Declining risk of surgical intervention following early medical abortion: A time trend analysis

A Danish nationwide cohort study of 86,437 early first-trimester medical abortions found gestational age, maternal age, previous deliveries, and history of induced abortion to be associated with the risk of subsequent surgical intervention []. However, the factors had very low predictive value, making the authors hypothesize that the risk of surgical intervention following early medical abortion mainly is influenced by other factors, including provider experience.

We evaluated the time trend in risk of surgical intervention following early medical abortions induced at a Danish gynecology clinic, where the procedure of medical abortion remained the same the entire study period, and where a single provider decided the necessity of surgical intervention.

Using information extracted from patient records, we assessed the annual risk of surgical intervention following early medical abortion in a Danish gynecology clinic (Gynecology Clinic Hvidovre) during the years 2008 (the first year of the clinic) to 2012. Throughout the study period, induction regimen as well as pre- and postabortion care did not change, and the necessity of surgical intervention was decided by the same provider (TN), a physician specialized in gynecology and obstetrics with a PhD in ultrasonography. Before the study period, the provider had worked as a chief physician at a Danish gynecology department. Considering that most first-trimester abortions induced at Danish hospitals were surgical at that time, the provider had limited experience in medical abortion prior to the year of 2008 [].

Abortions were induced ≤63 days of gestation with mifepristone, 200 mg, administrated orally in the clinic, followed 48 hours later by misoprostol, 0.8 mg, administrated vaginally at home. We assessed information on gestational age, maternal age, previous deliveries, and history of induced abortion from patient records and national health registers [,].

The provider offered follow-up ultrasonographic examination weekly for 3 weeks to all patients. These follow-up consultations were scheduled at day of mifepristone administration. If a woman presented with a persistent gestational sac, she was offered a repeated dose of mifepristone and misoprostol. If the gestational sac persisted following the second medical induction, or if the woman declined a repeated dose of medication, she was offered uterine manual vacuum aspiration on site.

Women would also be offered repeated medication or surgical intervention on site, if the provider found it necessary through clinical judgment in accordance with the national evidence-based recommendations on the management of early medical abortions []. The guideline stated that no objective parameter could be used to ensure the necessity of surgical intervention outside the clinical presentations of a persistent gestational sac or health-threatening hemorrhage [].

We manually extracted information on surgical intervention from the patient records. To account for the potential of women undergoing surgical intervention outside study site during nonbusiness hours of clinic, we also extracted information on surgical interventions provided by Danish hospitals within 8 weeks from mifepristone administration in the clinic, using The National Patient Register []. Indication for surgical intervention was only available for those occurring at study site.

We assessed the association between surgical intervention by the provider and treatment year using multiple logistic regression adjusting for gestational age, maternal age, previous deliveries, previous medical abortions, and previous surgical abortions. We performed a second analysis including all surgical interventions, including those by hospital providers outside business hours of the clinic. We also performed subanalyses that only included women without prior medical abortion. We assessed changes in surgical intervention indications over time using χ² test for trend. We performed all analyses using SAS (version 9.4, The SAS Institute, Cary, NC).

The Danish Data Protection Agency and The Danish Patient Safety Authority approved this study.

We identified 1372 medical abortions induced at study site before 63 days of gestation during the study period. Of these, 93 (6.8%) underwent surgical intervention within 8 weeks from mifepristone administration, 63 (68%) decided and done by the provider at study site and 30 (32%) decided and executed outside study site. Characteristics of the abortions are outlined in Table 1.

The proportion of surgical interventions decided and done by the provider at study site declined from 10.2% in 2008 to 2.6% in 2012, p < 0.001 (Fig. 1). The observed decline was independent of time changes in the distribution of gestational age, maternal age, previous deliveries, and history of induced abortion, the adjusted OR of surgical intervention for 1-year increase in calendar time being 0.71 (0.59–0.86). When including the surgical interventions decided and done outside the study site, the decline was still significant, OR 0.81 (0.70–0.95).

A subanalysis of abortions induced in women with no history of medically induced abortions provided similar ORs (surgical interventions at study site: OR 0.71 (0.58–0.88); all surgical interventions: OR 0.84 (0.71–0.99)).

The decline by time in the proportion of surgical interventions decided by the provider at study site was driven by a decline in the proportion of surgical interventions that were based on clinical judgment and undergone by women with no persistent gestational sac (p < 0.001, Fig. 1). The proportion of surgical interventions that were indicated due to a persistent gestational sac remained rather stable throughout the study period (p = 0.316, Fig. 1). No surgical intervention at study site was indicated due to health-threatening hemorrhage.

Median number of days (first-third quantile) from first medical administration to surgical intervention increased with around 1.5 week throughout the study period, from 17 days (14–24 days) in 2008 to 28 days (21–28) in 2012.

A total of 61 (4.4%) abortions received a repeated dose of mifepristone and vaginal misoprostol following the primary induction. Of these, 37 (60.7%) were followed by surgical intervention on site, while the remaining 24 (39.3%) abortions did not undergo further interventions following the secondary dose of medication.

The proportion of medical abortions receiving repeated medication prior to surgical intervention did not change significantly by time (p = 0.895), the average annual proportion being 59.4% of those undergoing surgical intervention. The proportion of medical abortions receiving a repeated dose of medication without further intervention did not change significantly by time (p = 0.658), the average annual proportion being 1.7%.

This study of 1372 early medical abortions showed a 75% decline in risk of surgical intervention over a 5-year-long time period independent of gestational age, maternal age, previous deliveries, and history of induced abortions in a study setting where a single gynecologist decided the necessity of surgical intervention, and where the medical regimen and the pre- and postabortion care did not change. Further, time from first medical administration to surgical intervention increased with around 1.5 week throughout the study period.

This 75% decline in risk of surgical intervention by time and the 1.5-week-increase in watchful waiting time may represent the role of increasing provider experience. Ongoing pregnancy and health-threatening hemorrhage are rare events following early medical abortion [,]. Most surgical interventions following early medical abortion are based on clinical judgment, and provider experience is therefore likely to influence risk of surgical intervention. A study by Hedley et al. pooled data from 4 published studies on medical abortion and found differences in risk of surgical intervention among the 4 study sites despite controlling for risk factors such as maternal age, gestational age, education, parity, history of spontaneous and induced abortion, and maternal weight []. They interpreted their result as an illustration of differences in experience among providers at the different sites. Considering that women in the current study were followed up by 3 physical consultations including ultrasonographic examination, it is likely that the provider in this study, over time, gained more experience in assessing the relevance and severity of bleeding symptoms and ultrasonographic features of the endometrium. Thus, the findings may not be generalizable to settings with different types of clinical follow-up (serum-hCG, remote assessment, or self-assessment), where the provider is not exposed to bleeding symptoms and ultrasonographic features of the endometrium in the same manner as in this study.

Clinical judgment of the necessity of surgical intervention following early medical abortion is not only influenced by the provider but also by the woman receiving the abortion. In the study by Allen et al., 31% of surgical interventions of early medical abortions were indicated due to patient request []. The woman's perception of the bleeding that naturally occurs during an early medical abortion, her way of reporting this perception to the provider, as well as her eager to either end the bleeding symptoms or avoid surgical intervention may all influence the provider decision to intervene. If women, throughout the study period, got more comfortable with the natural course of an early medical abortion or more prone to avoid surgery, this could also explain study findings.

Medical abortions are increasingly performed worldwide []. Thus, it should be of priority to conduct research investigating the potential beneficial effect of implementations (A) enhancing provider experience or (B) ensuring that receiving women have realistic expectations to the medical abortion procedure.

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

The study was funded by The Foundation for the Professional Development of Specialist Practices, which had no role in the scientific process.