Decline in IBD Incidence in Québec

We read with interest Dr. Kaplan's editorial, “Pitfalls and Perils of Using Administrative Databases to Evaluate the Incidence of Inflammatory Bowel Disease Overtime,” in which he described the advantages and disadvantages of using administrative health data in inflammatory bowel disease (IBD) research and addressed some methodological issues that can bias IBD incidence estimates.1 Kaplan concluded that the decline in IBD incidence observed in Québec was due to flaws in study methodology that produced a mixing of prevalent and incident cases. We agree that a major bias would be introduced if, in a given study, different washout periods were used before disease incidence is identified. This situation would occur if the washout period started from the study baseline, disregarding the fact that subjects enrolled in the health plan at a later point in time would not benefit from the same length washout period. To avoid this bias, the washout period needs to be implemented at the individual level, as was done in the Québec IBD cohort. The 2-year washout period was applied across the board before the first IBD contact that contributed to the case-defining algorithm, regardless of the time of enrollment in the health plan. Kaplan also raised the issue of the lack of validation for the washout period in Québec, assuming that in other provinces, the washout had been implicitly validated along with the case-identifying algorithm itself. However, none of the cited studies indicates that a validation of a washout period was conducted, which would have entailed testing several periods of varying lengths to identify the one that best fits the data.2 One article addressing IBD incidence in Ontario children clearly stated that the 3-year “lookback” period was selected based on the expert opinion of the clinicians conducting the research.3 This approach is common in IBD and other diseases, wherein the length of the washout period is arbitrarily defined by the investigators based on clinical experience and the characteristics of the disease.4,–6 A second methodological issue Kaplan presented as a possible explanation for our results is the use of the last of 4 IBD physician contacts to define incidence instead of the first, as occurred in all other provinces. This is a critical issue, because a subject cannot be counted as an incident case before meeting all IBD case-defining criteria (4 contacts). Considering IBD diagnosis at the first IBD contact rather than at the case-defining one may cause biases of up to 53% in incidence estimates.7 Furthermore, using the first rather than the last contact induces immortal time bias in prospective studies, in that the time required to fulfill all criteria is inherently immortal because every subject with IBD must survive long enough (until the fourth IBD contact) to be declared a case. For example, a study examining the effect of a treatment or intervention occurring after the first IBD contact would show increased survival in IBD cases, since by definition, all cases survive the time it takes to reach the fourth IBD contact, but no such benefit would be seen in the comparison group. Immortal time bias has been thoroughly addressed in the context of pharmacoepidemiology.8 Finally, Kaplan mentioned the lack of validation in Québec of the Alberta case-identifying algorithm as potential explanation for the declining trend in incidence we observed for the 2001–2008 period. However, a declining incidence for Crohn's disease was also observed in Québec for the 1998–2000 period, using an algorithm developed in Manitoba and validated in Québec.9 Furthermore, when the same Manitoba algorithm was applied to our current data, not only was the number of cases identified similar but the declining trend in incidence also remained significant (data not shown). In conclusion, we believe our findings reflect a true decline in incidence rates in Québec. This decline could be part of a change in the epidemiologic pattern for IBD in North America, especially in the context of the decline in Crohn's disease incidence observed among middle age adults in Manitoba,10 the overall decline in IBD incidence in Alberta11 and Nova Scotia,12 and the stabilization of incidence rates in the United States during the past 3 decades.13

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.