Labor Induction Termination of Pregnancy

Labor induction abortion is an alternative to surgical abortion in the second trimester, and provides access to abortion services when there are no providers trained in second trimester surgical abortion techniques. Induction of labor is also indicated in some cases of pregnancy termination for fetal abnormalities, such as when an intact fetus is important for diagnosis or for the patient's grieving process. However, an intact fetus is not necessary solely to confirm prenatal diagnosis of a genetic abnormality; accurate chromosomal analysis can be performed in almost every case after a dilation and evacuation (D&E) procedure.¹

Although the majority of second trimester abortions performed in the United States are performed surgically,² second trimester induction of labor is performed more frequently as gestational age increases. In the late second trimester and third trimester, labor induction is the primary method of elective termination in cases of lethal fetal abnormalities. When termination of a desired pregnancy is necessary for maternal indications, gestational age and the likelihood of fetal survival are factors in the choice of a treatment regimen.

The incidence of induction abortion is higher in many other countries where D&E is not commonly practiced either as a matter of medical policy or because of the lack of facilities or providers.

Assessment of medical abortions is not uniform. "Success" of medical abortion is usually defined as the proportion of women who abort at either 24 or 48 hours. However, it can also refer to procedures that are completed with the original regimen of medications, without resorting to a secondary medical procedure. Success is usually defined as the passage of the fetus, but in some reports it refers to unassisted passage of both fetus and placenta. Abortion time, abortion interval, or time to abortion is usually meant to refer to the time from the initiation of medication to the passage of the fetus, although in a few instances it may refer to passage of the placenta. 

In the United States, approximately 97,000 second trimester abortion procedures were reported to the Centers for Disease Control (CDC) in 2004.² This number is an underestimate of the total number of second trimester abortions performed in the United States, because it comes from only 46 states, and the District of Columbia. California, which possibly has the largest number of abortions, is not included. Of reported abortions, 11.7% occurred over 12 weeks of gestational age. Of all the procedures with gestational age and type of procedure reported, 89% were performed surgically, 0.6% used intrauterine instillation techniques, and 10.8% used another method. Other methods include early medical abortion (8.1%), and induction abortion and other techniques (including hysterotomy and hysterectomy), which total 2.7%.² The incidence of instillation methods has fallen over the past 10–15 years from 1–2%.^{3, 4} Induction abortion performed with oral or vaginal medication without intrauterine instillation, as is commonly performed now, does not have a separate category and may reported as 'other'.

Intra-amniotic instillation

Instillation methods are not frequently used at present, largely because alternative methods, including surgical options, have fewer side effects, lower risk of complications, and more rapid completion times. These invasive procedures require special skills to minimize risks such as a bloody tap and maternal intravascular injection, introduction of infection, and rupture of membranes.

Hypertonic saline

One of the first described instillation methods was hypertonic saline. Early use resulted in some maternal deaths attributable to poor technique, a free-flow system (rather than a drip infusion), and use of concentrations exceeding 20%.⁵ Major complications included maternal hypernatremia and coagulopathy, which was clinically evident in three of 1000 cases.⁶

When used alone, intra-amniotic hypertonic saline has a long latent period until the onset of contractions. Only 20% of women will abort within the first 24 hours and 81% will abort by 48 hours, with a mean time to abortion of 30 hours.⁷ Addition of oxytocin to this regimen improves the efficacy and expulsion time. In a series of 5000 saline inductions at 14–24 weeks' gestation⁶ using a drip intra-amniotic infusion of 20% saline and intravenous oxytocin at less than 50 mU per minute, average time to abortion ranged from 22.5 to 25 hours; 99.6% of patients were successfully induced and retained placenta occurred in 13% of cases. Significant complications included hemorrhage in 2.3%, transfusion in 0.5%, clinical coagulopathy in 0.3%, and cervical laceration in 0.3%. Two patients who received high doses of oxytocin (more than 150 milliunits per minute) had uterine rupture, one of which required hysterectomy.

A similar series from the Joint Program for the Study of Abortion (JPSA), a multicenter prospective observational study conducted by the United States CDC,⁸ found that among 4778 patients from 13 to 24 weeks' gestation treated with intra-amniotic hypertonic saline and intravenous oxytocin, mean induction interval was 25.6 hours, the rate of retained placenta was 14.8%, and the rate of transfusion was 1.7%.

Urea

Urea initially appeared to be a potentially safer agent than hypertonic saline for intra-amniotic instillation. Unlike hypertonic saline, urea is relatively safe when an inadvertent intravascular injection occurs, because urea rapidly traverses cell membranes and is an osmotic diuretic. When used alone, urea has a long instillation to abortion interval. As such, various additional agents have been evaluated and used in combination with urea to provide a clinically acceptable method of second trimester abortion.

Initial experience most commonly combined intra-amniotic urea, 80–90 g, with intravenous oxytocin.^{9, 10, 11} Early studies found varying success rates of 85–97% within 48 hours and an average time to expulsion of 19–29 hours, with the differences most likely caused by differences in the gestational ages of the treated population and variable amounts of oxytocin.^{9, 10} Higher success rates were reported with high-dose oxytocin (332 mU per minute);⁹ however, these rapid oxytocin infusions were associated with oliguria caused by the antidiuretic effect of high-dose oxytocin. The oxytocin-induced oliguria was reversible several hours after discontinuing the oxytocin.¹⁰

Intra-amniotic prostaglandins

Instillation of prostaglandin made sense as a medical method of labor induction abortion because the amniotic fluid would act as a reservoir, slowly allowing the prostaglandin to cross the sac to stimulate the myometrium. Natural prostaglandins such as PGE₂ and PGF_2α were investigated in large multicenter and multinational studies in the 1970s.

In one study of 700 women at 12–22 weeks' gestation (mean: 17.1 weeks), 5 mg of intra-amniotic PGE₂ were instilled in 500 patients, and 10 mg were instilled in 200 patients; intravenous oxytocin at 100 mU per minute was begun 6 hours later.¹² The abortion rates at 24 hours were 90% for those who received 5 mg instillations and 88% for 10 mg instillations, and mean times to abortion were 15.5 and 14.5 hours, respectively. Vomiting was reported in 39%, diarrhea in 5%, hemorrhage in 1.5%, transfusion in 0.7%, genital injury in 0.3%, and infection in 0.4%.

The World Health Organization⁷ reported a comparative analysis of 1503 women who had abortions at 13–22 weeks' gestation with either intra-amniotic PGF_2α or hypertonic saline. The regimens consisted of 25 mg of intra-amniotic PGF_2α with a repeat dose 6 hours later or hypertonic saline (20% NaCl) 200 mL as a single instillation. The abortion rates at 24 hours were 61% and 20% and at 48 hours were 86% and 80%, respectively. The mean times to abortion were 19.7 hours for PGF_2α and 30.0 hours for saline. Approximately 33% of patients in both groups required curettage for retained placenta or incomplete abortion. Although side effects like diarrhea (15% versus 1.3%) and vomiting (54% versus 19%) were more common after instillation of PGF_2α, there were no statistically significant differences in serious complications.

Grimes and associates¹³ reported data from the JPSA comparing side effects and complications between 1241 PGF_2α instillation abortions and 10,013 using hypertonic saline. In this series, the dose of PGF_2α ranged from 25 to 50 mg. Intravenous oxytocin was added in 61% of women receiving PGF_2α and 90% of those with saline. The major complication rate was 2.9% for the PGF_2α abortions and 1.8% for saline.  Specifically, the risks of hemorrhage, retained placenta, infection, convulsions, and re-admission were increased with PGF_2α abortions. Thus, although PGF_2α appears more effective than hypertonic saline at successfully terminating pregnancy, it is also associated with more side effects and complications.

Prostaglandin analogues, specifically 15-methyl PGF_2α, have a prolonged duration of action because they are not inactivated by the 15-dehydrogenase enzyme. Subsequently, their development led to investigations of whether they would be as effective for labor induction abortion. The World Health Organization¹⁴ performed three studies, including two randomized trials, comparing a single instillation of 2.5 mg of 15-methyl PGF_2α to 40 mg and 50 mg of PGF_2α. The abortion rates for the women receiving 15-methyl PGF_2α were similar in all three studies: 72–76% at 24 hours and 93–96% at 48 hours. These rates were significantly better than the rates of either 40 mg PGF_2α (54% at 24 hours and 82% at 48 hours) or 50 mg PGF_2α (68% at 24 hours and 87% at 48 hours). Use of 15-methyl PGF_2α resulted in significantly lower rates of hemorrhage (1–2% versus 3–4%), vomiting (1.3–1.5% versus 1.7–2.1%), and diarrhea (0.4–0.5% versus 1.2–1.3%). Cervical injury occurred equally in both groups (2.9%); the only serious complication in the entire study was an extension of one of these lacerations into the lower uterine segment. Thus, 15-methyl PGF_2α seems to be a better agent than the natural PGF_2α at effecting labor induction abortion in the doses tested.

Intra-amniotic medication combinations

In efforts to further improve efficacy of second trimester inductions and to decrease the induction to abortion interval, combinations of intrauterine agents have been investigated.

Hypertonic saline plus prostaglandin F_2α

In a study of 508 women from 14 to 24 weeks' gestation (mean: 19.9 weeks), 20 mg of PGF_2α was added to an instillation of hypertonic saline. Ninety-six per cent of women delivered by 24 hours, the mean instillation to abortion time was 16.1 hours, and 24% of the patients required curettage for retained placenta.¹⁵ Such high success and rapid delivery did result in undesirable side effects. Nausea, vomiting, diarrhea, and fever occurred in 48%, 33%, 1%, and 3%, respectively. Serious side effects included coagulopathy in 1%, hemorrhage in 2%, and transfusion in 2%. One case of uterine rupture and one case of sepsis also occurred.

Urea plus prostaglandin F_2α

The addition of intra-amniotic PGF_2α to intra-amniotic urea and intravenous oxytocin does result in more complete and rapid expulsion; however, the trade-off is a higher rate of side effects and complications.⁹ Nausea and vomiting occur in approximately 73% of patients treated with intra-amniotic urea and PGF_2α and intravenous oxytocin as compared to 53% of those who receive just the intra-amniotic urea and intravenous oxytocin. Similarly, diarrhea occurs in 5% and 1%, respectively. The hemorrhage (9–10% and 5%, respectively) and cervical laceration (3% and 1%, respectively) rates are also greater.

The combination of urea with 15-methyl PGF_2α, has also been evaluated.¹⁶ In a small series (n = 62) of women at 16–27 weeks' gestation (mean: 21.0 weeks), laminaria were inserted immediately before instillation of 80 g urea and 2 mg of 15-methyl PGF_2α. The success rate at 24 hours was 97% with a mean time to abortion of 13.2 hours. Curettage was necessary for incomplete abortion or retained placenta in 61% of cases. Vomiting occurred in 34%, diarrhea in 13%, hemorrhage in 2%, and infection in 2%. This series suggests that use of 15-methyl PGF_2α, combined with intra-amniotic urea and intracervical laminaria, has similar overall efficacy and speed to a similar regimen using PGF_2α¹⁷ but 15-methyl PGF_2α is associated with fewer side effects and does not require intravenous high-dose oxytocin to obtain this outcome.

Extra-amniotic instillation

Prostaglandins

Extra-amniotic PGE₂ instillation through the cervical os was first described in the early 1970s. A series of 1608 patients was treated with 1.5 mg or 2.5 mg of PGE₂ gel (injected via a catheter through the internal os into the extra-amniotic space), plus intravenous oxytocin beginning 6 hours after gel administration. Abortion occurred within 24 hours in approximately 78% of those women who received 1.5 mg and 90% who received 2.5 mg.¹² The mean time to abortion was also significantly different, 15.7 hours versus 14.0 hours, respectively. Vomiting was reported in 45%, diarrhea in 17%, transfusion in 0.5%, infection in 0.2%, and a cervical laceration in 0.06%. Only 35% had a complete abortion and did not require surgical evacuation for retained placenta. A small study¹⁸ used a lower concentration of PGE₂ gel (0.75 mg) with reapplication in 5 hours followed by intravenous oxytocin administration. Success rates were 80% at 24 hours and the mean time to abortion was 17.6 hours. Vomiting was reported in 30% and diarrhea in 5%. Retained placenta occurred in 20% of patients. This suggests that a lower dose of PGE₂ gel may be used with similar efficacy to a 2.5-mg dose, although the time to delivery is slightly longer in exchange for fewer side effects. Even lower doses have been tested but have unacceptable success rates and longer times for delivery.¹⁹

Ethacridine lactate

This method is seldom described in the English literature; historically, it has been described more commonly in German, Chinese, and Japanese reports. The most common places in which this method is used are some Eastern European and Asian countries. Ethacridine lactate is a yellow dye with antiseptic properties. Ingemanson²⁰ instilled a 0.1% solution of ethacridine lactate into the extra-amniotic space at a dose of 10 mL per gestational week to a maximum of 150 mL. No oxytocin was administered. In 29 women at 13–16 weeks' gestation, the abortion rate was 45% at 24 hours, 72% at 48 hours, and 93% at 72 hours. In 24 women at 17–20 weeks' gestation, the abortion rate was 29% at 24 hours and 96% at 48 hours. In both groups, the rate was significantly greater than that for a comparison group of women who received extra-amniotic hypertonic saline (20% NaCl). Additionally, no gastrointestinal side effects were noted. Ten per cent of the ethacridine lactate patients had significant temperature elevation and 1% had infection.

Two separate reports from Ankara, Turkey, have also described limited experience with this method compared with other induction techniques. Inan and colleagues²¹ used the same dosing method for ethacridine lactate as described, except the maximum dose was 200 mL. The authors found an abortion rate of approximately 80% at 24 hours in groups of 48 and 49 women, who received the ethacridine lactate alone or in combination with intravenous oxytocin, respectively. Yapar and associates²² used even more 0.1% ethacridine lactate (250–300 mL) combined with intravenous oxytocin if needed (only used in 50%) in 82 patients from 14 to 28 (mean: 19.4) weeks' gestation. The abortion rate was 69% at 24 hours and 99% at 48 hours with a mean time to delivery of 16 hours. Notable side effects were coagulopathy in 2% and hemorrhage secondary to uterine atony in 2%.

The clinical applicability of ethacridine is limited by the long abortion time. In a retrospective study comparing it to misoprostol, misoprostol was faster and more effective.²³ In a retrospective study, misoprostol was more efficient than ethacridine, with 24 hour abortion rates of 92.6% vs. 76.2%.²⁴ A summary of Chinese publications concluded that mifepristone and misoprostol were faster and had higher success rates, although gastrointestinal side effects were more common with mifepristone/misoprostol.^{25, 26}

A concentrated high-dose oxytocin regimen was first described by Winkler and associates²⁷ in a retrospective evaluation of 22 subjects. Beginning with a dose of 100 units per 500 mL of 5% dextrose in normal saline (D5NS), the 500 mL is infused over 3 hours, and then 1 hour is allowed for diuresis to preclude water intoxication. The dose of oxytocin is increased 50 units per 500 mL of D5NS until delivery is achieved, to a maximum of 300 units. This method was compared with a cohort of 59 women receiving vaginal PGE₂ (20 mg every 4 hours for 24 hours). Both groups included women at 17–24 weeks' gestation, with a mean of 21.3 weeks for the oxytocin group and 19.3 weeks for the PGE₂ group. This method effected delivery in 91% of women receiving the oxytocin regimen, with a mean time to delivery of 8.2 hours. The PGE₂ group had a similar rate of efficacy (93%) but had a significantly longer mean time to delivery (13.1 hours). As compared with the PGE₂ group, those receiving the oxytocin infusion had less nausea (46% versus 9%), vomiting (37% versus 9%), fever (64% versus 0%), diarrhea (20% versus 0%), and live births (8.5% versus 0%). These side effects occurred despite premedication of women in the prostaglandin group with antipyretic, antiemetic, and antidiarrheal agents.

This evaluation was followed by the same group with a prospective randomized trial using the same regimens in a similar population.²⁸ In this trial, all women with intact membranes and a closed cervix had hygroscopic dilator (Dilapan) placement 4 hours before initiation of treatment. Approximately 20% of the subjects in both groups had a fetal demise. The abortion rates by 24 hours were similar, 86% and 80% for the PGE₂ and oxytocin groups, as were the mean times to delivery, 13.2 and 12.6 hours, respectively. As with the preliminary study, women who received prostaglandin had more nausea (62% versus 33%), vomiting (54% versus 18%), fever (21% versus 0%), diarrhea (24% versus 0%), and live births (18% versus 3%). 

Yapar and colleagues²² reported their experience in Turkey with a less concentrated regimen of oxytocin in 36 women at 14–28 weeks' gestation (mean: 20.2 weeks). This regimen started with 5 units per 500 mL of D5NS infused over 3 hours, followed by 1 hour of D5NS. This 4-hour cycle is repeated, increasing the oxytocin concentration by 5 units each time to a maximum of 30 units. The abortion rates were 92% at 24 hours and 97% at 48 hours.

Thus, use of a high-dose oxytocin regimen appears, in several small studies, to be at least as efficient as intravaginal PGE₂ with fewer side effects.

Prostaglandin E₂

In 1978, vaginally administered PGE₂ became available for abortion in the second trimester. These suppositories are not stable at room temperature and require refrigeration. Typical doses are a single suppository (20 mg) every three or four hours. Abortion rates are approximately 50% at 12 hours^{29, 30} after initial administration and 81–96% at 24 hours.^{27, 28, 29, 30, 31, 32} In nonrandomized studies, addition of oxytocin and pretreatment with laminaria do not appear to shorten the time to delivery.^{27, 29} The average interval to delivery is approximately 12–14 hours,^{27, 28, 29, 30, 31, 32, 33} but earlier gestations respond more slowly than those of 16 weeks or more.^{29, 30}

Vaginally administered PGE₂ was a potentially attractive alternative to both intra-amniotic instillations and extra-amniotic administration of prostaglandins, because of the relatively rapid time to delivery and ease of administration. Unfortunately, side effects occur at a very undesirable rate despite pretreatment with antipyretic, antiemetic, and antidiarrheal agents. Vomiting is reported in approximately 30–80% of women, diarrhea in 15–39%, and pyrexia (temperature greater than 38^oC in 21–64%.^{27, 28, 29, 30, 32, 33} Between 42% and 50% of women will need a curettage for retained placenta or incomplete abortion.^{27, 33} These side effect rates appear similar to those with extra-amniotic PGE₂ gel administration. Live birth has been reported to occur in 1–21% of vaginal PGE₂ inductions.^{27, 28, 29, 30, 31} Rare side effects such as severe hypotension^{28, 33} and hypertension^{27, 33} have also been reported with the use of PGE₂ suppositories, but not at a rate greater than with intra-amniotic PGF_2α.³³

One more recent randomized trial compared intramuscular (IM) 15-methyl PGF_2α (0.25 mg every three hours) to intravaginal PGE₂ suppositories (20 mg every three hours).³¹ Of the 26 women at 14–24 weeks' gestation who received the IM regimen, only 69% were delivered at 24 hours compared with 96% for the PGE₂ group. The mean times to abortion were 21.2 and 12.2 hours, respectively. This intramuscular regimen does not provide a clinically acceptable rate of abortion given other options available to clinicians.

Prostaglandin E₁ analogue: gemeprost

Gemeprost, a PGE₁ analogue, is a vaginally administered prostaglandin analogue more commonly used in Europe. Like PGE₂, these suppositories are not stable at room temperature and require refrigeration. Typical doses were initially a single suppository (1 mg) every 3 hours with abortion rates of approximately 80% at 24 hours after initial administration and 95% at 48 hours.^{34, 35} In the 1990s, some investigators questioned whether a regimen with less frequent administration would provide equal efficacy with fewer side effects. Two prospective randomized studies performed in the United Kingdom compared vaginal gemeprost 1 mg every 3 hours for a maximum of five doses in 24 hours to 1 mg every 6 hours for four doses in 24 hours.^{36, 37} The regimens were repeated 24 hours after the initial treatment and intravenous oxytocin was started if the abortion had not occurred within 36–48 hours. These two studies found that although more frequent dosing was associated with a somewhat more rapid abortion interval, the cumulative abortion rates within 24 hours, the overall rate of side effects and the rate of surgical intervention for incomplete abortion were the same in both groups.^{36, 37} These data suggest that a 6-hour interval between dosing for gemeprost is similar to the 3-hour regimen, thereby allowing fewer dosings and lower costs.

Prostaglandin E1 analogue: misoprostol

Vaginal misoprostol, another PGE₁ analogue, has been evaluated in second trimester abortion. This preparation has the advantage of being stable at room temperature and inexpensive. It is absorbed orally, vaginally, sublingually, buccally and rectally. In an initial study of primarily early second trimester patients administered an 800 µg vaginal dose followed by 400 µg at 18 and 24 hours if needed, 102 of 128 (80%) aborted within the first 18 hours, with a mean interval to abortion of 11.8 hours.³⁸ No severe side effects were reported and pain was minimal, with only 20% of patients requiring paracetamol. All patients had curettage after expulsion of the fetus, so the rate of retained placenta was unable to be evaluated.

Since that report, multiple randomized controlled trials have investigated the use of misoprostol for second trimester induction of labor. Comparing or combining data from these studies is difficult for several reasons. First, different studies use different definitions of successful induction. Some studies define success as complete abortion, such that no curettage is required. This definition is similar to the definition used for successful medical abortion in the first trimester.³⁹ Other studies define success as delivery of the fetus within a prespecified time frame, usually 24 or 48 hours. Also, different studies managed patients with prolonged inductions (undelivered at 24 or 48 hours) differently. In some studies, all patients undelivered at 24 or 48 hours underwent D&E, and in other studies they were managed with alternative medications. Combining these reports is further complicated by differences in the mean gestational age of patients enrolled, by the use of additional interventions (such as fetal injections with intracardiac KCl and intracervical laminaria), and by the inclusion of patients with spontaneous fetal demise. All of these factors may have significant impact on outcomes. Table 1 provides these main outcomes for randomized controlled trials of misoprostol for second trimester induction of labor, when compared with other agents.

Table 1. Selected randomized controlled trials of misoprostol compared to other agents for second trimester induction of labor

PV, vaginal administration
^aSuccess defined as delivery of fetus within 48 hours
^bAbortion time (interval) presented in hours, some studies present mean, others median
^cVaginal administration of misoprostol
^dData only available for delivery of fetus within 24 hours
^ep <0.05
^fTime not stated

Summarizing these data, randomized trials have been conducted comparing misoprostol with PGE₂ suppositories alone,³² high-dose oxytocin combined with PGE₂ suppositories,⁴⁴ gemeprost,^{40, 41, 42} intra-amniotic PGF_2α,⁴⁵ and intra-amniotic 15-methyl PGF_2α.⁴³ In these comparisons, misoprostol was as effective as vaginal PGE₂ alone,³² intra-amniotic PGF_2α,⁴⁵ and equal to or more effective than gemeprost.^{40, 41, 42} In contrast, misoprostol was less effective than intra-amniotic 15-methyl PGF_2α or high-dose oxytocin combined with vaginal PGE₂.^{43, 44} However, these findings may be strongly influenced by the dose of misoprostol studied. Generally, higher doses and more frequent dosing of misoprostol have been associated with greater success rates and a decreased induction to abortion interval.

Although several randomized controlled trials have investigated different dosing regimens for misoprostol,^{40, 47, 48, 49, 50, 51} the ideal dosing for second trimester induction of labor remains unclear. These studies suggest that the ideal dose of vaginal misoprostol for second trimester induction is between 200 and 600 µg and the ideal dosing interval is between 3 and 12 hours. One study comparing 400 µg of misoprostol administered vaginally every 3 hours versus every 6 hours found greater success rates (91% versus 76%) and more rapid time to delivery (15 versus 19 hours) with more frequent dosing.⁴⁸ Two subsequent studies evaluating misoprostol 400 µg vaginally every 4–6 hours found that 76–85% of women delivered within 24 hours,^{50, 51} and the time to delivery was 19.6 hours (mean)⁵¹ and 15.1 hours (median).⁵⁰ Compared with a 600 µg dose, the 400 µg dose was associated with less vomiting (28% versus 37%), diarrhea (2% versus 10%), and fever.⁵⁰ However, the incidence of fever using the 400 µg dose every 4 hours was still high (25%).⁵¹ Despite the presence of fever, none of the patients had an infection diagnosed, and all fevers resolved after delivery.

Vaginal misoprostol is more effective than oral misoprostol;^{52, 46, 53} this is not unexpected as vaginal misoprostol results in more sustained serum levels.⁵⁴ However, sublingual and buccal misoprostol have also been widely used, as they are more convenient to administer and may be preferred by patients. The phamacokinetic profiles of sublingual and buccal misoprostol are similar to that of vaginal misoprostol.^{55, 56, 57} One randomized study of vaginal versus sublingual use showed no difference in abortion times.⁵⁸ A meta-analysis of studies of vaginal versus sublingual administration concluded that there was not a clinical difference in the routes of administration.⁵⁹ However, a large equivalence trial of vaginal versus sublingual administration showed that when medications were administered every 3 hours, that vaginal use was associated with a higher rate of complete abortion. They noted that multiparous women had a similar rate of abortion in each group.⁶⁰ The difference in overall rates appeared to be related to the effect of nulliparous women, who had longer abortion times with sublingual use. There is less information about buccal use. One study comparing repeat doses of buccal misoprostol (the first dose was administered vaginally) to all-vaginal misoprostol showed no difference in abortion times.⁶¹

The most dramatic improvement in second trimester induction abortion technology has come with the introduction of mifepristone. The induction to abortion interval has been dramatically reduced when mifepristone is used before prostaglandin analogues.⁶²

The classic dose of mifepristone is 600 mg administered orally, followed 36–48 hours later by hospital admission for administration of prostaglandin. Initially, the prostaglandin used was gemeprost, 1 mg every 3 hours for a maximum of five doses in 24 hours. The largest experience published with this regimen is a multicenter trial of 267 gestations between 12 and 24 weeks (mean: 16.6 weeks) from 20 centers.⁶³ Two per cent of women aborted just from the mifepristone, 100% aborted within 75 hours of beginning treatment, and the mean induction to abortion time was 7 hours after gemeprost administration. Vomiting occurred in 6% and diarrhea in 1% of women; these rates are much lower than those seen with other methods. Surgical evacuation for retained placenta or incomplete abortion was performed in 53% of patients.

Similar to other mifepristone dosing studies, a randomized trial demonstrated equal efficacy and similar rates of side effects and complications with 200 mg and 600 mg doses of mifepristone used 36–48 hours before misoprostol⁶⁴ (see Table 2). In this study, 11% of women had a retained placenta and one woman in each group required a transfusion.⁶⁴ Because mifepristone is an expensive drug, these data suggest that a decreased dose of mifepristone can be used with equal efficacy and much reduced cost.

As with early first trimester abortions, misoprostol has also been evaluated in combination with mifepristone. The advantages of misoprostol over gemeprost are lower cost and no need for refrigeration of the medication. Initially, two randomized controlled trials compared misoprostol with gemeprost when used after mifepristone for second trimester induction of labor.^{4, 65} These studies both demonstrated that, when administered 36–48 hours after mifepristone, misoprostol and gemeprost are equally effective, with similar induction to abortion intervals and side effect rates (Table 2).

Table 2. Randomized controlled trials of mifepristone and misoprostol for second trimester induction of labor

PV, vaginal administration; SL, sublingual administration; PO, oral administration
^aPercentage of women delivering the fetus within 24 hours
^bAbortion interval (time to abortion) presented in hours, some studies present mean, others present median
^cp <0.05

Three randomized controlled trials then compared a variety of misoprostol dosing regimens. In each study, misoprostol was initiated 36–48 hours after mifepristone (see Table 2). El-Rafaey and colleagues⁶⁶ used a single loading dose of 600 µg of vaginal misoprostol in all women and then compared repeat misoprostol doses of 400 µg every 3 hours, administered either orally or vaginally (to a maximum of 5 doses). No significant differences were found in success rates, time to abortion, side effects, and complications. Ho⁶⁷ compared 200 µg of misoprostol dosed every 3 hours, administered either orally or vaginally. He found a higher success rate and more rapid time to abortion using the vaginal route of administration. Ngai⁷⁰ compared 200 µg of misoprostol administered vaginally to 400 µg administered orally, given at the same interval, and found these doses to be equivalent. Randomized studies of Tang et al. and Hamoda et al. showed that sublingual administration had results equivalent to vaginal administration.^{69, 71}

In a large case series, Hamoda et al.⁷² reported 1002 women from 13 to 21 weeks' gestation. Between 36 and 48 hours after mifepristone, women were given an initial dose of misoprostol 800 µg inserted vaginally. They were given up to four repeat doses of oral misoprostol 400 µg. If, at the end of all five doses, they had not aborted, they rested overnight and then repeated the same schedule of misoprostol. Of these women, 97.1% aborted after the first cycle of up to five misoprostol doses, and 99% abortion after the second set. The median time to abortion was 6.6 hours for nulliparous women and 5.9 hours for multiparous women. Earlier pregnancies had shorter abortion times. Eight women (0.8%) had surgical abortion because the pregnancy was not passed. The rate of surgical intervention because of retained placenta was 7%. Seven women (0.7%) required a blood transfusion, and one woman at 14 weeks required a hysterectomy for intractable bleeding. Complications were not related to gestational age or parity, but retained placenta was more common in older women.

Most studies have used a mifepristone-to-misoprostol interval of 36–48 hours, but a shorter interval might be more convenient. Kapp et al. ⁷³ used a 24 hour interval between mifepristone and buccal misoprostol 400 µg every 6 hours, and had a median induction time of 10 hours. The gestational ages in this study were 18–23 weeks, resulting in a longer abortion time than other studies. Heikinheimo et al. randomized women to 1 and 2 day intervals between mifepristone and the start of misoprostol; they used mifepristone 200 mg, and misoprostol 400 µg vaginally every 3 hours for all doses.⁷⁴ They found that the median time from misoprostol initiation to abortion was 7.25 hours for the 1 day group and 6.2 hours for the 2 day group (p <0.05), however, the proportion of women who had aborted by 12 hours was the same. These findings were confirmed by Nilas et al.⁷⁵ and Mentula et al.⁷⁶ Two studies had contradictory findings on the incidence of placental removal.^{76, 77} Chai et al.⁷⁸ randomized women to receive mifepristone right at the time of starting misoprostol, compared to 36 hours later, and found, not surprisingly, that abortion times were shorter in the 36 hour group.

These trials suggest that 200 mg of mifepristone followed 24–48 hours later by misoprostol is very effective at inducing abortion within 24 hours after the first misoprostol dose. Higher doses of misoprostol have been associated with greater efficacy and shorter induction to abortion intervals. Data primarily from the United Kingdom have shown the highest complete abortion rates, using a vaginal loading dose of 600–800 µg of misoprostol followed by 400 µg orally every 3 hours. The efficacy of this regimen, and rates of transient fetal survival, for women late in the second trimester is unknown, because most of these trials only included women at 20 weeks' gestation or less. In contrast to earlier agents, parity is also a predictor of shorter induction time in multiple studies.^{48, 79, 71, 67, 74} For studies which include women in early second trimester, induction times are shorter than those of women later in second trimester. 

Mifepristone has been approved in several countries for use in the second trimester, and appears to be widely used where mifepristone is available.^{80, 81, 82, 83}

There is a wide variety of approaches to management of the placenta. Recommendations were made for placental removal 30 minutes after prostaglandin E₂⁸⁴ or 2 hours after saline infusion,⁸⁵ based on an increase in bleeding after the recommended time period. Using misoprostol for induction, the practice of surgical removal of the placenta after a set period of time, usually 1–2 hours, is documented in myriad articles. However, Leader documented that about half of women deliver the placenta within 2 hours and did not show an increase in bleeding when women were observed past the 2 hour mark.⁸⁶ This same study also concluded that routine misoprostol administration after fetal delivery was not beneficial. Green⁸⁷ in a series of 233 concluded that there was no increase in bleeding for at least 4 hours. The placenta was delivered within an hour in 59% of women, and the rate of operative removal was 6%. In women undergoing misoprostol-only abortion, a randomized trial of placebo, misoprostol, and oxytocin administered after fetal expulsion was done. Use of oxytocin resulted in a 90% placental expulsion rate within 1 hour.⁸⁸

Comparatively low rates of intervention for placental removal have been reported multiple times using mifepristone and misoprostol, with rates ranging from 5 to 11%.^{72, 89, 90} At this point it is not clear whether the low rates reported with mifepristone reflect a pharmacologic effect or practice patterns. In a small study which did state the time to placental delivery after mifepristone abortion, Kapp reported that one out of 32 women (3%) required placental removal.⁷³

Some practitioners perform curettage routinely after delivery of the placenta, whether the placenta was assisted or not. Operative management of the placenta may be considered as a complication, annoyance, or routine. If placental removal is done in the operating room with general anesthesia, the risk, expense, and professional time are greatly increased over procedures in a treatment room with local anesthesia and mild analgesia.  

It is unclear if pretreatment with osmotic dilators decreases the induction interval. Studies of older induction methods found a decrease in abortion interval when laminaria were placed approximately 18 hours before treatment with prostaglandins.⁵⁰ Two randomized studies found no decrease in time to abortion when laminaria were placed at the time of initiating misoprostol.^{37, 51} In addition, women who had laminaria required more pain medication.⁵¹ There are no published randomized studies of the use of osmotic dilators in advance of starting induction, although Hern reported good results in a large observational series.⁹¹ Mazouni, in a retrospective study of women have abortion with mifepristone and misoprostol, found that women who had laminaria inserted 24 hours prior to induction had markedly reduced abortions times.⁹² Carbonell-Esteve et al.⁹³ use osmotic dilators (Dilapan, a synthetic dilator) in most women undergoing mifepristone–misoprostol, but did not report a comparison group.

Transient fetal survival was very unlikely after installation of saline or urea, which are directly feticidal. Rates of transient fetal survival after misoprostol abortion range from 0% to 50% when the gestational age is greater than 18–20 weeks. The incidence of transient fetal survival is not stated in many articles, and it is uncertain how often the rate is zero.

Similarly, it is unclear whether induced fetal demise (feticide) before second trimester induction of labor decreases the induction to abortion interval. Most studies that include women with fetuses close to the limits of viability induce fetal demise to prevent transient fetal survival, but randomized trials of feticide in induction abortion have not been done. There is limited documentation about the effect of feticide on abortion outcome. Elimian et al. in a retrospective study of women having PGE₂ abortion, found shorter abortion times in women who had undergone feticide by intracardiac KCl injection.⁹⁴ Silva et al. reported that abortion time was not altered by feticide,⁹⁵ however, they induced fetal demise with KCl intracardiac injection just prior to induction. More prolonged fetal demise, for example 24 hours or more, may have an effect on abortion times, but studies are lacking. 

Side effects, mainly gastrointestinal, are common in second trimester abortion. They may be related to the agent used for induction, or to the process of abortion itself, or to medications used for analgesia. Nausea and vomiting are most common with prostaglandin analogues but can occur with any agent. The incidence of nausea and vomiting is frequently not reported. Chills and temperature elevation are also common after prostaglandin administration, with some reports as high as 40%. As an example, one large study of misoprostol-only procedures reported nausea, vomiting and diarrhea in 16%, 12%, and 24% of women; they also noted shivering in 38% and fever in 35%.⁶¹ However, clinical infection appears to be uncommon.

Hemorrhage occurs in a small percentage of women, with the incidence of transfusion ranging from 0 to 3%. The threshold for use of transfusion is dependent on local practice and may vary greatly from location to location. However, one study which measured hemoglobin before and after abortion found no difference in mean levels.⁹⁶ Cervical injury is thought to have been more common with PGF_2a procedures. It is rarely reported now.

Failure of medical abortion, requiring a surgical procedure, is very uncommon with misoprostol or gemeprost procedures, with or without mifepristone. However, the failure may be more common in women with prior cesarean section.⁹⁷

Uterine rupture is a rare but catastrophic complication that can result in hysterectomy. Reports of uterine rupture are virtually all case reports of one or two incidents, and there appear to be at least 50 cases in the literature to date. Uterine rupture has been reported with almost all agents used to cause medical abortion, including high-dose oxytocin,⁹⁸ hypertonic saline,⁶ PGF_{2a ,}¹⁵ ethacridine,⁹⁹ and gemeprost¹⁰⁰. Uterine rupture has been reported in women undergoing second trimester abortion with misoprostol.^{101, 102, 103, 104} Women were at various gestational ages, and various doses of misoprostol were used; most received multiple doses. Some of the women had prior cesarean sections and some had not. Mifepristone, by producing cervical softening prior to misoprostol, might be expected to decrease risk, but there have been reports of rupture after use of mifepristone as well.^{93, 105, 106} The absolute risk of uterine rupture with misoprostol induction of labor in the second trimester is unknown. Prior hysterotomy is suspected to be a risk factor for uterine rupture, as many of the case reports involve women with scarred uteri. However, two series with a combined total of 1147 women with unscarred uteri and 157 with scarred uteri had no ruptures in either group^{107, 108} and in another series rupture occurred in one of 216 women with unscarred uteri and none of 108 women with scarred uteri.¹⁰⁹ Goyal presented a comprehensive review of published cases of second trimester abortion using misoprostol, although about half the cases used other agents such as oxytocin in addition.¹¹⁰ He concluded that the risk of rupture for women with a prior cesarean delivery was 0.28% (95% CI 0.08, 1.0) compared with the rate for unscarred uteri, 0.04% (95% CI 0.02–0.20). Bergella et al. estimated the risk of rupture to be 0.4% after cesarean section.¹¹¹ 

Comparison of labor induction methods with D&E is difficult because the medical literature comparing modern methods of both techniques is sparse. Generally, physicians within a community tend to perform either one type of procedure or the other, making comparisons difficult.

One of the first reports to compare labor induction to D&E was published in 1984 by Kafrissen and associates¹¹² from the United States CDC. In this study, 2805 patients at 13–24 weeks' gestation (95% between 17 and 24 weeks) were treated with intra-amniotic urea combined with PGF_2α. Eighty per cent of women also received intravenous oxytocin, and 41% received intracervical laminaria. The comparative group of D&E included 9572 patients, of which 91% were between 13 and 16 weeks' gestation and 9% between 17 and 24 weeks. Laminaria were used for cervical dilation in only 23% of the D&E procedures. Overall, the serious complication rate was 1.03% for the instillation abortion and 0.49% for D&E, with a relative risk of 2.1. When the risk was adjusted for age, race, parity, medical conditions, and follow-up, the relative risk remained significantly elevated (1.9; 95% CI 1.2–3.1). Instillation abortion patients were more likely to have fever, retained products, endometritis, and cervical injury requiring repair. However, women having D&E were more likely to have a perforation (0.2% versus 0%). The incidences of coagulopathy and death were rare for both procedures (1–2 per 10,000) and not significantly different. The authors separately evaluated abortions performed at 17–20 weeks' gestation and found no differences in complications rates between techniques.

Autry and associates¹¹³ compared complication rates among 158 women undergoing second trimester induction abortion and 139 undergoing D&E, in a retrospective, non-randomized study. The patients had a mean gestational age of 20.3 weeks in the induction group and 18.4 in the D&E group (p <0.001). Of the induction group, 125 women were treated with misoprostol and the remaining women were managed with other techniques. Complications included failed medical abortion (requiring dilation and curettage), hemorrhage requiring transfusion, infection requiring intravenous antibiotics, retained products of conception requiring evacuation, organ damage that required additional surgery (including uterine perforation), cervical laceration that required repair, and readmission to the hospital. Using this definition, 29% (45/158) of women in the induction group had at least one complication compared with 4% (5/139) in the D&E group (p <0.001). One woman in each group had hemorrhage requiring transfusion, two women in the induction group and none in the D&E group had organ damage requiring surgery (including uterine rupture or perforation), and two women in the induction group versus three in the D&E group had a cervical laceration requiring repair. The most frequent complications (33) in the induction group were related to placental management or change in medication (11). When only serious complications are concerned, the complication rates are much closer (6% versus 4%).

Bryant et al.¹¹⁴ also performed a retrospective cohort study, including women with fetal demise. Induction was done with misoprostol, without mifepristone. The gestational age was higher in the induction group (20 weeks) than the D&E group (18 weeks). Overall complications were more common in the induction group (24% vs. 3%), but most complications were interventions for placental removal. When complications other than placental removal were considered, the complication rates were the same (3% vs. 3%).

Since Autry's retrospective study, techniques of induction have changed significantly, becoming safer, and more efficient and effective. Pretreatment with mifepristone has resulted in shorter procedures, with apparently less chance of retained placenta and failed procedure. There is one randomized trial which enrolled 122 women at 13–20 weeks of pregnancy and randomized them to receive either medical abortion with mifepristone and misoprostol or surgical abortion using gemeprost cervical priming.¹¹⁵ Immediate medical complications were similar in both groups. However, at 14-day follow-up, women in the medical group were more likely to report heavier bleeding and that they had had more pain with the abortion itself. This study was hampered by poor follow-up at 14 days.

Induction techniques have been modified and combined with surgical evacuation methods. One technique was described by Hern.⁹¹ In a series of 832 abortions, he placed serial osmotic dilator insertions over several days and performed feticide. He then ruptured membranes and placed vaginal misoprostol. Most women delivered within several hours, either by manually assisted fetal expulsion or with surgical assistance to collapse fetal parts during delivery. Women who did not deliver spontaneously had a surgical extraction. This allowed the abortion to remain an outpatient procedure. There were no major complications in this series. Other variations of this technique have been reported.

The data comparing these two procedures as currently practiced are extremely limited. There are no studies comparing combination methods such as described by Hern to either induction or D&E. There is very little information about women at gestations over 20 weeks, and no information about the experiences of women who choose to abort an anomalous fetus. Further work will be necessary to compare modern methods of labor induction and D&E, including overall risk and cost.

Labor induction abortion provides a safe method of terminating pregnancies in the second trimester. With the introduction of new prostaglandin analogues and mifepristone, there have been major changes in efficacy, side effect profiles, and clinical practice. Procedures with prostaglandin E1 analogues have the shortest induction times. Prostaglandin E₂ procedures may be slightly longer, but are accompanied by a higher rate of side effects. Prostaglandin F and non-prostaglandin methods have the longest induction times. The time to induction is influenced by parity, with nulliparous women having longer inductions, and by gestational age, with higher gestation associated with longer inductions.

Complications of procedures with current methods include bleeding and the need for transfusion, which varies but is commonly reported around 0.5%. Infection is uncommon. A more serious concern is uterine rupture, which is documented by multiple case reports. It is unclear whether rupture is associated with a particular agent. Although rupture appears to be more common with a uterine scar, neither the absolute nor relative risk has been determined. Retained placenta or the use of operative procedure occurs in most series of 5–10% of women using mifepristone techniques and in some series of misoprostol abortion without mifepristone.

Mifepristone followed 36–48 hours later by a vaginal prostaglandin analogue (primarily misoprostol) appears to be the most effective and efficient method of labor induction. With this regimen, all medications are administered either orally or vaginally, and generally no invasive techniques are required. This combination of medications results in higher overall success rate and the shortest interval from start of induction to abortion than other techniques. However, the published experience with mifepristone and misoprostol for second trimester induction abortion is still relatively small compared with the thousands of women reported in the medical literature who have had abortions with instillation techniques. To date, most of the clinical trials of mifepristone for second trimester induction abortion have been conducted in the United Kingdom or Asia, and most include women at a mean gestational age of 15–17 weeks. Guidelines for second trimester abortion have been published by WHO.¹¹⁶ The Royal College of Obstetricians and Gynecologists has similar guidelines,¹¹⁷ as has the Society of Family Planning.¹¹⁸ The FIGO Working Group on the Prevention of Unsafe Abortion and its Consequences has also recently published guidelines.¹¹⁹

 Table 3. Recommendations for abortions using mifepristone and misoprostol or gemeprost

^aIf abortion does not occur in the first 24 h, may be repeated the following day if necessary
^bIf abortion does not occur in the first 24 h, gemeprost 1 mg may be administered q3h on the following day

Further studies must be performed to better understand the relative differences in second trimester pregnancy termination techniques, taking into account gestational age. In addition to comparing efficacy, complications, and side effects, patient preferences must be taken into account. The advent of shorter procedures means that for many women, induction can be an outpatient procedure. 

Whether a second trimester abortion is performed medically by labor induction or surgically by D&E, appropriate training and clinical skill is important to maximize safety and efficacy and minimize side effects and complications.