Placebo-controlled randomized trials are the reference standard for assessing the effectiveness and safety of therapeutic agents. The purpose of a placebo is to separate results attributable to patient expectations, conditioning, spontaneous improvement, and regression to the mean from those dependent on the therapeutic agent in question.12 However, leaving a patient in pain with no analgesic treatment except placebo poses ethical and practical dilemmas.17 The World Medical Association Declaration of Helsinki prohibits offering patients an intervention that is less effective than the best proven,26 and the prospect of receiving a placebo may discourage patients from taking part in clinical pain trials and lead to a high dropout rate, complicating the interpretation of trial results.
One solution to this issue is to offer study subjects rescue medication, also known as escape or reliever medication, to be taken if the blinded study drug, active or placebo, fails to relieve their pain adequately. Another option is to allow participants to continue their usual analgesic treatment during the trial. These 2 approaches are not mutually exclusive. However, analyzing and interpreting the results of a trial that allows varying intake of supplementary analgesics can be complicated. If the drug under investigation is effective, one would expect the consumption of supplementary analgesics in the placebo group to be higher than that in the active group, thereby supporting the effectiveness of the experimental drug. However, a higher analgesic consumption in the placebo group will likely reduce the difference in pain between the 2 groups, potentially concealing the true effect of the experimental drug. Furthermore, the use of pain medication on top of the placebo or experimental drug may complicate the interpretation of tolerability and safety assessments.9
Although rescue medication is commonly used, the methodology applied to its use and analysis has received very little attention. The Consolidated Standards of Reporting Trials (CONSORT) guidelines make no mention of rescue medication,20,22 and the Initiative on Methods, Measurement, and Pain Assessment in Clinical Trials (IMMPACT)9 only recommends that if “rescue analgesia is permitted, it is important to record and report the amount used by subjects, which may be greater in the placebo group in a trial of an efficacious treatment and should be considered in analyzing and interpreting the data.” Proper reporting facilitates critical appraisal, interpretation of study results, and appropriate replication.
However, the extent to which reports of trials of pharmacotherapy for low back or neuropathic pain contain adequate information regarding rescue medication and concomitant analgesics has not been explored. Despite their potential to affect the results of such trials, a scientific discussion regarding the methodology for use of supplementary analgesics is generally missing. Therefore, the aim of the current review was to explore how rescue medication and prestudy analgesics have been handled and reported in placebo-controlled trials of pharmacotherapy for neuropathic pain and low back pain.
2.1. Trial selection
To ensure a representative sample of trials, we reviewed trials included in 2 recent systematic reviews and meta-analyses, one of pharmacotherapy for neuropathic pain11 and one of pharmacotherapy for low back pain.2 The neuropathic pain review included placebo-controlled double-blind trials of oral and topical pharmacotherapy of at least 21 days' duration and ≥10 participants per treatment arm. We identified trials using the reference list in the supplementary appendices 7 and 11.11 The low back pain review included randomized controlled trials and previous systematic reviews of therapies for nonradicular or radicular low back pain. We identified individual trials of pharmacotherapy from the reference list in appendices C and D of the source document.3 Trials in which the experimental treatment or placebo was added to existing therapy were excluded. The references for the included studies are given in Appendix 1 (available online at http://links.lww.com/PAIN/A870).
2.2. Data extraction
Descriptive and outcome data were extracted using a pilot-tested data collection form. The pilot testing was performed on 32 randomly selected trials that were independently scored by 3 investigators (L.G., A.J.H., and E.H.). Disagreements were resolved by a consensus discussion, and the manual was modified for clarity. The remaining 233 trials were each scored independently by 2 investigators. Inconsistencies were resolved by discussion. Data were obtained from the published articles or through associated appendices, when available.
2.2.1. Descriptive data
Descriptive data included trial design (parallel group or crossover, inert or active placebo), number of subjects randomized, and trial duration. An active placebo was defined as a substance mimicking the side effects of the investigational drug with no therapeutic effect on the condition being treated. Trial duration was recorded in days, excluding periods of drug washout or titration. If a minimum level of pain was required for inclusion in the trial, the required level was recorded. Mean baseline pain intensity reported on a continuous or ordinal scale was recorded on a 0 to 10 scale if appropriate, or converted to a 0 to 10 format. When study arms contained unequal numbers of participants, or pain levels were reported separately for each arm, the mean baseline pain intensity was weighted according to the number of subjects in each arm.
2.2.2. Rescue medication
Trials were classified according to whether use of rescue medication was permitted. In trials investigating the pain-relieving effects of capsaicin, we did not consider that drugs administered to alleviate local pain caused by capsaicin at the application site were rescue medication. If no mention of rescue medication was made in the article's methods or results section, we presumed that it had not been used. If rescue medication was permitted, we recorded whether it was specified as an outcome measure and, if so, whether it was primary, secondary, or explorative. We classified rescue medication as paracetamol (acetaminophen), nonsteroidal anti-inflammatory drug (including aspirin), weak opioid (tramadol, codeine, dihydrocodeine, and dextropropoxyphene), strong opioid, or other. We recorded how the consumption of rescue medication was quantified (by dosage or use/no use), how it was assessed (by self-report or pill count), and whether it was analyzed statistically.
2.2.3. Prestudy analgesics
Participants' use of their usual prestudy analgesics was coded as stopped (if stopped before initiation of the study drug), allowed (if no change in dose or frequency was required), or not specified. The code allowed with limitations was used if the patients were allowed to continue with some specified drugs but not others. Trials that reported drug washout before the start of the assigned treatments and provided no further information regarding prestudy analgesics were coded as stopped. Low-dose aspirin (≤325 mg/day) used for the prevention of cardiovascular disease was not considered as prestudy analgesic. If continued use of prestudy analgesics was allowed, or allowed with limitations, we recorded whether the intake of prestudy analgesics during the trial was reported. We also recorded any restrictions on medications not classified as analgesics but with pain-relieving potential, such as benzodiazepines, antidepressants, and anticonvulsants. We did not code a restriction if this was based on contraindications or interactions with the study drug.
Data were analyzed using IBM SPSS v.25.0 (IBM Corp, Armonk, NY) and RStudio v1.1.463 (RStudio, Inc, Boston, MA)/R version 3.5.1 (R Foundation for Statistical Computing, Vienna, Austria).
We included 265 trials consisting of 83 trials of pharmacotherapy for low back pain presented in 82 articles, and 182 trials of pharmacotherapy for neuropathic pain presented in 178 articles (Fig. 1). The trials were conducted between the years 1966 and 2015.
3.1. Trial characteristics
Trial characteristics are presented in Table 1. The median number of trial participants was 93 days (interquartile range 40-248) and the median trial duration was 56 days (interquartile range 30-84), with the low back pain trials having shorter duration than the neuropathic pain trials. A parallel group design was used in 73% of the trials, and a crossover design in 26%, the latter being more common among the neuropathic pain trials. A minimum level of pain for inclusion was required in 197 (74%) trials, of which the majority required at least 4 on a scale of 0 to 10, or moderate pain on a verbal scale. At baseline, the mean (SD) pain intensity was 6.4 (0.9) (178 trials). Eighteen trials reported baseline pain by the median (range of medians, 4.9-8.6), 30 trials did not report pain intensity on a numeric scale, and 39 trials did not report baseline pain.
3.2. Handling of rescue medication and prestudy analgesics
In 113 (43%) trials, the usual analgesics were stopped before study initiation (Table 2); 67 trials (25%) allowed patients to continue their usual analgesics, and 59 (22%) allowed them to continue but with certain limitations. The remaining 26 (10%) trials did not specify the handling of prestudy analgesics.
Rescue medication was permitted in 117 trials (44%) and explicitly prohibited in 10 (4%), with a similar frequency in the low back pain and neuropathic pain trials. The remaining 138 (52%) trials made no mention of rescue and were considered not to have used rescue medication. The most common rescue drug was paracetamol, either alone or in combination with other analgesics or nonsteroidal anti-inflammatory drugs (Table 3). Stand-alone paracetamol was more commonly used in the neuropathic pain trials (73% vs 22%), whereas strong opioids were more common in the low back pain trials (28% vs 4%).
Seventy-two trials (27%) permitted rescue medication but stopped prestudy analgesics, whereas 41 (16%) allowed participants to take rescue medication and to continue their usual analgesics. Six trials (2%) stopped prestudy analgesics and explicitly prohibited rescue medication. Thirty-five trials (13%) stopped prestudy analgesics, but did not specify whether or not rescue medication was permitted. For 22 trials (8%), no information regarding rescue medication or prestudy analgesics was given.
Nearly half of all trials imposed restrictions on the use of nonanalgesic medications such as antidepressants, benzodiazepines, or anticonvulsants (Table 1). Of the 113 trials that stopped prestudy analgesics, 69 (61%) also imposed restrictions on concomitant nonanalgesic medication.
The utilization of rescue medication increased over time (Fig. 2): 18% of trials (14/76) conducted before 2000 permitted rescue compared with 55% (103/189) conducted after 2000.
Table 4 shows the reporting features of the 117 trials that permitted rescue medication. Of these, 44 (38%) explicitly defined rescue medication as an outcome measure. Sixty-two (53%) trials did not specify whether it had been used as outcome and for 11 (9%) trials, the text was unclear.
More than one-third (45/117) of trials did not specify whether the consumption of rescue medication was quantified by dosage or use/no use, and more than half (67/117) did not specify whether it was assessed by patient self-report or pill count. In total, 44 (38%) trials made no mention of rescue medication in the results sections. Of those trials that reported rescue medication consumption, 27 did so using a dichotomous variable (yes/no), whereas 42 trials reported consumption by the actual dosage taken during the study.
Just under half of the trials (49%, 57/117) performed a statistical analysis of rescue medication consumption, of which 40 specified the statistical test that had been applied. In total, 22 (19%) trial reports included complete information as to whether rescue medication was used as outcome, which drug(s) were used, how the consumption was assessed and measured, and reported and analyzed its use (Fig. 3).
Of the 126 trials that allowed patients to continue their usual analgesics, 70 (56%) did not report on the actual intake during the trial. Of the 72 trials that permitted rescue medication but disallowed prestudy analgesics, 48 (67%) did not report the quantity of rescue medication taken.
This review shows that rescue and concomitant analgesics are commonly allowed in trials of pharmacotherapy for low back pain and neuropathic pain. However, procedures and results regarding their use were poorly described and incompletely reported.
Many patients under consideration for enrollment in pharmacological pain trials are already taking pain medication. From a methodological viewpoint, minimizing noise from analgesics not under study would be desirable. However, restricting access to pain medication may make patients unwilling to participate, cause them to leave the study prematurely because of a lack of efficacy of the trial medication, or lead to undisclosed self-medication. Reduced attrition rates in the experimental arm may signal efficacy, but reasons why participants drop out can be difficult to identify, and a large number of dropouts will complicate the analyses. Furthermore, restricting the supply of analgesics may undermine the external validity of the trial: results obtained from patients who accept the prospect of receiving placebo with no or little additional pain medication may not be valid for all patients with the same condition.
The use of rescue medication enables patients to diminish their pain, but with a risk of falsely minimizing differences between the active and placebo arms. If patients in the placebo arm use more rescue medication than those in the active arm, there is a risk of wrongly assuming noneffectiveness of the experimental drug. About half of the trials in this review permitted rescue medication. The issue of interpreting and analyzing data from pain trials that use rescue analgesics has generally received little attention, which may explain why less than half the trials included in this review performed a formal analysis of rescue medication consumption, and even fewer reported the statistical method that was applied. Assumptions regarding statistical testing, such as sample-size calculations, were generally not specified. White et al.28 discussed a statistical approach for estimating differences between the investigational drug and placebo in trials involving rescue medication. They also suggested that if the amount of rescue medication is low and roughly equal in each group, it may suffice to report the extent and time pattern of rescue medication consumption in each randomized group. Incorporating rescue medication into a composite outcome was shown to improve responsiveness in trials of pharmacotherapy for postoperative pain6,23 and osteoarthritis,21 and has also been proposed in trials of treatments for allergic rhinoconjunctivitis.4 With regard to prestudy analgesics, it has been suggested that post hoc subgroup analyses should be performed by calculating the primary efficacy outcomes for patients either taking or not taking concomitant analgesics.10
Concurrent use of several supplementary analgesics may complicate further the interpretation of trial results. It is not self-evident how the influence of, for example, paracetamol compared with hydrocodone should be assessed. Several of the trials that stopped prestudy analgesics and allowed rescue medication permitted more than one drug as rescue. Another 16% allowed both prestudy analgesics and rescue medication. However, in the trials reviewed, the investigators' decisions regarding the handling of concomitant analgesics and rescue medication were generally not explained, justified, or discussed.
There is evidence that subjects who take concomitant pain medication report smaller effects of the experimental treatment than subjects who do not take concomitant pain medication.5,27 Katz et al.15 showed that trials that restricted concomitant or rescue analgesics were more likely to be associated with a positive result. In the present review, more than half of the trials allowed patients to continue their usual analgesics. Allowing the continued use of prestudy medications may add noise to the data, increasing the variability of the treatment effect estimate. This could make it more difficult to detect a significant effect of the active drug. Two patients who at baseline report equal pain intensity—one taking strong opioids and the other taking no analgesics—are unlikely to be directly comparable.
Most trials that allowed participants to continue their usual analgesics required that background treatment remained stable during the trial. However, compliance with this requirement was seldom reported. We assume that details regarding concomitant medication were recorded, but why they were not reported is unknown. The fact that 10% of the trial reports did not mention prestudy analgesics adds to our general finding of poor reporting. An exception is Kieburtz et al.16; based on the World Health Organization's three-step analgesic ladder, they developed an algorithm to assess whether their subjects' analgesic medications had increased, decreased, or remained stable between baseline and the last week of treatment. It is particularly discouraging that one-third of the trials that permitted rescue did not report the actual intake of rescue medication. Some reported consumption as yes/no, meaning that one single dose would be weighted the same as multiple doses daily over several weeks. Only one-third of the trials provided really useful data by reporting rescue consumption as a continuous variable. Poor handling and reporting preclude valid interpretation and critical appraisal of both individual and aggregate trial results. For example, improper handling of data about rescue medications contributed to the production of unreliable summary estimates in a meta-analysis of sodium–glucose cotransporter 2 inhibitors as treatment for type 2 diabetes.25 Poor reporting also hinders appropriate trial replication and prevents clinicians from achieving similar outcomes in the treatment of individual patients.
4.1. Comparison with other studies
We are not aware of other studies that have evaluated the handling and reporting of concomitant and rescue medication in chronic pain trials. Singla et al.24 discussed rescue regimen and methodological challenges in trials of acute postoperative pain, and called for a broader scientific discussion of rescue therapy as an experimental design.
Our findings of poor reporting are consistent with numerous previous evaluations of randomized controlled trials.7,18 There are often insufficient details about treatments to allow their use by clinicians.1 A review of cancer chemotherapy trials published in major oncology journals found that only 11% of 262 trials provided complete details of the trial treatments.8 Improvements such as extensions of the CONSORT statement have been suggested.13
4.2. Strengths and limitations
A strength of this study is that the trials evaluated were part of 2 previous systematic reviews and meta-analyses that formed the basis for treatment recommendations for low back pain and neuropathic pain. Some of the trials were published several years ago and may not reflect current practice. There is some evidence that reporting and methodologic quality have improved over recent decades.7 We did not relate our results to journal metrics such as impact factor. The original systematic review on neuropathic pain excluded some trials because they were of short treatment duration or did not use pain as the primary outcome,11 and the low back pain review excluded some trials because of inadequate study design, not being reported in English, small sample size, and other factors.2 Finally, although reporting quality is often associated with methodologic quality, it should be acknowledged that poorly reported trials may have been well conducted.14,19
4.3. Future recommendations
Investigators, journal editors, and reviewers need to address the issues of concomitant and rescue analgesic medication in the design, conduct, and reporting of placebo-controlled trials of pharmacotherapy for pain. We recommend that all trial reports explicitly state whether or not rescue medication was permitted and how prestudy analgesics were handled. Furthermore, a preliminary checklist for investigators of trials using rescue medication or allowing patients to continue their prestudy analgesics is presented in Table 5. This preliminary checklist will have to be refined and updated according to contributions from the medical community and the expanding body of methodological research. It is currently not clear whether a statistical analysis is always necessary or appropriate, but all trial reports should include a discussion about whether use of concomitant or rescue medication may have influenced the results and, if so, in which direction. An extension of the CONSORT statement regarding the reporting of concomitant and rescue medication would be helpful.
In this review of 265 placebo-controlled trials of pharmacotherapy for low back pain and neuropathic pain, ambiguous procedures and incomplete reporting of the use of concomitant and rescue analgesics were widespread. These shortcomings hinder the interpretation, critical appraisal, and replication of trial results. Thus, improvements, particularly regarding reporting quality, are warranted.
Conflict of interest statement
The authors declare no conflicts of interest.
Appendix A. Supplemental digital content
Supplemental digital content associated with this article can be found online at http://links.lww.com/PAIN/A870.
Supplemental video content
Video content associated with this article can be found online at http://links.lww.com/PAIN/A871.
. Altman DG, Moher D, Schulz KF. Improving the reporting of randomised trials: the CONSORT Statement and beyond. Stat Med 2012;31:2985–97.
. Chou R, Deyo R, Friedly J, Skelly A, Weimer M, Fu R, Dana T, Kraegel P, Griffin J, Grusing S. Systemic pharmacologic therapies for low back pain
: a systematic review for an American College of Physicians Clinical Practice GuidelineSystemic Pharmacologic Therapies for Low Back Pain
. Ann Intern Med 2017;166:480–92.
. Chou R, Deyo R, Friedly J, Skelly A, Hashimoto R, Weimer M, Fu R, Dana T, Kraegel P, Griffin J, Grusing S, Brodt E. Noninvasive treatments for low back pain
. Comparative effectiveness review no. 169. (Prepared by the Pacific Northwest evidence-based practice center under contract no. 290-2012-00014-I.) AHRQ publication no. 16-EHC004-EF. Rockville, MD: Agency for Healthcare Research and Quality, 2016.
. Clark J, Schall R. Assessment of combined symptom and medication scores for rhinoconjunctivitis immunotherapy clinical trials. Allergy 2007;62:1023–8.
. Clifford DB, Simpson DM, Brown S, Moyle G, Brew BJ, Conway B, Tobias JK, Vanhove GF, Group NCS. A randomized, double-blind, controlled study of NGX-4010, a capsaicin 8% dermal patch, for the treatment of painful HIV-associated distal sensory polyneuropathy. J Acquir Immune Defic Syndr 2012;59:126–33.
. Dai F, Silverman DG, Chelly JE, Li J, Belfer I, Qin L. Integration of pain score and morphine consumption in analgesic clinical studies. J Pain 2013;14:767–77 e768.
. Dechartres A, Trinquart L, Atal I, Moher D, Dickersin K, Boutron I, Perrodeau E, Altman DG, Ravaud P. Evolution of poor reporting and inadequate methods over time in 20 920 randomised controlled trials included in Cochrane reviews: research on research study. BMJ 2017;357:j2490.
. Duff JM, Leather H, Walden EO, LaPlant KD, George TJ. Adequacy of published oncology randomized controlled trials to provide therapeutic details needed for clinical application. J Natl Cancer Inst 2010;102:702–5.
. Dworkin RH, Turk DC, Peirce-Sandner S, Baron R, Bellamy N, Burke LB, Chappell A, Chartier K, Cleeland CS, Costello A, Cowan P, Dimitrova R, Ellenberg S, Farrar JT, French JA, Gilron I, Hertz S, Jadad AR, Jay GW, Kalliomäki J, Katz NP, Kerns RD, Manning DC, McDermott MP, McGrath PJ, Narayana A, Porter L, Quessy S, Rappaport BA, Rauschkolb C, Reeve BB, Rhodes T, Sampaio C, Simpson DM, Stauffer JW, Stucki G, Tobias J, White RE, Witter J. Research design
considerations for confirmatory chronic pain clinical trials: IMMPACT recommendations. PAIN 2010;149:177–93.
. European Medicines Agency Guideline on adjustment for baseline covariates. London, United Kingdom: European Medicines Agency, 2013.
. Finnerup NB, Attal N, Haroutounian S, McNicol E, Baron R, Dworkin RH, Gilron I, Haanpää M, Hansson P, Jensen TS, Kamerman PR, Lund K, Moore A, Raja SN, Rice AS, Rowbotham M, Sena E, Siddall P, Smith BH, Wallace M. Pharmacotherapy for neuropathic pain
in adults: a systematic review and meta-analysis. Lancet Neurol 2015;14:162–73.
. Finniss DG, Kaptchuk TJ, Miller F, Benedetti F. Biological, clinical, and ethical advances of placebo effects. Lancet 2010;375:686–95.
. Hoffmann TC, Glasziou PP, Boutron I, Milne R, Perera R, Moher D, Altman DG, Barbour V, Macdonald H, Johnston M, Lamb SE, Dixon-Woods M, McCulloch P, Wyatt JC, Chan AW, Michie S. Better reporting of interventions: template for intervention description and replication (TIDieR) checklist and guide. BMJ 2014;348:g1687.
. Huwiler-Müntener K, Jüni P, Junker C, Egger M. Quality of reporting of randomized trials as a measure of methodologic quality. JAMA 2002;287:2801–4.
. Katz N. Methodological issues in clinical trials of opioids for chronic pain. Neurology 2005;65(12 suppl 4):S32–49.
. Kieburtz K, Simpson D, Yiannoutsos C, Max M, Hall C, Ellis R, Marra C, McKendall R, Singer E, Dal Pan G. A randomized trial of amitriptyline and mexiletine for painful neuropathy in HIV infection. Neurology 1998;51:1682–8.
. Lewis JA, Jonsson B, Kreutz G, Sampaio C, van Zwieten-Boot B. Placebo-controlled trials and the declaration of Helsinki. Lancet 2002;359:1337–40.
. Mason L, Moore RA, Derry S, Edwards JE, McQuay HJ. Systematic review of topical capsaicin for the treatment of chronic pain. BMJ 2004;328:991.
. Mhaskar R, Djulbegovic B, Magazin A, Soares HP, Kumar A. Published methodological quality of randomized controlled trials does not reflect the actual quality assessed in protocols. J Clin Epidemiol 2012;65:602–9.
. Moher D, Hopewell S, Schulz KF, Montori V, Gøtzsche PC, Devereaux P, Elbourne D, Egger M, Altman DG. CONSORT 2010 explanation and elaboration: updated guidelines for reporting parallel group randomised trials. J Clin Epidemiol 2010;63:e1–37.
. Parkes MJ, Callaghan MJ, Tive L, Lunt M, Felson DT. Responsiveness of single versus composite measures of pain in knee osteoarthritis. J Rheumatol 2018;45:1308–15.
. Schulz KF, Altman DG, Moher D. CONSORT 2010 statement: updated guidelines for reporting parallel group randomised trials. BMC Med 2010;8:18.
. Silverman DG, O'Connor TZ, Brull SJ. Integrated assessment of pain scores and rescue morphine use during studies of analgesic efficacy. Anesth Analgesia 1993;77:168–70.
. Singla NK, Meske DS, Desjardins PJ. Exploring the interplay between rescue drugs, data imputation, and study outcomes: conceptual review and qualitative analysis of an acute pain data set. Pain Ther 2017;6:165–75.
. Stack CB, Localio AR, Griswold ME, Goodman SN, Mulrow CD. Handling of rescue and missing data affects synthesis and interpretation of evidence: the sodium-glucose cotransporter 2 inhibitor example. Ann Intern Med 2013;159:285–8.
. Temple R, Ellenberg SS. Placebo-controlled trials and active-control trials in the evaluation of new treatments. Part 1: ethical and scientific issues. Ann Intern Med 2000;133:455–63.
. Tominaga Y, Koga H, Uchida N, Wanibe M, Hirose K, Matsumura T, Okamoto A, Richarz U, Etropolski M. Methodological issues in conducting pilot trials in chronic pain as randomized, double-blind, placebo-controlled studies. Drug Res 2016;66:363–70.
. White IR, Bamias C, Hardy P, Pocock S, Warner J. Randomized clinical trials with added rescue medication
: some approaches to their analysis and interpretation. Stat Med 2001;20:2995–3008.