Defining chronic pain in epidemiological studies: a systematic review and meta-analysis.

The objective was to document the operational definitions applied in epidemiological studies of chronic pain and to examine whether pain definitions and other methodological factors are systematically related to prevalence estimates. MEDLINE, EMBASE, and PsychINFO were searched for original research reports with study samples of at least 1000 individuals, excluding studies of less than 5 out of 15 selected body regions and studies solely concerned with specific pain conditions. Meta-analyses and meta-regressions were applied with random effects models; covariates were geography, sampling year, survey method, sampling frame, participation rate, percentage women of all participants, pain duration, and pain location. Of 6791 hits, 86 studies were included in the syntheses. The phrasing, content, and combinations of the chronic pain definition criteria were highly inconsistent, with virtually no 2 studies from independent research groups using the exact same criteria. Prevalence estimates ranged from 8.7% to 64.4%, with a pooled mean of 31%. Huge heterogeneity was shown in all forest plots. Prevalence estimates were significantly related to survey method (β = -10.8 [95% confidence interval: -17.2 to -4.4]), but it only counted for a small fraction of the between-studies variation in the estimates. There were also interaction effect of survey method by sex (female-male prevalence ratio [95% confidence interval]: questionnaire = 1.20 [1.16 to 1.25], and interview = 1.38 [1.29 to 1.47]). The other covariates investigated were not significantly related to the prevalence estimates. Researchers and clinicians should be aware of the probability that interview survey method of collecting data may give lower chronic pain reporting than questionnaire survey method and that this effect may be stronger in men than women.