Approximately 1 in 5 adults suffer from chronic pain worldwide, and another 1 in 10 adults are diagnosed with chronic pain each year.20 Given its impact and the limitations of long-term opioid use for chronic, noncancer pain,6,42 there is a pressing need for new analgesics.
Despite recent increases in academic and industry support for basic and translational pain research,36 only a fraction of novel drugs that have shown promise in animal models have advanced to regulatory approval.5,23 For example, although a number of promising targets have been identified in preclinical testing, including NK1 receptor antagonists, TRPV1 antagonists, and NMDA receptor antagonists, they have failed to exhibit efficacy in clinical trials.30,33,53 Although the reasons for these failures vary, some commentators have posited that inadequate design,1,33 conduct,11,34 and reporting9,39 of preclinical experiments may explain some of the challenges encountered in translational pain research. These arguments echo those raised by commentators in other disease areas, such as cancer and stroke.4,43,50
Sound preclinical study design and reporting takes on special importance in pain research, since studies require at least a transient induction of unrelieved suffering in animals. Animal findings may also inform clinical decision-making in pain management. Off-label prescription is common in pain, and owing to the scarcity of large randomized trials for various pain states,19,21 the rationale for prescription is often based on small clinical studies8,46,52 and preclinical research.17,31
Here, we report a systematic review and meta-analysis of preclinical studies of the blockbuster drug pregabalin (Lyrica). In the United States, pregabalin is approved for the management of fibromyalgia, neuropathic pain associated with spinal cord injury, diabetic peripheral neuropathy and postherpetic neuralgia, and as adjunctive therapy in the treatment of partial seizures in adults. In the European Union, pregabalin is also approved for the treatment of generalized anxiety disorder, and its label extends to all neuropathic pain. Pregabalin has also been tested against a large number of pain indications in preclinical experiments. Our primary objective was to explore the reported quality and design characteristics of studies testing the efficacy of pregabalin in behavioral models of pain. We also tested the relationship between design features and reported effect sizes.
2.1. Literature search
To identify all in vivo animal studies testing the efficacy of pregabalin (“efficacy studies”), we queried Ovid MEDLINE In-Process & Other Non-Indexed Citations and Ovid MEDLINE (dates of coverage from 1948 to 2018), Embase Classic and Embase database (dates of coverage from 1974 to 2018), and BIOSIS Previews (dates of coverage from 1969 to 2018) on April 10, 2018, using a search strategy described previously (see supplemental methods, available at http://links.lww.com/PAIN/A911).18 Search results were pooled into an EndNote library, and duplicates were removed.
2.2. Study selection
Screening was performed at both the title/abstract level and at the full-text level. Inclusion criteria for preclinical studies were (1) original reports, (2) full-text articles, (3) English language, (4) contained at least one experiment assessing a behavioral measure of pain using a neuropathic, inflammatory, disease model, or acute etiology in live, nonhuman animals, (5) used pregabalin in a control, comparator, or experimental context, and (6) published before January 2018. As we were interested in a big-picture view of the relationship between design practices and preclinical effect size, we lumped all behavioral pain experiments together, regardless of the etiology/measure used.
For inclusion in the quantitative meta-analysis, additional criteria for preclinical studies were (1) reported sample size, (2) included time-course data for experiments that lasted at least 120 minutes, (3) provided prebaseline (ie, before injury) and baseline (ie, before drug) measurements, and (4) measured variance as SD or SE of the mean.
2.3. Data extraction
All included preclinical studies were evaluated at the study-level (extraction elements included broad study-level questions, eg, conflict of interest disclosure, funding type, etc.), but only those with eligible experiments (defined as those evaluating the effect of pregabalin on behavioral measures of pain) were advanced to experiment-level extraction.
We defined internal validity as the extent to which an experiment is able to isolate the pharmacological intervention as the cause of measured effects. We defined construct validity as the strength of the relationship between experimental systems and the human scenarios they were intended to simulate. We extracted design elements addressing internal and construct validity based on consensus design practices identified in a systematic review of validity threats in preclinical research.24 In accordance with other theoretical treatments of validity, we defined external validity as the extent to which a causal relationship can hold up over variations in models, outcomes, settings and treatments.24,45 As such, and in contrast to internal and construct validity (which are assessed at the experiment level), addressing threats to external validity involves conducting replication studies under varied experimental conditions.45 As in previous work,25,32 we operationalized external validity using an index that summed the number of etiology/measure combinations and species tested for a given study. For example, if a single study contained 2 experiments, one testing the efficacy of pregabalin against a chronic constriction injury (a neuropathic etiology) using von Frey fibers in rats and a second testing pregabalin against spontaneous pain behaviors after intraplantar formalin injection in mice (an inflammatory etiology), it was scored a 3 (1 point for the second species, 1 point for the second etiology, and 1 point for the second measure). Etiology and measure classifications were made based on a review article of animal models in pain.33 For example, experiments that applied a noxious stimulus to a body part leading to nocifensive withdrawal, or to other simple behaviors, were deemed to be of an acute etiology. Those experiments testing the effect of an inflammatory agent (eg, formalin, capsaicin, carrageenan, complete Freund's adjuvant, etc.) on behavioral responses were classified as inflammatory. Neuropathic etiologies were those that used surgical techniques such as nerve transection, ligation, compression, and/or neuritis to produce partial damage to nerves. Finally, those experiments that modeled a clinical syndrome (eg, causalgia, painful diabetic neuropathy, post herpetic neuralgia, chemotherapy-induced neuropathic pain, arthritis, etc.) were classified as disease etiologies.
We also extracted all necessary data for calculating effect sizes. We chose to extract only those studies that included raw, time-course data, so that the percentage of maximum possible effect (% MPE) of pregabalin could be calculated for each experiment at all reported doses. For each experiment, data presented graphically were extracted using the graph digitizer software GraphClick (Arizona Software). A training set of studies was extracted by 2 independent coders using Numbat Meta-Analysis Extraction Manager7; discrepancies in double coding were reconciled through discussion, and coding criteria were refined. The agreement rate on the training set exceeded 80%. The remaining studies were then extracted by a single coder.
2.4. Analysis and statistics
The analgesic efficacy of pregabalin was expressed as the % MPE. It represents the percentage difference between the measured response for each experimental group (pregabalin arm) and complete analgesia (ie, a return to prebaseline outcome measures at all time points after drug administration). As such, comparisons with control groups were not made because animals in the pregabalin arm were treated as their own controls. We calculated % MPE by first measuring the area under the time-course curve (AUC) using the trapezoidal rule for all reported doses of pregabalin at 120 minutes after drug administration. To calculate AUC, a multiplier for the linear expression of the trapezoidal rule was calculated. Because the trapezoidal rule estimates the MPE as a linear combination of the individual estimates on the time-course curve, the same linear combination was used to estimate the variance of the calculated MPE assuming independent measurement of error.
Percentage of MPE was then calculated as the AUCexperimental/AUCcomplete × 100. Therefore, a % MPE of 0% signifies no analgesia, whereas a % MPE of 100% signifies complete analgesia. In theory, % MPE can exceed 100% as treated animals can perform better (ie, have lower pain sensitivity) than their prebaseline levels.
For experiments testing multiple doses of pregabalin, we took a weighted average of the outcomes and created a pooled effect size (except in dose–response analyses).
Pooled effect sizes were calculated using a random-effects model using the DerSimonian and Laird method,15 in R version 3.3.3. This method represents a generic inverse-variance approach to meta-analysis that incorporates an assumption that different studies are estimating different, yet related, intervention effects.26 Exploratory subgroup meta-analyses were performed to test the impact of pain assay and validity elements on pooled effect sizes for experiments. P-values were calculated by a 2-sided independent-group t test and provide evidence of heterogeneity of intervention effects (variation in effect estimates beyond chance); statistical significance was set at P ≤ 0.05. As this analysis was exploratory, we did not adjust for multiple hypothesis testing.
Publication bias was evaluated using a funnel plot with Duval and Tweedie's trim and fill method of estimating missing studies and adjusting the post estimate using R v3.3.3.
Dose–response relationships were investigated for all models in aggregate; individual effect sizes (% MPE) were plotted against reported dose (mg/kg). To avoid the confounding effects of discontinuous dosing, we included only experiments that tested a single, bolus dose of pregabalin.
All extraction criteria and analyses were prespecified in a timestamped protocol (supplemental methods, available at http://links.lww.com/PAIN/A911).
3.1. Study characteristics
Our literature search identified 1119 citations (Fig. 1). Of these, 244 studies assessed the efficacy of pregabalin in nonhuman animals; 40 studies were excluded from further evaluation because they tested pregabalin in nonpain indications (eg, epilepsy). Our final sample thus included 204 preclinical pain studies, corresponding to 531 unique experiments assessing the efficacy of pregabalin in behavioral models of pain, in a total of 29,186 nonhuman animals.
Properties of studies included in our qualitative synthesis are shown in Table 1. More than half of all studies (53%; n = 109) were conducted in Asia, and 50% (n = 103) were supported by a combination of public and private funding. Most studies (92%; n = 184) were published after pregabalin received its first regulatory approval in the United States, in 2004. More than half of the identified efficacy experiments (62%; n = 126) used pregabalin as an active comparator in a study testing a different drug.
3.2. Design elements addressing validity threats
Adherence to consensus design practices aimed at reducing threats to internal validity was low (Fig. 2A). Twenty-eight percent of experiments (n = 151) reported using randomized allocation to treatment, whereas 33% (n = 177) reported using blinded outcome assessment. Four percent of preclinical experiments (n = 23) reported the performance of a sample size calculation, and 7% (n = 35) addressed the attrition of animals during experiments. Of the 531 experiments in our sample, 23% (n = 123) evaluated pregabalin dose–response relationships.
Design elements aimed at maximizing the correspondence between experimental setup and clinical scenarios (construct validity) were also variable (Fig. 2B). Experiments relied disproportionately on male animals (75%; n = 400) and most commonly tested either a neuropathic etiology (45%; n = 237) or disease model (39%; n = 206). Nearly half of all experiments (48%; n = 253) measured efficacy using a mechanical pain measure.
There were limited attempts to test the robustness of findings (external validity), and little variation occurred in testing within single reports (Fig. 2C). Nearly three-quarters of all studies (73%; n = 149) tested 2 or fewer etiology/measure combinations, and most experiments were conducted in a single species. Only 2 studies (<1%) tested the efficacy of pregabalin in more than one species.
3.3. Effect sizes in preclinical studies
Four hundred twenty-seven experiments (80%) were excluded from the quantitative synthesis because they did not report elements required for meta-analysis (eg, sample size, a measure of variance, etc.; see Fig. 1 for details). Exclusions were most frequently due to the absence of relevant time-course data (42%; n = 178). The 104 experiments included in our quantitative analysis used a reported 5284 animals.
All experiments demonstrated some analgesic effect for pregabalin, regardless of the etiology/measure tested. The efficacy of pregabalin ranged from 24% of complete analgesia to 136%, and the overall pooled % MPE was 71% (95% confidence interval [CI] = 66%-75%; Fig. 3).
3.4. Impact of validity practices on effect sizes
Implementation of internal validity practices did not demonstrate a consistent relationship in terms of lower validity producing higher pooled effects sizes. For example, experiments that reported a precise sample size had larger pooled effect sizes than experiments that did not (73% vs 65%, P > 0.05; Fig. 4A), although the difference was not significant. There were also no significant differences in pooled effect sizes between experiments that reported using randomization (71% vs 70%, P > 0.05), blinded treatment allocation (71% vs 71%, P > 0.05), or blinded outcome assessment (71% vs 70%, P > 0.05) and those that did not.
There were also no clear impacts on effect sizes of experiments using greater construct valid designs. There were no significant differences in pooled effect sizes between experiments testing male animals vs female animals (71% vs 69%, P > 0.05; Fig. 4B). Experiments that tested an acute pain etiology had larger pooled effect sizes than other etiologies, although the difference was not statistically significant (99% vs 69%, P > 0.05).
Those experiments that reported a private, for-profit funding source had larger pooled effect sizes than those that did not report a sponsor (77% vs 67%, P > 0.05; Fig. 4C), although the relationship was not statistically significant.
For external validity, there was no clear relationship between external validity scores and pooled effect sizes (Fig. 4D).
3.5. Impact of publication bias
The funnel plot, presented in Figure 5, suggested asymmetry. Trim and fill analysis suggested a 27% overestimation of effect size across all pain states, with an adjusted % MPE of 51% (95% CI = 46%-56%) compared with an unadjusted % MPE of 71% (95% CI = 66%-75%).
3.6. Dose–response effects
We were unable to observe a dose–response relationship over 3 orders of magnitude (0.1-400 mg/kg) for all experiments testing a single, bolus dose of pregabalin (P > 0.05; R2 = 0.03; Fig. 6).
Our systematic review of preclinical pain studies demonstrates that the blockbuster status of pregabalin in the clinic was accompanied by a large volume of animal testing. In total, over 531 experiments testing the efficacy of pregabalin in behavioral models of pain were identified.
We observed limited implementation of practices that have been endorsed elsewhere1,24 that may address basic threats to validity. First, preclinical studies rarely reported the performance of power calculations to determine sample size, and randomization and blinded treatment allocation were reported less than one-third of the time. Second, model systems were often not matched well with clinical scenarios. For example, researchers relied predominantly on male animals, whereas findings from epidemiological and clinical studies demonstrate that women are at substantially higher risk for many common pain conditions.3 Furthermore, recent recommendations from funding bodies, including the NIH and CIHR, stress the importance of sex-balancing in biomedical research.10,48 In addition, most preclinical studies relied on stimulus-evoked measurements, which do not measure pain, but rather the hypersensitivity that often accompanies it.33 By contrast, clinical pain is very often of a spontaneous nature.2,41 For example, in a study comparing the clinical pain symptoms of 1235 neuropathic pain patients, all of whom reported spontaneous pain, percentages of patients with sensory gain only (including both mechanical and thermal hypersensitivity) ranged from 5% to 30%, and including patients with sensory loss raised these percentages to 75% at most.29 Third, few models were used to demonstrate analgesic activity. These practices are similar to those reported in a large systematic review of animal models of chemotherapy-induced peripheral neuropathy,12 and in other disease areas such as stroke,27,44,49 cancer,4,25 and ALS.38 Finally, our analyses suggest publication bias that may have led to an overestimation of effect of 27%; this is in keeping with figures reported in cancer and stroke.25,32,44 We also did not observe a dose–response relationship in preclinical studies, a finding that is in keeping with other meta-analyses of preclinical studies.25 Exposures are rarely mentioned in preclinical papers, and this limits the interpretation of effect because it is not possible to put the dose into context with in vitro data and also cannot be related to the clinic.
We also found that pregabalin showed strong analgesic properties, regardless of the pain models tested. The pooled effect size for all preclinical experiments was 71% of prebaseline measurements. One way of interpreting this is that pregabalin analgesic properties were highly reproducible within and across different pain models (ie, effects in one model supported inferences that effects would be observed in a different model). However, pregabalin preclinical studies do not seem to show inferential reproducibility22 with respect to clinical generalizability. For example, in experiments testing acute, nociceptive pain etiologies, pregabalin demonstrated robust pooled effect sizes despite the fact that clinical efficacy in acute and chronic, non-neuropathic pain is virtually nil.16,35 If preclinical studies were predictive of clinical response, one might expect reported effect sizes would suggest better discrimination between human pain states that respond to pregabalin and those that do not. Another interpretation is that although sensitivity to detect potential analgesia was high during the preclinical development of pregabalin, it may have come at the cost of specificity to detect differences in efficacy between pain states. Moreover, pregabalin is increasingly recognized as being only moderately effective for the treatment of neuropathic pain, further calling into question the robustness of preclinical findings. For example, a recent systematic review suggests that more than half of those treated with pregabalin will not have achieved worthwhile pain relief.14 In fibromyalgia, pregabalin offers moderate pain relief to only a minority of people (number need to treat to benefit one person = 7.2-11).13
Although rigorous preclinical design is required for internal validity, we did not detect a clear relationship between experimental design and effect sizes. For example, our analyses did not show a consistent relationship between the use of techniques such as randomization and smaller effect sizes; this is consistent with some reports25,32 but not with others.27,28 That this relationship has not consistently held up in studies suggests that other factors may play a greater role in explaining discordances between effects observed in nonhuman animals and those in patients. For example, latent environmental factors, such as housing, diet, and experimenter sex, may affect stress levels in the preclinical testing environment, influencing both behavioral and nonbehavioral outcomes.34,47 Although standardization is conventionally used to reduce within-experiment variation, some commentators have suggested that environmental heterogenization, ie, systematically increasing within-experiment variation relative to between-experiment variation, might be the key to reducing spurious results and improving preclinical reproducibility.40,51 Moreover, the assessment of pain in behavioral models is also complicated by the fact that many drugs, including pregabalin, cause sedation, which can mimic the effects of analgesics in conventional assays of pain-stimulated behaviors (eg, tail or paw withdrawal). Drugs can decrease the expression of pain-stimulated behaviors not only by reducing sensory sensitivity to the putative pain stimulus, but also by reducing motor competence to emit the measured pain behavior. Although the sedative effects of drugs can be tested using behavioral assays, such as the rotarod test, they were rarely reported in publications. In our sample, only 41 studies (20%) included both an analysis of sedation and efficacy, limiting our ability to systematically assess potency ratios for sedation and efficacy. Assays of pain-depressed behavior, including burrowing, nesting, or pain-related depression of locomotion, should also be used as they may reduce analgesia false-positives.37 Likewise, pregabalin has demonstrated anxiolytic properties, which might also confound pain experiments. Future work will examine the frequency of sedation testing in the preclinical development of pregabalin.
Our systematic review has a number of limitations. First, we relied on what authors reported in publications. It is possible that certain experimental practices, such as blinding and randomization, were more widely used but simply not reported; alternatively, studies may have stated that they were randomized when they were in fact not. This analysis also relied on published reports, and restriction of searches to the English language may have excluded some articles; this analysis likely does not reflect the totality of preclinical testing of pregabalin. Second, preclinical effect sizes were calculated based on the availability of time-course data, which means that many experiments were performed and reported but could not be analyzed. Third, because this article takes a big-picture view of the preclinical testing of pregabalin, we lumped all behavioral pain experiments together when testing the impact of validity practices, regardless of etiology or measure, as we did not expect qualitative interactions between the types of behavioral pain experiments. Nevertheless, we observed a high level of variability between and within types of pain experiments, and our pooled results should be interpreted with caution. Fourth, we calculated effect size and its associated variance using group-level data within each experiment. In so doing, experiment length was standardized, but one consequence of using the trapezoidal rule to calculate AUC is that it assumes that measurement error at successive time points is independent. If the measurement error is correlated, the SE of the AUC will be underestimated. Furthermore, we ignored clustering at the study level; for example, experiments performed within a single study were likely to be correlated with respect to validity items; that is, investigators were likely to report blinding across all experiments within their study. Finally, there are many preclinical experimental procedures (eg, habituation or handling) that are critical for validity, particularly in behavioral pain experiments,34 but they were not captured in our analysis.
This work contributes to a better understanding of the techniques that investigators are using in preclinical pain research, and the impact of various experimental design practices on effect sizes. Although some study design practices, such as more widespread use of randomization and blinding, may be relatively easy to implement, our analysis suggests that such changes alone are unlikely to dramatically improve the concordance between effects observed in behavioral models of pain and those in clinical settings.
Conflict of interest statement
The authors have no conflicts of interest to declare.
Appendix A. Supplemental digital content
Supplemental digital content associated with this article can be found online at http://links.lww.com/PAIN/A911.
Supplemental video content
A video abstract associated with this article can be found at http://links.lww.com/PAIN/A912.
The authors thank Amanda K. Hakala, BSc, for her help coding the pregabalin training set.
This work was supported by the CIHR Grant EOG 111391 to J. Kimmelman and the Louise and Alan Edwards Foundation's Edwards PhD Studentships in Pain Research to C.A. Federico.
. Andrews NA, Latremoliere A, Basbaum AI, Mogil JS, Porreca F, Rice AS, Woolf CJ, Currie GL, Dworkin RH, Eisenach JC, Evans S, Gewandter JS, Gover TD, Handwerker H, Huang W, Iyengar S, Jensen MP, Kennedy JD, Lee N, Levine J, Lidster K, Machin I, McDermott MP, McMahon SB, Price TJ, Ross SE, Scherrer G, Seal RP, Sena ES, Silva E, Stone L, Svensson CI, Turk DC, Whiteside G. Ensuring transparency and minimization of methodologic bias in preclinical pain research: PPRECISE considerations. PAIN 2015;157:901–9.
. Backonja MM, Stacey B. Neuropathic pain symptoms relative to overall pain rating. J Pain 2004;5:491–7.
. Bartley EJ, Fillingim RB. Sex differences in pain: a brief review of clinical and experimental findings. Br J Anaesth 2013;111:52–8.
. Begley CG, Ellis LM. Drug development: raise standards for preclinical cancer research. Nature 2012;483:531–3.
. Borsook D, Hargreaves R, Bountra C, Porreca F. Lost but making progress—Where will new analgesic drugs come from? Sci Transl Med 2014;6:249sr243.
. Busse JW, Wang L, Kamaleldin M, Craigie S, Riva JJ, Montoya L, Mulla SM, Lopes LC, Vogel N, Chen E, Kirmayr K, De Oliveira K, Olivieri L, Kaushal A, Chaparro LE, Oyberman I, Agarwal A, Couban R, Tsoi L, Lam T, Vandvik PO, Hsu S, Bala MM, Schandelmaier S, Scheidecker A, Ebrahim S, Ashoorion V, Rehman Y, Hong PJ, Ross S, Johnston BC, Kunz R, Sun X, Buckley N, Sessler DI, Guyatt GH. Opioids for chronic noncancer pain: a systematic review and meta-analysis. JAMA 2018;320:2448–60.
. Carlisle BG. Numbat meta-analysis extraction manager, 2017. Available at: https://www.bgcarlisle.com/blog/numbat3/
. Chou R, Fanciullo GJ, Fine PG, Adler JA, Ballantyne JC, Davies P, Donovan MI, Fishbain DA, Foley KM, Fudin J, Gilson AM, Kelter A, Mauskop A, O'Connor PG, Passik SD, Pasternak GW, Portenoy RK, Rich BA, Roberts RG, Todd KH, Miaskowski C. Clinical guidelines for the use of chronic opioid therapy in chronic noncancer pain. J Pain 2009;10:113–30.
. Clark JD. Preclinical pain research: can we do better? Anesthesiology 2016;125:846–9.
. Clayton JA, Collins FS. Policy: NIH to balance sex in cell and animal studies. Nature 2014;509:282–3.
. Currie GL, Delaney A, Bennett MI, Dickenson AH, Egan KJ, Vesterinen HM, Sena ES, Macleod MR, Colvin LA, Fallon MT. Animal models of bone cancer pain: systematic review and meta-analyses. PAIN 2013;154:917–26.
. Currie GL, Angel-Scott HN, Colvin L, et al. Animal models of chemotherapy-induced peripheral neuropathy: a machine-assisted systematic review and meta-analysis. PLoS Biology 2019;17:e3000243.
. Derry S, Cording M, Wiffen PJ, Law S, Phillips T, Moore RA. Pregabalin
for pain in fibromyalgia in adults. Cochrane Database Syst Rev 2016;9:Cd011790.
. Derry S, Bell RF, Straube S, Wiffen PJ, Aldington D, Moore RA. Pregabalin
for neuropathic pain in adults. Cochrane Database Syst Rev 2019;1:Cd007076.
. DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin trials 1986;7:177–88.
. Fabritius ML, Strom C, Koyuncu S, Jaeger P, Petersen PL, Geisler A, Wetterslev J, Dahl JB, Mathiesen O. Benefit and harm of pregabalin
in acute pain treatment: a systematic review with meta-analyses and trial sequential analyses. Br J Anaesth 2017;119:775–91.
. Fallon M, Hoskin PJ, Colvin LA, Fleetwood-Walker SM, Adamson D, Byrne A, Murray GD, Laird BJ. Randomized double-blind trial of pregabalin
versus placebo in conjunction with palliative radiotherapy for cancer-induced bone pain. J Clin Oncol 2016;34:550–6.
. Federico CA, Carlisle B, Kimmelman J, Fergusson DA. Late, never, or nonexistent: the inaccessibility of preclinical evidence for new drugs. Br J Pharmacol 2014;171:4247–54.
. Federico CA, Wang T, Doussau A, Mogil JS, Fergusson D, Kimmelman J. Assessment of Pregabalin
Postapproval Trials and the Suggestion of Efficacy for New Indications: A Systematic Review. JAMA Intern Med 2019;179:90–7.
. Goldberg DS, McGee SJ. Pain as a global public health priority. BMC Public Health 2011;11:770.
. Goodman CW, Brett AS. Gabapentin and pregabalin
for pain—is increased prescribing a cause for concern? N Engl J Med 2017;377:411–14.
. Goodman SN, Fanelli D, Ioannidis JP. What does research reproducibility mean? Sci Transl Med 2016;8:341ps312.
. Hay M, Thomas DW, Craighead JL, Economides C, Rosenthal J. Clinical development success rates for investigational drugs. Nat Biotechnol 2014;32:40–51.
. Henderson VC, Kimmelman J, Fergusson D, Grimshaw JM, Hackam DG. Threats to validity
in the design and conduct of preclinical efficacy studies: a systematic review of guidelines for in vivo animal experiments. PLoS Med 2013;10:e1001489.
. Henderson VC, Demko N, Hakala A, MacKinnon N, Federico CA, Fergusson D, Kimmelman J. A meta-analysis of threats to valid clinical inference in preclinical research of sunitinib. eLife 2015;4:e08351.
. Higgins JPT, Thomas J, Chandler J, Cumpston M, Li T, Page MJ, Welch VA (editors). Cochrane Handbook for Systematic Reviews of Interventions version 6.0 (updated July 2019). Cochrane, 2019. Available from www.training.cochrane.org/handbook
. Macleod MR, O'Collins T, Howells DW, Donnan GA. Pooling of animal experimental data reveals influence of study design and publication bias. Stroke 2004;35:1203–8.
. Macleod MR, van der Worp HB, Sena ES, Howells DW, Dirnagl U, Donnan GA. Evidence for the efficacy of NXY-059 in experimental focal cerebral ischaemia is confounded by study quality. Stroke 2008;39:2824–9.
. Maier C, Baron R, Tolle TR, Binder A, Birbaumer N, Birklein F, Gierthmuhlen J, Flor H, Geber C, Huge V, Krumova EK, Landwehrmeyer GB, Magerl W, Maihofner C, Richter H, Rolke R, Scherens A, Schwarz A, Sommer C, Tronnier V, Uceyler N, Valet M, Wasner G, Treede RD. Quantitative sensory testing in the German Research Network on Neuropathic Pain (DFNS): somatosensory abnormalities in 1236 patients with different neuropathic pain syndromes. PAIN 2010;150:439–50.
. Mao J. Current challenges in translational pain research. Trends Pharmacol Sci 2012;33:568–73.
. Mathieson S, Maher CG, McLachlan AJ, Latimer J, Koes BW, Hancock MJ, Harris I, Day RO, Billot L, Pik J, Jan S, Lin CC. Trial of pregabalin
for acute and chronic sciatica. N Engl J Med 2017;376:1111–20.
. Mattina J, MacKinnon N, Henderson VC, Fergusson D, Kimmelman J. Design and reporting of targeted anticancer preclinical studies: a meta-analysis of animal studies investigating sorafenib antitumor efficacy. Cancer Res 2016;76:4627–36.
. Mogil JS. Animal models of pain: progress and challenges. Nat Rev Neurosci 2009;10:283–94.
. Mogil JS. Laboratory environmental factors and pain behavior: the relevance of unknown unknowns to reproducibility and translation. Lab Animal 2017;46:136–41.
. Moore RA, Straube S, Wiffen PJ, Derry S, McQuay HJ. Pregabalin
for acute and chronic pain in adults. Cochrane Database Syst Rev 2009:Cd007076.
. National Institutes of Health, U.S. Department of Health and Human Services. Estimates of funding for various research, condition, and disease categories (RCDC), 2018. Available at: https://report.nih.gov/categorical_spending.aspx
. Negus SS. Core outcome measures in preclinical assessment of candidate analgesics. Pharmacol Rev 2019;71:225–66.
. Perrin S. Preclinical research: make mouse studies work. Nature 2014;507:423–5.
. Rice AS, Cimino-Brown D, Eisenach JC, Kontinen VK, Lacroix-Fralish ML, Machin I, Mogil JS, Stohr T. Animal models and the prediction of efficacy in clinical trials of analgesic drugs: a critical appraisal and call for uniform reporting standards. PAIN 2008;139:243–7.
. Richter SH, Garner JP, Wurbel H. Environmental standardization: cure or cause of poor reproducibility in animal experiments? Nat Methods 2009;6:257–61.
. Scholz J, Mannion RJ, Hord DE, Griffin RS, Rawal B, Zheng H, Scoffings D, Phillips A, Guo J, Laing RJ, Abdi S, Decosterd I, Woolf CJ. A novel tool for the assessment of pain: validation in low back pain. PLoS Med 2009;6:e1000047.
. Schuchat A, Houry D, Guy GP Jr. New data on opioid use and prescribing in the United States. JAMA 2017;318:425–6.
. Sena E, van der Worp HB, Howells D, Macleod M. How can we improve the pre-clinical development of drugs for stroke? Trends Neurosci 2007;30:433–9.
. Sena ES, van der Worp HB, Bath PM, Howells DW, Macleod MR. Publication bias in reports of animal stroke studies leads to major overstatement of efficacy. PLoS Biol 2010;8:e1000344.
. Shadish W, Cook T, Campbell D. Experimental and quasi-experimental designs for generalized causal inference. Boston: Houghton Mifflin, 2002.
. Shanthanna H, Gilron I, Rajarathinam M, AlAmri R, Kamath S, Thabane L, Devereaux PJ, Bhandari M. Benefits and safety of gabapentinoids in chronic low back pain: a systematic review and meta-analysis of randomized controlled trials. PLoS Med 2017;14:e1002369.
. Sorge RE, Martin LJ, Isbester KA, Sotocinal SG, Rosen S, Tuttle AH, Wieskopf JS, Acland EL, Dokova A, Kadoura B, Leger P, Mapplebeck JC, McPhail M, Delaney A, Wigerblad G, Schumann AP, Quinn T, Frasnelli J, Svensson CI, Sternberg WF, Mogil JS. Olfactory exposure to males, including men, causes stress and related analgesia in rodents. Nat Methods 2014;11:629–32.
. Tannenbaum C, Schwarz JM, Clayton JA, de Vries GJ, Sullivan C. Evaluating sex as a biological variable in preclinical research: the devil in the details. Biol Sex Differ 2016;7:13.
. Tsilidis KK, Panagiotou OA, Sena ES, Aretouli E, Evangelou E, Howells DW, Al-Shahi Salman R, Macleod MR, Ioannidis JP. Evaluation of excess significance bias in animal studies of neurological diseases. PLoS Biol 2013;11:e1001609.
. van der Worp HB, de Haan P, Morrema E, Kalkman CJ. Methodological quality of animal studies on neuroprotection in focal cerebral ischaemia. J Neurol 2005;252:1108–14.
. Voelkl B, Vogt L, Sena ES, Wurbel H. Reproducibility of preclinical animal research improves with heterogeneity of study samples. PLoS Biol 2018;16:e2003693.
. Wiffen PJ, Derry S, Moore RA, Aldington D, Cole P, Rice AS, Lunn MP, Hamunen K, Haanpaa M, Kalso EA. Antiepileptic drugs for neuropathic pain and fibromyalgia—an overview of Cochrane reviews. Cochrane Database Syst Rev 2013;11:Cd010567.
. Yezierski RP, Hansson P. Inflammatory and neuropathic pain from bench to bedside: what went wrong? J Pain 2018;19:571–88.