As many as 5% to 13% of pregnancies end with serious pregnancy complications, such as preterm birth, preterm labor, prenatal hemorrhage, and gestational hypertension, and these complications can cause maternal and neonatal morbidity and mortality (
1–
4). The prevalence of pregnancy complications has risen over the past decade, most commonly affecting women from racial-ethnic minority groups and low-income, unmarried, and psychosocially stressed women (
5–
8).
Preterm birth rates are estimated at 9.6% overall and 13% among African-American women (
1). Preterm labor occurs in up to 12% of all pregnancies and generally precedes preterm birth. Risk factors for preterm labor include high stress (
2), lack of social support (
9,
10), low income, intimate partner violence, and long hours of standing at work, as well as smoking and drug and alcohol use (
1,
11).
Gestational hypertension complicates between 5% and 10% of all pregnancies (
2,
12). Seventeen percent of women with gestational hypertension develop preeclampsia, a life-threatening condition that is characterized by increased proteinuria and hypertension, a plummeting of platelets, and risk of stroke and organ failure (
13). Women with low resources and gestational hypertension are at greater risk of progressing to preeclampsia perhaps because of reduced detection through routine prenatal care (
14).
Prenatal hemorrhage is another common pregnancy complication that can threaten the life of the mother and the neonate. The two most common types of antepartum hemorrhage are placenta previa (placenta covering the cervical opening resulting in vaginal bleeding) and placental abruption (separation of the placenta from the uterine wall). Placenta previa affects .4–.5% of all pregnancies. In severe or untreated cases, early delivery may be the only way to stop the bleeding, resulting in fetal death or other neonatal complications in the case of prematurity or risk to the mother’s life in the case of spontaneous vaginal delivery with a major previa (
4). Placental abruption occurs in as many as .6%−1.0% of all U.S. pregnancies, and 10% to 20% of all perinatal deaths are related to placental abruption (
15).
The contribution of preconception chronic psychiatric and general medical conditions to poor obstetrical outcomes has become a focus in recent decades (
6,
16,
17). Depression is a risk factor for preterm birth (
6,
7,
18) and gestational hypertension (
19,
20). Posttraumatic syndrome stress disorder (PTSD) during pregnancy puts women at higher risk of ectopic pregnancy, spontaneous abortion, preterm contractions, and excessive fetal growth (
17), as well as preterm birth (
21). Schizophrenia is associated with preterm birth and placental abnormalities (
22). Recent research on preconception mental health has highlighted that psychiatric disorders are modifiable risk factors and that treatment is likely to reduce the incidence of adverse pregnancy complications and birth outcomes (
23,
24).
Ten percent of pregnant women in the United States use one or more psychotropic medications during pregnancy (
25). Although psychotropic medication use may also result in adverse pregnancy outcomes (
22), studies in this area have shown inconsistent findings of pregnancy complications as the result of prenatal psychotropic exposure (
22,
26–
28).
Existing literature on the associations of psychiatric disorders and psychotropic medication use with pregnancy complications is limited mostly to clinical reviews. Few longitudinal population-based studies have examined mental health status before and during pregnancy, as well as pregnancy outcomes (
24). To the best of our knowledge, only Witt and colleagues (
23) measured mental health status as a prepregnancy condition. Although that study provided the first evidence of a positive association between poor preconception mental health status and pregnancy complications in a nationally representative sample, the study used self-rating of mental health and was limited to measuring pregnancy complications as a dichotomous variable, grouping all complications together. Other studies that examined diagnosis-based behavioral health status also have been limited to one or two specific psychiatric diagnostic groups and pregnancy complications.
This study aimed to identify whether pregnancy complications differ between women diagnosed as having a psychiatric disorder prior to pregnancy and women without such a diagnosis and whether various psychiatric diagnoses and types of psychiatric treatment are associated with various pregnancy complications.
Methods
Data Source and Study Sample Selection
The study used Pennsylvania Medicaid claims data to select the study sample. The University of Pennsylvania’s Institutional Review Board approved this study. Women in a case group and a matched control group were selected by using the following criteria. [A figure in an online supplement to this article illustrates the sample selection process.] First, all women who gave birth between January 1, 2007, and December 31, 2009, and who were enrolled in Medicaid from one year prior to their pregnancy until their delivery date were identified. All women with diagnostic-related group (DRG) codes of 0370 (cesarean section with complication), 0371 (cesarean section without complication), 0372 (vaginal delivery with complicating diagnosis), and 0373 (vaginal delivery without complicating diagnosis) were identified as women who gave birth. The index date for giving birth was the delivery claim date. Pregnancy duration for normal delivery and for delivery with preterm birth was calculated by subtracting 280 days and 215 days, respectively, from the delivery date. A total of 27,332 women in Pennsylvania who gave birth for the observation duration were identified. Second, women with a psychiatric diagnosis who had one or more behavioral health claims or psychotropic medication prescriptions during the year prior to their pregnancy were assigned to the case group (N=4,965), and a matched control group (N=4,965) was selected on the basis of delivery month and year, age at delivery, race-ethnicity, and chronic general medical illness status during one year prepregnancy.
Dependent Variables: Pregnancy Complications
ICD-9 codes were used to identify pregnancy complications coded as dichotomous variables: antepartum hemorrhage, 641.xx; pregnancy hypertension, 642.30, 642.31, 642.33, 642.40, 642.41, 642.43, 642.50, 642.51, 642.53, 642.60, 642.61, 642.63, 642.70, 642.71, 642.73, 642.90, 642.91, and 642.93; preterm labor, 644.00, 644.03, 644.10, and 644.13; and preterm birth, 644.20 and 644.21 and DRG code 0379.
Independent Variables
Demographic variables included age at delivery (≤20, 21–30, and 31–40) and race-ethnicity (African American, Hispanic, white, and other).
Outpatient psychotherapy use.
Provider specialty codes and procedure codes of the behavioral health claims were used to identify outpatient psychotherapy use. The behavioral health claims with provider specialty codes of psychiatric outpatient, outpatient practitioner, and psychiatric behavioral consultant and the claims with medical procedure codes of psychiatric interview/exam, interactive psychiatric diagnosis interview, individual psychotherapy office/outpatient facility visit, family psychotherapy with patient present, group psychotherapy, behavioral health counseling and therapy, therapeutic behavioral service, and community-based wraparound service (therapeutic support staff) were recoded as “yes” for use of outpatient psychotherapy.
Psychotropic medication use.
Psychotropic medication use during one year prepregnancy and during pregnancy were identified by using the National Drug Code in pharmacy claims data and recoded as “yes” for use of psychotropic medication if an individual had one or more claims for the drug classes of antidepressants, anticonvulsants, antipsychotics, and benzodiazepines. Both outpatient psychotherapy use and psychotropic medication use were identified at two periods: one year prepregnancy and during pregnancy.
Psychiatric diagnosis and chronic illness status.
Psychiatric diagnosis and chronic general medical illness status were identified by using ICD-9 codes and recoded as dichotomous variables (having the condition, 1; not having the condition, 0). Psychiatric diagnosis categories included bipolar disorder (ICD-9 codes 296.00–296.06, 296.40–296.46, 296.50–296.56, 296.60–296.66, 296.80–296.82, and 296.89), depression (296.20–296.26 and 296.30–296.36), other mood disorder (296.xx, 300.xx, and 311.xx, except for the codes classified as bipolar disorder and depression), substance abuse disorder (292.xx, 303.xx, 304.xx, and 305.xx), PTSD (309.81), schizophrenia (295.xx), adjustment disorder (309.xx), and other psychiatric disorder (all other ICD-9 codes between 290 and 319).
Chronic illness status was defined as having cardiac disease (ICD-9 codes 398.xx, 411.xx–414.xx, 424.xx, 425.xx, 427.xx, 428.xx, and 446.xx), diabetes (250.xx, 271.xx, and 277.xx), hypertension (401.xx–405.xx), or epilepsy (345.xx).
Prenatal care visit.
Prenatal care visit was recoded into a dichotomous variable of prenatal care visit (yes versus no) and number of prenatal care visits.
Statistical Analysis
First, descriptive analysis of the study sample, including demographic characteristics, chronic general medical illness status, and prenatal care visits, was conducted. Second, five logistic regression models compared the odds of case and matched control groups having pregnancy hypertension, antepartum hemorrhage, preterm labor, and preterm birth and of having one or more of the pregnancy complications, adjusting for age at delivery, race-ethnicity, and chronic illness status one year prior to and during pregnancy. Third, five logistic regression models identified significant prepregnancy factors associated with pregnancy complications among women with psychiatric disorders (the case group). Finally, five logistic regression models to identify the relationship between use of outpatient psychotherapy, psychotropic medication use, or prenatal care visit during pregnancy and pregnancy complications were conducted.
Results
Table 1 presents data on the demographic characteristics of the study sample. Most were age 21–30 years (60.0%) and white (53.1%), 1.4% had a chronic general medical illness in the year before pregnancy. Overall, 98.4% had a prenatal care visit during their pregnancy and the mean number of prenatal care visits was 11.8.
Overall, 43.9% had at least one pregnancy complication (
Table 2). Preterm labor was most common (32.7%). Women who received treatment for a psychiatric disorder prior to pregnancy had greater odds than those who had not received such treatment of having at least one pregnancy complication (odds ratio [OR]=1.48). Compared with the control group, their odds of having antepartum hemorrhage were 1.50 times higher, their odds of experiencing preterm labor were 1.45 times higher, and their odds of experiencing preterm birth were 1.61 times higher.
Among women with psychiatric disorders, the odds of having any pregnancy complication were 1.19 times higher for African Americans than for whites and 1.39 times higher for Hispanics than for whites (
Table 3). Compared with women ages 31–40, those ages ≤20 or 21–30 had, respectively, 1.51 times and 1.38 times higher odds of having a pregnancy complication. Compared with women without the indicated diagnosis, the odds of having a complication were 1.17 times higher among women with depression, 1.20 times higher among women with other mood disorders, 2.43 times higher among women with smoking disorder, and 1.26 times higher among women with other psychiatric disorders.
The odds of having pregnancy hypertension were lower for Hispanics (OR=.53) than for whites; higher (OR=2.78) for women who had a chronic general medical illness prior to pregnancy, compared with those who did not; and lower (OR=.70) for women with a substance abuse disorder, compared with those without such a disorder. Use of anticonvulsants prior to pregnancy was associated with higher odds (OR=1.64) of having pregnancy hypertension.
The odds of having antepartum hemorrhage were 1.24 times higher for women with other mood disorders, compared with those without this diagnosis, and lower for Hispanics (OR=.71) than for whites.
The odds of preterm labor were 1.30 times and 1.60 times higher, respectively, for African Americans and Hispanics than for whites. For women ages ≤20 and 21–30, the odds of having preterm labor were 1.72 times and 1.45 times higher, respectively, than for women ages 31–40. Compared with women without the indicated diagnosis, the odds of having preterm labor were higher for women with depression (OR=1.17), other mood disorder (OR=1.20), smoking disorder (OR=2.10), or other psychiatric disorder (OR=1.22).
The odds of having a preterm birth were 1.78 times higher and 2.52 times higher for African Americans and Hispanics, respectively, than for whites, and 3.07 times higher for persons of other race-ethnicity than for whites. The odds of having a preterm birth were 1.63 times higher for women with PTSD than women for those without PTSD.
Having a prenatal care visit was associated with 2.43 higher odds of having one or more of the pregnancy complications and 2.69 higher odds of having preterm labor (
Table 4). Outpatient psychotherapy use during pregnancy was associated with 1.40 higher odds of having a preterm birth. Use of psychotropic medication during pregnancy was not significantly associated with having pregnancy complications.
Discussion
This study found that women with a psychiatric disorder prior to conception were more likely to have pregnancy complications, including pregnancy hemorrhage, preterm labor, and preterm birth, after the analysis controlled for age, race-ethnicity, and chronic general medical illness before and during pregnancy. Women receiving outpatient psychotherapy and prenatal care during pregnancy or using anticonvulsants before pregnancy had a greater risk of pregnancy complications. The association with outpatient psychotherapy is likely due to the fact that individuals with enduring psychiatric symptoms may require longer-term outpatient psychotherapy. The association between receipt of prenatal care and greater risk of having a pregnancy complication is most likely due to an increased need for surveillance of the pregnancy and fetal development among women with a known pregnancy complication. Use of anticonvulsants in prepregnancy was associated with higher odds of having pregnancy hypertension. The study sample included 10 women who had epilepsy prior to pregnancy and who used anticonvulsants before pregnancy. However, only one person who had an epilepsy diagnosis and who also used anticonvulsants in the year before pregnancy had pregnancy hypertension. To our knowledge, no previous study has reported an association between use of anticonvulsants prior to pregnancy and pregnancy hypertension. The finding regarding use of anticonvulsants prior to conception warrants further study to identify the causal pathway that leads to pregnancy complications.
As found in previous research, depression before conception was associated with increased risk of preterm birth (
6,
7,
18) and increased risk of pregnancy hypertension (
19,
20). PTSD was one of the most potent predictors of preterm birth, as found in previous research (
21). Of the disorders documented by medical claims for individuals in this cohort, PTSD had the strongest association with preterm birth. Smoking disorder also increased the risk of preterm labor. Previous studies have reported that smoking causes preterm labor and increases the risk of spontaneous preterm delivery, which generally follows preterm labor (
29,
30). Because this study was based on administrative data, only those with a smoking disorder diagnosis were identified, which likely is a proxy for a more severe or established smoking habit. If all smoking in the sample were identified, the observed risk of pregnancy complications associated with smoking would be greater (
31).
The findings regarding outpatient psychotherapy and prenatal care and poor pregnancy outcomes were likely confounded by selection factors. However, the findings regarding mental health conditions and poor pregnancy outcomes may be less likely to have been confounded and provide support for trials of interventions to reduce risk among pregnant women with mental disorders. Effective treatments for disorders such as PTSD and nicotine dependence could limit exposure and reduce risk of suboptimal pregnancy outcomes. Numerous preconception interventions to modify diet, alcohol consumption (
32), and risky behaviors have shown positive effects on pregnancy health. Integrated care that combines lifestyle and physical activity interventions with psychiatric treatment will be an important next step.
This study highlights the importance of assessing patients’ plans for conception or risk of unintended pregnancy when the prescription of psychotropic medications, particularly anticonvulsants, is being considered. Patients who plan to conceive in the near future should be informed of the risks and benefits of medication. Those with a high risk of unintended pregnancy (that is, those who are sexually active with a male partner and use no birth control method), should be counseled on the use birth control methods, including long-acting reversible contraceptives.
The findings confirm the persistent disparities in pregnancy complications by age and race-ethnicity (
33). Younger women and those from racial-ethnic minority groups were at much higher risk of preterm labor and racial-ethnic minority groups were at significantly higher risk of preterm birth.
The study had some limitations. The data were from Pennsylvania only, and the generalizability of the findings is unknown. The observation of pregnancy complications was limited to between 2007 and 2009. Although few studies have used data from Medicaid-enrolled pregnant women with psychiatric disorders since 2009, it is important to update the data and examine whether the risk of pregnancy complications has decreased. Psychiatric disorders and the use of psychotropic medications have both been shown to be associated with pregnancy complications in some but not all studies (
29,
34,
35), and few studies have successfully evaluated the impact of medications separately from the underlying psychiatric disorder or severity of symptoms. To make a true determination, a randomized controlled trial would be necessary; however, this is not feasible because of current ethical standards.
Conclusions
Although the findings that outpatient psychotherapy and prenatal care were associated with a higher risk of a pregnancy complication are counterintuitive, the findings of an association between having a psychiatric disorder and a higher risk of a pregnancy complication suggest the importance of population-based preconception interventions for women with psychiatric disorders and of prenatal monitoring to reduce the risk of pregnancy complications among women with psychiatric disorders. Although programs such as the Maternal Infant Health Program (
36) have been shown to reduce risk of low birth weight and preterm birth in the general population, it is not known whether these types of programs have the same effects for women with psychiatric disorders. Testing these programs in this particularly at-risk group and integrating infant health programs with evidence-based psychiatric case management and therapeutic interventions could do much to reduce the risk of pregnancy complications among women with psychiatric disorders.
The association between anticonvulsant medication use at the time of conception and pregnancy complications is concerning and adds to the conflicting information regarding the effects of psychotropic medication treatment on pregnancy outcomes. The results beg for careful public health surveillance to identify the consequences of anticonvulsant use by women of reproductive age, because about 50% of pregnancies in the United States are unplanned (
37). National efforts, such as the State Prenatal Quality Collaborations and the Pregnancy Risk Assessment Monitoring System, should be engaged as part of these surveillance efforts (
38). Continued careful surveillance will allow women with psychiatric disorders to make informed decisions about what is in their best interest and the interest of the fetus.