To the Editor:
In contrast to numerous case-control studies detailed elsewhere,1,2 the only conventional cohort study published to date found no reduction in ulcerative colitis (UC) risk after appendectomy (standardized incidence ratio [SIR] = 0.87; 95% confidence interval [CI] = 0.69–1.07) (n = 84).1 Hallas et al3 report a relative risk measure similar to an SIR, with a value of 1.06 (95% CI = 0.95–1.19) for the postappendectomy period (n = 330). Thus, available cohort-based evidence favors no association between appendectomy and subsequent UC risk. Hallas et al conclude differently based on within-cohort comparisons of UC rates before and after appendectomy.
The main effect measure computed by Hallas and colleagues is neithera SIR (as claimed p.174) nor a ratio of post-to-preappendectomy SIRs (p.175), but a post-to-preappendectomy ratio of directly (not indirectly, p.174) age- and sex-standardized rates; in other words, it is a standardized rate ratio (SRR) using Denmark 1990 as the standard. Considering the characteristic age-incidence curves for UC that differ between the sexes,4 age may be inadequately controlled by use of 10-year bands (age standardization) and the broad age categories (0–39, 40–59, 60+ years) common to men and women in the Poisson model.
For meaningful comparison of 2 directly standardized rates, ratios of stratum-specific rates in the compared populations should be constant.5 No information is provided by Hallas et al as to whether this requirement is met, and Table 2 suggests otherwise. The dramatic finding for persons aged <15 years at start of follow-up in 1981 illustrates the discordance. Taken literally, appendectomy should reduce UC risk by 87% for persons born 1966+ (regardless of age at operation) but by only 16% for persons born before 1966. The “protection” was seen only for persons who had appendectomies after age 40 (all born before 1966) (Table 2). The authors acknowledge this “asymmetry.” (p.177)
The preappendectomy rate used as denominator in the main effect measure is a directly standardized UC incidence rate among persons surviving on average 6.3 years until appendectomy. Such conditioning on the future, a nontrivial requirement among older subjects, further calls into question the value of the main findings. Moreover, there seem to be logical inconsistencies. While appendectomy without appendicitis conferred no “protection” (SRR = 0.97; 95% CI = 0.73–1.31), appendectomy with appendicitis did (0.68;0.55–0.85) (Table 2). These results suggest that appendicitis, not appendectomy, is protective.
The authors have missed an opportunity to clarify the subject. Their report will encourage continued speculations to explain an apparent inverse relationship, when that finding seems to rest on methodologically flawed studies. The repeated suggestion of a protective effect might even revitalize undue speculations that appendectomy may be a useful prophylactic measure among first-degree relatives of UC patients.6
The Danish Epidemiology Science Centre; Statens Serum Institut; Copenhagen, Denmark; firstname.lastname@example.org
1. Frisch M, Johansen C, Mellemkjær L, et al. Appendectomy and subsequent risk of inflammatory bowel diseases. Surgery
2. Frisch M, Biggar RJ. Appendectomy and protection against ulcerative colitis. N Engl J Med
3. Hallas J, Gaist D, Sørensen HT. Does appendectomy reduce the risk of ulcerative colitis? Epidemiology
4. Langholz E, Munkholm P, Nielsen OH, Kreiner S, Binder V. Incidence and prevalence of ulcerative colitis in Copenhagen county from 1962 to 1987. Scand J Gastroenterol
5. Esteve J, Benhamou E, Raymond L. Descriptive epidemiology. Statistical Methods in Cancer Research, Volume IV
. 1st ed. Lyon, France: International Agency for Research on Cancer (IARC), 1994.
6. Schattner A. Appendicectomy in ulcerative colitis. Lancet