Predicting postoperative analgesia outcomes: NNT league tables or procedure-specific evidence?{dagger}

A. Gray1, H. Kehlet2,*, F. Bonnet3 and N. Rawal4

1 Medical Department, IdeaPharma Ltd, Cranfield, UK. 2 Section for Surgical Pathophysiology, The Juliane Marie Centre, Copenhagen, Denmark. 3 Service Anesthesie Reanimation, Hôpital Tenon, Paris, France. 4 Department of Anaesthesia and Intensive Care, University Hospital, Örebro, Sweden

* Corresponding author: Section for Surgical Pathophysiology, The Juliane Marie Centre 4074, Rigshospitalet Blegdamsvej 9, 2100 Copenhagen, Denmark. E-mail: henrik.kehlet{at}rh.dk

Accepted for publication November 16, 2004.


    Abstract
 Top
 Footnotes
 Abstract
 Introduction
 Are NNT values for...
 Are NNT values for...
 How can we optimize...
 References
 
Number needed to treat (NNT) values have been recommended and used to assess efficacy of analgesics for acute pain management. However, the data analysed come from a variety of procedures, which may potentially hinder the interpretation of the NNT value for specific procedures. We reanalysed available NNT data with acetaminophen in relation to the magnitude of surgical injury. Acetaminophen was less effective for pain relief after orthopaedic procedures than after dental procedures. The relative risk ratio for more than 50% pain relief, compared with placebo, was only 1.87 compared with 3.77 (P<0.05). Although NNT can give a valuable overview of efficacy, this concept is not necessarily applicable to all types of surgery. We suggest that estimates of NNT should be related to specific surgical procedures.

Keywords: analgesia, postoperative ; analgesics, non-opioid, acetaminophen ; pain, postoperative ; pharmacology, acetaminophen ; potency, analgesic


    Introduction
 Top
 Footnotes
 Abstract
 Introduction
 Are NNT values for...
 Are NNT values for...
 How can we optimize...
 References
 
A prerequisite for the successful management of postoperative pain, and for optimal patient recovery, is the use of dynamic analgesic protocols that aim to maximize benefit whilst minimizing adverse events.1 The need for accessible information on analgesics has led to the construction of analgesic league tables, whereby the efficacy and adverse events of analgesics are ranked.2 Relative benefit is expressed as the number needed to treat (NNT): the number of patients that must receive an analgesic to observe an effect in one more patient than in a placebo comparator group. Adverse events are expressed as the number needed to harm (NNH): the number of patients that must receive a treatment to observe a side-effect or a complication in one more patient than in the placebo group. The data to calculate these values are derived by pooling the results of multiple studies examining the efficacy and safety profiles of analgesic interventions in different surgical procedures, and are based on the proportion of patients with at least a 50% maximal pain response (≥50%maxTOTPAR) over 4–6 h after surgery.

League tables have been constructed to provide clinicians with an overview of comparative analgesic efficacy and harm, assisting in the planning of postoperative pain management protocols. However, they have a significant limitation: available NNT data have not identified statistically significant differences between analgesics with efficacies as disparate as those of acetaminophen and morphine,2 which is not clinically intuitive. This raises the fundamental question of whether pooling data of analgesic effects from different procedures and in different patient groups limits their interpretability, by creating an average value with a wide margin of error that lacks applicability to particular clinical scenarios.


    Are NNT values for analgesics applicable to specific surgical procedures?
 Top
 Footnotes
 Abstract
 Introduction
 Are NNT values for...
 Are NNT values for...
 How can we optimize...
 References
 
An assumption in the construction of analgesic league tables is that different pain models are comparable, and that benefit and harm can be extrapolated from one model to another. Although such extrapolation is not unusual in clinical practice, it was not until comparatively recently that this approach was scrutinized. One of the first publications to examine this issue reviewed the effects of aspirin vs placebo in the postoperative, postpartum and dental impaction settings (Table 1).3


View this table:
[in this window]
[in a new window]
 
Table 1 Percentage maximum pain relief of aspirin and placebo in different pain models. Figures are mean (standard error). After Cooper, 19913

 
A number of important observations can be made from this analysis. Firstly, the placebo response differs substantially between the pain models: for example, the response in the postpartum setting is over twice as large as that in dental pain. Secondly, the response to aspirin and the margin of efficacy over placebo differ between pain models. From a clinical perspective, it is interesting to note that aspirin is more effective in postpartum women than in the other two pain models, with the highest proportion of maximum relief obtained. However, the high placebo response means that the margin of efficacy over placebo is around half of that seen in the postoperative setting. A final observation is that the standard error for the aspirin response is larger in the postoperative setting than in the two remaining pain models. This suggests that the pooling of data from a number of different surgical procedures creates a more heterogeneous outcome compared with analysing data from single settings, such as dental impaction. In his publication of these data, Cooper indicated that there were ‘some clinically relevant differences among the models’.3

A systematic review of the efficacy of acetaminophen, NSAIDs and their combination in postoperative pain was also able to examine relative effects in different pain models.4 Although the reviewers were unable to draw firm conclusions because of methodological problems in some of the individual studies examined, they found evidence that acetaminophen and NSAIDs had equivalent efficacy in major surgery, but that in dental pain NSAIDs had an advantage over acetaminophen for pain scores. This issue has been re-examined recently using data for aspirin, acetaminophen and ibuprofen, with pooled data from two groups of studies: dental extraction and postoperative pain.5 Using ≥50%maxTOTPAR to calculate NNT values, the authors observed that the relative benefit of these agents did not differ significantly between the two pain models examined. However, in common with the observations of Cooper, the efficacy of placebo and active agents differed, albeit not to the point of statistical significance, between the postoperative and dental pain settings. The relative benefits of aspirin, acetaminophen and ibuprofen were consistently lower in the postoperative than in the dental pain setting (Table 2). Furthermore, as with the analysis of Cooper, the range of benefits within the postoperative group was wider than in the dental group, a finding that again suggests differences between the procedures pooled in the postoperative group.


View this table:
[in this window]
[in a new window]
 
Table 2 Comparison of analgesic treatment outcomes following dental procedures or surgery. A comparison of the percentage of patients who achieved ≥50% pain relief with aspirin, acetaminophen or ibuprofen compared with placebo. This shows the relative efficacy of the active compared with the placebo treatment. Note that the efficacy of all the agents relative to placebo is greater in dental procedures than in the postsurgical setting. Data from the analysis of Barden et al.5

 
From these data, it can be hypothesized that the effects of analgesics differ significantly between pain models. In order to examine this hypothesis further, we analysed a large database for the effects of acetaminophen on postprocedure pain (Fig. 1). 6 The first clue from these data, which pooled results from studies of postoperative, postpartum, dental extraction and oral surgery pain, is that the data are highly significantly heterogeneous (P<0.00001), a sign that they lack comparability. Although it has been argued that conventional heterogeneity tests may be unhelpful,7 these data are also highly heterogeneous according to the updated criteria of Higgins and colleagues (I2=75.4%).8 When the effects of acetaminophen are analysed on a procedure-specific basis, with a single study from the 23 originally included removed as it contained data from mixed procedures, the reason for this heterogeneity becomes clear: the effect of acetaminophen in the postoperative setting, derived from orthopaedic procedures, differs significantly from that in the dental extraction model (Fig. 2). These data strongly support what many clinicians know to be intuitive: what works in one procedure may have a different effect or no significant effect at all in another. They also demonstrate that, in averaging effects from different procedures, the strength of evidence from one particular model (for example, dental pain, a commonly used model to examine efficacy) can falsely weight the average in a particular direction.



View larger version (35K):
[in this window]
[in a new window]
 
Fig 1 Single-dose acetaminophen for postprocedure pain (postoperative, postpartum, dental extraction and oral surgery pain studies). Number of patients with ≥50% pain relief.1537

 


View larger version (13K):
[in this window]
[in a new window]
 
Fig 2 Relative risk (RR) of achieving ≥50% maximal pain relief for acetaminophen vs placebo in different pain models. FE, fixed effects meta-analysis model; RE, random effects meta-analysis model. Data derived from the meta-analysis of Barden et al.6

 

    Are NNT values for analgesics interpretable for clinical practice?
 Top
 Footnotes
 Abstract
 Introduction
 Are NNT values for...
 Are NNT values for...
 How can we optimize...
 References
 
In seeking measures of analgesic outcome that have validity for clinical trials and are also suitable for combining inmeta-analyses, a number of methods have been used. The calculation of the potential maximum analgesic benefit in a given trial and the proportion of this maximum achieved by individual patients [the area under the pain curve (AUC), %maxTOTPAR] provides a useful measure of analgesic benefit over a period of time.9 10 These are also used to calculate NNT values.9 10 A mathematical model can be used to calculate the proportion of patients achieving a ≥50%maxTOTPAR without the need for individual patient data.9 10 This measure not only gives an impression of analgesic benefit during the postoperative period, but also takes into account the lack of a normal distribution in pain response.11

However, the shape of the analgesic benefit curve, not merely the area beneath it, can provide clinically important information. For example, similar AUCs may be achieved with agents with different times of onset, peak analgesic efficacy and duration of action. However, these specific details have significant implications for clinical practice. Furthermore, a 50% maximal pain response may not provide clinically relevant information. Guidelines suggest that a visual analogue scale (VAS) score of <30/100 mm is an appropriate target for analgesia, and a minimum clinically relevant drop in pain intensity has been considered to be 13 mm.12 It is also likely that, in order to achieve a clinically relevant drop in pain intensity, a larger decrease is required in patients with an initially higher pain intensity than in those with a lower pain intensity.13

NNT values based on a 50% response do not therefore provide information on whether the change was from high- or low-intensity pain (for example, from a VAS score of 90 to 45 or from 40 to 20). Additionally, they provide no information on reduction in time or overall need for supplementary analgesia. An analgesic intervention may provide little additional benefit in VAS scores, but may decrease supplementary opioid use; this is an important finding that cannot be contained within an NNT value. Thus, NNT values derived using this method may eliminate the details of analgesic benefits that are important for clinicians in making prescribing decisions.


    How can we optimize postoperative analgesia prescribing decisions?
 Top
 Footnotes
 Abstract
 Introduction
 Are NNT values for...
 Are NNT values for...
 How can we optimize...
 References
 
The use of NNT values can therefore be criticized on a number of counts. Firstly, a credible NNT value needs to be derived from at least 500 patients,14 which has demanded the pooling of data from heterogeneous studies. Despite this recommendation, of the 50 NNT values quoted, most (32/50) are based on fewer than 500 patients, with eight based on fewer than 100.2 By pooling the data, an average effect of a given analgesic is created, with a wide margin of error, which ignores the specific effects of analgesics in different procedures. This creates a league table with many overlapping values, with the propensity to confuse clinicians and lead to extrapolations of efficacy that are inappropriate for all procedures. Secondly, the calculation of NNT values, which are derived through mathematical modelling and not from individual patient data, removes specific information concerning the pattern of effect of interventions, depriving clinicians of valuable information. Lastly, although these values may provide an accessible method of comparing different agents, they are not clinically intuitive for many physicians, as they provide little indication of what change in pain the patient is likely to experience with a given analgesic, and they give no information concerning the placebo responses in specific procedures, which are markedly different.

In conclusion, using calculated outcome measures may not provide the most useful or reliable information on analgesic efficacy for clinical practice. It may be that the impact of analgesics vs placebo on pain intensity scores at multiple time points—one of the most basic measures of efficacy—still provides the most useful marker for clinicians. This needs to be supplemented by data on the effects on reducing supplementary analgesic consumption, an important endpoint that has utility in aiding clinicians to reduce postoperative opioid consumption. However efficacy information are presented, it is clear that average values derived by pooling data from different procedures can provide misleading information to clinicians. This reinforces the need to examine procedure-specific outcomes wherever possible, to ensure that postoperative pain management protocols are optimized, although further work is needed to define the boundaries of procedure-specificity.


    Footnotes
 Top
 Footnotes
 Abstract
 Introduction
 Are NNT values for...
 Are NNT values for...
 How can we optimize...
 References
 
{dagger} Declaration of interest. The authors are members of the Procedure Specific Postoperative Pain Management (PROSPECT) Group, which conducts procedure-specific systematic reviews of the literature using the Cochrane Protocol, supplements these with evidence from other procedures and from clinical practice, and produces guidelines for the management of postoperative pain. The PROSPECT Group is funded by an unrestricted educational grant from Pfizer Inc., who provided financial support for the writing of this article. Back


    References
 Top
 Footnotes
 Abstract
 Introduction
 Are NNT values for...
 Are NNT values for...
 How can we optimize...
 References
 
1 Kehlet H, Dahl JB. Anaesthesia, surgery, and challenges in postoperative recovery. Lancet 2003; 362: 1921–8[CrossRef][ISI][Medline]

2 Moore A, Edwards J, Barden J, McQuay H. Bandolier's Little Book Of Pain. Oxford: Oxford University Press, 2003

3 Cooper SA. Single-dose analgesic studies: the upside and downside of assay sensitivity. In: Max M, Portenoy R, eds. Advances in Pain Research and Therapy. New York: Raven Press, 1991; 117–24

4 Hyllested M, Jones S, Pedersen JL, Kehlet H. Comparative effect of paracetamol, NSAIDs or their combination in postoperative pain management: a qualitative review. Br J Anaesth 2002; 88: 199–214[Abstract/Free Full Text]

5 Barden J, Edwards JE, McQuay HJ, Andrew Moore R. Pain and analgesic response after third molar extraction and other postsurgical pain. Pain 2004; 107: 86–90[CrossRef][ISI][Medline]

6 Barden J, Edwards J, Moore A, McQuay H. Single dose oral paracetamol (acetaminophen) for postoperative pain. Cochrane Database Syst Rev 2004, CD004602

7 Higgins J, Thompson S, Deeks J, Altman D. Statistical heterogeneity in systematic reviews of clinical trials: a critical appraisal of guidelines and practice. J Health Serv Res Policy 2002; 7: 51–61[CrossRef][Medline]

8 Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses. Br Med J 2003; 327: 557–60[Free Full Text]

9 Moore A, Moore O, McQuay H, Gavaghan D. Deriving dichotomous outcome measures from continuous data in randomised controlled trials of analgesics: use of pain intensity and visual analogue scales. Pain 1997; 69: 311–5[CrossRef][ISI][Medline]

10 Moore A, McQuay H, Gavaghan D. Deriving dichotomous outcome measures from continuous data in randomised controlled trials of analgesics: verification from independent data. Pain 1997; 69: 127–30[CrossRef][ISI][Medline]

11 Moore A, McQuay H, Gavaghan D. Deriving dichotomous outcome measures from continuous data in randomised controlled trials of analgesics. Pain 1996; 66: 229–37[CrossRef][ISI][Medline]

12 Gallagher EJ, Liebman M, Bijur PE. Prospective validation of clinically important changes in pain severity measured on a visual analog scale. Ann Emerg Med 2001; 38: 633–8[CrossRef][ISI][Medline]

13 Averbuch M, Katzper M. Severity of baseline pain and degree of analgesia in the third molar post-extraction dental pain model. Anesth Analg 2003; 97: 163–7[Abstract/Free Full Text]

14 McQuay HJ, Edwards JE, Moore RA. Evaluating analgesia: the challenges. Am J Ther 2002; 9: 179–87[CrossRef][Medline]

15 Bentley KC, Head TW. The additive analgesic efficacy of acetaminophen, 1000 mg, and codeine, 60 mg, in dental pain. Clin Pharmacol Ther 1987; 42: 634–40[ISI][Medline]

16 Berry FN, Miller JM, Levin HM, Bare WW, Hopkinson JH 3rd, Feldman AJ. Relief of severe pain with acetaminophen in a new dose formulation versus propoxyphene hydrochloride 65 mg and placebo: a comparative double-blind study. Curr Ther Res Clin Exp 1975; 17: 361–8[ISI][Medline]

17 Bjune K, Stubhaug A, Dodgson MS, Breivik H. Additive analgesic effect of codeine and paracetamol can be detected in strong, but not moderate, pain after Caesarean section. Baseline pain-intensity is a determinant of assay-sensitivity in a postoperative analgesic trial. Acta Anaesthesiol Scand 1996; 40: 399–407[ISI][Medline]

18 Cooper SA, Erlichman MC, Mardirossian G. Double-blind comparison of an acetaminophen-codeine-caffeine combination in oral surgery pain. Anesth Prog 1986; 33: 139–42[Medline]

19 Cooper SA, Reynolds DC, Reynolds B, Hersh EV. Analgesic efficacy and safety of (R)-ketoprofen in postoperative dental pain. J Clin Pharmacol 1998; 38 (2 Suppl): 11S–18S[Abstract/Free Full Text]

20 Cooper SA, Schachtel BP, Goldman E, Gelb S, Cohn P. Ibuprofen and acetaminophen in the relief of acute pain: a randomized, double-blind, placebo-controlled study. J Clin Pharmacol 1989; 29: 1026–30[Abstract/Free Full Text]

21 Edwards JE, McQuay HJ, Moore RA. Combination analgesic efficacy: individual patient data meta-analysis of single-dose oral tramadol plus acetaminophen in acute postoperative pain. J Pain Symptom Manage 2002; 23: 121–30[CrossRef][ISI][Medline]

22 Hersch EV, Levin LM, Cooper SA. Ibuprofen liquigel for oral surgery pain. Clin Ther 2000; 22: 1306–18[CrossRef][ISI][Medline]

23 Kiersch TA, Halladay SC, Hormel PC. A single-dose, double-blind comparison of naproxen sodium, acetaminophen, and placebo in postoperative dental pain. Clin Ther 1994; 16: 394–404[ISI][Medline]

24 Laska EM, Sunshine A, Zighelboim I, et al. Effect of caffeine on acetaminophen analgesia. Clin Pharmacol Ther 1983; 33: 498–509[ISI][Medline]

25 Lehnert S, Reuther J, Wahl G, Barthel K. [The efficacy of paracetamol (Tylenol) and acetyl salicylic acid (aspirin) in treating postoperative pain]. Dtsch Zahnarztl Z 1990; 45: 23–6[Medline]

26 McQuay HJ, Carroll D, Frankland T, Harvey M, Moore A. Bromfenac, acetaminophen, and placebo in orthopedic postoperative pain. Clin Pharmacol Ther 1990; 47: 760–6[ISI][Medline]

27 Mehlisch DR, Frakes LA. A controlled comparative evaluation of acetaminophen and aspirin in the treatment of postoperative pain. Clin Ther 1984; 7: 89–97[ISI][Medline]

28 Mehlisch DR, Jasper RD, Brown P, Korn SH, McCarroll K, Murakami AA. Comparative study of ibuprofen lysine and acetaminophen in patients with postoperative dental pain. Clin Ther 1995; 17: 852–60[CrossRef][ISI][Medline]

29 Mehlisch DR, Sollecito WA, Helfrick JF, et al. Multicenter clinical trial of ibuprofen and acetaminophen in the treatment of postoperative dental pain. J Am Dent Assoc 1990; 121:257–63[ISI][Medline]

30 Moller PL, Norholt SE, Ganry HE, et al. Time to onset of analgesia and analgesic efficacy of effervescent acetaminophen 1000 mg compared to tablet acetaminophen 1000 mg in postoperative dental pain: a single-dose, double-blind, randomized, placebo-controlled study. J Clin Pharmacol 2000; 40: 370–8[Abstract/Free Full Text]

31 Rubin A, Winter L Jr. A double-blind randomized study of an aspirin/caffeine combination versus acetaminophen/aspirin combination versus acetaminophen versus placebo in patients with moderate to severe post-partum pain. J Int Med Res 1984; 12: 338–45[ISI][Medline]

32 Sakata RK, Lauzi J, Kuniyoshi HS, Ono MT. Comparative double-blind study with a single dose of acetaminophen, dipyrone and placebo in the treatment of postoperative pain. Rev Bras Cir 1986; 76: 301–4

33 Santos Pereira E. Comparative study of paracetamol, dipyrone and placebo for the treatment of postoperative orthopaedic pain. [Estudo comparativo entre acetaminofen, dipirona e placebo no tratamento da dor pos-operatoria em ortopedia]. Folha Med 1986; 92: 99–105

34 Schachtel BP, Thoden WR, Baybutt RI. Ibuprofen and acetaminophen in the relief of postpartum episiotomy pain. J Clin Pharmacol 1989; 29: 550–3[Abstract/Free Full Text]

35 Seymour RA, Kelly PJ, Hawkesford JE. The efficacy of ketoprofen and paracetamol (acetaminophen) in postoperative pain after third molar surgery. Br J Clin Pharmacol 1996; 41: 581–5[CrossRef][ISI][Medline]

36 Winnem B, Samstad B, Breivik H. Paracetamol, tiaramide and placebo for pain relief after orthopedic surgery. Acta Anaesthesiol Scand 1981; 25: 209–14[ISI][Medline]

37 Winter L, Appleby F, Ciccone PE, Pigeon JG. A double blind comparative evaluation of acetaminophen, caffeine and the combination of acetaminophen and caffeine in outpatients with post-operative oral surgery pain. Curr Ther Res 1983; 33: 115–22[ISI]





This Article
Abstract
Full Text (PDF)
All Versions of this Article:
94/6/710    most recent
aei144v1
E-Letters: Submit a response to the article
Alert me when this article is cited
Alert me when E-letters are posted
Alert me if a correction is posted
Services
Email this article to a friend
Similar articles in this journal
Similar articles in ISI Web of Science
Similar articles in PubMed
Alert me to new issues of the journal
Add to My Personal Archive
Download to citation manager
Search for citing articles in:
ISI Web of Science (1)
Disclaimer
Request Permissions
Google Scholar
Articles by Gray, A.
Articles by Rawal, N.
PubMed
PubMed Citation
Articles by Gray, A.
Articles by Rawal, N.