The most challenging result in our study is the emerging of socioeconomic inequalities in age at first delivery, number of pregnancies, number of alive newborns, and diagnosis of fertility problems in women born after 1950. Breastfeeding, suffering abortions or dead newborns, consumption of oral contraceptives or hormonal replacement therapy were associated with socioeconomic level in women born before 1950 but not in women born after that year.
Stratifying our analysis according to being born before or after 1950 was not arbitrary. It is noteworthy that women born in that year reached their sexual maturity around 1970, which would be considered some kind of social milestone for Spanish women. Spanish women incorporated to labor market after 1970, the labor activity rate in 1976 being 55.1% for women aged 2024 (i.e., born after 1950) and only 29.7% for women aged 2559 (i.e., born before 1950) [18]. Secondly, sexual behavior began to change in most western countries in the 60s, involving usage of contraceptives, family planning and women taking more control of their sexual / reproductive lives. The Spanish society, however, had their sexual habits ruled by the dictatorship that ended in the 70s, including for instance late initiation of sexual relationships, one partner only, or inclusion of adultery in the penal laws until 1976 [19]. Therefore, sexual freedom and generalized accessibility to contraceptives was reached by Spanish women from 1976 on, with some delay respect to women in other western countries. Thirdly, a deep economic crisis affecting Western countries (the so-called first oil shock) began in 1973. This eventually led to high unemployment rates and to deep falls in birth rates, which were more intense in Spain: the average number of children per women fell from 2.90 in 1970 to 2.22 in 1980 and 1.36 in 1990 (for comparison, figures in the UK were 2.43 in 1970 and 1.83 in 1990). In the same 20-year period, the birth rate fell in Spain from 19.5 in 1970 to 15.3 in 1980 and 10.3 newborns per 1000 inhabitants in 1990 (in the UK: 16.2 in 1970 and 13.9 in 1990) [20]. Fourthly, the public health service increased its coverage from 1970 to 1980, reaching universal coverage in 1988 [21]. Finally, the continued social trend toward reduced family size instead of that in the past, in rural areas above all, where children were needed and deemed to be suitable for farming or helping at home [22].
That social context could easily explain changes when the socioeconomic inequalities decreased in women born after 1950. For instance, hormonal contraceptives were marketed in Spain from 1964 on, but their indications initially included keeping the ovary in rest, controlling menstrual cycle, and treating dysmenorrhea and acne, while their contraceptive effect was considered as an undesirable side effect [23]. For years, they were more accessible for highly educated women living in urban areas [24]. When the public health service included contraception in its portfolio, women could access it without socioeconomic differences, leading to the results we have described for women born after 1950. Hormonal replacement therapy appeared much later than hormonal contraceptives, but its usage was amplified via private practicing doctors. This eventually resulted in higher consumption for women in higher socioeconomic levels; in 2001, when most women born after 1950 had not reached menopause, the Womens Health Initiative study found an association between hormonal replacement therapy and several cancers [25] and other chronic diseases [26], leading to a dramatic decrease in hormone replacement therapy in Spain [27].
The emergence of new socioeconomic disparities in age at first delivery and number of pregnancies in women born after 1950, however, is challenging. These inequalities appeared or were intensified in an era of universal coverage of the public health service, with free access to contraceptive methods and widely available information about them. In this regard, it is noteworthy that universal health coverage does not imply -by itself- equity in health assistance. For instance, the OECD have noticed that despite the fact that most OECD countries have achieved universal health coverage, people from the most socially disadvantaged groups tend to have worse access to health services. Possible reasons include lack of awareness of health services, poorer quality of care and co-payments for care [28,29,30]. In this regard, Spanish women aged 2544 (most of their fertile age) declared having unmet needs for health care in higher percentages if their income are in the lower quintile, although the gradient associated with income level was rather mild (5.84, 3.13, 4.15, 3.52 and 3.11% for Q1 to Q5 in 2001, [31]. Admittedly, these data on unmet needs in 2001 are too late for explaining our results, but National Health Surveys carried out before 2001 did not include any question on unmet needs. Apart from this explanation, we can only speculate on whether cultural issues associated with lower both socioeconomic and educational levels or differential access to work market could have prevented women in such levels to decrease their fertility as women in higher levels did. For instance, it could be possible that women in higher socioeconomic level incorporated earlier to the work market and, thus, delayed their decisions on having their first child. It is noteworthy, in this regard, that socioeconomic-level associated inequalities were mainly in drugs usage (hormonal contraceptives and hormone replacement therapy) in women born before 1950, which could be associated with inequities in accessing health care. The main inequalities in women belonging to later generations, however, seem to be associated with their own decisions (number and age of pregnancies), not with access to medical care. Along the same lines, a recent review carried out in 2019 showed how social determinants play an important role in the stage of breast cancer in diagnosis and survival [32].
Selection of population controls is a main strength of this study. Women aged 2085years were enrolled in 12 Spanish provinces after being selected from general practitioners roasters; they can provide a representative sample of the Spanish women, given the almost universal coverage of the national health system in Spain.
Some limitations of the study should also be noted. Firstly, reproductive variables were self-reported, which could lead to recall bias; however, as women were not aware of the hypotheses of this study, we would expect that recall bias -if exists- could be non-differential. Secondly, one of the SE indicators we have use -the Urban Vulnerability Index- is ecological by nature, which makes it possible the occurrence of ecological bias. In this regard, aggregate deprivation indexes have been found good proxies of individual income but less efficient to measure education or occupational category [33].
