Journal of

Oral Rehabilitation

Journal of Oral Rehabilitation 2010 37; 766770

Improving TMD classication using the Delphi technique

M. T. JOHN
Department of Diagnostic & Biological Sciences, School of Dentistry; Division of Epidemiology and Community Health, School of Public Health, University of Minnesota, Minneapolis, MN, USA

SUMMARY The classication of temporomandibular disorders (TMD) is still controversial. Consensus methods such as the Delphi technique, a method that polls experts anonymous opinion in an iterative process with controlled feedback and statistical aggregation of group response, could be valuable to improve this challenging topic. The article illustrates the application of the Delphi technique for deciding whether the terms myalgia or myofascial pain should be used in a TMD classication system and discusses the techniques potential for TMD classication in general. In three Delphi rounds, 14 TMD experts from the Division of TMD and

Orofacial Pain of the University of Minnesota reached a consensus about which TMD diagnoses should be included in a TMD classication system. They preferred the term myofascial pain over myalgia. The Delphi technique has the potential to provide answers to complex questions in TMD classication, e.g., TMD nomenclature and range as well as scope of conditions included in a future TMD classication system. KEYWORDS: temporomandibular disorders, classication, consensus, Delphi technique, nomenclature Accepted for publication 18 April 2010

Introduction
There is no agreed classication system for temporomandibular disorders (TMD), and widely used classication systems such as the one suggested by the American Academy of Orofacial Pain (1) or the Research Diagnostic Criteria for Temporomandibular Disorders (2) differ substantially in the number of the conditions and the criteria for each TMD subtype. Clearly, a classication accepted by many TMD stakeholders would facilitate communication among clinicians, researchers and patients. However, deriving a new classication system is a complex task. Even if knowledge synthesis using systematic reviews and meta-analyses should be an integral part of such an effort, a group of experts covering a broad range of expertise will need to be involved as it is commonly done in health status classication projects. These experts interact with each other and ideally reach a consensus about the topic. Consensus methods are available to facilitate this process (3). One of the most widely used methods is
2010 Blackwell Publishing Ltd

the Delphi technique which is dened as a process of reaching a consensus by the anonymous solicitation and comparison of the views of experts (4) is a structured process providing controlled feedback regarding the polled (expert) opinions. This technique offers advantages for decision making in groups or committees which are often dominated by individuals with vested interests. In particular, participants can change their opinions in anonymity in an iterative process guided by statistical measures summarizing the group opinion. However, not only conceptual advantages for reaching a consensus but also practical benets when organizing the consensus effort such as the involvement of many experts with global representation and reasonable burden using email internet make the technique a promising tool for future TMD classication efforts. I will illustrate the application of the Delphi technique for deciding whether the terms myalgia or myofascial pain should be used in a TMD classication system. I will discuss the techniques potential to improve TMD classication in general.
doi: 10.1111/j.1365-2842.2010.02109.x

IMPROVING TMD CLASSIFICATION USING THE DELPHI TECHNIQUE

importance and the condence ratings (question 1 and 3) added as extra columns in the excel sheet. For question 2, the proportion of positive answers, i.e., the per cent of panelists recommending the inclusion of the diagnosis in TMD classication, was given. Provided with the ndings from the rst Delphi round, the panelists answered again the three questions. This process asking panelists to score the questions again while considering the ndings from the previous round was continued until a priori set criterion for consensus among the panelists was reached. This denition of a consensus among the panelist was based on the correlation of the importance ratings, i.e., a consensus was deemed present when the panelists answers correlated, that is, agreed with each other. As a criterion measure for a consensus, Graham et al. (7) suggested using Cronbachs alpha (8) as a measure of internal consistency of the panel. Cronbachs alpha is a function of the average inter-correlation among the items (in this study the panelists). Cronbachs alpha should therefore reect the extent of consensus within the group for the importance of that item (7), and higher alpha values from round to round would therefore indicate a more homogenous group opinion. We used a criterion of alpha 090 to stop the consensus process. This criterion is based on guidelines who recommend this level of minimum reliability of clinical application (9). In addition, we examined the correlation of importance ratings between each panelist and the groups of panelists and the standard deviations for ratings for each diagnosis as indicators for a consensus process. Each panelists responses should correlate at least moderately with the group [according to Cohen correlations between 030 and 049 are considered medium and correlations between 050 and 100 are high (10)], and correlations should be similar across panelists. In a consensus process, standard deviations should decrease from round to round, indicating that panelists decreased the variability of their ratings for diagnoses.

767

Methods
The present study to determine whether TMD experts prefer to term myalgia over myofascial pain was part of a larger feasibility study to explore what diagnoses should be included in a (future) TMD classication system (5). In this larger study, TMD experts were provided with a list of diagnoses from the RDC TMD Validation Project (6) that could be derived by a gold standard examiner (an experienced clinician using patient-reported, clinical-examination-derived, and radiological data) for diagnosing patients with TMD. These diagnoses were considered a comprehensive set of possible diagnoses for patients with TMD.

TMD experts, the study subjects The subjects of the Delphi study were 13 dentists and 1 physical therapist (panelists in the remainder of the manuscript) who were presently or formerly involved in TMD patient care and were associated with the Division of TMD and Orofacial pain of the University of Minnesota.

Delphi technique The panelists received by email a detailed explanation about the study and an invitation to participate. An email followed with an attached excel sheet that contained the set of 47 TMD diagnoses of the RDC TMD validation project as rows. In the rst round of the Delphi study, panelists were requested to ll in three columns with the following three questions: 1 How important to you are the following diagnoses in a TMD classication system on an 11-point rating scale from 0 completely unimportant to 10 extremely important? 2 Should the diagnosis be included in a TMD classication [0 - no, 1 - yes]? 3 How condent are you with your recommendation about the diagnosis on a 11-point rating scale from 0 not condent at all to 10 extremely condent? The panelists should ll in the requested information and sent the excel le back by email. When asked about additional diagnoses for TMD that would be useful in TMD classication, panelists added 3 diagnoses, creating a nal set of 50 diagnoses. In the second round of the Delphi process, panelists received medians and interquartile ranges for the
2010 Blackwell Publishing Ltd

Terms myalgia and myofascial pain used in TMD classication Among the set of 50 diagnoses were the terms myalgia and myofascial pain. The ratings for these two diagnoses will be analysed in this study to illustrate the Delphi technique. The importance ratings (question 1) were used in the consensus denition and the monitoring of the consensus process. The inclusion ratings

768

M. T. JOHN
(question 2) will be used to characterize the group opinion together with the condence ratings (question 3) that indicated how comfortable the panelists were with their opinions.
Table 2. Proportion of experts suggesting the use of the term myalgia or myofascial pain in a future TMD classication system Proportion (95% condence interval) Delphi round Myalgia 62 (3192)% 62 (3192)% 54 (2285)% Myofascial Pain 77 (50100)% 85 (62100)% 92 (76100)%

Results
Process of achieving a consensus among panelists Panelists reached the consensus criterion after three rounds, and individual panelists correlated with the group opinion between 031 and 067 (Table 1). These correlations were medium or high according to guidelines (10) and met a priori expectations. The mean of all standard deviations for individual diagnoses decreased in round 1 from 26 over 20 to 14 in round 3. Except for 4 diagnoses, the magnitude of the standard deviation decreased for all diagnoses. These ndings were interpreted as evidence for a successful consensus process in the group of panelists.

1 2 3

myalgia (Table 2). With more Delphi rounds, the proportion of panelists supporting myofascial pain increased, whereas the proportion of panelists supporting the term myalgia decreased. The panelists condence in their ratings was high. Mean condence ratings in all three rounds were 79. Whereas the condence for the use of the term myalgia in TMD classication decreased from 82 (95% CI: 70 93) to 79 (95% CI: 6394), the condence for the term myofascial pain increased from 86 (95% CI: 72 99) to 98 (95% CI: 95100).

Use of the term myalgia or myofascial pain in TMD classication The consensus process started with more panelists favouring the term myofascial pain compared to
Table 1. Correlations between individual panelists and the group opinion and Cronbachs alphas as a measure of the homogeneity of group opinion for three Delphi iteration rounds Panelist-group correlation Panelist 1 2 3 4 5 6 7 8 9 10 11 12 13 14 Cronbachs alpha for panelists (lower limit of 95% condence interval) 1. round 052 052 039 067 065 031 051 051 064 054 055 048 068 048 077 (067) 2. round 069 076 066 085 044 044 069 062 072 077 059 080 074 068 088 (083) 3. round 070 087 077 096 055 059 078 092 083 083 076 054 079 093 093 (091)

Discussion
This study illustrated the use of the Delphi technique for TMD classication. It demonstrated that a group of TMD experts is able to reach a consensus about TMD nomenclature. TMD experts preferred the term myofascial pain compared with the term myalgia in TMD classication.

Myalgia and myofascial pain what is the difference? Myalgia simply means muscle pain. It is a non-specic and descriptive term whereas myofascial pain is dened more narrowly. According to the International Association for the Study of Pain, myofascial pain depends upon the demonstration of a trigger point (tender point) and reproduction of the pain by manoeuvres which place stress upon proximal structures or nerve roots (11). According to this denition, myofascial pain is a subcategory of myalgia. Clark supports this view when pointing out that For local and regional myalgia, if some additional anatomic features are added, such as taut bands, trigger points within the taut band, and referred pain sensations upon sustained compression of the trigger point, then the term myalgia can be changed to myofascial pain. (12) Accordingly, the original (2) as well as the revised RDC TMD (6, 13)
2010 Blackwell Publishing Ltd

IMPROVING TMD CLASSIFICATION USING THE DELPHI TECHNIQUE

present myofascial pain as a diagnosis under muscle disorders. Okeson presents myofascial pain under muscle pain with other entities such as myositis (14). On the other hand, the International Classication of Diseases, ninth revision (ICD-9) presents myalgia and myositis (unspecied) together under code 7291 (15), a code frequently used in TMD practice. That myofascial pain and myalgia exist as separate diagnoses that are on par is suggested by the classication system of the American Academy of Orofacial Pain (1). It lists local myalgia, centrally mediated myalgia, and myofascial pain as separate disorders under masticatory muscle disorders. The ambiguous use of myofascial pain is emphasized by the International Association of the Study of Pain in their Core Curriculum for Professional Education in Pain. Here it is pointed out in chapter 33 Muscle and myofascial pain that the term myofascial pain includes a general denition that refers to all muscle pain and a specic denition that refers to pain caused by myofascial trigger points (16). The panelists ratings in this study reect the preference of myofascial pain in the literature. The TMD experts favoured myofascial pain over myalgia and they were condent in their opinion. Whereas the consensus about including certain diagnoses into TMD classication was reached after three rounds, extrapolating the observed trend that myofascial pain became more and myalgia less popular during the study, a fourth round would probably have been useful to polarize the groups opinion even more for the specic myofascial pain myalgia question. This situation highlights the fact that there is no predetermined number of rounds in a Delphi process. Usually studies include 2 to 3 rounds (17), but up to 7 rounds was reported in a review (18). published Delphi studies from 2000 onward (19). None of them was used for temporomandibular disorders a eld where challenges in the diagnosis and classication of conditions still remain after publication of RDC TMD version 2. Anderson et al. mention four major areas for improvement of physical (axis I) conditions: (i) differentiating TMD from other pain conditions, (ii) nomenclature, (iii) range and scope of conditions and (iv) standards for diagnostic sensitivity and specicity in future investigations (20). The present study is an example how to improve nomenclature, but the Delphi techniques characteristics of anonymity, iteration, controlled feedback and statistical aggregation of group response (18) would also be useful for the other areas. In addition, application of the Delphi technique is not restricted to improve the classication of physical TMD conditions. Some of the challenges in TMD pain-related disability and psychological status (axis II) characterization such as determining the number and character of axis II constructs as mentioned by Anderson et al. (20) could also benet from the Delphi technique. As a conclusion, the technique is a promising approach to complex questions in TMD classication, nally aiming for integration of TMD pain into orofacial pain classication.

769

Acknowledgments
Im grateful to Mrs. Karen Decker, PT and Drs. Eric Schiffman, Cori Herrman, Ana Velly, Richard Madden, Donald Nixdorf, Gary Anderson, Jim Fricton, Mariona Mulet, Estephan Moana, Lauriant Azangue, Radwa Sobieh, and Patricia Fernandes for their participation in the study.

References
Future use of the Delphi technique in improving TMD classication The Delphi method is widely used. Above and beyond areas such as forecasting where it was developed originally, applications in medicine are now abundant (3). It has been estimated that over 1400 publications demonstrating the use of the technique in a health care setting have been published up to 2005 (17). Applications involve forecasting, consensus in clinical problemsolving, education, development of clinical guidelines, and education (17). However, in dentistry the technique is not widely used a review presented 21
2010 Blackwell Publishing Ltd

1. de Leeuw JR. Orofacial pain: guidelines for assessment, diagnosis, and management. Chicago: Quintessence, 2008. 2. Dworkin SF, LeResche L. Research diagnostic criteria for temporomandibular disorders: review criteria, examinations and specications, critique. J Craniomandib Disord. 1992;6: 301355. 3. Jones J, Hunter D. Consensus methods for medical and health services research. BMJ. 1995;311:376380. 4. MSN Encarta. Delphi technique. 2009. Available at: http:// encarta.msn.com/dictionary_561533021/delphi_technique.html, accessed 14 December 2009. 5. John MT, Schiffman E. Application of the Delphi technique to derive a TMD taxonomy. J Dent Res. (AADR abstracts). 2008;87:1229.

770

M. T. JOHN
6. Schiffman EL, Truelove EL, Ohrbach R, Anderson GC, John MT, List T et al. The research diagnostic criteria for temporomandibular disorders. I: overview and methodology for assessment of validity. J Orofac Pain. 2010;24:724. 7. Graham B, Regehr G, Wright JG. Delphi as a method to establish consensus for diagnostic criteria. J Clin Epidemiol. 2003;56:11501156. 8. Cronbach LJ. Coefcient alpha and the internal reliability of tests. Psychometrika. 1951;16:297334. 9. Bland JM, Altman DG. Statistics notes: Cronbachs alpha. BMJ. 1997;314:572. 10. Cohen J. Statistical power analyses for the behavioral sciences. New York: Academic press, 1977. 11. Task Force on Taxonomy of the International Association for the Study of Pain. Classication of chronic pain: descriptions of chronic pain syndromes and denitions of pain terms. Seattle, Wash.: IASP Press; 1994. 12. Clark GT. Classication, causation and treatment of masticatory myogenous pain and dysfunction. Oral Maxillofac Surg Clin North Am. 2008;20:145157, v. 13. Schiffman EL, Ohrbach R, Truelove EL, Tai F, Anderson GC, Pan W et al. The research diagnostic criteria for temporomandibular disorders. V: methods used to establish and validate revised Axis I diagnostic algorithms. J Orofac Pain. 2010;24:6378. 14. Okeson JP. The classication of orofacial pains. Oral Maxillofac Surg Clin North Am. 2008;20:133144. 15. The International classication of diseases, 9th revision, clinical modication: ICD-9-CM. Ann Arbor, Mich.: Commission on Professional and Hospital Activities; 1986. 16. Charlton JE. (ed). Core curriculum for professional education in pain. Seattle: IASP Press, 2005. 17. Thangaratinam S, Redman CWE. The Delphi technique. The Obstetrician & Gynaecologist. 2005;7:120125. 18. Rowes G, Wright G. The Delphi technique as a forecasting tool: issues and analysis. Int J Forecast. 1999;15:353375. 19. Cramer CK, Klasser GD, Epstein JB, Sheps SB. The Delphi process in dental research. J Evid Based Dent Pract. 2008;8:211220. 20. Anderson GC, Gonzalez YM, Ohrbach R, Truelove EL, Sommers E, Look JO et al. The research diagnostic criteria for temporomandibular disorders. VI: future directions. J Orofac Pain. 2010;24:7988.

Correspondence: Mike T. John, DDS, MPH, PhD, Associate Professor, Department of Diagnostic & Biological Sciences, School of Dentistry; Adjunct Associate Professor, Division of Epidemiology and Community Health, School of Public Health, University of Minnesota, Minneapolis, MN, USA. E-mail: mtjohn@umn.edu

2010 Blackwell Publishing Ltd