Related Experiment Videos
A capture-recapture estimate of inflammatory bowel disease prevalence: the Florence population-based study
D Palli1, G Masala, G Trallori
1Analytical Epidemiology Section, CSPO, Firenze, Italy. md0632@mclink.it
Background:
A recent population-based epidemiological study identified the patients with a diagnosis of ulcerative colitis or Crohn's disease resident in the metropolitan Florence area in the period 1978-1992 and defined their vital status on 31 December, 1992.
Aims:
To estimate the completeness of Inflammatory Bowel Disease prevalent case ascertainment in the study area.
Subjects And Methods:
In a Registry, specifically developed during the study, 767 patients fulfilled the criteria for the definition of prevalent cases as of 31 December, 1992. At the same time, we had access to an independent source of potential patients: the Regional Health Department kept a list of all Inflammatory Bowel Disease patients allowed free access to specific health care provided only to selected diagnostic categories. We then compared the two different sources and used a capture recapture analysis to estimate the number of cases missed by both sources.
Results:
A total of 331 patients were reported by both sources, 436 were found only in the Registry while a large group of potential cases not present in our Registry was found only in the List. After careful confirmation of the diagnosis for each individual patient and contact with his/her physician, we identified an additional 102 cases that were included in the final population series of 869 prevalent cases. This capture-recapture analysis led to a revised estimate of 1,003 prevalent cases, suggesting that 134 patients had been missed by both sources, resulting in an Inflammatory Bowel Disease prevalence rate of 186 per 100,000.
Conclusions:
According to this method our previous study underestimated the true prevalence of 13.4% (95% confidence interval: 9.8-16.6%). On account of some degree of negative dependence between the two sources the loss was probably in the lower range of this interval. Completeness of case ascertainment should be evaluated and discussed in all studies designed to provide population-based estimates for health care planning.