Edlow, Andrea G. MD, MSc; Hou, Melody Y. MD, MPH; Maurer, Rie MA; Benson, Carol MD; Delli-Bovi, Laurent MD; Goldberg, Alisa B. MD, MPH
Intrauterine fetal demise has historically been associated with higher rates of maternal morbidity and mortality compared with other pregnancy outcomes. Many of the studies examining maternal complications of intrauterine fetal demise date back to the 1950s and 1960s when intrauterine fetal demise went undetected for long periods of time as a result of few prenatal visits and infrequent clinical use of ultrasonography.1–3 The recognition of an increased risk of coagulopathy in cases of intrauterine fetal demise delivered remote from demise, as well as the development of new and safer techniques for removal of a demised fetus in the second trimester, has led to active uterine evacuation soon after recognition of intrauterine fetal demise.
Since the shift toward active management, few studies have examined maternal outcomes in cases of intrauterine fetal demise.4–7 Some recent analyses suggest that despite the shift toward active management, intrauterine fetal demise may still be associated with higher rates of maternal morbidity7 and mortality.8 The maternal morbidity associated with induction of labor compared to dilation and evacuation for management of intrauterine fetal demise is unknown. Data comparing induction of labor and dilation and evacuation for second-trimester induced abortion are also scarce.9–11
Our primary objective was to estimate whether uterine evacuation for second-trimester intrauterine fetal demise, defined as pregnancy loss between 14 and 24 weeks of gestation, is associated with an increased risk of maternal morbidity compared with induced second-trimester abortion. Our secondary objective was to compare the maternal morbidity of second-trimester dilation and evacuation with that of induction of labor in the second trimester.
MATERIALS AND METHODS
We conducted a retrospective cohort study with approval by the Partners Healthcare Institutional Review Board. To investigate maternal morbidity and mortality in women found to have intrauterine fetal demise in the second trimester, we compared the maternal outcomes of two cohorts: 1) women diagnosed with a fetal demise by ultrasonography between 14 and 24 weeks of gestation at Brigham and Women's Hospital from January 1, 2003, to May 8, 2009, who subsequently underwent uterine evacuation by dilation and evacuation or induction of labor at Brigham and Women's Hospital or at a Brigham and Women's Hospital-affiliated private outpatient facility dedicated to providing dilation and evacuation services (Women's Health Services); and 2) women undergoing induced abortion for any indication at Brigham and Women's Hospital between 14 and 24 weeks of gestation by either dilation and evacuation or induction of labor during the same time period.
The intrauterine fetal demise cohort was identified through a comprehensive radiology database that was queried for all diagnoses of “demise” from 14–24 weeks of gestation during the study period of interest. To be included in the intrauterine fetal demise cohort, demise had to be diagnosed before labor induction and could not be an incidental finding after induction of labor was started for another indication. Similarly, iatrogenic demises after feticidal injection as part of an induced abortion were not included in the intrauterine fetal demise cohort. We identified the induced abortion cohort by procedure type and date of procedure using the following methods: 1) for each woman with an intrauterine fetal demise who underwent dilation and evacuation as the mode of uterine evacuation, the schedule of Brigham and Women's Hospital Family Planning Clinic was queried for the first eligible patient undergoing induced second-trimester abortion by dilation and evacuation on the same day, or next closest day, as the procedure for intrauterine fetal demise. The rationale for selecting the first eligible “control” patient in this manner was to achieve standardization of clinical procedures as much as possible and to ensure a similar health care provider pool. The Family Planning Clinic is an ambulatory clinic within Brigham and Women's Hospital dedicated to providing outpatient gynecologic procedures, including uterine evacuation for first- and second-trimester induced and spontaneous abortions. The Family Planning Clinic schedule also includes patients who, as a result of medical complexity, history of anesthesia issues, or other considerations, underwent dilation and evacuation in the main hospital operating room; and 2) for each woman with an intrauterine fetal demise who chose induction of labor, control subjects were selected in a similar fashion from the population of all women who underwent induction-termination in the labor and delivery unit at Brigham and Women's Hospital during the study period. Indication for pregnancy termination was not a criterion for selection of control subjects.
The indications for induced abortion in the induction of labor and dilation and evacuation control cohorts included: undesired pregnancy with no known fetal or maternal conditions informing decision for termination, fetal anomaly or fetal condition incompatible with life, preterm premature rupture of membranes, and severe maternal medical conditions in which pregnancy could jeopardize the mother's health or life.
Exclusion criteria were the same for both cohorts and included: 1) multiple gestations; 2) placentation abnormalities including placenta previa, accreta, increta, and percreta; 3) any maternal condition that would preclude eligibility for either dilation and evacuation or induction; and 4) uterine evacuation that was not completed at either Brigham and Women's Hospital or Women's Health Services, so that accurate and complete documentation of the procedure and all complications was available. Patients were not excluded from either the intrauterine fetal demise or the control cohort for maternal conditions including preterm premature rupture of membranes, severe preeclampsia, abruption, and chronic medical conditions, including hypertension, systemic lupus erythematosus or other immunologic conditions, thrombophilias or bleeding disorders, renal dysfunction, immunosuppression, or diabetes.
Estimated gestational age for both cohorts was based on earliest documented ultrasonography. For many women, the earliest documented ultrasonography was at the time of presentation with intrauterine fetal demise or for induced abortion. Ultrasonographic dating of gestational age was based on biparietal diameter corrected for occipitofrontal diameter in the majority of cases; if only biparietal diameter was available, this measurement was used to estimate gestational age.12 Gestational age was based on femur length in situations in which biparietal diameter or biparietal diameter corrected for occipitofrontal diameter was either unavailable (eg, anencephaly) or of questionable accuracy (eg, partial collapse of skull or overlapping sutures noted on ultrasonographic report).
Both Brigham and Women's Hospital and Women's Health Services provide abortion services up to 24 weeks gestational age. Abortion services are routinely offered into the early part of the 23rd week of gestation to ensure that with standard error in ultrasonographic dating, procedures remain under 24 weeks. At both Brigham and Women's Hospital and Women's Health Services, procedures are performed by obstetrics and gynecology residents supervised by family planning fellows or attendings or by family planning fellows and attendings without resident involvement. There is significant overlap in the health care provider pool at these two sites. At both Brigham and Women's Hospital and Women's Health Services, patients undergoing dilation and evacuation for a fetus of 14-week size or greater routinely received laminaria for cervical ripening on the day before the procedure. All patients routinely received antibiotics to presumptively treat Chlamydia (either 7 days of 100 mg doxycycline twice daily or a single dose of 1 g azithromycin) starting on the day of laminaria placement. Patients between 18 and 24 weeks may have received 2 days of cervical ripening with laminaria at the discretion of the health care provider. Starting in November 2006, as a result of a cluster of infectious complications, all patients receiving 2 days of laminaria received ceftriaxone (or levofloxacin for known penicillin or cephalosporin allergy) prophylaxis in addition to the aforementioned oral regimen for treatment of Chlamydia. Starting in 2007, at Brigham and Women's Hospital, patients receiving 2 days of laminaria also received intraamniotic digoxin injection. Paracervical vasopressin was routinely administered at both Brigham and Women's Hospital and Women's Health Services before dilation and evacuations at 14–24 weeks, and uterotonic agents were administered during or after the procedure as needed. All procedures were performed with ultrasonographic guidance and conducted under intravenous general anesthesia by a staff anesthesiologist or nurse anesthetist.
For patients who underwent an induction of labor in the second trimester for intrauterine fetal demise, the most commonly used misoprostol protocol involved the initial administration of 400 micrograms of vaginal misoprostol followed by 200 micrograms of vaginal misoprostol every 3–4 hours until delivery. For some patients, a high-dose oxytocin regimen was used at the discretion of the health care provider. Laminaria were often, but not uniformly, placed 12–24 hours before misoprostol or oxytocin induction, depending on gestational age and health care provider preference.
For labor induction terminations for anomalies, preterm premature rupture of membranes, or hemolysis, elevated liver enzymes, low platelets syndrome, feticidal injection was performed on the day before induction, most commonly with intraamniotic digoxin, but occasionally with fetal intracardiac potassium chloride. With the exception of feticidal injection, patients undergoing induction of labor for reasons other than intrauterine fetal demise received the same drug regimen as patients with intrauterine fetal demise. Patients who developed fever during the induction process were treated for presumed chorioamnionitis at the discretion of the health care provider based on the health care provider's suspicion for true chorioamnionitis and not prostaglandin-induced fever. Typically patients are not treated for chorioamnionitis during misoprostol induction unless fever is more than 101°F.
Data collection included maternal demographic information; obstetric and substance use history; size of pregnancy at the time of diagnosis of demise by radiologic criteria (biparietal diameter corrected for occipitofrontal diameter or crown rump length or femur length if applicable); gestational age by best obstetric dating (eg, gestational age based on earliest ultrasonography) at the time of diagnosis of demise if different from size of pregnancy by radiographic criteria; maternal tobacco use; presence of fetal anomalies; obstetric complications of current pregnancy; maternal medical comorbidities; induction agent; laminaria administration, feticidal injection, or both when relevant; estimated blood loss when available; fetal autopsy findings when available; and length of hospital stay.
The following events were defined as major maternal complications: death; need for transfusion; infection requiring treatment with intravenous antibiotics (not prophylaxis); intensive care unit stay; sepsis, shock, or systemic inflammatory response syndrome; uterine perforation; unplanned procedure (including same day or interval suction curettage, hysterotomy, hysterectomy, laparotomy, uterine artery embolization, or other unplanned procedure); and thrombotic event (amniotic fluid embolus, stroke, deep venous thrombosis, or pulmonary embolus) diagnosed within 6 weeks of dilation and evacuation or induction of labor. A dichotomous outcome variable for any of these major maternal complications was created.
Computerized medical records and paper charts were examined for both the intrauterine fetal demise and control cohorts. Data were abstracted from charts by two individuals, including the primary author (A.G.E.), using a standardized data sheet. The primary author reviewed every completed data sheet to ensure accuracy and uniform abstraction technique.
Numeric variables were compared using either Student's t tests or Wilcoxon rank-sum tests. Categorical variables were compared using either Fisher's exact tests or chi square tests. Data were missing for more than 5% of records for the following variables: race (16%), drug use (12%), smoking (10%), and history of cesarean delivery (7%). The pattern of missing data appeared to be random, so we applied the multiple imputation method for analysis.13 Five imputed data sets were created for multiple imputation. Maternal age (30 years old or older compared with younger than 30 years old) and gestational age (20 weeks or greater compared with less than 20 weeks) were dichotomized.
Univariable logistic regression was performed to assess the group effect and procedure effect on maternal morbidity. Multivariable logistic regression was also performed adjusting for potential confounders, including maternal age, parity, gestational age, and smoking status. There were no significant baseline differences between groups with respect to race, obstetric complications, and maternal medical comorbidities, so the multivariable logistic regressions did not adjust for these variables. We did choose to adjust for baseline differences that were not statistically significant between groups but that we believed might be associated with morbidity. A P value, odds ratio (OR), and its 95% confidence intervals (CIs) were reported for each finding. SAS 9.1 was used for the analysis. Assuming maternal morbidity of 11% for intrauterine fetal demise and 1% for induced second-trimester abortion,10 at least 94 patients were needed in each group to maintain 80% power and a two-sided α of 0.05.
Of 253 patients diagnosed with fetal demise on Brigham and Women's Hospital ultrasonography at 14–24 weeks of gestation, 121 women met inclusion criteria. Of these, 82 underwent dilation and evacuation, 37 underwent induction of labor, one woman underwent spontaneous vaginal delivery at home with spontaneous placental delivery in the hospital, and one woman underwent spontaneous vaginal delivery at home and required dilation and evacuation for retained placenta. The two women who delivered spontaneously were included in the induction of labor cohort for analysis, because they had indicated intent to pursue induction of labor. The control cohort included 121 women who underwent induced abortion (82 by dilation and evacuation and 39 by induction of labor).
The intrauterine fetal demise and induced abortion cohorts did not vary significantly with respect to race, parity, prior cesarean delivery, obstetric complications in the current pregnancy, chronic medical conditions, or drug use (Table 1). The intrauterine fetal demise cohort had a greater incidence of preterm premature rupture of membranes and chronic hypertension. Women in the intrauterine fetal demise cohort were significantly older, less likely to smoke, had a lower gestational age, and were less likely to have fetal anomalies than were women in the induced abortion cohort. The median (quartile 1, quartile 3) number of days from diagnosis of intrauterine fetal demise to initiation of procedure was 1 (0, 3).
There were no maternal deaths in the study population. Maternal morbidity associated with uterine evacuation for intrauterine fetal demise compared with induced abortion is depicted in Table 2. Evacuation of second-trimester intrauterine fetal demise was not associated with increased risk of maternal morbidity compared with second-trimester induced abortion on crude analysis (25 of 121 women undergoing uterine evacuation for intrauterine fetal demise had associated maternal morbidity compared with 27 of 121 women undergoing induced abortion; OR 0.91; 95% CI 0.49–1.68; P=.75). Multivariable logistic regression was performed to adjust for potential confounders, including maternal age, parity, gestational age, and smoking status (Table 3). However, the association between uterine evacuation for intrauterine fetal demise and maternal morbidity remained nonsignificant (adjusted OR 1.15; 95% CI 0.57–2.32). Blood transfusions were increased in the intrauterine fetal demise group (n=7) compared with the induced abortion group (n=1), although this association did not achieve significance (P=.07).
Although the numbers of women with any individual marker of morbidity were too small to adequately power statistical comparisons, there were no cases of thrombotic events; intensive care unit stay; or sepsis, shock, or systemic inflammatory response syndrome among women having second-trimester induced abortion, whereas there were eight instances of these morbidities in the intrauterine fetal demise cohort, representing six patients. In a post hoc analysis excluding intravenous antibiotic use as an indication of major maternal morbidity, the rate of morbidity for the induced abortion cohort fell to 7.4%, the rate for the intrauterine fetal demise cohort fell to 19.8%, and intrauterine fetal demise became associated with a significant increase in maternal morbidity (crude OR 3.08; 95% CI 1.37–6.94; P=.005).
Higher gestational age was significantly associated with maternal morbidity. The mean gestational age among women who had any maternal morbidity was 137.6±19.3 days (19.6 weeks), whereas the mean gestational age among women with no morbidity was 127.8±18.3 days (18.3 weeks) (P<.001). Gestational age 20 weeks or greater was associated with significantly greater odds of having any maternal morbidity compared with gestational age less than 20 weeks (OR 2.59; 95% CI 1.39–4.84; P=.002). A stratified analysis examining maternal morbidity associated with intrauterine fetal demise compared with induced abortion by gestational age 20 weeks or greater or less than 20 weeks found that at gestational ages 20 weeks or greater, 18 of 57 women undergoing induced abortion had maternal morbidity compared with 10 of 30 undergoing a procedure for intrauterine fetal demise (OR 1.08; 95% CI 0.42–2.78; P=.87). At gestational ages less than 20 weeks, 9 of 64 women undergoing induced abortion had maternal morbidity compared with 15 of 91 undergoing a procedure for intrauterine fetal demise (OR 1.21; 95% CI 0.49–2.96; P=.68).
We examined the relationship between the presence of maternal morbidity and the placental and fetal pathology findings at autopsy for the intrauterine fetal demise group. Placental and fetal pathologic findings included presence or absence of abruption; placental infarcts; severe chorioamnionitis, funisitis, or both; anomalies; and cord accident. We found that only the presence of severe chorioamnionitis, funisitis, or both was significantly associated with the presence of any maternal morbidity (P<.001).
Demographic characteristics of patients in the induction of labor and dilation and evacuation cohorts (pooling women with intrauterine fetal demise and women who underwent a second trimester induced abortion) are listed in Table 4. Overall, induction of labor was more morbid than dilation and evacuation in both crude and adjusted analyses (crude OR 6.27; 95% CI 3.23–12.17; OR adjusted for maternal age, gestational age, parity, and smoking status, 5.36; 95% CI 2.46–11.69). Among patients undergoing induction of labor, there was no increased morbidity for women with an intrauterine fetal demise (OR 0.43; 95% CI 0.17–1.07). Among patients undergoing dilation and evacuation, there was no significant increase in morbidity for those with an intrauterine fetal demise (OR 2.17; 95% CI 0.77–6.10). Maternal morbidity associated with dilation and evacuation compared with induction of labor is depicted in Table 5. The results of the multivariable logistic regression analysis for maternal morbidity by procedure type (dilation and evacuation compared with induction of labor) are depicted in Table 6.
The increased morbidity observed in the induction of labor cohort can be attributed to increased odds of intravenous antibiotic treatment. We performed a post hoc analysis excluding intravenous antibiotic use from the definition of morbidity and found that 11 of 78 patients in the induction of labor group had any morbidity compared with 13 of 164 patients in the dilation and evacuation group. This result was not significant (P=.13). Thirty-two of the 52 patients who had any morbidity were treated with intravenous antibiotics for presumed infection. Of these, 11 had evidence of severe histologic chorioamnionitis or funisitis on the placental pathology report. Two of the patients treated with intravenous antibiotics had sepsis, shock, or systemic inflammatory response syndrome, and both of these patients also had evidence of severe histologic chorioamnionitis. One patient underwent dilation and evacuation and the other induction of labor.
We performed an analysis to examine the effect of procedure on morbidity within the intrauterine fetal demise and the second-trimester induced abortion cohorts. Within the intrauterine fetal demise cohort, including intravenous antibiotic use in the definition of morbidity, odds of morbidity were 2.92 times higher in women who underwent induction of labor. Thirteen of 39 women (33%) who underwent induction of labor experienced maternal morbidity compared with 12 of 82 (15%) of women who underwent dilation and evacuation. The association was statistically significant (OR 2.92; 95% CI 1.18–7.21; P=.02). In the induced abortion cohort, 21 of 39 women (54%) with induction of labor experienced maternal morbidity compared with six of 82 women with dilation and evacuation (7%). The odds of morbidity were 14.78 times higher undergoing induction of labor for induced abortion compared with dilation and evacuation (95% CI 5.21–41.92; P<.001).
Uterine evacuation for second-trimester intrauterine fetal demise is not significantly associated with increased overall maternal morbidity compared with induced abortion in our study population. However, when we excluded intravenous antibiotic use as an indication of major maternal morbidity, we found that intrauterine fetal demise was associated with significantly more morbidity than induced abortion. The higher transfusion rate in the intrauterine fetal demise cohort is also concerning. Induction of labor was more morbid than dilation and evacuation in the second trimester for the overall cohort as well as within each group. Increased use of intravenous antibiotics to treat presumed infection in the induction of labor cohort comprised the majority of morbidity, but the rate of serious infection was low overall.
Our findings should be viewed in the context of prior studies. The few studies examining maternal outcomes in the setting of active management of intrauterine fetal demise suggest persistence of increased maternal morbidity and mortality.5,7,8,14 A recent analysis suggested that maternal mortality rates in the setting of intrauterine fetal demise may be as high as 96.3 per 100,000 outcomes compared with an overall maternal mortality rate of 5.59 per 100,000.8 Prior studies have been limited by their largely descriptive or cross-sectional nature, lack of control groups, and potential overestimation of maternal morbidity by including primarily intrauterine fetal demises in the third trimester when options for removal of a demised fetus may be limited to induction of labor or abdominal delivery.4–7 Gestational age in prior studies ranged from 24 to 40 weeks, and uterine evacuation was achieved by either vaginal delivery or hysterotomy in all studies except one.7 The complications seen in these studies, including perineal and lower genital tract lacerations, reflect the higher gestational age and mode of evacuation.
Evidence comparing maternal morbidity associated with induction of labor compared with dilation and evacuation for the management of intrauterine fetal demise is sparse. Data comparing induction of labor and dilation and evacuation using modern techniques for second-trimester induced abortion are also limited.9–11 One retrospective study found that dilation and evacuation was associated with significantly lower complication rates than induction of labor for second-trimester induced abortion.11 Rates of chorioamnionitis and endometritis observed in the induction of labor cohort of two small retrospective studies were much lower than in our study, at 9%10 and 1.4%.11 Given the difficulty with randomization to dilation and evacuation or induction of labor for second-trimester induced abortion,9 retrospective and nonrandomized studies such as this one are likely to provide the best data available about maternal morbidity.
Our study has several limitations. Retrospective data collection may have led to ascertainment bias both in patient identification and in data acquisition. Selection bias could have contributed to a difference in morbidity between induction of labor and dilation and evacuation, particularly because the induction of labor group had a higher mean gestational age. We do not believe that this difference was the result of health care provider avoidance of dilation and evacuation at higher gestational ages. Rather, there may have been more routine offering of, and easier access to, induction of labor for patients with intrauterine fetal demise, fetal anomalies, or pregnancy complications, which are typically conditions diagnosed later in pregnancy. To minimize selection bias, all women included in the study needed to be clinically eligible for either dilation and evacuation or induction of labor.
Although our expected rate of maternal morbidity in the second-trimester induced abortion group was 1%, our observed event rate was 22%. Much of the observed morbidity in the induced abortion cohort was attributable to the use of intravenous antibiotics for treatment of presumed infection. Because intravenous antibiotic use may not truly represent serious morbidity, and because some of these diagnosed infections may have been prostaglandin-induced fevers instead of true infections, we performed a post hoc analysis excluding the use of intravenous antibiotics from the composite maternal morbidity score. With this exclusion, the rate of maternal morbidity observed in the second-trimester abortion cohort fell to 7.4%, a rate consistent with previously reported complication rates for second-trimester abortions performed in a hospital setting.10 After making this adjustment, the complication rate among women undergoing evacuation for intrauterine fetal demise was significantly higher than for women undergoing second-trimester induced abortion.
A strength of this study is the presence of a rigorously selected control group. We did not exclude patients from either cohort based on maternal conditions, which makes it likely that patients in both cohorts were drawn from a similar population. Selecting the induced abortion cohort by procedure type and date helped ensure standardization of methods and health care provider pool. Our selection of women undergoing second-trimester induced abortion as control participants also has clinical relevance. In the United States, women undergoing second-trimester induced abortion most often have dilation and evacuation procedures in an outpatient setting with very low rates of morbidity, whereas women with a second-trimester intrauterine fetal demise may receive more heterogeneous care. In this study, we estimated the morbidity of each group to help assess whether these patients can and should be managed similarly.
Although our initial analysis found that dilation and evacuation was associated with less morbidity than induction of labor for all women requiring second-trimester uterine evacuation regardless of indication, when intravenous antibiotic use for treatment of fever and presumed infection was excluded as a criterion for morbidity, there was no difference in morbidity between procedures. Thus, it remains unclear whether it is better for a woman with an intrauterine fetal demise to delay uterine evacuation to access a dilation and evacuation procedure or whether she should proceed with induction of labor if that service is more readily available. Perhaps in situations in which induction of labor is more readily available and intravenous antibiotics are accessible, patients should not delay uterine evacuation to access a dilation and evacuation. Future larger studies should explore specific aspects of morbidity, including the need for transfusion of blood products, to help health care providers select the safest setting (ie, outpatient clinic or hospital-based operating room) for the management of women with second-trimester intrauterine fetal demise.
© 2011 by The American College of Obstetricians and Gynecologists. Published by Wolters Kluwer Health, Inc. All rights reserved.