In this analysis of use of the so-called third generation pills that appeared in the 1980s, the authors examine the mechanisms responsible for the emergence of new inequalities. The continual renewal of contraceptive products—the emergency pill, the female condom, improved contraceptive implants, new types of IUD—ensures a wide relevance for the analysis presented here.

Since contraception was legalized in France in 1967, recourse to medical contraception has increased continuously and the proportion of French women now using these methods is among the highest in the world (Leridon and Toulemon, 2003). The present situation is due specifically to the trend in use of the contraceptive pill, whereas use of the intrauterine device (IUD) has levelled off since 1988 following a period of strong growth (see Table 2 in the introductory article). In 1978, of 100 women who were sexually active and not seeking to get pregnant, 40 reported using the contraceptive pill, compared with 50 in 1988 and 60 in 2000, corresponding to around 5 million pill users today. This recomposition of contraceptive practice in favour of medically prescribed methods may well have generated new forms of inequality, comparable to the inequalities observed in other fields of health (Leclerc et al., 2000). In addition, the rapid spread of oral contraception has coincided with a far-reaching change in the products available. In 1967, there was only one type of contraceptive pill, the “first generation combined contraceptive pill”, but in 1974 the “second generation pill”, containing lower doses of oestrogen, appeared on the market (see box on next page), and 1982 saw the arrival of the “third generation” oral contraceptives. The latter contain the same dose of oestrogen as the second generation pills but a new progestogen that has fewer androgenic effects (weight gain, hair growth, acne). Since then, new products containing lower doses of oestrogen have regularly come on to the market. Prescribed with increasing frequency and not refunded by the social security, these third generation contraceptive pills are supposed to be better tolerated by women because the type of progestogen they contain has fewer androgenic effects and because some have a smaller oestrogen component (Spira, 1993).

Using data from the COCON survey (Bajos et al., 2003), which is presented in the introduction to this dossier, we first try to identify possible trends in social inequality in access to different contraceptive methods. Second, we analyse the effect of social class on oral contraceptive use in particular, in order to capture and interpret the logics behind any social inequalities in access to the third generation contraceptive pill. If such social inequalities exist, bearing in mind that these new pills are supposed to be better tolerated, we need to examine whether the women using them are indeed more satisfied. This forms the subject of part three of the article.

The analysis of access to third generation, non-reimbursed, oral contraceptives was conducted on 1,041 women using a combined oestrogen-progesterone pill.

Main brands of pill used in France

For the satisfaction study, three groups of users were identified according to the chemical composition of the pill. For Group A this was first or second generation pills containing between 30 and 40 μg of ethinyl estradiol (n= 592), all of which are reimbursed. Group B comprised users of third generation pills that also contain between 30 and 40 μg of ethinyl estradiol (n= 184) but are not reimbursed. Group C was the users of low oestrogen third generation pills (between 15 and 20 μg ethinyl estradiol; n= 206), which are not reimbursed. Women who reported using the Cycléane brand without further details (n= 59) had to be excluded, as the ethinyl estradiol dose can be 20 or 30 μg. In all, the satisfaction analysis was conducted on 982 women.

We examined whether the type of pill used had a specific effect, “all other things being equal”, on the level of satisfaction reported. Given the diversity of variables linked to satisfaction, three intermediate multivariate analysis models were estimated to provide input for a final model. Only this final model is presented here.

Model one included selected characteristics of the women:

The second model included variables pertaining to the prescription context:

Model three included characteristics of the partner relationship:

The final model included the variables significantly associated with satisfaction in at least one of these three models. The interactions between pill type and the analytical variables listed above were thus tested for.

Table 1
Social and demographic variables associated with contraceptive practice in 1978,1988 and 2000 (results of likelihood ratio tests applied to logistic regressions)

At the start of the third millennium, the pattern of inequalities in contraceptive use is somewhat modified (Table 1). For the first time, overall access to contraception is no longer socially determined, if we except the fact that women over age 25 are slightly less likely than other women to use contraception. This age effect, which had disappeared in 1988, is probably the result of a large increase in medical contraceptive coverage, here meaning the pill, among younger women: 69% of women aged 20-24 reported using a medical contraceptive method in 2000, against 40% in 1978 (Table 2 in the article by Clémentine Rossier and Henri Leridon in this issue). Use of the pill has increased at all ages, although is still lower among older women and married women. These data probably reflect the increase in the average duration of pill use, and the growing recourse to progestogen-only pills that tend only to be prescribed for older women and in particular for those who smoke or who have medical contra-indications for oestrogens.

However, the diffusion of hormonal contraception has not occurred evenly across the country. The proportion of women who have adopted it is higher in the Paris region than in smaller urban centres or in rural areas (Leridon et al., 2002). This pattern may be linked to an effect of contraceptive provision, and more specifically to the unequal distribution of gynaecologists in France, since specialists are more likely — as we shall see — to prescribe new contraceptive products. Use of the IUD has hardly changed quantitatively since the 1990s (see Table 2 of the introductory article). This method seems to be increasingly restricted to women over age 40 and those who have completed their families. It is as if the medical norms relative to prescription of the IUD have become stricter over the last decade, excluding women who have yet to have children, because of concerns about infection and possible sterility, and reserving it for women in whom the possibility of secondary sterility is judged “acceptable”. However, this risk is associated primarily with sexually transmitted diseases, and although STDs are certainly more common among young women because on average they have more partners, they are not all, far from it, affected by them.

In 2000, although there is no longer any evidence of a selection effect operating for educational level, use of the IUD remains low among manual workers. This social category thus appears today as having lagged behind in the general tendency towards the medicalization of contraception.

Finally, differences in the use of condoms were recorded in 2000. In 1978 and at present, the percentage of women who report using this method for contraceptive purposes is low (8% in 1978 and 7% in 2000). But the HIV prevention campaigns do seem to have had an effect. The condom is currently used primarily by young women and women not in a couple, whereas in 1978 there were no clear differences between female users (Toulemon and Leridon, 1992). In addition, the differences by educational level have increased, and the socially most advantaged categories, who were the most receptive to the campaigns promoting the condom to combat STDs and HIV, also seem to have more easily adopted it as a contraceptive.

The inequalities in access to contraception in general and to medical contraception in particular observed in the 1970s and 1980s have thus decreased substantially, initially for use of the pill, and subsequently for use of the IUD. Analysis of recourse to the third generation contraceptive pill that is not reimbursed by the social security will allow us to reveal shifts in the pattern of social inequalities affecting access to new oral contraceptive products.

This diffusion of the third generation pills seems to have affected different social groups in different ways (Table 2). The distribution of pill users according to pill type shows that women under 20 and those with two or more children are less likely to use a third generation pill.

Table 2
Type of pill used according to demographic, social and reproductive characteristics of pill users (%)

There are marked social differences: 32% of women in households where monthly income per person is below 560 euros use a third generation pill, but the proportion is 51% where income per person is above 990 euros. The proportion also varies according to whether or not the woman has a mutuelle(private complementary health insurance) allowing full refund of prescription costs (41% of those with complementary insurance as against 26% of those without). Among women with no qualifications or only a lower secondary certificate (BEPC), 22% used a third generation pill, compared with 56% among those with higher education. Correlatively, women in managerial employment were more likely to use these new pills than women at the other end of the social scale (64% against 17% among manual workers).

Women in managerial positions are also more likely to go to a gynaecologist for their contraception (90% against 50% among manual workers), and gynaecologists are much more likely than general practitioners to prescribe third generation pills. The multivariate analysis (Table 3) confirms the specific effect of these variables [1], with the exception of age and income (though if we include socio-occupational category rather than educational level in the model, a high income remains associated with use of a third generation pill).

Table 3
Variables associated with use of a third generation contraceptive pill (as opposed to a first or second generation pill) Results from a logistic regression model (pill users, n=1,041)

The typical profile of the third generation pill user is thus a woman who is highly educated, has managerial-level employment and complementary health insurance, and who goes to a gynaecologist for her contraception: 70% of pill users in the group of women with these characteristics use a third generation pill. A priori, this is the classic pattern for the social diffusion of new health products, which raises the question of the reasons for such disparities. One hypothesis is that people from the most advantaged social backgrounds are more attentive to all of the side-effects associated with these low-dose pills and in addition have the financial resources necessary for this contraception. However, since a prescription is needed to obtain these products, the question also arises of whether prescribing obeys a particular logic. The first point to note is that the gynaecological supervision of women varies greatly according to their social circumstances (Table 4). Thus, while 58% of manual workers had consulted a gynaecologist about their reproductive health in the previous 12 months, the proportion reached 85% for managers. Now, contraception specialists (gynaecologists) are much more likely than general practitioners to prescribe third generation pills (46% of cases against 21%, p< 0.01) (Table 2); also, the distributors of these new products have launched major advertising campaigns aimed at specialist doctors. The prescription behaviour of gynaecologists seems to vary depending on the social profile of the women, which is not so for general practitioners. Thus, the prescription of third generation pills by the latter is independent of educational level and socio-occupational category, whereas gynaecologists prescribe more non-reimbursed pills to women who are in managerial employment and are the most highly qualified (Table 5).

Table 4
Gynaecological supervision in previous 12 months by socio-occupational category

Table 5
Percentage of pill-users taking a third generation contraceptive pill by socio-occupational category, educational level and gynaecological supervision in previous 12 months

The overwhelming impression is thus that the socially most advantaged women have a more demanding attitude towards new products, one that is echoed, or even initiated, by the contraception specialists.

In total, 86% of women who used the pill reported that this method suited them “completely”, 10% “fairly well” and 4% “not very well or not at all” (Table 6). A high level of satisfaction with the method of contraception is also found in other studies, women who were not very satisfied having usually already changed method (den Tonkelaar and Oddens, 2001). Our data show that the level of satisfaction increases with duration of use.

Table 6
Percentage of woman who report being very satisfied with their contraception by socio-demographic, relational, contraceptive and health characteristics

TABLE 6 (CONT.)
PERCENTAGE OF WOMAN WHO REPORT BEING VERY SATISFIED WITH THEIR CONTRACEPTION BY SOCIO-DEMOGRAPHIC, RELATIONAL, CONTRACEPTIVE AND HEALTH CHARACTERISTICS

A slightly higher proportion of women using first or second generation pills (88%) than using third generation pills of the same (84%) or lower dose (80%) reported that this contraception suited them completely (p= 0.06). The duration of use of the current brand was longer for second generation pills (5.4 years) than for high-dose (4.7 years) or, in particular, for low-dose (2.6 years) third generation pills (p< 0.0001). The link observed between duration and satisfaction is probably due to a selection effect over time, but it remains when we limit the analysis to pills that had been on the market for more than 3 years so as to allow for any bias introduced by the changing contraceptive supply (result not shown).

On the whole, younger pill users, those who were not in a couple and those with no children were most likely to report being completely satisfied with their method of contraception (Table 6). Since the IUD is rarely prescribed to young or nulliparous women (Leridon et al., 2002), it is possible that the higher degree of satisfaction among these women reflects the fact that they are not really given the choice of another method, especially as this result persists when duration of use is taken into account. Other social characteristics (educational level, income, complementary insurance) are not associated with satisfaction — and hence do not appear in Table 6 — with the exception of socio-occupational category: women in intermediate professions and managerial positions less often reported being satisfied, perhaps reflecting their higher social expectations and their possession of the “cultural competence” needed to express these in a survey initiated by a health research institute.

The pill users who wanted a child straight away or within the year were not more likely to be dissatisfied with their contraceptive method. This contrasts with the findings of Luker for the United States (Luker, 1975), which showed that, in the 1970s, American women who were ambivalent in their attitude to the desire for children tended to be less satisfied with their method of contraception and less able to accept its side-effects.

As has been observed among German women (Oddens, 1999), women who have not experienced contraceptive failure in general, and with the pill in particular, are more likely to report being completely satisfied with their contraception. The same is true for women who feel that they have taken charge of the decision concerning their contraceptive method. Other studies have also shown the importance of the prescription context for women’s degree of satisfaction with their contraception. A review of the literature by Delbanco and Daley (1996) concluded that the more doctors involved women in the choice of method, the more the women were satisfied with their contraception and the fewer failures they experienced. Our multivariate analysis (Table 7) confirms these findings, except concerning the effects of age, marital status and having become pregnant while on the pill.

Table 7
Characteristics linked to satisfaction with contraception used Multivariate analysis model (n=982)

In total, the women using first or second generation pills were not less satisfied than the others. Conversely we observed that women using low-dose third generation pills reported being significantly less satisfied.

Except for headaches, French users of low-dose third generation pills do not report more side-effects (weight gain, long or heavy periods, nausea, breast pain) than those using a higher dose third generation pill or a first or second generation pill (Table 8).

Table 8
Percentage of women reporting side-effects by pill type

Women who suffer frequent headaches are slightly less likely than other women to report being completely satisfied with their contraception (83% as against 88%, p< 0.05). Controlling in the model for the reporting or not of headaches does not affect the relationship between pill type and satisfaction. The lower level of satisfaction among users of low-dose third generation pills would thus seem not to be because they are more prone to headaches. It might derive from different pill prescribing practices depending on the “sensitivity” of the women, either as expressed by the women or anticipated by the prescribers. Our hypothesis is that there may be such a “prescription bias”, originating in a tendency of doctors to prescribe a lower-dose third generation pill to women for whom they believe it is particularly suited, in view of the weaker anticipated side-effects. If these women are less likely to be satisfied with their contraceptive method, irrespective of the product used, this would lead to a higher percentage of dissatisfied users of third generation pills compared with users of first and second generation pills.

We tested for this prescription bias by selecting women who were using the pill for the first time. After controlling for duration of use, a variable that is strongly linked to satisfaction, we found that among women using the pill for the first time (n= 128), those using a first or second generation pill were more likely to be very satisfied with their method than those using a third generation pill containing the same (p= 0.03) or lower (p= 0.001) oestrogen dose. This specific effect of pill type remained after controlling for variables linked to satisfaction in the multivariate model presented above (Table 7) and for reporting of headaches (the only side-effect more common among users of third generation pills): the new users of third generation pills remained significantly less satisfied than other women (result not shown).

We cannot, however, exclude the possibility that this lower level of satisfaction reflects not only a prescription bias but also the fact that these pills really are less popular with women. To test this we would need to be able to measure the satisfaction these users felt if they were using another type of pill, which would require a survey protocol different from that of the COCON survey, of the randomized trial variety.

The data from the COCON survey, which reflect the point of view of women and not those of prescribing doctors or manufacturers of hormonal contraception, do not therefore allow us to conclude that women are more satisfied with third generation pills, at least not on the basis solely of the indicator of satisfaction used here. Several authors have drawn attention to the conceptual shortcomings of this type of indicator (Hall and Dornan, 1988; Williams, 1994). However, these limitations have little impact when, as in this article, we are working within a comparative perspective. Our concern here has been to identify a specific effect of pill type on the satisfaction felt by women, rather than to analyse the level of satisfaction that women feel towards their method of contraception.

The long-term effects on women’s health of second and third generation pills are very similar. Third generation pills are associated with a slightly increased risk of venous thromboembolism (Jick et al., 2000) but this risk affects only a very small number of women. On the other hand, their protective effect against myocardial infarction compared with second generation pills is still a subject of debate (Spitzer et al., 2002). Given that no link has so far been established between pill type and user satisfaction it seems that the social inequalities in access to third generation pills are not accompanied by health inequalities. The effects on the body remain to be assessed, which will be possible through analysis of follow-up data on the cohort in an epidemiological perspective, allowing further progress to be made on this question and providing material for the debate over social security reimbursement of third generation pills.

For the present, giving women access to an inexpensive generic low-oestrogen dose third generation pill remains a topical issue. The more varied the contraception made available, in terms of pill type and more generally of methods of contraception, the greater chance women will have of finding the method that is best suited to their social, relational and sexual situation as well as to their physiology.