Our study provided an opportunity to investigate the influence of determinants on abortion trends using complete abortion reporting in a unique situation - legislation and access to abortion have not changed, but political, economic and social changes have been significant during the last two decades in Estonia. Over the 16-year period studied (1996–2011), trends indicated a considerable decline in the number of induced abortions and a slow but consistent decline in the number of repeat abortions undergone in Estonia. The declining abortion rate was mainly attributed to the younger cohorts. The proportion of women undergoing repeat abortions in 2004–2011, compared to 1996–2003, decreased within all socio-demographic subgroups. However, this decrease was not significant among women with a lower level of educational attainment and students. The percentage of non-Estonians undergoing repeat abortions, and of those obtaining third and subsequent abortions was higher than that of Estonians. Most women did not use any contraceptive method prior to their first or subsequent abortions.
Validity of EAR data
Before these findings are discussed further, we must comment on the completeness and validity of EAR data. For this investigation, we presumed that because it is relatively easy to obtain an abortion in Estonia, virtually no illegally performed abortions occurred. All pregnancies should be documented in case records and diagnosed by ultrasound or hCG-test eliminating procedures called “menstruation regulation” or “miniabortion”, thus ensuring that missing data are minimal within the EAR dataset. When comparing abortion rates between countries, the question of complete reporting will always arise particularly since the accuracy and consistency of reporting impacts directly on the abortion rate reliability [3, 10, 11]. It has been argued that the decline in induced abortion rates seen in former Soviet countries since 1995 might be overestimated because abortions were increasingly being performed in the private sector so may not have been included in reported statistics [3, 10, 17]. Therefore, the reliable data collection conducted by the EAR may partly explain why the abortion rate in Estonia, especially compared to other post-Soviet countries, has been reported as the highest in the European Union .
In Estonia, the number of repeat abortions, and especially the number of third and higher-order abortions, is high. For instance, in 2011, among all induced abortions, the proportion of women undergoing fourth and subsequent abortions was 17.9% in Estonia, compared to 5.6% in Sweden and <1% in England and Wales [4, 5]. A high percentage of women who have undergone repeat abortions reflects a high historical abortion rate. Our analysis revealed that the overall decline in abortion rate observed during the study period was mainly attributed to younger cohorts. Moreover, this observation must be viewed alongside fertility rate: in 1996, the highest fertility rate (i.e. 101 live births per 1000 women of fertile age) was seen among women aged 20–24, but had halved by 2011 to 55.7 and the peak in fertility rate has shifted to women aged 25–35 . The average age of mothers during first births increased from 23.1 in 1996 to 26.3 in 2011 . Currently, the highest percentage of women undergoing repeat abortions are those in their 30’s and 40’s who have either had several abortions in their lifetime, or had a greater number of repeat abortions when younger and showed a lower acceptance of effective contraceptive methods compared to younger women . Thus, we can assume that when older cohorts of women “age out” of their reproductive years, a more rapid decline in the number of repeat abortions is likely to happen in future years. The opposite trend has been observed in many countries during the last 30 years [10, 19] and was forecasted by Tietze and Jain: “The proportion of repeat abortions among all legal abortions increases over time as more women in the population have had a first abortion and are, therefore, at risk of having a repeat abortion, until a steady state is reached” . According to the literature, pregnancy unacceptance and contraceptive failure have both been associated with either being single or student [21, 22]. This may be one reason why the decrease in the percentage of women undergoing repeat abortions among students and single women was not marked compared to other subgroups, although it should be noted that students and single women represent a small proportion of the overall study population.
Influence of age, parity and ethnicity
Age is the main predictor for repeat abortion because on the one hand, older women have had more years of exposure to risk of pregnancy and on the other, teenagers who have had one abortion, are at greater risk of having another . It has been argued that women's attitudes and behaviour towards induced abortion are established at a young age and persist during a woman's fertile age .
Parity is another key characteristic for repeat abortions because women obtaining repeat abortions are more likely to indicate that they don’t want to have more children [6, 19]. In our population 90% of repeat abortions were obtained by parous women. However, following the traditional Western European pattern, unwanted pregnancies were terminated before the childbearing commenced, and this can already be seen in abortion statistics during two periods: 23.9% in 1996–2003 and 26.7% in 2004–2011 (p < 0.0001) of all abortions were obtained by nulliparous.
Although less influential than age and parity, ethnic origin has also been associated with repeat abortions . Estonia is ethnically diverse: 34.8% of its population in 1996 and 28.8% in 2011 was composed of non-Estonians, the vast majority of which were Russians . In our dataset, 96.2% of non-Estonians were Russians during 1996–2003 and 97.5% during 2004–2011. Over the last 20 years, the overall abortion rate has been substantially higher and fertility rate lower among non-Estonians compared to Estonians . Non-Estonians were overrepresented among women of fertile age obtaining repeat abortions in 1996 and in 2011 (46.0% and 38.7%, respectively). However, the decrease in repeat abortions among non-Estonians was almost comparable with that among Estonians, while the percentage of non-Estonians obtaining the third or higher order abortions was markedly higher compared to Estonians. One likely explanation, derived from previous studies , is that the Russian-speaking women prevented unintended pregnancies by using less reliable contraceptive methods and, in contrast to Estonians, having an abortion increased the risk of them not using contraception in the future.
Contraceptive patterns before pregnancy termination have been explored in a number of studies [6, 19, 20] and, according to these data, the variations reflect the differences in overall contraceptive practices across countries. The proportion of women who did not use any contraception prior to their first, second, third and fourth or subsequent abortion, accounted for almost two thirds of the overall population in our study; while this increased over the 16-year period, the use of unreliable contraceptive methods decreased. This might be due to an actual decrease in the use of unreliable contraceptive methods among abortion patients, or these trends may also be explained by an improved knowledge of fertility control. For instance, once the use of the withdrawal or rhythm method fails, women no longer appeared to perceive these approaches as contraceptive methods. The use of condoms was the most frequently reported failed method of contraception prior to abortion and this finding is in accordance with previously published results. Condom use has been shown to have the highest failure rate among contraceptive methods, especially among adolescents, students, single women, and those with no children [19, 21, 22]. The use of hormonal contraceptive methods prior to abortion was considerably lower than in studies from the US, Finland, France and Denmark [6, 19, 21, 22]. Use of long-acting reversible contraceptive methods, like intrauterine device (IUD) (e.g. copper intrauterine device, levonorgestrel intrauterine system), compared with user-dependent methods, are associated with a lower risk of repeat abortion . This was in agreement with the findings in our study in which IUD users represented the smallest proportion of women having their first or a higher-order abortion. The decrease in the number of IUD users among abortion patients might be the result of an increased use of the reliable levonorgestrel intrauterine system during the study period . However, our findings about pre-abortion contraception are in discordance with the data from other studies from developed countries where the majority of women obtaining their first-time or repeat abortion failed to use a contraceptive method at the time of conception [6, 19, 21, 22, 24]. This may be due to variations in study design or may reflect different contraceptive patterns, but is more likely to be influenced by a high overall abortion rate. We can conclude that the majority of abortions in Estonia did not follow contraceptive failures, but occurred because of contraceptive non-use. This confirms that the availability of and access to contraception is not enough to lower the incidence of unintended pregnancies.
A major impact on the abortion rate is the quality of health care services [24, 25]. Although there is no robust evidence that contraceptive counselling improves contraceptive adherence and, therefore, reduces the risk of repeat abortion [26, 27], there are data to support the fact that having contraception choice, empowers women to make their own decisions and if made at the time of abortion, are important in preventing unintended pregnancies in the future [6, 21, 26]. An elegant prospective study from Finland showed that immediate initiation of any contraceptive method after abortion, but especially long-acting methods, was linked to a lower risk of repeat abortion . A population-based study in 2004 found that only 24.0% of Estonian women reported receiving pre- or post-abortion contraceptive counselling . In contrast in France, 79.6% of women declared they had received information about contraception before or after abortion .
Study strengths and limitations
Our study’s main limitation was related to the surveillance system not permitting personal identification numbers. This did not allow to link with other data and an analysis of associations between different socio-demographic factors and repeat abortion. Except for age and ethnicity, there were no population-based data about mean annual numbers of other socio-demographic characteristics in Estonia and therefore we couldn’t obtain an exact estimate of how the trends were relative to the population. We are also aware that reliance on the self-reporting of sensitive issues like previous abortions may cause underreporting. However, one study has validated that there is a high degree of completeness in the reporting of recent abortions in Estonia  and convinced us that underestimation is minimal. It should also be noted that the Abortion Card only offers five options for contraceptive methods, which includes hormonal contraceptive pills but not transdermal and vaginal hormonal contraceptive methods despite their availability since the early 2000’s. It cannot be guaranteed that misclassification of these hormonal contraceptive methods may have occurred. However, it is our assumption that under the methods named “other” mostly unreliable contraceptive methods such as the rhythm and withdrawal methods and spermicides were classified. Finally, a cross-sectional, population-based study showed that only 6.5% of women who need contraception reported contraception non-use at the time of their last sexual intercourse . This discrepancy with EAR data might be due to a different study sample, different study design, and data collection methods or reflect actual variation in contraceptive practices among women who undergo induced abortion in Estonia. We can only assume that the EAR data largely reflects the actual patterns of contraceptive use at the time of conception.
Despite these limitations, our study provided a unique opportunity to utilise a large, reliable, registry-based dataset to provide a detailed and comprehensive overview of the area of reproduction in Estonia.