Abstract
OBJECTIVE--To review a series of conceptual and methodological problems encountered in surveys primarily devoted to pain disorders. CRITERIA FOR INCLUSION AND EXCLUSION OF ARTICLES--Published reports were systematically collected by electronic database searches (Medline), citations in existing publications, and through personal contacts. Relevant articles from clinical and epidemiological research on pain were included and special attention was given to epidemiological research on back pain. CONCLUSIONS--Surveys of pain disorders should be based on a multidimensional pain model that includes nociceptive input, pain perception, suffering, and pain behaviour as major components. Because of the limited applicability of diagnostic procedures or genuine "non-specificity" of pain states, or both, epidemiological surveys may result in a considerable proportion of cases without an identifiable pathophysiological basis. Staging and grading procedures for pain disorders (as distinguished from classification) may comprise various aspects of pain perception: regional distribution, pain intensity, temporal characteristics, sensory qualities, and dimensions of cognitive-emotional appraisal. Description of temporal development and chronification (staging) should refer to different components of the multidimensional pain model. Explicit a posteriori procedures for grading are preferable to implicit grading based on question wording. Evidence from several sources suggests that localistic concepts of pain may be misleading. Identification of complex pain syndromes should be one primary target for epidemiological pain surveys. Of the many factors that may impair the reliability and validity of data collected in pain surveys, recall biases seem to deserve special attention.