121 patients, and four different NSAIDs were included. The generation of the allocation sequence and the use of methods to conceal the allocation were not described in any of the studies. The studies were double‐blind but it was not clear whether the blinding was effective. Methods for collecting adverse effects were not described. The NSAIDs were preferred more often than paracetamol by the patients or the investigator. In the largest trial, 20 out of 54 patients (37%) preferred ibuprofen and 7 out of 54 (13%) paracetamol. Investigators preference (as established by joint tenderness, grip strength and joint circumference) was 17 out of 35 for diclofenac versus 5 out of 35 for paracetamol in another trial. However, because of the weaknesses in the trials, no firm conclusion can be drawn.

Authors' conclusions

When considering the trade off between the benefits and harms of non‐steroidal anti‐inflammatory drugs and paracetamol/acetaminophen, it is not known whether one is better than the other for rheumatoid arthritis. But people with rheumatoid arthritis and the researchers in the study did prefer non‐steroidal anti‐inflammatory drugs more than acetaminophen/paracetamol. There is a need for a large trial, with appropriate randomisation, double‐blinding, test of the success of the blinding, and with explicit methods to measure and analyse pain and adverse effects.

Plain language summary

Paracetamol versus nonsteroidal anti‐inflammatory drugs for rheumatoid arthritis

In rheumatoid arthritis (RA), the joints are swollen, stiff and painful. Nonsteroidal anti‐inflammatory drugs (NSAIDs) such as ibuprofen are often recommended to ease the pain and swelling in the joints. Paracetamol (also known as acetaminophen) is another type of medication to relieve pain in RA.

This Cochrane review found only four old and small trials of poor quality that have compared the two types of drugs. There were a total of 121 patients in the four trials. In each trial, the patients tried both types of drugs, one after the other, in different periods of the trial.

In the largest trial, of 54 patients, where each drug was tested twice, 20 patients preferred ibuprofen on both occasions, and 7 paracetamol.

In the trials, each drug was used for only 4‐7 days and side effects from the drugs were poorly reported. It is therefore not clear whether NSAIDs are better than paracetamol.

Background

Rheumatoid arthritis is the most common inflammatory arthritis, affecting about one percent of the general population (Wolfe 1968). Many patients with rheumatoid arthritis want pain relief above all else (Gibson 1985). Nonsteroidal anti‐inflammatory drugs (NSAIDs) are usually preferred for simple analgesics such as paracetamol. It is not clear, however, what the trade‐offs between benefits and harms are. Analgesic agents with little or no anti‐inflammatory activity might be less effective than NSAIDs, but they might also have less side effects. Recently, a randomised double‐blind trial found no statistically significant difference or clinical relevant difference in the mean reduction in pain score when treated with either nonsteroidal antiinflammatory drugs, paracetamol or a combination of both for treating pain after musculoskeletal injury (Woo 2005).

Paracetamol (acetaminophen) is the active metabolite of phenacetin and is responsible for its analgetic effect. It is a weak prostaglandin inhibitor in peripheral tissues (Katzung 1998). NSAIDs appear to offer mainly symptomatic relief. In animal models they reduce inflammation and the pain it causes, but have doubtful effect on the progression of bone and cartilage destruction (Katzung 1998).

Objectives

To compare paracetamol with NSAIDs in patients with rheumatoid arthritis.

Methods

Criteria for considering studies for this review

Types of studies

Randomised, double‐blind clinical trials of at least 3 days duration on each drug were included.

Types of participants

Patients with a diagnosis of rheumatoid arthritis according to the criteria of the American College of Rheumatology (Arnett 1988), or similar (Ropes 1958).

Types of interventions

Experimental: Paracetamol. 
 Control: Any NSAID.

Types of outcome measures

Primary outcome measure: Pain after treatment, preferably day‐time pain during activity. 
 Secondary outcome measures: Joint tenderness (often assessed as Ritchie's index) (Ritchie 1968) and grip strength.

Search methods for identification of studies

PubMed (1966 to August 2007) and EMBASE (1980 to August 2007) were searched: see Appendix 1 for full search strategy. 
 
 One author (TW) assessed the abstracts for identification of eligible studies, scanned reference lists from trials, and contacted principal investigators and manufacturers of paracetamol in Denmark to identify any unpublished trials.

Data collection and analysis

We assessed the studies for methodological quality independently based on previously published criteria (Schulz 1995a, Schulz 1995b): 
 a. Was the study described as a randomised and was the method of randomisation well described and appropriate? 
 b. Was the study described as double‐blind and was the method of double‐blinding described and appropriate? 
 c. Was there a description of withdrawals and dropouts and how did they influence on the result?

Data extraction and analysis 
 We intended to use standard methods for meta‐analysis, but this was not possible because of lack of the required data. We therefore describe the primary authors' findings for each trial. We extracted data on effects and adverse effects independently; disagreement was resolved by discussion.

Results

Description of studies

Seven randomised trials were identified. We excluded two trials because the duration of trial therapy was only 4 hours (Brooks 1982) and 6 hours (Hardin 1979), and another trial because it was not randomised but had used alternation (Hajnal 1959).

Four old and small studies involving a total of 121 patients were included. In three studies (Munno 1982; Solomon 1974; Solomon 1977), the criteria of the American Rheumatism Association for classical or definite rheumatoid arthritis (Ropes 1958) were fulfilled; in the fourth study (Thompson 1968), the diagnosis was not made according to formal criteria. The mean age was about 55 years (Munno 1982; Solomon 1974; Solomon 1977) and two thirds were females (Munno 1982; Solomon 1977). Duration of the disease was only reported in one study (Munno 1982), and disease severity was not reported in any of the studies.

All studies had a cross‐over design and three were designed as sequential analyses (Solomon 1974; Solomon 1977; Thompson 1968). The patients were treated for four days (Munno 1982) or seven days (Solomon 1974; Solomon 1977; Thompson 1968) in each period; in one trial, the patients were treated with each drug twice (Thompson 1968). There was no background treatment in one of the studies (Solomon 1974); in another (Munno 1982), patients continued their basic therapy with quinoline derivatives or gold salts, and in two studies (Solomon 1977; Thompson 1968), it was not stated if there was any background treatment.

The NSAIDs used were tolmetin (Munno 1982), diclofenac (Solomon 1974), bumadizone (Solomon 1977) and ibuprofen (Thompson 1968). In none of the studies was the dose of any of the trial drugs motivated by reference to dose‐response studies; it was therefore unclear if the drugs were given in equipotent doses as regards the analgesic effect.

Risk of bias in included studies

The generation of the allocation sequence and the use of methods to conceal the allocation were not described in any of the studies. All studies were double‐blind. In two studies (Thompson 1968; Munno 1982), identical capsules were used, and in two, the drugs had "identical form" (Solomon 1974; Solomon 1977). In two studies (Munno 1982; Thompson 1968) it was not clear whether there had been more patients randomised than those stated because there was no description of drop‐outs, or a statement that there were none.

Effects of interventions

Four cross‐over studies, involving 121 patients were included. In a study of 54 patients, where each drug was tested twice, 20 patients preferred ibuprofen on both occasions, and 7 paracetamol (Thompson 1968). In a study of 12 patients, 6 preferred tolmetin and 3 paracetamol (Munno 1982). In the other two studies, the investigator's preference was established by a global assessment of three criteria: Ritchie's articular index, grip strength and joint circumference. There were 17 such preferences for diclofenac and 5 for paracetamol in a trial of 35 patients (Solomon 1974), and 7 preferences for bumadizone and none for paracetamol in a trial of 20 patients (Solomon 1977).

In one trial (Solomon 1974), there were no drop‐outs because of adverse effects on diclofenac but 6 on paracetamol; gastrointestinal symptoms were reported in one versus 6 patients. In another trial, there were 13 patients out of 53 with side effects on bumadizone (summary results for three trials) and 7 out of 20 on paracetamol (Solomon 1977). In a trial of 54 patients, there were no side effects (Thompson 1968), and in a trial of 12 patients, one patient tolerated paracetamol suppositories badly (Munno 1982).

Discussion

The included trials indicated that NSAIDs are more effective than paracetamol in the chosen doses. However, the trials have important methodological weaknesses and various biases could have occurred. The randomisation methods were not described and it has been shown that such trials overestimate the effect by about 20%, on average, when the effect in measured as an odds ratio (Pildal 2007). In none of the trials was it described in detail how double‐blinding was ensured. In two of the trials, identical capsules were used (Munno 1982; Thompson 1968), but capsules can easily be opened by the patients and the identity of the trials drugs could therefore have been revealed. In the other two trials, no information was offered on the blinding method (Solomon 1974; Solomon 1977). Trials that are not effectively double‐blinded may yield larger estimates of effects (Pildal 2007), especially for subjective outcomes as in the trials we reviewed. Further, the trials were very small and significant results from small trials are often wrong (Peto 1976). Finally, there was no description of the methods used to collect adverse effects.

In two trials (Solomon 1974; Solomon 1977), the examiner evaluated the results and decided whether or not the data should be included in the statistical analysis. This is a bias‐prone procedure, in particular because there was no clear rule as to how the data should be combined to give an overall "investigator's preference". Finally, there was underreporting; many more variables were measured than those accounted for in the papers, and it has been shown that variables are more often reported fully when they favour the treatment of interest (Chan 2004). We chose pain as the primary outcome measure beforehand because, apart from a global evaluation of the effect, it is not only the most meaningful variable when pain‐relieving drugs are assessed, it is also the most sensitive one (Gøtzsche 1990). Pain was measured in two studies but there were no results (Solomon 1974; Solomon 1977).

All trials had titles that were quite uninformative for the purpose of this review. Three trials (Munno 1982; Solomon 1974; Solomon 1977) were conducted with the purpose of motivating patients to use a new specific NSAID for rheumatoid arthritis.

It was not clear why the authors selected the doses they did. In one study where six patients dropped out because of adverse effects while they were on paracetamol, the drug was given in a dose of 6 g daily which is more than recommended. In a more recent study, in osteoarthritis (Bradley 1991), number of patients with side effects were very similar with 4 g paracetamol (19 of 61 patients), as with ibuprofen (16 of 62 patients on 1200 mg, and 24 of 61 patients on 2400 mg), which, in addition, is better tolerated than most other NSAIDs (Gøtzsche 2002).

A weakness of the cross‐over design is the possibility of carry‐over and period effects. A further weakness is that a sequential analysis can only use preference for one or the other drug and will discard neutral preferences. This may give a false impression that one drug is far better than another. In one trial, for example (Thompson 1968), there were 53 included patients, but only 27 patients gave a preference for either of the drugs.

It is often thought that NSAIDs have an inherent advantage over paracetamol in patients with rheumatoid arthritis because of their anti‐inflammatory effect. However, such effects have only been demonstrated in animal experiments. In a meta‐analysis, joint size measured with jeweller's rings was very similar with NSAIDs as with placebo (Gøtzsche 1990). It is therefore not clear whether the effect of NSAIDs in rheumatoid arthritis and other musculoskeletal disorders is anything more than a pain‐relieving effect.

Authors' conclusions

Implications for practice.

It is not clear whether the trade‐offs between benefits and harms of NSAIDs are preferable to those of paracetamol.

Implications for research.

There is a need for a large trial, with appropriate randomisation, double‐blinding, test of the success of the blinding, and with explicit methods to measure and analyse pain and adverse effects, preferably with more than two arms, allowing more than one dose on each drug.

What's new

Date	Event	Description
7 July 2008	Amended	Converted to new review format. CMSG ID: C062‐R

History

Protocol first published: Issue 3, 2002
 Review first published: Issue 1, 2004

Date	Event	Description
26 November 2003	New citation required and conclusions have changed	Substantive amendment

Appendices

Appendix 1. Search strategy for PubMed and EMBASE

Search strategy: 
 No. Request 
 1 rheumatoid arthritis 
 2 paracetamol 
 3 acetaminophen 
 4 #2 or #3 
 5 #1 or #4

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Munno 1982.

Methods	Generation of allocation: NS. Concealment: NS. Blinding: Double‐blind (identical capsules and suppositories). Analysis blinded: NS. Cross‐over design. Wash‐in: 4 d.. Wash‐out: 3 d.. Intention to treat: NS. Duration of treatment: 4 days.
Participants	N = 12. Excluded: NS. Not clear whether there were more than 12 patients as there was no information on drop‐outs.
Interventions	Experimental: Paracetamol 2 g/day. Control: Tolmetin 1200 mg/day. One capsule twice daily, one suppository at bedtime.
Outcomes	Primary: Grip strength. Secondary: Ritchies index. ESR. Any side effect. Patient preference.
Notes	Grip strength measurements based on only 9 patients. Six patients preferred tolmetin, 3 paracetamol, 3 had no preference. "Bad tolerability" for paracetamol suppositories in one patient. Tolmetin grip strength values at baseline 1501.50 mmHg (SE 172.00) and final 1851.44 mmHg (SE 172.77), % variation compared to baseline: +23.3. Paracetamol grip strength values at baseline 1430.28 mmHg (SE 163.72) and final 1423.44 mmHg (SE 170.70), % variation compared to base‐line: ‐0.5.
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Unclear risk	B ‐ Unclear

Solomon 1974.

Methods	Generation of allocation: NS. Concealment: NS. Blinding: Double‐blind ("identical form"). Analysis blinded: NS. Sequential analysis. Cross‐over design. Wash‐in: Yes, length not stated. Wash‐out: No. Intention to treat: NS. Duration of treatment: 7 days.
Participants	N = 35. Excluded: NS
Interventions	Experimental: Paracetamol 6 g/day. Control: Diclofenac 150 mg/day.
Outcomes	Primary: Investigator's preference established by a simple majority of the 3 principal criteria: Ritchie articular index. Grip strength. Joint circumference. Secondary: Degree of pain. Degree of early morning stiffness. Ability to walk. Global feeling of well‐being.
Notes	17 preferences for diclofenac, 5 for paracetamol. There were 6 drop‐outs because of adverse effects associated with paracetamol and 3 drop‐outs for reasons not related to the trial. Side effects in 6 patients on diclofenac, and in 11 on paracetamol; gastrointestinal symptoms in 1 vs 6. Data: NS.
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Unclear risk	B ‐ Unclear

Solomon 1977.

Methods	Generation of allocation: NS. Concealment: NS. Blinding: Double‐blind ("identical form"). Analysis blinded: NS. Sequential analysis. Cross‐over design. Wash‐in: 1 d. Wash‐out: No. Intention to treat: NS. Duration of treatment: 7 days.
Participants	N = 20. Exclusions: NS. No drop‐outs.
Interventions	Experimental: Paracetamol 6 g/day. Control: Bumadizone calcium 660 mg/day.
Outcomes	Primary: Investigator's preference established by a global assessment of the 3 principal criteria: Ritchie articular index. Grip strength. Joint circumference. Secondary: Pain. Degree of morning stiffness. Ability to walk. Global feeling of well‐being.
Notes	7 preferences for bumadizone, none for paracetamol. Side effects of bumadizone only stated for all patients in three trials; there were 13 of 53 patients on bumadizone and 7 of 20 on paracetamol. Data: NS.
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Unclear risk	B ‐ Unclear

Thompson 1968.

Methods	Generation of allocation: NS. Concealment: NS. Blinding: Double‐blind (capsules). Analysis blinded: No. Sequential analysis. Four period cross‐over design. Wash‐in: 7 d.. Wash‐out: No. Intention to treat: NS. Duration of treatment: 7+7 days for each drug.
Participants	N = 54. Excluded = 1. Not clear whether there were more than 54 patients as there was no information on drop‐outs.
Interventions	Experimental: Paracetamol 3 g/day. Control: Ibuprofen 600 mg/day.
Outcomes	Primary: Patient preference. Secondary: Grip strength.
Notes	Each patient received each drug twice. One patient withdrawn because of lack of compliance. "No side effects". 20 patients preferred ibuprofen in both periods, 7 paracetamol. No data on grip (35 patients of 50 patients had better grip with ibuprofen, but small differences). Data: NS.
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Unclear risk	B ‐ Unclear

NS: not stated.

Characteristics of excluded studies [ordered by study ID]

Study	Reason for exclusion
Brooks 1982	Duration of trial therapy only 4 hours.
Hajnal 1959	Duration of trial therapy only 6 hours.
Hardin 1979	Not a randomised trial; alternation was used.

Contributions of authors

Both reviewers contributed to the protocol and the review. TW assessed the abstracts for eligible studies, scanned reference lists from trials, and contacted principal investigators and manufacturers of paracetamol in Denmark to identify unpublished trials. TW wrote first drafts of the protocol and the review. Guarantors: both authors.

Sources of support

Internal sources

• Rigshospitalet, Denmark.

External sources

• No sources of support supplied

Declarations of interest

None known.

Edited (no change to conclusions)