During the 1990s in France, the condom became the almost universal method of protection at first sexual intercourse. Has this dramatic progress, linked to AIDS prevention campaigns, slowed or decreased the use of the contraceptive pill ? In this article, based on the contraceptive histories of women aged 18-44, collected in the COCON survey, Clémentine Rossierand Henri Leridonshow that this is not the case. The diffusion of condom use initially reduced the proportion of unprotected first sexual intercourse to a very low level. While the use of the pill tends to be delayed slightly, it has become more common than during the 1980s. At the very beginning of sexual life the condom not only protects against AIDS, it also paves the way for other types of contraception.

In Figures 1A and 1B we have assembled data from five surveys carried out on the general population which included information on condom use at the time of first sexual intercourse. The five surveys are : Analyse des comportements sexuels des jeunes (ACSJ – Analysis of young people’s sexual behaviour) carried out in 1994 on a sample of young people aged 15-18 (Lagrange and Lhomond, 1997), the aforementioned ESFE survey (Toulemon and Leridon, 1995), Baromètre Santé (Health Barometer) 2000 sponsored by the Comité français d’éducation pour la santé (CFES – French Committee on Education for Health) (Guilbert et al., 2001), Connaissances, attitudes, croyances et comportements face au VIH/Sida (Knowledge, Attitudes, Beliefs, and Practice regarding HIV/AIDS – KABP 2001) carried out by the Observatoire régional de santé d’Île-de-France (Regional Surveillance Site – ORS, 2001), and finally the COCON survey. In each of these sources, the proportion of respondents who declared having used a condom at the time of first intercourse was calculated by year in which first intercourse occurred, then smoothed over three years (using the method of moving averages with weights equal to 0.25, 0.50 and 0.25), for the years that allowed observation without truncation in each source. Figure 1A shows the results according to answers given by women ; and Figure 1B shows the average response of both men and women when both sexes were interviewed.

Natural methods (rhythm method, withdrawal) were hardly ever used any more by the end of the 1980s. Hence, the sustained increase in the use of medical methods during the 1990s (essentially the pill among the youngest women) cannot be attributed to a substitution effect. To explain the trend in the context of increasing condom use we analyse the period data from the four successive surveys (1978, 1988, 1994, and 2000) at the observable ages in those surveys (18 years or more). Subsequently we test the quality of the complete contraceptive histories collected in 2000 by comparing them to the data from the period surveys. Finally we use the latter data to provide a more detailed description of the stages involved in the onset of sexuality and in the use of the various contraceptive methods among the cohorts that experienced the major changes of the 1980s and 1990s.

Our analysis of the period data pertains to women aged between 18 and 29. The lower limit is imposed by the structure of the samples of the 1978, 1988, 1994, and 2000 surveys. The upper limit corresponds to the end of the transition period between the beginning of sexuality and entry into a stable relationship initiating the phase of family building. Indeed, at the ages of 18-24, 30% of the women live with a partner, compared to 65% at 25-29 years, and 76% at 30-34 years ; but before age 30 fewer than one out of two unions is formalized by marriage (Cassan et al., 2001). The average age of women at their first marriage was 28 in 2000, and their age at first birth was close to 29 in the same year (Beaumel et al., 2002).

Specific analyses of the 1978 survey (Sardon, 1986) demonstrated the necessity of complementing open-ended questions on contraception along the lines of “Do you or your partner currently use a contraceptive method to avoid pregnancy ?” by a series of follow-up questions in the case of negative reply, to avoid an incorrect reporting of the “natural” methods that are not always perceived as “contraceptive methods” by their users. Simultaneous use of several methods should also be taken into account. In that instance, in order to end up with a simple classification, we define a “principal method”, essentially on the basis of its theoretical effectiveness. The rank ordering used in the surveys from 1978 to 1994, in descending order, was the following : sterilization, pill, IUD, rhythm method, condom, withdrawal, other or not specified (Sardon, 1986). The respective position of the pill and the IUD in this list is unimportant as hardly any woman uses both methods simultaneously. We use the same classification to compare the results of the 2000 survey to those at earlier dates. But when presenting detailed results from the 2000 survey we use a slightly different rank ordering which includes the simultaneous use of the pill and the condom and rates the latter’s effectiveness more highly : sterilization, pill and condom (simultaneously), pill (only), IUD, condom, rhythm method, withdrawal, others.

The 1990s were marked by a substantial rise in contraceptive prevalence among 18-19 year-olds, which increased by 12 points between 1988 and 2000, from 51.6% to 63.3% (Table 1). The change can be broken down into a 10 point increase for the pill, 7 points for the condom (as the main method of contraception), and a 5 point decline for female local methods (diaphragm, ovules, spermicide gel, etc.) and natural methods together. The increase in the use of the pill appears therefore to have largely exceeded the decline in the use of these other methods.

The rise in contraceptive practice is even more marked among women aged 20-24 than among the younger ones, with an increase of 20 points between 1978 and 2000 (Table 2). By decomposing the evolution according to the method used, a linear increase in the use of the pill during the 1980s and 1990s is observed. The same is true for the condom, but only since the late 1980s, a fact which can clearly be linked to the beginning of AIDS prevention campaigns ; the increase remains modest, however. The use of natural methods declines markedly until the early 1990s and then stabilises at a low level. Overall, non-medical methods are replaced by medical methods (in effect the pill) within this age group during the 1980s. In the following decade, however, the increase in the use of the pill no longer corresponds to a substitution.

During the 1990s, the contraceptive behaviour of women aged 25-29 changed in the same way as that of women aged 20-24 (Table 3). They postponed their pregnancies and increasingly used the pill and the condom, the increase in the use of the latter being particularly apparent at the end of the period (4% in 1988, 5.3% in 1994, and 7.3% in 2000). Moreover, condom use varies by cohort. Women who began having sexual relations at the time when AIDS appeared reached 25-29 in the mid-1990s, and it was precisely at that time that condom use increases within this age group.

Changes in the contraceptive practices of young women during the 1990s were a result of four independent but intertwined factors. First of all, although fertility remained stable among women aged 18-19, it declined at 20-24, and to a lesser extent at 25-29, as a result of the continuous rise in age at first birth. Second, this postponement of first births was not accompanied by a rise in the age at first intercourse, as we will see later, even though it is possible that reporting on sexual activity in the surveys has changed. Third, the tendency to replace natural methods with medical methods came to an end during the 1990s, initially among the youngest and later among the other women, with the near disappearance of natural methods. Finally, the impact of AIDS prevention campaigns can be clearly measured from the late 1980s, with the rise in condom use. This increase concerned those aged under 25 in particular as the condom is most frequently used at the beginning of new relationships, which are more frequent at young ages. This effect was combined with a cohort effect, as the spread of the condom occurred mainly among the youngest women at the time when the AIDS epidemic erupted.

Complete contraceptive histories were collected in the COCON survey carried out in 2000 ; the interview method is described in detail in the box below (p. 12). As with any biographical data we must first examine their reliability.

The sample interviewed in 2000 is representative of the female resident population of France at that date. But it is not necessarily representative of women living in France in 1988, for example, and this can cause problems of comparability with the sample interviewed in 1988. Besides the age truncation (women aged 18-44 in 2000 were aged 6-32 in 1988), which can be accounted for in a simple way, a selection process may have taken place. Let us ignore selection through mortality (some women present in 1988 may have died by 2000) as the probability of dying is very low at these ages and its association with contraceptive practice is even lower. Migration could have a more important effect : the population in 2000 included immigrants who had entered France since 1988 and, conversely some women present in 1988 may have left France before 2000. The data on migratory movements over the last decades show that immigration has been higher in France than emigration, as all the censuses show positive net migration. It might be thought that this positive balance masks a high turnover, with migrants remaining in France for a short period only and then being replaced by others in greater numbers. In fact, X. Thierry (2001) has shown that the departure rate after arrival in France does not exceed 35% in the first year, and it may be assumed that it is lower thereafter. Nevertheless, some women interviewed in 2000 report on their situation in 1988, when they were not yet residents in France and could therefore not have been part of the universe covered by the ERN survey. As contraceptive use was significantly lower in their native countries, they should less often report having used contraceptive methods, and this would lead to an under-estimation of past practices in the contraceptive histories. We will see that our results go in the opposite direction.

We consider that the data on reported behaviour at the time of each of the period surveys are reliable. Moreover, their quality was verified in 1978, 1988, and 1994 by comparing reported use of the pill and the IUD with the results of sales figures (de Guibert-Lantoine and Leridon, 1999). We may therefore compare the proportions of users at each age observed in each of the three surveys with the values obtained for the same years in the contraceptive histories collected by the COCON 2000 survey [1]. When there is a difference between the data taken from the period surveys and those from the histories, we consider that the latter over- or under-estimate the use of the various methods.

In their retrospective reporting, the respondents tend to underestimate the periods of non-contraception and of use of natural methods, probably by omitting to report the shortest of these episodes. This results in an over-estimation of periods in which medical methods were used (which include de facto some of the preceding episodes) and in an overall over-estimation of contraceptive use. In their analysis of comparable data collected in 1988, Toulemon and Leridon (1992a) had reached identical conclusions. For the condom, the observed over-estimation is no doubt strongly associated with the normative attitudes that have developed in the wake of HIV prevention campaigns.

The continuous decline in fertility before age 30 as well as, to a lesser extent, the progressive replacement of natural methods by medical methods which ends at this time, explain the continued increase in the prevalence of the pill during the 1990s among women aged between 20 and 29. However, this interpretation cannot be applied to women under 20, whose fertility was already very low at the end of the 1980s and who no longer used natural methods at the time.

Women aged 18-19 reported greater sexual activity in 2000 than in 1988. Does this evolution partly explain the increase in the use of the pill ? It does not necessarily mean that first intercourse occurred at an earlier age in the successive cohorts (Table 4). The median age at first intercourse (which by definition marks the beginning of the contraceptive history) shows only a slight drop between the 1956 and 1980 birth cohorts. The data agree with those of the Analyse des comportements sexuels en France (ACSF) survey taken in 1992 (Spira et al., 1993 ; Bozon and Leridon, 1993), which show that the drop was marked up to the cohorts born around 1956-1960 and that the median age stabilised around 18 years after that (see Figure 3). Analysis of the data by year of first intercourse (first line of Table 6) also gives an impression of great stability during the last twenty-five years, with a median age slightly over 18 years.

A study of the first two episodes of the contraceptive history will help us better understand the simultaneous increase in use of the pill and of the condom among the youngest women.

In order to test whether the progression from the condom to the pill is less frequent when young women have only short relationships with their partners, we selected women whose first relationship lasted less than one year, and compared those who had first intercourse in 1980-1984 and in 1995-2000. Among the women who started their sex life between 1980 and 1984, the first relationship lasted less than 12 months in 38.4% of cases ; 51% of first relationships initiated fifteen years later lasted less than 12 months (this may include first relationships of less than 12 months still under way at the time of the survey).

Complete contraceptive histories are rarely collected, as they are believed a priori to be unreliable. We have shown that the quality of the retrospective data collected during the COCON survey allows a satisfactory analysis of the rank ordering of the various methods in each period and of their evolution over time. As the period surveys on contraceptive use carried out in France at regular intervals did not include information on the behaviour of women aged less than 18 or 20, we have used the contraceptive histories collected in 2000 to study the first stages of contraceptive practice and the effects of the emergence of AIDS on this practice.