Description of the condition
Acute pain occurs as a result of tissue damage either accidentally due to an injury or as a result of surgery. Acute postoperative pain is a manifestation of inflammation due to tissue injury. The management of postoperative pain and inflammation is a critical component of patient care and is important for cost-effective use of healthcare resources. Good postoperative pain management helps to achieve a satisfied patient who is in hospital or at home and unable to carry out normal activities for a minimal amount of time.
Description of the interventions
Analgesics used for relief of postoperative pain include so called 'mild' or step 1 (WHO 2010) analgesics, such as paracetamol, and non-steroidal anti-inflammatory drugs (NSAIDs) such as ibuprofen and celecoxib, 'moderate' or step 2 analgesics, which are weaker opioids such as codeine, and 'strong' or step 3 analgesics, which are strong opioids such as oxycodone and fentanyl.
Paracetamol has become one of the most used antipyretic and analgesic drugs worldwide, and is often also used in combination with other stronger analgesics. NSAIDs as a class are the most commonly prescribed analgesic medications worldwide and their efficacy for treating acute pain has been well demonstrated (Moore 2003). Opioids as a class have long been used to treat pain during and immediately after surgery, because they can be given parenterally, and because dose can be titrated to effect for immediate pain relief. Oral opioids are less often used alone, but are used in fixed-dose combination with drugs like paracetamol or ibuprofen (McQuay 1997).
This overview will consider only oral administration of analgesics. Parenteral administration by intravenous, intramuscular, or subcutaneous injections is useful for some drugs immediately following surgery, particularly for patients unable to swallow or where oral intake is not possible for other reasons (McQuay 1997). Most postoperative patients can swallow and oral administration is clearly the least technically demanding and cheapest method of drug delivery, especially when the benefits of injection over oral administration have not been demonstrated, as with NSAIDs (Tramer 1998).
Acute pain trials
Postoperative (after surgery) pain relief is part of a package of care that covers the preoperative (before surgery), perioperative (during surgery), and postoperative periods and involves using the best evidence at all times (Kehlet 1998). This overview involves only one aspect of one part of the patient journey, namely how well different oral drug interventions work to relieve pain. The choice of particular oral drug intervention depends on the clinical and operational circumstances and how any choice fits in with local care pathways. This overview only examined the efficacy of oral drugs: how to use them effectively in the relief of postoperative pain is a question not addressed here.
Clinical trials measuring the efficacy of analgesics in acute pain have been standardised over many years. To show that the analgesic is working, it is necessary to use placebo (McQuay 2005; McQuay 2006). There are clear ethical considerations in doing this. These ethical considerations are answered by using acute pain situations where the pain is expected to go away, and by providing additional analgesia, commonly called rescue analgesia, if the pain has not diminished after about an hour. This is reasonable, because not all participants given an analgesic will have acceptable pain relief. Approximately 18% of participants given placebo will have adequate pain relief (Moore 2006), and up to 50% may have inadequate analgesia with active medicines. Therefore the use of additional or rescue analgesia is important for all participants in the trials.
Trials have to be randomised and double-blind. Typically, in the first few hours or days after an operation, patients develop pain that is moderate to severe in intensity, and will then be given the test analgesic or placebo. Pain is measured using standard pain intensity scales immediately before the intervention, and then using pain intensity and pain relief scales over the following four to six hours for shorter-acting drugs, and up to 12 or 24 hours for longer-acting drugs. Half the maximum possible pain relief or better over the specified time period (at least 50% pain relief) is typically regarded as a clinically useful outcome. For patients given rescue medication it is usual for no additional pain measurements to be made, and for all subsequent measures to be recorded as initial pain intensity or baseline (zero) pain relief (baseline observation carried forward). This process ensures that analgesia from the rescue medication is not wrongly ascribed to the test intervention. In some trials the last observation is carried forward, which gives an inflated response for the test intervention compared to placebo, but the effect has been shown to be negligible over four to six hours (Moore 2005). Patients usually remain in the hospital or clinic for at least the first six hours following the intervention, with measurements supervised, although they may then be allowed home to make their own measurements in trials of longer duration.
Important characteristics of an analgesic include the proportion of patients who experience clinically useful levels of pain relief at a given dose, the duration of useful pain relief (which informs dosing intervals), and the drug's tolerability. Single dose studies can provide us with information on the number needed to treat (NNT) for at least 50% maximum pain relief over four to six hours compared with placebo and the proportions of participants achieving that outcome, the NNT to prevent (NNTp) use of rescue medication and the proportions needing rescue medication, the median (or mean) time to use of rescue medication, and the number needed to harm (NNH) for one or more adverse events, and the proportions experiencing adverse events. Additional information may also be available for the occurrence of serious adverse events and adverse event withdrawals, although the numbers of events captured in single dose trials are usually too few to allow statistical analysis.
How the intervention might work
Non-steroidal anti-inflammatory drugs
NSAIDs reversibly inhibit the enzyme cyclooxygenase (prostaglandin endoperoxide synthase or COX), now recognised to consist of two isoforms, COX-1 and COX-2, mediating production of prostaglandins and thromboxane A2 (Fitzgerald 2001). Prostaglandins mediate a variety of physiological functions such as maintenance of the gastric mucosal barrier, regulation of renal blood flow, and regulation of endothelial tone. They also play an important role in inflammatory and nociceptive (pain) processes. However, relatively little is known about the mechanism of action of this class of compounds aside from their ability to inhibit cyclooxygenase-dependent prostanoid formation (Hawkey 1999). Aspirin is a special case, in that it irreversibly blocks COX-1.
Paracetamol lacks significant anti-inflammatory activity, implying a mode of action distinct from that of NSAIDs. Despite years of use and research, however, the mechanisms of action of paracetamol are not fully understood. Paracetamol has previously been shown to have no significant effects on COX-1 or COX-2 (Schwab 2003), but has recently been considered as a selective COX-2 inhibitor (Hinz 2008). Significant paracetamol-induced inhibition of prostaglandin production has been demonstrated in tissues in the brain, spleen, and lung (Botting 2000; Flower 1972). A 'COX-3 hypothesis' wherein the efficacy of paracetamol is attributed to its specific inhibition of a third cyclooxygenase isoform enzyme, COX-3 (Botting 2000; Chandrasekharan 2002; PIC 2008) now has little credibility and a central mode action of paracetamol is thought to be likely (Graham 2005).
Opioids bind to specific receptors in the central nervous system (CNS), causing reduced pain perception and reaction to pain, and increased pain tolerance. In addition to these desirable analgesic effects, binding to receptors in the CNS may cause adverse events such as drowsiness and respiratory depression, and binding to receptors elsewhere in the body (primarily the gastrointestinal tract) commonly causes nausea, vomiting, and constipation. In an effort to reduce the amount of opioid required for pain relief, and so reduce problematic adverse events, opioids are commonly combined with non-opioid analgesics, such as paracetamol.
Why it is important to do this overview
Knowing the relative efficacy of different analgesic drugs at various doses, under standard conditions, can be helpful. Choice of drug for an individual patient will depend on relative efficacy and a number of other factors including availability, cost, tolerability, and individual considerations, such as the patient's history and contraindications to a particular medication, and their ability to remedicate orally. A large number of systematic reviews of individual oral analgesics versus placebo in acute postoperative pain have been completed, using identical methods. An overview is required to facilitate indirect comparisons between individual analgesics, providing estimates of relative efficacy which can help to inform treatment choices.
To provide an overview of the relative analgesic efficacy of oral analgesics that have been compared with placebo in acute postoperative pain in adults, and to report on adverse events associated with single doses of these analgesics. This will be done using a number of different outcomes and ways of expressing results, which have been set by informed discussions with various groups of healthcare professionals, and using reviews newly published or updated Cochrane Reviews that incorporate these methods to give the best presentation of results.
Criteria for considering reviews for inclusion
All Cochrane Reviews of randomised controlled trials (RCTs) of single dose oral analgesics for acute postoperative pain in adults (≥ 15 years).
Search methods for identification of reviews
We searched the Cochrane Database of Systematic Reviews for relevant reviews. See Appendix 1 for the search strategy. A series of Cochrane Reviews have been conducted by the same team, covering analgesics identified in the British National Formulary.
Data collection and analysis
Two review authors independently carried out searches, selected reviews for inclusion, carried out assessment of methodological quality, and extracted data. Any disagreements were resolved by discussion, involving a third review author if necessary.
Selection of reviews
Included reviews assessed RCTs evaluating the effects of a single oral dose of analgesic given for relief of moderate to severe postoperative pain in adults, compared to placebo, and included:
- a clearly defined clinical question;
- details of inclusion and exclusion criteria;
- details of databases searched and relevant search strategies;
- patient-reported pain relief; and
- summary results for at least one desired outcome.
Data extraction and management
We extracted data from the included reviews using a standard data extraction form. We used original study reports only if specific data were missing.
We collected information on the following:
- number of included studies and participants;
- drug, dose, and formulation (if formulation is an issue);
- pain model (dental, other surgical).
We sought relative risk (RR) and numbers needed to treat to benefit (NNT), to prevent an event (NNTp), and to harm (NNH) or calculated these for the following outcomes:
- ≥ 50% maximum pain relief over four to six hours (patient-reported): this outcome encapsulates both degree of pain relief and duration of the effect, and is a dichotomous measure of success over a defined period following drug ingestion;
- use of rescue medication (or mean or median if appropriate, for example for time to remedication);
- patients suffering one or more adverse events; and
- withdrawal due to an adverse event.
We also sought information on the proportions of individuals with the outcomes listed above, and median or mean time to use of rescue medication. We collected information concerning serious adverse events if present.
Assessment of methodological quality of included reviews
Quality of included reviews
All included reviews were carried out according to a standard protocol which satisfied the criteria specified in the 'assessment of multiple systematic reviews' (AMSTAR) measurement tool (Shea 2007) for rigorous methodological quality.
Each review was required to:
- provide an a priori design;
- carry out duplicate study selection and data extraction;
- carry out a comprehensive literature search;
- include published and unpublished studies irrespective of language of publication;
- provide a list of studies (included and excluded);
- assess and document the scientific quality of the included studies;
- use the scientific quality of the included studies appropriately in formulating conclusions;
- use appropriate methods to combine the findings of studies; and
- state conflicts of interests.
For each review we assessed the likelihood of publication bias by calculating the number of participants in studies with zero effect (relative benefit of one) that would be needed to give a NNT too high to be clinically relevant (Moore 2008). In this case we considered a NNT of ≥ 10 for the outcome 'at least 50% maximum pain relief over four to six hours' to be the cut-off for clinical relevance. We used this method because statistical tests for presence of publication bias have been shown to be unhelpful (Thornton 2000).
Quality of evidence in included reviews
All included reviews used only primary studies that were both randomised and double-blind, so minimising the risk of bias from these items. All used patients with at least moderate pain intensity at baseline, providing a sensitive assay of analgesic efficacy. All used standard methods and reported standard outcomes, or provided data from which they could be calculated using validated methods. For studies in acute pain lasting up to six hours, it has been shown that use of last observation carried forward rather than baseline observation carried forward does not significantly influence results (Moore 2005).
We assessed the strength of evidence for each outcome according to the total number of participants contributing data and the methodological quality of, and degree of clinical heterogeneity (pain condition mix) in, the primary studies, as reported in the reviews. We also considered the number of additional participants needed in studies with zero effect (relative benefit of one) required to change the NNT for at least 50% maximum pain relief to an unacceptably high level (in this case the arbitrary NNT of 10) (Moore 2008). Where this number was less than 400 (equivalent to four studies with 100 participants per comparison, or 50 participants per group), we considered the results to be susceptible to publication bias and therefore unreliable.
We used information on the selected efficacy outcomesto draw up comparisons of analgesic efficacy, using indirect comparison of different drugs from almost identical clinical trial conditions, with placebo as a common comparator (Glenny 2005; Song 2003). The trials used in these reviews have a high level of clinical and methodological homogeneity, having, for more than 50 years, used consistent validated methods of measuring pain in patients with established pain of at least moderate severity, over at least four to six hours, and with placebo as a common comparator. Some of these data have been used to demonstrate the superiority of indirect over direct comparison in circumstances where there are large amounts of indirect data and small amounts of direct data (Song 2003). The one potential source of clinical heterogeneity is the case mix, namely dental versus other surgery, and while this has previously been shown to have minimal effect on some descriptors, like NNT, it can result in differences in other descriptors, like percentage of participants obtaining an outcome (Barden 2004). This is addressed by examining results for dental and other surgery separately and together, where there are sufficient data. Any differences between different analgesics for harmful outcomes are highlighted, but single dose studies are not designed to reliably demonstrate such differences.
Comparative results are expressed as:
- patients achieving at least 50% maximum pain relief, as a percentage and as NNT, compared with placebo;
- duration of analgesia, as mean or median duration, and percentage remedicating by various times after dosing; and
- adverse events - given the nature of the studies, especially their short duration, the outcome most often reported was percentage reporting at least one adverse event.
The overview included 35 separate Cochrane Reviews investigating 38 analgesics or analgesic combinations given as single oral doses in acute postoperative pain conditions (Aceclofenac 2009; Acemetacin 2009; Aspirin 1999; Celecoxib 2008; Codeine 2010; Dexibuprofen 2009; Dextropropoxyphene ± Paracetamol 1999; Diclofenac 2009; Diflunisal 2010; Dihydrocodeine 2000; Dipyrone 2010; Etodolac 2009; Etoricoxib 2009; Fenbufen 2009; Fenoprofen 2011; Flurbiprofen 2009; Gabapentin 2010; Ibuprofen 2009; Indometacin 2004; Ketoprofen and Dexketoprofen 2009; Lornoxicam 2009; Lumiracoxib 2010; Mefenamic acid 2011; Meloxicam 2009; Nabumetone 2009; Naproxen 2009; Nefopam 2009; Oxycodone ± Paracetamol 2009; Paracetamol + Codeine 2009; Paracetamol 2008; Piroxicam 2000; Rofecoxib 2009; Sulindac 2009; Tenoxicam 2009; Tiaprofenic acid 2009). In total there were 448 studies, combining the number of studies in the individual reviews. However, many studies had both placebo and active comparators; ibuprofen, for example, was used as an active comparator in many of them. The number of unique studies was probably closer to 350.
All of the reviews used the same methodological approach and the same primary outcome of NNT for at least 50% maximum pain relief over four to six hours compared with placebo. The sum of the number of participants in the reviews was 50,456, but there will have been double-counting of placebo participants both within reviews (comparison of different drug doses separately against placebo) and between reviews (drugs like ibuprofen are commonly used as an active comparator for new test analgesics and placebo arms will be counted in reviews of both analgesics). In these circumstances the number of unique participants is more likely to be of the order of 45,000.
Description of included reviews
Included reviews each had the same structure and organisation, and used identical methods based on criteria established by extensive analysis and validation, using individual patient data (see Table 1). They all used the same criteria and typically these were as follows.
- Adult participants with established pain of at least moderate intensity (Collins 1997).
- Single dose oral administration of analgesic or placebo (with additional analgesia available, typically after 60 to 120 minutes).
- Randomised, double-blind studies.
- Pain assessed by patients using standard pain intensity and pain relief scales.
- Study duration of four hours or more.
- Searching included electronic searches, plus databases created by handsearching the older literature, now part of CENTRAL. Searching also included different retail names for drugs.
- No language restriction on included papers.
- Assessment of study quality according to established criteria and minimum criteria for inclusion.
Methodological quality of included reviews
All the reviews:
- had a priori design;
- performed duplicate study selection and data extraction;
- had a comprehensive literature search;
- used published and any unpublished studies included irrespective of language of publication, though not all reviews contacted companies or researchers for unpublished trial data;
- provided a list of included and excluded studies;
- provided characteristics of included studies;
- assessed and documented the scientific quality of the included studies;
- the scientific quality of the included studies was used appropriately in formulating conclusions, because only studies with minimal risk of bias were included (a particular issue was trial size, but conclusions were not drawn from inadequate data sets, based on previously established criteria (Moore 1998));
- used appropriate methods to combine findings of studies and importantly provided analyses according to drug dose; and
- conflict of interest statements were universal.
The reviews all used validated methods for conversion of mean to dichotomous data (Moore 1996; Moore 1997b; Moore 1997c), providing the number and proportion of participants with the clinically-relevant outcome of at least 50% maximum pain relief. Remedication is common within a few hours with placebo, therefore the method of imputing data after withdrawal is potentially of importance to the measurement of treatment effect. In the case of the primary outcome of the reviews, that of NNT for at least 50% maximum pain relief compared with placebo over four to six hours, the imputation method had been shown not to make any appreciable difference (Moore 2005), though use of last observation carried forward tended to overestimate treatment effect compared with baseline observation carried forward over longer periods (Moore 2005).
Effect of interventions
To assess the effects of interventions, we used a four-step process.
- Note drugs for which no acute pain data could be found.
- Note drug/dose combinations with inadequate amounts of information, where inadequate is defined as fewer than two studies and 200 participants - with limited flexibility around 200 participant limit).
- Note drug/dose combinations with data but no evidence of effect, or with evidence of no effect.
- Note drug/dose combinations with high susceptibility to publication bias, as defined in the Methods section.
Any remaining results would be of effective drug/dose combinations, backed by high-quality data not subject to bias, of sufficient size for random chance effects to be unimportant, and not susceptible to publication bias.
All extracted information on all reviews is available in Table 1.
1. Drugs for which Cochrane Reviews found no information
2. Drugs for which Cochrane Reviews found inadequate information (< 200 patients in comparisons, in at least two studies)
We found only limited information for six drugs, namely:
- Dexibuprofen 200 and 400 mg (176 participants with the two doses in one study) (Dexibuprofen 2009).
- Dextropropoxyphene 130 mg (50 participants in one study) (Dextropropoxyphene ± Paracetamol 1999).
- Diflunisal 125 mg (120 participants in two studies) (Diflunisal 2010).
- Etoricoxib 60 mg (124 participants in one study) (Etoricoxib 2009).
- Fenbufen 400 mg and 800 mg (46 participants with the two doses in one study) (Fenbufen 2009).
- Indometacin 50 mg (94 participants in one study) (Indometacin 2004).
3. Drugs for which Cochrane Reviews found no evidence of effect or evidence of no effect
There was evidence for lack of effect for three drug/dose combinations, with no difference between active drug and placebo:
- Aceclofenac 150 mg (217 participants in one study) (Aceclofenac 2009).
- Aspirin 500 mg (213 participants in two studies) (Aspirin 1999).
- Oxycodone 5 mg (317 participants in three studies) (Oxycodone ± Paracetamol 2009).
4. Drug/dose combinations for which Cochrane Reviews found evidence of effect, but where results were potentially subject to publication bias
Summary table A shows the drug/dose combinations in all types of surgery, and in dental and other postoperative pain situations separately, where our judgement was of high susceptibility to publication bias. These tended to have larger (less effective) NNTs, small numbers of participants, or both. The appropriateness or otherwise of this categorisation is discussed below, but these results are the least reliable of those available from the reviews. For gabapentin, the NNT was above 10, and based on a relatively small number of participants.
Summary table A: Results potentially subject to publication bias
5. Drug/dose combinations for which Cochrane Reviews found evidence of effect, where results were reliable and not subject to potential publication bias
Reliable results are presented by pain condition for the primary outcome of NNT compared with placebo for at least 50% maximum pain relief over four to six hours: firstly all types of surgery together, then dental pain only, and finally by other painful conditions. The results contain all available data. Some of the data are from doses of drugs not typically used clinically, such as 180/240 mg etoricoxib or ibuprofen 100 mg, or from drugs not commonly available in many parts of the world, like rofecoxib. All data are presented so that readers can use that which is relevant for them.
All types of surgery
For all types of surgery, the results judged to be reliable are shown in Summary table B. Overall, about 45,000 participants contributed data. For lornoxicam 4 mg only 151 participants from two studies provided data, but more than 400 participants would have been needed in zero effect studies to overturn the result; our judgement was that this result was on the borderline of being reliable. For codeine 60 mg, although the NNT was above 10, it was based on over 2400 participants and we deemed that a reliable result.
The number of participants was high (above 2000) with ibuprofen 400 mg and 200 mg, aspirin 600/650 mg, paracetamol 975/1000 mg, and rofecoxib 50 mg. Results with high numbers of participants and low (good) NNTs were particularly robust, with almost 20,000 participants needed in zero effect studies to overturn the result for ibuprofen 400 mg and over 13,000 to overturn that for rofecoxib 50 mg.
NNTs varied from as low as 1.5 for high doses of etoricoxib, to as high as 12 for codeine 60 mg. The majority of drug/dose combinations had NNTs below 3. A listing by rank order is shown in Figure 1. Higher doses of the same drug tended to have lower (better) NNTs, though this was not particularly evident with paracetamol.
|Figure 1. All types of surgery: NNT for at least 50% maximum pain relief over four to six hours compared with placebo, by rank order.|
Summary table B: Results judged to be reliable in all types of surgery
In practice this means almost exclusively the third molar extraction model, with minor differences in the number of teeth removed, and the extent of any bone involvement during surgery. Results judged to be reliable are shown in Summary table C; overall, about 29,000 participants contributed data.
For etodolac 400 mg, and ketoprofen 50 and 100 mg, fewer than 200 participants provided data, but many more than 400 participants would have been needed in zero effect studies to overturn the result; our judgement was that this result was on the borderline of being reliable. For codeine 60 mg, although the NNT was above 10, it was based on over 1146 participants and we deemed that a reliable result.
The number of participants was high (above 2000) with ibuprofen 400 mg and 200 mg, aspirin 600/650 mg, paracetamol 975/1000 mg, and rofecoxib 50 mg. Results with high numbers of participants and low (good) NNTs were particularly robust, with about 18,000 participants needed in zero effect studies to overturn the result for ibuprofen 400 mg, and over 13,000 to overturn that for rofecoxib 50 mg.
NNTs varied from as low as 1.5 for high doses of etoricoxib to as high as 21 for codeine 60 mg. The majority of drug/dose combinations had NNTs below 3. A listing by rank order is shown in Figure 2. Higher doses of the same drug tended to have lower (better) NNTs, though this was not particularly evident with paracetamol.
|Figure 2. Dental pain: NNT for at least 50% maximum pain relief over four to six hours compared with placebo, by rank order.|
Both Summary of results C and Figure 2 give all results for a particular dose of a particular drug, irrespective of drug formulation. There can be important differences between formulations, and examples of this are shown in Summary table C for sodium and potassium salts of diclofenac, and soluble and standard formulations of ibuprofen. These results show that, based on reasonable and reliable evidence, formulation has a major impact on efficacy in acute pain for diclofenac (Diclofenac 2009) and ibuprofen (Ibuprofen 2009).
Summary table C: Results judged to be reliable in painful dental conditions
Other painful conditions
This grouping included all acute postoperative pain that is not dental; it includes conditions like episiotomy, orthopaedic, and abdominal surgery, where the pain is of at least moderate in intensity and oral analgesics are indicated. There were insufficient data to allow further subgrouping according to type of surgery. Results judged to be reliable are shown in Summary table D; overall, about 7000 participants contributed data.
For diflunisal 500 mg fewer than 200 participants provided data, but more than 400 participants would have been needed in zero effect studies to overturn the result; our judgement was this result was on the borderline of being reliable.
The number of participants was above 1000 with aspirin 600/650 mg, ibuprofen 400 mg, and paracetamol 975/1000 mg. NNTs varied from as low as 2.1 for dipyrone 500 mg and flurbiprofen 50 mg to as high as 5.6 with paracetamol 1000. A listing by rank order is shown in Figure 3. Higher doses of the same drug tended to have lower (better) NNTs, though this was not particularly evident with paracetamol or ibuprofen.
|Figure 3. Other painful conditions: NNT for at least 50% maximum pain relief over four to six hours compared with placebo, by rank order.|
Summary table D: Results judged to be reliable in other painful conditions
6. Percentage of patients achieving target of at least 50% maximum pain relief
These results are described in Summary tables B, C, and D for each drug/dose combination. There was very wide variation between drugs even in the same painful condition, and where there were consistent responses with placebo. Figure 4 shows that in dental pain, while some drugs achieved a high level of pain relief in over 60 to 70% of participants, in others it was as low as about 30%. The response with placebo in dental pain averages about 10% to 15%, but tends to be higher in other surgical conditions.
|Figure 4. Percentage of patients achieving at least 50% maximum pain relief (dental pain).|
7. Time to remedication
A number of reviews reported the mean of the mean or median time to remedication, a useful secondary outcome indicating the duration of effective analgesia before the pain intensifies to the point where additional analgesia is required. For placebo, averaging over all reviews, the mean time to remedication is two hours; trials typically have a one to two-hour period before which additional analgesia is not allowed, to allow time for any analgesic to work. For active drugs in dental pain, the mean duration varied between below three hours for codeine 60 mg and oxycodone 5 mg, up to 20 hours for etoricoxib 120 mg (Figure 5; Appendix 3).
|Figure 5. Mean time to remedication in painful dental conditions.|
8. Percentage remedicated with time
We collected information on the percentage of patients who had remedicated with active treatment and placebo at various times after the start of therapy and this is reported in Appendix 3. This was sparsely reported in a small subsection of studies. In brief, typically 70% to 90% of participants given placebo had used rescue medication by six hours, and this tended to increase further at longer durations, though it never reached 100%. With analgesics, the numbers remedicating at six hours were always lower than with placebo.
9. Experience of adverse events
Adverse event reporting in acute pain studies is known to be heavily influenced by the methods used (Edwards 2002). Most reviews reported no serious adverse events and the only common report was that of participants experiencing at least one adverse event during the period of the study. These results are shown in Summary table E. The usual finding was no difference in adverse event rates between active and placebo groups (Figure 6). Statistical differences were found only for aspirin 600/650 mg (NNH 44), codeine + paracetamol 60/650 mg (NNH 6.0), diflunisal 1000 mg (NNH 7.7), dihydrocodeine 30 mg (NNH 7.4), and oxycodone ± paracetamol combinations (NNH 3.5 to 4.5).
Summary table E: Participants experiencing at least one adverse event (AE)
Summary of main results
We have reliable efficacy estimates of 46 drug/dose combinations in all types of surgery: 45 in painful dental conditions (overwhelmingly following third molar extraction) and 14 in other postoperative conditions. These estimates of efficacy have all been obtained using essentially the same clinical trial methods since they were first set out (Beecher 1957), and both trial and review methods have been standardised based on good evidence. The original philosophy concerning acute pain trials has been tested subsequently in a number of analyses using individual patient data (Moore 1997a; Moore 2005; Moore 2011) and those and other analyses also underpin the trials and reviews. This makes the results of studies comparable and that has previously included finding no significant difference between different pain models (Barden 2004).
We also know that there are a number of drugs for which there are no available trial data on how effective they are in acute pain (Acemetacin 2009; Meloxicam 2009; Nabumetone 2009; Nefopam 2009; Sulindac 2009; Tenoxicam 2009; Tiaprofenic acid 2009), as well as drug/dose combinations with inadequate evidence of benefit, or definite evidence of no benefit.
Placebo responses in the different meta-analyses - the percentage achieving at least 50% maximum pain relief with placebo over four to six hours - were consistent, with most falling between 5% and 15%, especially with larger numbers of participants given placebo for dental conditions (Figure 7) and all postoperative conditions (Figure 8). For other postoperative conditions the numbers of participants given placebo tended to be small and the range of responses somewhat higher (Figure 9). The degree of variability is what is expected by the random play of chance (Moore 1998).
|Figure 7. Plot of percent with outcome with placebo versus number of participants given placebo - dental only.|
|Figure 8. Plot of percent with outcome with placebo versus number of participants given placebo - other conditions only.|
|Figure 9. Plot of percent with outcome with placebo versus number of participants given placebo - all types of surgery.|
The efficacy results with adequate evidence show a range of values, whether measured relative to placebo in terms of a number needed to treat (NNT) for at least 50% maximum pain relief over four to six hours, in terms of the percentage of participants obtaining this level of benefit, or in terms of time before additional analgesia is required. Some drugs could be shown to not have any beneficial effects at some doses. Adverse events in these short-duration studies were generally not different between active drug and placebo, with a few exceptions, principally opioids.
The results also show clearly that even the most effective drugs fail to deliver good analgesia to a proportion of patients, meaning that a degree of analgesic failure is to be expected. Figure 4 shows that with many interventions, it is to be expected in more than half of patients treated.
There was also an interesting relationship between efficacy over four to six hours and duration of analgesia measured by mean time to remedication (Figure 10). Drugs with short duration of action tended to have higher (worse) NNTs, while drugs with longer duration of action had universally lower (better) NNTs, typically of two or below in those where mean remedication time was eight hours or longer. While not unexpected, this relationship implies that drugs with longer effects are likely to be more useful and effective in clinical practice.
|Figure 10. Plot of NNT over four to six hours versus mean time to remedication.|
Overall completeness and applicability of evidence
The 35 Cochrane Reviews cover almost all oral analgesics, although throughout the world many different combination analgesics can be found, typically without any published clinical trials. The review found that for seven drugs there were no clinical trial data and for a further six drugs there was inadequate information for any reliable basis of efficacy. In both these cases there are probably unpublished clinical trials. The authors' (unpublished) experience is that obtaining clinical trial data for older drugs is difficult and often impossible - though not always, as the eventual publication of 14 unpublished clinical trials of tramadol in a meta-analysis demonstrated (Moore 1997a). None of the drugs or doses for which this was a concern are used commonly in treating acute pain.
Some reviews appear not to be recent; all had been updated since 2008, but without finding any new studies and so they have kept their original citation dates (Aspirin 1999; Dextropropoxyphene ± Paracetamol 1999; Dihydrocodeine 2000; Piroxicam 2000). Additional searches for these drugs revealed no new studies since the reviews were completed. For other drugs, like etoricoxib, one or two additional studies have very recently been published, but do not materially change the conclusions.
There are no Cochrane Reviews for some commonly used drugs. These include tramadol, though there is an extant protocol for this, tramadol + paracetamol, and the combination of ibuprofen + paracetamol, a recently released combination, and one where these commonly-available drugs are frequently taken together. Non-Cochrane reviews are available for these (Edwards 2002; Moore 1997a; Moore 2011), which used the same methods and standards as the Cochrane Reviews, but results of these have not been included in the comparative figures. For completeness, results for these non-Cochrane reviews are shown in Summary table F.
The results for tramadol 50 mg in dental pain and for tramadol 100 mg in other painful conditions are clearly not reliable, as they are subject to potential publication bias. Results for higher doses of tramadol, tramadol and paracetamol, and ibuprofen and paracetamol are reliable. It is worth noting that reviews of tramadol indicated high rates of adverse events, though they were not reported in ways comparable to Cochrane Reviews (Edwards 2002; Moore 1997a).
Summary table F: Data from non-Cochrane reviews
Acute pain studies using a single dose of analgesic and with limited duration represent a poor test of adverse events, which can also often be complicated by proximity to anaesthesia. They are particularly limited in speaking to serious adverse events that might occur following long-term use of any of the drugs in this review. Moreover, the populations of postoperative patients participating in these studies will have tended to be younger and without many of the comorbid conditions that can occur. The aim of the studies was solely to test whether the drugs were analgesics.
Quality of the evidence
The quality of the evidence was good, using standard reviews examining standard clinical trials designed to measure the analgesic efficacy of drugs in sensitive assays in acute painful conditions. The overview process further removed any results likely to be the object of potential publication bias, so that only reliable results remained. This leaves a very large body of efficacy results presented both by all types of surgery, and split by the main painful conditions of dental pain and other (non-dental) painful conditions.
These results report a clinically useful level of pain relief over a sensible period, and with the common comparator of placebo. Though indirect comparisons are often criticised, this is one circumstance where indirect comparison can be justified because of the clinical homogeneity of trials and outcomes, and because data like these have been tested and indirect comparison found to be a reasonable approach (Song 2003).
Potential biases in the overview process
No obvious biases in the overview process exist, for the reasons given above. One possible concern would be if placebo responses varied extensively, as that would indicate a lack of clinical homogeneity, and some potential biases with high placebo responses in some studies or reviews limiting the measurement of efficacy of NNT, which measures absolute risk difference (Moore 2011). Figure 7, Figure 8 and Figure 9 show the placebo responses according to review and number of participants given placebo for dental studies, other postoperative studies, and all combined.
Small data sets are clearly more variable than larger, as would be expected (Moore 1998). However, with few exceptions placebo response rates were within limited ranges, typically between 5% and 20% for dental pain and 15% to 30% for other painful conditions.
Most studies in the individual reviews will have been sponsored or conducted by manufacturers. This is not likely to be a source of any bias, since specific analyses have been conducted on some of the larger data sets to demonstrate that no industry bias exists in like-for-like comparisons (Barden 2006).
Agreements and disagreements with other studies or reviews
The only other overview of this type known to exist for acute pain studies is a non-Cochrane overview in dental pain (Barden 2004). The general methods used were similar and there were no major differences.
Other important issues
This overview has brought together information on a very large number of participants and studies that have had one single aim, namely to test whether a particular drug at a particular dose had analgesic properties. The basic design of the individual studies was developed in the 1950s and 1960s, and rigorously tested at the time when randomised and double-blind studies were needed for objective assessment of analgesic efficacy (Houde 1960). Even the earliest studies emphasised large individual variability, and the variability in treatment groups of small size (Keats 1950).
These methods of analgesic testing have, with little change, become the standard way of demonstrating that a drug is an analgesic, and are typically performed early in the development of any new pain-relieving drug. A number of relatively recent individual patient analyses have examined various aspects of their design, conduct, and reporting (Barden 2004; Barden 2006; Moore 2005; Moore 1997a; Moore 2011). All of these investigations confirmed the success of the model, though adverse event reporting was inadequate (Edwards 1999). Other individual patient analyses of the postoperative period have demonstrated that patient satisfaction is highly correlated with good pain relief, showing the value of the outcome of at least 50% maximum pain relief (Mhuircheartaigh 2009).
While the reviews in this overview provide an excellent assessment of analgesic efficacy, both in the fact of the effects and often in its magnitude, there remains a distinction between measurement in trials and effectiveness in the clinic, and for different types of acute pain. Relative efficacy is, however, maintained between different painful conditions. For example, in dental pain ibuprofen 400 mg (NNT 2.3) is better than paracetamol 1000 mg (3.2) and aspirin 1000 mg (4.2). In migraine the same pattern is seen (Derry 2010; Kirthi 2010; Rabbie 2010), while NSAIDs are better than paracetamol for osteoarthritis (Towheed 2006). Information about analgesic efficacy from individual systematic reviews and overviews can be incorporated into schema for effective management of acute pain (Frampton 2009), or into other acute painful conditions.
It is the case that many of the individual studies used both a placebo and an active comparator. However, the actual drug and dose of active comparator varied so widely that useful direct comparisons between any two drugs was not available. Despite the fact that indirect comparisons have been shown to be reliable where sufficient high-quality data existed (Song 2003), one further step might be taken. That step would involve the use of network meta-analysis to confirm the assessment of relative efficacy in the overview, and to explore further methodological issues in this highly standardised and homogeneous data set (Caldwell 2005; Salanti 2008).
Implications for practice
The major implication for practice is the knowledge that there is a body of reliable evidence about the efficacy of 46 drug/dose combinations in acute pain. These results include information of immediate practical relevance including the percentage of patients likely to benefit in the short term, and comparative information about the likely duration of effect - a matter of pragmatic importance. However, not every patient will achieve good pain relief even with the most effective drugs, and analgesic failure is to be expected with a single dose, or perhaps with particular drugs in particular patients. Failure to achieve good pain relief should be actively and regularly sought and rectified.
Acute pain treatment is often part of a complex of interactions between patient, condition, and desired outcome; the overview helps by presenting evidence from which rational choices and decisions can be made. The evidence linking short-term benefit with longer duration of action is particularly important in this regard.
The overview also, and importantly, demonstrates where there are major absences of evidence. Where there is no evidence of efficacy, the drugs in question should probably not be used to treat acute pain.
Implications for research
Possibly the main implication for research is methodological. There will be few circumstances where such a body of information exists in such a clinically homogenous data set and it might appear to be an ideal opportunity to test new methods in meta-analysis, like network meta-analysis.
Support for this review came from the Oxford Pain Research Trust.
Appendix 1. Search strategy for Cochrane Reviews
- (postoperative):ti,ab,kw or (post NEXT operative):ti,ab,kw
- (pain):ti,ab,kw or (painful):ti,ab,kw or (analgesi*):ti,ab,kw
- (1 AND 2) in Cochrane Database of Systematic Reviews
Appendix 2. Results for remedication in individual reviews
Protocol first published: Issue 9, 2010
Review first published: Issue 9, 2011
Contributions of authors
SD and RAM carried out searches, selected reviews for inclusion, carried out assessment of methodological quality, and extracted data. HJM and PW acted as arbitrators. All authors were involved in discussing the results writing and approving the overview.
RAM/SD will be responsible for updating the overview.
Declarations of interest
All authors have received research support from charities, government and industry sources at various times. RAM, HJM, and PW have consulted for various pharmaceutical companies in the past. RAM and HJM have received lecture fees from pharmaceutical companies related to analgesics and other healthcare interventions.
Sources of support
- Oxford Pain Research Trust, UK.
- No sources of support supplied
Medical Subject Headings (MeSH)
MeSH check words