This study aimed to achieve consensus from an international group of experts in neuropathic pain epidemiology on (1) the components of a questionnaire instrument for identifying neuropathic pain in epidemiology research, and (2) when this pain could be termed “refractory”. Lack of consensus in the first of these has hindered epidemiology research to date, and contributed to the considerable variance in published prevalence rates. This was not intended to supplant the clinical definition of (possible) NeuP , but to allow this to be partly operationalised in population-based research. There has previously been no consensus on refractoriness, despite several publications purporting to report on refractory neuropathic pain.
Identifying neuropathic pain
In summary, we found that a case definition instrument for neuropathic pain should certainly include five questions about pain characteristics, and one eliciting a clinical history consistent with neuropathic pain (Table 2). An additional six items produced mixed responses from the expert group. Given the need for brevity and focus in most population research questionnaires, we propose that only the six items with >70% consensus be included in a standard case definition instrument for epidemiology research. The five characteristics of pain quality identified by survey respondents are all featured within the three most widely used screening tools for neuropathic pain (S-LANSS, DN4, painDETECT) [9, 31]. The remaining characteristic (a relevant patient history) does not feature in these screening tools but is part of the diagnostic algorithm advocated by Treede et al .
We examined these five characteristics of pain quality against the cut-off scores within each of the screening tools to determine which combination(s) might be used as a case definition tool for epidemiological use:
· DN4: cut-off is score of 4, equivalent to any four items .
· S-LANSS: cut-off is score of 12 or more. Of the five possible combinations using four items each, three combinations reach a positive score. The remaining two combinations score 11 .
· painDETECT: cut-off score is 13 or more. If three items are rated at least 3 (strongly) and one item 4 (very strongly), a positive score is reached .
Based on this analysis, we propose to test the validity of a case definition tool for epidemiological use using any four characteristics of pain quality in combination with questions eliciting ‘a relevant patient history’. It is not proposed that this tool would replace detailed history and examination, which must remain as cornerstones in the clinical setting.
Determining refractory neuropathic pain
For neuropathic pain to be considered refractory, based on the consensus reported above, we propose that:
· It should have had a trial of treatment with at least four drugs of known effectiveness in neuropathic pain
· Each of these drugs should have been tried for at least three months or until adverse effects prevent adequate dosage
· Despite the above treatment, the intensity of pain should have reduced by less than 30%, or should remain at a level of at least 5 on a 0–10 scale; and/or it should continue to contribute significantly to poor quality of life.
We propose that the duration of neuropathic pain, by itself does, not determine whether it is refractory (though it is clearly relevant when considering the duration of any trials of treatment). We also propose that, while non-pharmacological treatments should be considered, and may be effective, neuropathic pain can be considered refractory even in their absence.
Strengths and limitations of the study
Delphi surveys are an established method of achieving consensus in complex areas and/or where consensus has not previously been able to be reached . For example previous studies have used Delphi methods to develop a standard definition for back pain for use in prevalence studies (a consensus of 28 experts in back pain research from 12 countries) , and to develop guidelines for the management of hip and knee osteoarthritis (a consensus of 16 experts from four medical disciplines and six countries) . The composition of the expert panel is important, and should comprise a reasonable number of individuals, representing a widespread range of experts in the relevant discipline(s).
The response rate to this international study was good, with participation from 25 (73.5% of eligible) invited experts in NeuP in round one, and from 19 (76%) of these experts participating in all three rounds. These included participants from a range of clinical and research disciplines and a number of countries worldwide, all of whom had been previously recognized through international peer selection as experts in their fields. Considerable weight can therefore be given to the items on which clear consensus was achieved. Although the nature of work in this area meant that most of the respondents knew members of the research team, it is unlikely that this would have contributed importantly to any bias in the responses: responses were anonymous, and there appeared to be no hesitation in expressing forthright views. It would have been possible to include more experts, increasing the number of responses and their sources. We were concerned to include individuals who had contributed importantly to the literature in this field, and our sample represents a good selection of these, but it is possible that including experts based in other countries, including the developing world, would have further enriched our responses.
It must be noted, however, that unanimity was not achieved for any item, and there must still be room for flexibility in designing and interpreting questionnaire aiming to identify refractory neuropathic pain. For example, whilst we have proposed that an adequate trial of at least four drugs is required before NeuP should be considered “refractory”, good consensus was nearly achieved (66.7%) on a trial of three drugs. Indeed, two of the experts approached stated their view that it was inadvisable to try to identify neuropathic pain through a questionnaire, preferring instead a clinical assessment. We recognise the distinction between diagnosing neuropathic pain using clinical assessment in order to initiate treatment, and estimating the likelihood of neuropathic pain in a population using proxy criteria in order to understand epidemiological factors. Whilst we respect the importance of rigorous clinical examination in providing a detailed assessment of neuropathic pain, we maintain that this is not practical for research in studies of the sample size required for accurate epidemiology. We are not proposing the above case definition questionnaire as a “gold standard” for neuropathic pain, but as a rigorous consensus, for research purposes only, which may now be tested in the field. Clinical examination is required to identify and assess “definite” neuropathic pain , (though this, too, must be based on consensus  rather than any “gold standard”). It is likely that our questionnaire will be capable of approximating “possible” neuropathic pain in large samples .
Implications of the findings
It is anomalous that, while there was strong consensus that “refractory” and “pharmacoresistant” neuropathic pain are different from each other, there was consensus (84%) on the definition of “refractory” neuropathic pain. Although the a priori level of 70% consensus was not achieved for the importance of trying non-pharmacological treatments, agreement approached this (63%), and free text comments supported it, and these treatments are therefore likely to represent the main difference between these two concepts. Psychological treatments and pain management programme, for example, may be effective in neuropathic pain [34, 35]. Although “pharmacoresistant” neuropathic pain is not itself a consensus term, as our “refractory” neuropathic pain is, it is based on international consensus of a treatment paradigm of efficacious drugs [14, 36].
These and other consensuses on pharmacological treatment, [36–39] provide an excellent basis on which clinicians can determine the order and nature of drugs to be used in neuropathic pain. Importantly, they also provide a framework for inquiring about treatment trials in epidemiology research. Whereas most of these focus on treatment in specialist settings, the recent guidelines from the National Institute for Health and Clinical Excellence (NICE) in the UK  focused on the non-specialist setting. The recommendations in the NICE guideline are broadly consistent with determining refractoriness at the point of specialist referral (that is, after a trial of three or four effective drugs).
It is important, as a far as possible, to use standard definitions and tools in research, including epidemiology. Good epidemiology provides a foundation basis for identifying resource, educational, treatment and prevention strategies, and for the subsequent evaluation of these. The use of standard tools allows comparison between different geographical areas, clinical and demographic subgroups, and periods of time. This study has allowed progress towards this for research on (refractory) neuropathic pain. Further work is now required to validate our proposed criteria in a general population study, and to review results of such a study in comparison with existing data, based on the range of instruments currently available. Our proposals represent a distilled compromise between the best features of existing instruments, with consensus from experts internationally.