Article Text

Download PDFPDF

Epidemiology supports oral contraceptives as a risk factor in Crohn's disease
  1. M ALIC
  1. 3720 Balboa St
  2. San Francisco, CA 94121, USA

    Statistics from

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Editor,—A recent clinical alert (Gut1999;44:311–12) commented on a study of risk factors for relapse in Crohn's disease.1 The author concluded that, unlike the established association with smoking, the link with oral contraceptive use is still controversial. To contribute to this discussion, I have investigated temporal trends in age and sex specific disease incidence and correlated them with the chronology of oral contraceptive use.

    The birth control pill was first introduced in 1960 and soon became widely accepted; 10 million American women were taking it by 1973.2 Concerns about side effects prompted further research, and by the mid-1970s most women taking were taking oral contraceptive pills containing 50 μg or less of oestrogen—a considerable decrease from the 100–150 μg in the pills of the 1960s.

    Most epidemiological studies of Crohn's disease (especially those from the USA) have shown a rapid rise in the incidence of disease between the early 1960s and early 1970s, followed by a plateau phase in the 1980s.3 In my analysis I have used the only two American studies4 5 for which detailed age and sex incidence distributions were available and correlated them to American data on oral contraceptive use.

    Without investigating the incidence trends in a group of 20–29 year old women (these are the most likely users of oral contraceptives), a correlation between the introduction and adoption of oral contraceptives and overall incidence trends of Crohn's disease would not be sufficient to establish oral contraceptive use as a risk factor. From 1964, both studies showed that there was a striking increase in incidence among the 20–29 year old age group, and an increased female to male incidence ratio. Unpublished data from the Olmsted County study5 showed a crude incidence of 26.8 cases per 100 000 person years for this age group in women between 1964 and 1973. This is the highest incidence among all age and sex groups in the entire study period (1940–93), and the highest incidence rise between consecutive time periods. The incidence rise in men for the same period was much less dramatic and the crude rate for the same age group was 17.2. Data for Baltimore showed4 a 9.46-fold incidence increase for for women aged 20–29 between the 1960–3 and 1973 surveys, and only a 2.33 increase for men of the same age group. Again, this jump in female incidence is the most abrupt and the highest incidence among all age and sex groups in all periods. Consistent with the introduction of oral contraceptives with a lower oestrogen content, incidence studies of Crohn's disease reported lessening of predominance in women aged 20–29 from the mid-1970s. Figure 1 shows changing female to male incidence ratio for this age group corresponding to oral contraceptive use.

    Figure 1

    Trends in the use of oral contraceptives and incidence of Crohn's disease as a female to male ratio in the 20–29 year age group.

    Although detailed data on incidence and oral contraceptive use were not available, we used two European studies as controls,6 7and, in the period 1960–5, both had the highest female to male incidence ratio, which corresponded to the rise in the USA.

    The above epidemiological findings are concordant with Timmer and colleagues' study1 and support their explanation that previous use of oral contraceptives is more strongly associated with relapse in Crohn's disease than current use. The change in oestrogen content of oral contraceptives may account for the contradictory findings of a link between oral contraceptive use and Crohn's disease, in studies published in the past 15 years.


    The author would like to thank Edward V Loftus Jr for contributing unpublished data from the Olmsted County study.