The relative risk of cervical cancer rises the longer a woman is on the pill. According to a meta-analysis based on data from 28 studies,1women who use the pill for 10 or more years are 2.2 times as likely as never-users to develop cervical cancer, whereas the relative risks among users for 5-9 years and fewer than five years are 1.6 and 1.1, respectively. Among women infected with human papillomavirus (HPV), the virus associated with virtually all cases of cervical cancer, long-term pill users are 2.5 times as likely as never-users to develop cervical cancer. The results were broadly similar among the studies, despite the wide variation in the individual variables that were statistically controlled for.
To assess the relationship between use of hormonal contraceptives (primarily the pill) and cervical cancer, the researchers pooled data from four cohort and 24 case-control studies published from 1986 through 2002 on users' risks of developing cervical cancer relative to that of never-users. The researchers examined relative risks of cervical cancer for three durations of pill use--fewer than five years (short-term), 5-9 years (medium-term) and 10 or more years (long-term). Wherever possible, they categorized use as current or recently discontinued (i.e., within eight years) versus discontinued further in the past (i.e., eight or more years ago). Each of the 28 studies controlled for one or more of the following variables: the woman's HPV infection status, number of sexual partners, history of cervical screening, smoking and history of barrier method use. Some studies focused exclusively on the developed or the developing world, on HPV-positive or HPV-negative women, on invasive or in situ cervical cancers, or on squamous cervical cancers or adenocarcinomas of the cervix. Only the most fully adjusted data were used in the meta-analysis.
The studies, about half of which were conducted in the developing world, included 12,531 women with invasive or in situ cervical cancer. Among controls, the proportion reporting ever-use of the pill ranged from 19% to 92% in the studies conducted in developed countries and from 25% to 65% in those carried out in developing countries; the proportion who took the pill for longer than five years was 18-62% in developed countries and 8-20% in developing countries.
For all 28 studies combined, the relative risk of cervical cancer rose with increasing duration of pill use: The risk of diagnosis among short-term pill users relative to never-users was 1.1, whereas the relative risk for medium-term users was 1.6, and that for long-term users was 2.2. The general finding of increasing risk with increasing duration of pill use was broadly similar among all 28 studies, irrespective of the control variables used.
Data from the 12 studies that assessed cervical cancer risk among HPV-positive women suggest that the relative risk rose with longer periods of use, but it was significantly elevated for long-term users only (2.5). The three studies that examined the association between injectable use and cervical cancer indicate that use of this method for five or more years was associated with a slight, marginally significant increase in cervical cancer risk (1.2).
Eleven studies had data on both how recently pill users had stopped their method and their total duration of use. These suggested that the risk of cervical cancer might decline after use is stopped. Among short-term users, women who were still using the pill or had stopped within the last eight years had a higher relative risk than those who had stopped at least eight years ago (1.4 vs. 1.1). Similarly, among long-term pill users, the relative risk of cervical cancer was 2.1 among women who were still on the pill or had gone off it more recently, whereas the relative risk among those who had stopped in the more distant past was 1.4 and only marginally significant.
According to the researchers, combining the studies' results indicated that cervical cancer risk rose with increasing duration of pill use "in virtually every way that the data were examined." They warn, however, that the variation in methodologies, study design and HPV measurement suggest that their summary indicators should be interpreted with caution. Further, none of the studies controlled simultaneously for all potentially confounding variables, and although the data suggest that risk may decline after women stop hormonal contraceptive use, the researchers acknowledge that their meta-analysis was "hampered by the lack of published data cross-classifying women by duration of use and time since last use." Since the question of whether and how long effects of hormonal contraceptive use persist is "critical," to answer it they recommend that the data be reanalyzed with uniformly defined variables.--L. Remez
1. Smith JS et al., Cervical cancer and use of hormonal contraceptives: a systematic review, Lancet, 2003, 361(9364):1159-1167.