Secondary Logo

Journal Logo

Contents: Contraception: Original Research

Induced Abortion in a Population Entitled to Free-of-Charge Long-Acting Reversible Contraception

Gyllenberg, Frida K. MD; Saloranta, Tuire H. MD; But, Anna MSc; Gissler, Mika MSocSci, PhD; Heikinheimo, Oskari MD, PhD

Author Information
doi: 10.1097/AOG.0000000000002966
  • Free

Long-acting reversible contraceptive (LARC) methods are the most effective contraceptive options for preventing unintended pregnancies.1 Nevertheless, these devices are used less than other short-acting methods.2,3 The provision of contraception, and particularly LARCs free of charge, has been advocated,4–6 because high upfront costs are considered a frequent barrier for initiation of these methods,7,8 yet scarcity of data on outcomes of no-cost programs contributes to difficulty in implementing such programs.

Both in the CHOICE study in St. Louis and in our previous study in Vantaa (Finland), provision of LARC methods free of charge correlated with increased initiation of long-acting methods.9,10 Furthermore, in the CHOICE study, the abortion rate among study participants was low compared with national levels.11 Similarly, in our previous study, the aggregated abortion rate decreased in the community by 16% after implementation of the no-cost LARC program.10 However, it remains unknown to what degree abortion rates individually differ among those initiating a LARC free of charge compared with the general population.

Our aim was to examine the rate of induced abortion (hereafter abortion) during a public program entitling all patients their first LARC method free of charge. We compared the abortion rate among females using public family planning clinic services in Vantaa and a control group of females not using the services and continued to compare the abortion rate among those who initiated a free-of-charge LARC method and those who did not.


There are four family planning clinics in Vantaa serving a population of approximately 220,000 residents. The clinics are part of the public health care services and provide care free of charge for the clients. The clinics provide counseling for residents in need of family planning services, including initiation and follow-up of hormonal contraceptives, initiation and removal of LARC methods, diagnostics and treatment of sexually transmitted infections among contraception users, counseling on fertility and sexual health as well as referrals for and follow-up visits after induced abortion. As of January 2013, every female in Vantaa has also been eligible to receive one LARC method free of charge at these clinics. In Finland, abortion can be requested on several grounds mentioned in the legislation. The social indication (ie, continuation of pregnancy represents an unbearable burden) contributes to 92% of all induced abortions.12 Illegal abortions are very rare, because the law is interpreted liberally, services are easily accessible, and the expenses largely are covered by public health care. All abortions are reported to the national registry, and the reporting is mandatory and based on law. There is also a national guideline on induced abortion (, which contributes to comparable classification and reporting.

Our study population consisted of all nonsterilized females, aged 15–44 years, who lived in Vantaa during 2013 and 2014 and were entitled to a LARC method free of charge. The formation of cohorts is presented in the flowchart (Fig. 1). We defined the four cohorts as follows: 1) LARC cohort (ie, all patients who initiated a free-of-charge LARC method at public family planning clinics during the years 2013–2014), 2) no LARC cohort (ie, all patients who visited a public family planning clinic during the years 2013–2014 but did not choose a LARC method), 3 and 4) control groups of females for cohorts 1 and 2. Cohorts 1 and 2 are together referred to as service users and 3 and 4 as nonservice users. The nonservice users were age-matched with a ratio of 1:3 from the study population of nonsterilized patients, entitled to a free-of-charge LARC method but who did not visit a public family planning clinic during the years 2013–2014. Data on visits and LARC method initiation at the family planning clinics were obtained from structured electronic health records of Vantaa. Exclusion criteria were age younger than 15 years or older than 45 years; sterilization before start of follow-up; and previous use of a LARC method. Only a patient's first LARC method was provided free of charge; thus, patients with previous use were not entitled to a free LARC. Previous users were identified in the health records according to at least one of the following criteria: initiation or removal of a LARC method during the years 2000–2013; removal of a LARC method after January 1, 2013, without a preceding insertion; or initiation of a LARC method, which was not free of charge in 2013 or 2014.

Fig. 1.:
Formation of the study cohorts. Eligible females using public family planning services and initiating a free-of-charge long-acting reversible contraceptive (LARC) method were identified from electronic health records. Females in the control group in a ratio of 1:3 were randomly picked from the population of eligible females.Gyllenberg. Free-of-Charge LARCs and Induced Abortion. Obstet Gynecol 2018.

The LARC methods provided free of charge during the study period included the 52-mg levonorgestrel-releasing intrauterine system (LNG-IUS), the 13.5-mg LNG-IUS, a 380-mm2 copper intrauterine device, and two contraceptive implants with either levonorgestrel or etonogestrel. Participants entered the study during the years 2013–2014. For participants in the LARC cohort, individual start of follow-up was defined as the date of a free-of-charge LARC initiation. For participants in the no LARC cohort, start of follow-up was set to the date of the first visit at a public family planning clinic. For nonservice users, the start of follow-up was set to the same date as that of the matched index participant. If a patient in any of the cohorts was pregnant on the start date of follow-up, her follow-up was set to start 30 days after the day the pregnancy ended. End of follow-up was defined as the first of the following dates: start of pregnancy ending in abortion, start of pregnancy ending in birth, date of move abroad from Vantaa, date of sterilization, or February 28, 2016. To include all pregnancies starting during follow-up, we followed national registers to identify births and abortions that occurred until December 31, 2016. Start of pregnancies ending in birth, move abroad, and sterilization were treated as censoring events.

We obtained data on outcomes, censoring events, and confounders from the Finnish national registers. In Finland, data from different sources can be combined through a computerized register linkage using unique personal identification codes assigned for every permanent resident in Finland since the late 1960s. Data on births, induced abortions, and sterilizations were obtained from the corresponding registers maintained by the National Institute of Health and Welfare. The Medical Birth Register was established in 1987 and includes data on both mothers and fetuses, including stillbirths if the gestational age is at least 22 weeks. It does not include information on pregnancies ending before 22 weeks of gestation, for example, miscarriages and ectopic pregnancies. The Register on Induced Abortions, available in electronic format since 1983, contains information of gestations ending in induced abortion, including gestational age at the time of abortion and hence start of gestation can be calculated. These registers are high quality and validated.13–15 Demographic data on cohort members including date of birth, marital status, and migration out of Vantaa were obtained from the Central Population Register of Vantaa. Data on the socioeconomic status and education were obtained from Statistics Finland.

The crude rate of abortions (per 1,000 woman-years) was calculated as the number of pregnancies ending in induced abortion divided by the woman-years accumulated during follow-up in each cohort. Univariate Poisson regression models were used with the log of woman-years as offset and assessed the difference in abortion rates as crude rate ratios with 95% CIs for service users compared with nonservice users and further for the LARC cohort compared with the no LARC cohort. The LARC and no LARC cohorts were also compared with females in corresponding age-matched control groups. Survival curves were assessed by Kaplan-Meier estimates.

The effects of the following well-established risk factors for unintended pregnancy and induced abortion were examined as potential confounders: age at the start of follow-up, marital status, socioeconomic status, level of education, and prior births and abortions.16,17 Lacking information about race or ethnicity, a binary variable of speaking either of the national languages, Finnish or Swedish, or another language as the mother tongue was recorded. Marital status was defined as a binary variable: married or not married, because cohabitation is not registered in Finland. Participants were allocated in four groups regarding socioeconomic status based on national standards by Statistics Finland: upper-level employees; lower-level employees; manual workers; and a fourth group consisting of farmers, students, unemployed, retired, and housewives for which socioeconomic status cannot be confirmed. The education variable was transformed from nine-level educational data into a binary variable with only elementary education (which in Finland encompasses 9 years of education) compared with more than elementary education. Previous births and previous abortions were both binary variables. The effects of potential confounders were assessed first in univariate analysis, which suggested that all these covariates were associated with the outcome of interest (P<.05). The final multivariate Poisson regression model was formed through a manual forward and backward selection process, in which the effect of each of the candidate covariates was examined. Selection of covariates was guided by the statistical significance of their effect on the outcome variable (P<.05) and with information loss measured by the difference in Akaike information criterion (ie, decreasing Akaike information criterion for the improved model). The assumption of equal mean and variance was met for the Poisson regression models.

Because a patient might have attended a family planning clinic before initiating a LARC method, the comparison between the LARC and no LARC cohorts could be susceptible to time-related bias (ie, immortal time bias18). Hence, to evaluate the potential effect of immortal time bias, we performed sensitivity analysis by adding the woman-years that accumulated for individuals in the LARC cohort from the first visit at the family planning clinic to the LARC initiation visit (400 woman-years) to the no LARC cohort woman-years. Furthermore, because the number of events in the LARC cohort was small compared with the number of covariates, we computed additional analyses for the LARC cohort comparisons with fully matched control participants and cohorts (one-to-one matching by age, socioeconomic status, marital status, previous pregnancy, and previous abortion).

Statistical significance was set at the 5% level. All analyses were conducted using R 3.3.2. The study was approved by the ethics committee of the Hospital District of Helsinki and Uusimaa (304/13/03/03/2015) and the register keeping organizations authorized the use of health register data in this study. As a register study, no informed consent was needed. We followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines in the study design and manuscript preparation.


During 2013–2014, 9,669 patients entitled to a free-of-charge LARC method visited one of the four public family planning clinics, which equals 20% of the entire eligible 15- to 44-year-old female population in Vantaa (Fig. 1). Of these, 2,035 (21% of all service users) patients initiated a free-of-charge LARC method and 7,634 patients visited a public family planning clinic without choosing a LARC method. A total of 28,729 patients were identified for the age-matched control groups: 5,981 were matched to the LARC cohort and 22,748 to the no LARC cohort. As shown in Table 1, patients in the LARC cohort were older and more frequently had a history of previous pregnancy and abortion than both no LARC cohort participants and control participants.

Table 1.:
Characteristics of Females Using and Not Using the Family Planning Services and the Long-Acting Reversible Contraception and No Long-Acting Reversible Contraception Study Cohorts

In the LARC cohort, 1,389 patients initiated an intrauterine method (68% of the LARCs that were initiated free of charge). Of the patients initiating an intrauterine method, 1,079 chose the 52-mg LNG-IUS (78% of the intrauterine methods, 53% of all LARCs), 124 chose the 13.5-mg LNG-IUS (9% and 6%, respectively), and 186 chose a copper intrauterine device (13% of the intrauterine methods and 9% of all free LARCs). Altogether 646 patients chose a contraceptive implant (32% of those starting a free LARC); nine chose the levonorgestrel-releasing implants and the remaining 637 chose the etonogestrel-releasing implant.

During the follow-up period between January 1, 2013, and February 28, 2016, 78,500 woman-years accumulated with a mean follow-up time of 2.0 years (SD 0.7) and a crude incidence rate of pregnancy ending in abortion of 12.3 abortions per 1,000 woman-years (95% CI 11.6–13.1). The woman-years in the four cohorts are shown in Table 2. As further shown in Table 2, a total of 996 patients underwent an induced abortion, and of the 259 abortions (26% of all abortions) occurring among patients using the family planning services, 16 (6%) occurred in the LARC cohort and 243 (94%) in the no LARC cohort. The crude incidence rates of pregnancy ending in abortion per 1,000 woman-years were 12.92 for service users and 12.60 for nonservice users. Among service users, the crude incidence rate per 1,000 woman-years was 3.88 in the LARC cohort and 15.25 in the no LARC cohort; in the corresponding matched nonservice user cohorts, the respective figures were 11.10 and 12.97. The probability of not having a pregnancy ending in abortion during follow-up in each of the four cohorts is presented in Figure 2.

Table 2.:
Number of Abortions, Follow-up Time in Woman-Years, and Abortion Incidence Among the Cohorts
Fig. 2.:
Survival curves by Kaplan-Meier estimates of the survival without pregnancy ending in an induced abortion among females in the four cohorts, P<.001. P value for overall difference among curves obtained by log-rank test. LARC, long-acting reversible contraceptive.Gyllenberg. Free-of-Charge LARCs and Induced Abortion. Obstet Gynecol 2018.

In the final model, adjusted rate ratios were estimated using a multivariate Poisson regression model controlling for the effect of continuous age, socioeconomic status, marital status, previous pregnancy, and previous abortion. Neither language nor level of education improved the model and were left out.

The unadjusted rate ratio of a pregnancy ending in an abortion was markedly lower in the LARC cohort compared with both the no LARC cohort and nonservice user control participants (Table 3). As further shown in Table 3, the rate ratio remained significantly lower after adjustment in the LARC cohort as compared with the no LARC cohort (rate ratio 0.20, 95% CI 0.11–0.32, P<.001) as well as compared with its nonservice user control cohort (rate ratio 0.26, 95% CI 0.14–0.43, P<.001). The additional analysis of the LARC cohort compared with the fully matched control cohorts yielded similar results for both the no LARC comparison (rate ratio 0.19, 95% CI 0.10–0.31, P<.001) and the nonservice user comparison (rate ratio 0.28, 95% CI 0.16–0.48, P<.001). There was no statistically significant difference in the abortion rates between the no LARC cohort and its nonservice user control cohort (rate ratio 1.01, 95% CI 0.87–1.18, P=.87).

Table 3.:
Crude and Adjusted Poisson Regression Estimates of the Rate Ratio of Abortion in the Service User and Long-Acting Reversible Contraception Cohorts Compared With Nonservice Users and No Long-Acting Reversible Contraception Cohorts

In the sensitivity analysis addressing potential immortal time bias, the crude incidence rate in the no LARC cohort was similar to that in the main results (14.88, 95% CI 13.12–16.80 vs 15.25, 95% CI 13.45–17.23).


We found that the abortion rate was more than 70% lower among patients who initiated a LARC method free of charge compared with patients not choosing a LARC method despite being entitled to one. This study provides further evidence of the effect size of providing free-of-charge LARC methods compared with matched cohorts in the general population. In the CHOICE study, participants were offered all contraceptives at no cost and the abortion rate among study participants varied between 4.4 and 7.5 per 1,000 patients compared with the aggregated regional level of 19.5 abortions per 1,000 patients.11 These results are in line with our study. Other studies on provision of LARC methods have been conducted in different study settings, for example, among patients in postabortion care. In such studies, initiation of a LARC method has been shown to decrease the need of repeat abortion by 50–60%,4,19,20 which is less than in our study.

We found no difference in abortion rates in the no-LARC group compared with the control group, suggesting that patients using family planning services are sexually active and fertile. Thus, service use without initiation of a LARC method did not decrease the rate of abortion among these patients.

Of all patients visiting a public family planning clinic in Vantaa during the years 2013–2014, 21% initiated a LARC method free of charge. This is a substantially smaller portion than in the CHOICE project where 75% initiated a LARC method when all contraceptive methods were provided without charge.9 However, in the real-life setting of this study, not all visits regarded initiation or switching contraceptive method, but visits also included contraceptive follow-up visits.

The strengths of this study include the real-life population setting facilitated by the use of national registers. These high-quality registers allow for detailed adjustments also at the population level. The public family planning clinics in Vantaa are well known and widely used by the population; one in four females younger than 25 years of age visits a clinic every year, and more than 90% of patients seeking an abortion referral use the services according to the structural health registers. To increase accessibility of LARC methods, same-day initiations at doctors' appointments are provided. Health care providers at public schools and maternity clinics are also aware of the public LARC program, and, because these services are used by almost 100% of adolescents and pregnant women, information on the free-of-charge program reached a considerable part of the population, especially young women and women with previous pregnancies.

The following limitations should be considered. In this observational study, we could not rule out all potential common underlying factors affecting both a patient's choice of contraceptive method and her risk of becoming unintentionally pregnant. Furthermore, we could not include cohabitation as a confounding sociodemographic background factor because cohabitation is not registered in Finland unless a couple specifically applies for a registered partnership. Information on breastfeeding was not available and could hence not be accounted for. Neither did we have information on race nor on ethnicity of cohort individuals, because these variables are banned in the Person Data Act. However, adding language in the stepwise evaluation of confounders did not improve the model. Also, the Finnish national registers only contain information on education and reproductive and work-related history obtained from Finnish institutions. This may cause missing information, especially among immigrant females. Despite this lack of data, we could control for several of the known risk factors of unintended pregnancy and decrease the influence of possible confounding. Given the low number of abortions in the LARC group, the full model with five covariates might be overfitted. Therefore, we repeated the crude comparison of no-cost LARC users with two fully matched control cohorts. This did not change the results. Finally, date of death was unavailable for nonservice users, and censoring at the date of death could hence not be performed for the main analyses. However, we had aggregated data on the all-cause mortality among females in Vantaa during the years 2013–2014, which was low and did not differ among the general population (0.56 deaths/1,000 females aged 15–44 years) compared with service users (0.50/1,000). Therefore, it is reasonable to assume that missing data on death had no influence on our results.

Regardless of the low national abortion rate in Finland (8.3 abortions/1,000 females aged 15–49 years in 2016),12 the abortion rate was significantly lower among LARC initiators than among the other study participants. We conclude that providing the population with the option of free-of-charge LARC methods is likely an effective means to decrease the abortion rate.


1. Winner B, Peipert JF, Zhao Q, Buckel C, Madden T, Allsworth JE, et al. Effectiveness of long-acting reversible contraception. N Engl J Med 2012;366:1998–2007.
2. Lindh I, Skjeldestad FE, Gemzell-Danielsson K, Heikinheimo O, Hognert H, Milsom I, et al. Contraceptive use in the Nordic countries. Acta Obstet Gynecol Scand 2017;96:19–28.
3. Kavanaugh ML, Jerman J, Finer LB. Changes in use of long-acting reversible contraceptive methods among U.S. women, 2009-2012. Obstet Gynecol 2015;126:917–27.
4. Rodriguez MI, Evans M, Espey E. Advocating for immediate postpartum LARC: increasing access, improving outcomes, and decreasing cost. Contraception 2014;90:468–71.
5. Fox J, Barfield W. Decreasing unintended pregnancy: opportunities created by the Affordable Care Act. JAMA 2016;316:815–6.
6. Cleland K, Peipert JF, Westhoff C, Spear S, Trussell J. Family planning as a cost-saving preventive health service. N Engl J Med 2011;364:e37.
7. Eisenberg D, McNicholas C, Peipert JF. Cost as a barrier to long-acting reversible contraceptive (LARC) use in adolescents. J Adolesc Health 2013;52(suppl):S59–63.
8. Kumar N, Brown JD. Access barriers to long-acting reversible contraceptives for adolescents. J Adolesc Health 2016;59:248–53.
9. McNicholas C, Madden T, Secura G, Peipert JF. The contraceptive CHOICE project round up: what we did and what we learned. Clin Obstet Gynecol 2014;57:635–43.
10. Gyllenberg F, Juselius M, Gissler M, Heikinheimo O. Long-acting reversible contraception free of charge, method initiation, and abortion rates in Finland. Am J Public Health 2018;108:538–43.
11. Peipert JF, Madden T, Allsworth JE, Secura GM. Preventing unintended pregnancies by providing no-cost contraception. Obstet Gynecol 2012;120:1291–7.
12. THL. Induced abortions 2016. Available at: Retrieved April 24, 2018.
13. Heino A, Niinimäki M, Mentula M, Gissler M. How reliable are health registers? Registration of induced abortions and sterilizations in Finland. Inform Health Soc Care 2018;43:310–319.
14. Gissler M, Ulander VM, Hemminki E, Rasimus A. Declining induced abortion rate in Finland: data quality of the Finnish abortion register. Int J Epidemiol 1996;25:376–80.
15. Gissler M, Teperi J, Hemminki E, Meriläinen J. Data quality after restructuring a national medical registry. Scand J Soc Med 1995;23:75–80.
16. Wellings K, Jones KG, Mercer CH, Tanton C, Clifton S, Datta J, et al. The prevalence of unplanned pregnancy and associated factors in Britain: findings from the third National Survey of Sexual Attitudes and Lifestyles (Natsal-3). Lancet 2013;382:1807–16.
17. Sihvo S, Bajos N, Ducot B, Kaminski M. Women's life cycle and abortion decision in unintended pregnancies. J Epidemiol Community Health 2003;57:601–5.
18. Targownik LE, Suissa S. Understanding and avoiding immortal-time bias in gastrointestinal observational research. Am J Gastroenterol 2015;110:1647–50.
19. Pohjoranta E, Mentula M, Gissler M, Suhonen S, Heikinheimo O. Provision of intrauterine contraception in association with first trimester induced abortion reduces the need of repeat abortion: first-year results of a randomized controlled trial. Hum Reprod 2015;30:2539–46.
20. Baldwin MK, Edelman AB. The effect of long-acting reversible contraception on rapid repeat pregnancy in adolescents: a review. J Adolesc Health 2013;52(suppl):S47–53.

Supplemental Digital Content

© 2018 by the American College of Obstetricians and Gynecologists. Published by Wolters Kluwer Health, Inc. All rights reserved.