Defining constipation to estimate its prevalence in the community: results from a national survey

Different definitions of constipation have been used to estimate its prevalence in the community but this creates difficulties when comparing results from various studies. This study explores the impact of different definitions on prevalence estimates in the same population and compares the performance of simple definitions with the Rome III criteria. The prevalence of constipation in a large nationally representative sample of community-dwelling adults was estimated using five simple definitions of constipation and compared with definitions based on the Rome III criteria. The sensitivity, specificity, and positive and negative predictive values, were calculated for each definition using the Rome III criteria as the gold standards for chronic and sub-chronic constipation. Prevalence estimates for the five simple definitions ranged from 9.4 to 58.9%, while the prevalence estimates using the Rome III criteria were 24.0% (95%CI: 22.1, 25.9) for chronic constipation and 39.6% (95%CI: 37.5, 41.7) for sub-chronic constipation. None of the simple definitions were adequate compared to the Rome III criteria. Self-reported constipation over the past 12 months had the highest sensitivity (91.1%, 95%CI: 88.8, 93.4) and negative predictive value (94.5%, 95%CI: 93.1, 96.1) compared to the Rome III criteria for chronic constipation but an unacceptably low specificity (51.3%, 95%CI: 48.8, 53.8) and positive predictive value (37.1%, 95%CI: 34.4, 39.9). The definition used to identify constipation within a population has a considerable impact on the prevalence estimate obtained. Simple definitions, commonly used in research, performed poorly compared with the Rome III criteria. Studies estimating population prevalence of constipation should use definitions based on the Rome criteria where possible.