Reduction of maternal mortality has become a priority in the United States and other nations.1,2 As a consequence of collaborative efforts among the American College of Obstetricians and Gynecologists, the Centers for Disease Control and Prevention (CDC), the Society for Maternal-Fetal Medicine, and others, common causes of pregnancy-related mortality such as hemorrhage, hypertension, venous thromboembolism, and infection are now targeted with specific action plans and safety bundles delineating appropriate clinical care.2 In February 2016, the Council on Patient Safety in Women's Healthcare released a safety bundle for maternal mental health, which summarizes screening strategies and suggested responses when disorders are identified.3
Despite these advances, there has been relatively little focus on maternal deaths resulting from self-harm, specifically those related to accidental overdose and suicide. This lack of focus on deaths from self-harm may be, in part, a result of barriers to screening and treatment for psychiatric disorders.4,5 The lack of recognition of deaths from self-harm as a significant contributor to maternal death may also be a result of the exclusion of accidental or incidental deaths occurring during pregnancy or in the first year postpartum from reporting to the CDC as pregnancy-related mortality.6 The first steps in prevention include gaining a better understanding of the characteristics of women with a maternal death resulting from self-harm.
The objectives of this study were to define rates of maternal deaths from self-harm in Colorado, describe the demographic and clinical characteristics of women who died, and assess risk factors for these deaths to identify opportunities for prevention.
MATERIALS AND METHODS
We conducted a case study of all women with pregnancy-associated deaths (death during pregnancy or within 1 year postpartum from any cause) from self-harm in Colorado from January 1, 2004, through December 31, 2012. Deaths resulting from suicide (regardless of cause, including suicide by poisoning) and drug overdoses (unintentional poisonings and poisonings of undetermined intent) were categorized broadly as deaths from “self-harm” for the purposes of this study, because it is often difficult for the coroner and Colorado Maternal Mortality Review Committee (a panel of experts including obstetrician–gynecologists, perinatologists, substance use disorder experts, injury prevention experts, psychiatrists, and a coroner) to accurately classify these deaths by intent even after reviewing all available documentation.7,8 This approach was also supported by prior literature demonstrating an overclassification of overdoses as accidental rather than suicide.7 Cases of overdoses are not classified as suicide by a medical examiner or coroner unless there is clear evidence indicating an intention to take one's own life and deliberate means. Often this must be corroborated by other evidence of suicide such as a suicide note, known suicidal ideations, or a prior attempt.7 When cases of maternal death by accidental overdose are reviewed by mental and behavioral health experts on the Colorado Maternal Mortality Review Committee, it is often concluded that the death may have been intentional despite a documented manner of death as accidental. Thus, these outcomes were combined for the purposes of this study.
Institutional review board approval was sought from both the Colorado multiple institutional review board (because data were to be analyzed at the University of Colorado) and the Colorado Department of Public Health and Environment institutional review board (because data were abstracted at the Colorado Department of Public Health and Environment). This project was deemed exempt by the Colorado multiple institutional review board because all participants were dead and approved by the Colorado Department of Public Health and Environment institutional review board.
Each year, Colorado maternal deaths are identified by the Colorado Department of Public Health and Environment from three sources: 1) International Classification of Disease, 10th Revision “O” codes on the death certificate, indicating that the cause of death was related to or aggravated by pregnancy, childbirth, or the puerperium; 2) selection of a pregnancy field on the Colorado death certificate (includes “pregnant at the time of death,” “not pregnant, but pregnant within 42 days of death,” and “not pregnant, but pregnant 43 days to 1 year before death”); and 3) matching of all state death certificates among females aged 8–65 years to the reported name of the mother on all live birth and fetal death certificates during the same year as the death or the year prior.
After identifying possible cases of maternal death by these methods, Colorado Department of Public Health and Environment staff and members of the Colorado Maternal Mortality Review Committee excluded cases that did not meet the accepted definition of a pregnancy-associated death (death during pregnancy or within 1 year postpartum from any cause).9 The Maternal Mortality Review Committee categorized all pregnancy-associated deaths and did not limit our review to the CDC definition of pregnancy-related mortality, which excludes deaths from accidental or incidental causes.6
Cases of maternal death from self-harm were identified from all maternal deaths in Colorado over the study time period. Deaths from self-harm were identified from death certificate documentation of a suicide or accident with a secondary classification of overdose. Accidental overdoses included poisonings from illicit drugs and medications. If the manner of death was documented as unknown or natural on the death certificate, we relied on the determination of the Colorado Maternal Mortality Review Committee that the death was a result of overdose or suicide.
For identified maternal deaths from self-harm, three written contacts were made by Colorado Department of Public Health and Environment staff to obtain records from the coroner, primary prenatal care provider, and hospital of delivery, death, or both. Records requested included: birth certificate, death certificate, labor and delivery records, prenatal and postpartum care medical records (including social work and psychiatric records, when available), autopsy report, toxicology testing, coroner's report, and law enforcement report (if applicable).
All available self-harm death records were reviewed by a medical professional trained in detailed data abstraction using a data abstraction instrument developed by the investigators. Data abstraction items included sociodemographics (race, ethnicity, maternal age, marital status, type of residence, insurance status, weeks of gestation or number of months postpartum at time of death), psychiatric history (documented psychiatric diagnosis, current psychopharmacotherapy, discussion about continuation of pharmacotherapy during pregnancy, prior suicide attempts, and current or prior substance use), social history (involvement of the father of the child, living situation, employment status), and clinical history (preterm delivery, adverse obstetric events).
Drugs identified on toxicology testing at the time of death were categorized by drug class as opioids, alcohols, benzodiazepines, other sedatives or hypnotics, antidepressants, amphetamines, muscle relaxants, cannabinoids, acetaminophen, anesthetics, antipsychotics, barbiturates, or phencyclidine. Illegal drugs (cocaine, heroin, phencyclidine) were combined into an illicit drug category for further descriptive analysis. Of note, medicinal marijuana was legal in Colorado for a portion of the study period. Marijuana was not fully legalized for recreational use until after the study period. However, given the overlap with legalization of medicinal marijuana, cannabinoids were not included as illicit drugs. Prescription opioids (including methadone), benzodiazepines, sedatives and hypnotics, antidepressants, and muscle relaxants were grouped as pharmaceutical medications.
All data were abstracted into a secure web-based Research Electronic Data Capture application.10
Data from live births were extracted from state birth certificate data by Colorado Department of Public Health and Environment staff. Although maternal deaths and live births are not mutually exclusive, statewide data on all live births over the study time period were collected to provide context for our detailed data abstraction including the frequency and percent of women in each category of residence (urban, rural, frontier, unknown), race, marital status, and insurance status.
Descriptive data (frequencies, percentages) for maternal deaths from self-harm were generated. Death ratios were calculated over the number of live births per year. Categorical demographic characteristics were summarized for women with a maternal death from drug overdose or suicide and for all women with a live birth in Colorado over the same time period. All analyses were completed in SAS 9.4, and graphics were created in GraphPad Prism 6.0.
There were 211 maternal deaths in Colorado over the 9-year study period (January 1, 2004, through December 31, 2012), of which 63 (30%) were classified as maternal death by self-harm (accidental overdose or suicide). Self-harm was the leading cause of maternal death in Colorado (Fig. 1). From 2004 to 2012, the overall pregnancy-associated mortality ratio was 34.4 (95% confidence interval [CI] 29.9–39.3) per 100,000 live births; the mortality ratio from accidental overdose was 5.0 (95% CI 3.4–7.2) per 100,000 live births and from suicide was 4.6 (95% CI 3.0–6.6) per 100,000 live births.
Detailed medical records were obtained for 94% (n=59) of women who died of accidental overdoses (n=31) or suicide (n=28). The manner of death on the death certificate was accidental with a subclassification of drug overdose in 26 women. There were five additional women for whom the manner of death on the death certificate was unknown that were subsequently classified as accidental overdose by the Colorado Maternal Mortality Review Committee. In all of these women, there was evidence of overdose by toxicology testing.
The manner of death was documented on the death certificate as suicide in 26 women. Among the women with suicide documented as the manner of death, the most common means were further subclassified as asphyxia by hanging (n=10), penetrating trauma (gunshot wound or stab wound, n=8), and intentional overdose (n=5). There were two women in whom suicide was not documented on the death certificate but the coroner and police report indicated that the death was the result of suicide; the Colorado Maternal Mortality Review Committee classified these as suicides and they were included in the study. No cases of infanticide were documented in association with maternal death from self-harm. Sociodemographic characteristics of the maternal deaths resulting from accidental overdose and suicide are summarized in Table 1.
Toxicology results were available from 50 (84.7%) of the decedents: a single drug or medication was identified in 12 (24.0%), two drugs or medications were identified in 10 (20.0%), and three or more drugs or medications were identified in 22 (44.0%). No drugs or medications were identified in toxicology testing from six decedents (12.0%). Pharmaceutical medications were most commonly identified, with 27 (54.0%) of the decedents having a positive toxicology test result for a medication and 14 (28.0%) for illicit drugs, including 10 with cocaine or cocaine metabolites and four with heroin (6-monoacetyl-morphine) or a heroin metabolite (morphine with known recent heroin use).
Among the 50 women with toxicology testing available, the most commonly detected class of drugs was opioids. In 21 (42.0%) women, pharmaceutical opioids were detected. In three women, heroin or heroin metabolites were detected with no other opioids. In almost all cases of a pharmaceutical opioid overdose, there was documentation of a recent prescription for opioids, a known opioid use disorder, or empty opioid pill bottles found at the scene of the death. There was only one woman in which the coroner's report documented that opioids were “bought on the street”; in all other cases, the women were thought to have been prescribed opioids. Classes of drugs detected on toxicology screening are reported in Figure 2.
Deaths were equally distributed throughout the first postpartum year (mean 6.21±3.3 months postpartum) with only six maternal deaths during pregnancy (Fig. 3). The deaths during pregnancy occurred on average at 19.2±11.6 weeks of gestation with the earliest occurring at approximately 5 weeks of gestation and the latest occurring at 35 weeks of gestation.
Eighty-one percent (n=48) of the 59 included women with maternal deaths from self-harm had prenatal records available for review by the study team. Three of the 59 women did not have prenatal care as documented at the time of delivery. The remaining eight women in the case study may have had prenatal care, but we were unable to obtain antenatal records from the primary obstetrician despite three attempts. Among the women with prenatal records available, the mean gestational age at initiation of prenatal care was 13.2±6.5 weeks of gestation. Women with prenatal care had a mean of 7.9 (95% CI 6.6–9.4) prenatal visits. Of the 46 women who had prenatal care and the opportunity to attend a postpartum visit (ie, did not die while pregnant), 20 (43.5%) attended a postpartum visit.
Among the 59 maternal deaths from self-harm, 16.9% (n=10) had a substance use disorder documented in the prenatal medical record or during hospitalization for delivery or death, of whom six were either currently in substance use disorder treatment or had a documented history of substance use disorder treatment. In 10 (16.9%) women, there was no documentation of screening for substance use disorder in the prenatal medical record. A urine toxicology screen was performed at a prenatal visit for only eight (13.6%) of the women.
Prior psychiatric diagnoses (Table 1) were documented in the clinical records of 54.2% (n=32) and prior suicide attempts in 10.2% (n=6). Of the six women with documentation of prior suicide attempts, five completed suicide and one died of an accidental overdose. Thirteen women (22.0%) did not have a known psychiatric or substance use disorder.
Depression was the most common documented psychiatric diagnosis (Table 1). Although approximately half (n=27) of the self-harm cases were documented to be taking psychopharmacotherapy at conception, 48% of them discontinued the medications as a result of the pregnancy. Self-discontinuation of the medication(s) occurred in nine women, and discontinuation at the recommendation of a physician occurred in four women. In the majority of women, psychopharmacotherapy consisted of selective serotonin reuptake inhibitors. Other notable prescribed therapies included: sleep aids, mood stabilizers, and other classes of antidepressants. Selective serotonin reuptake inhibitors were the drug class most commonly discontinued during pregnancy; 10 of the 13 women who discontinued medications stopped taking a selective serotonin reuptake inhibitor. Four of the 13 women who were documented to have stopped the medications during pregnancy had psychiatric medications on toxicology testing at the time of autopsy.
Social stressors were commonly documented in the medical records of women who died of self-harm including unemployment (n=38 [64.4%]); being single, divorced, or separated (n=24 [40.7%]); history of domestic violence (n=11 [18.6%]); unstable living situations such as homelessness (n=3 [5.1%]); and current domestic violence (n=3 [5.1%]). Based on the medical record or coroner report, the father of the child was noted to be involved in the majority (n=41 [69.5%]) of cases. Despite documentation of social stress in the majority of women, engagement with a social worker was only found in 21 (35.6%) cases either during the course of prenatal care or at the time of delivery.
Adverse obstetric events (preterm delivery, preeclampsia, fetal growth restriction, third- or fourth-degree laceration, stillbirth, neonatal intensive care unit admission, or maternal intensive care unit admission) occurred in 16 (27.1%) women. There was a preterm (less than 37 weeks of gestation) delivery rate of 10.2% (n=6) among the women; all of these neonates were admitted to the neonatal intensive care unit. The demographic characteristics of women with a maternal death are summarized along with the characteristics of all women with live births in Colorado over the same time period in Table 2.
Self-harm (accidental overdose and suicide) was the leading cause of maternal death in Colorado from 2004 to 2012 with a mortality ratio of 9.6/100,000 live births. Almost 90% of the deaths from self-harm occurred in the postpartum period. In the majority of women, substance use and psychiatric disorders were present. However, in 22% of women with maternal deaths, neither of these risk factors was present. Depression was the most common psychiatric diagnosis. Almost half of women on psychiatric medications in early pregnancy discontinued them during pregnancy or postpartum either on their own or at the recommendation of a care provider. Pharmaceutical opioids were the most common substance identified on toxicology screening at the time of autopsy.
Colorado's maternal death ratio from suicide (4.6/100,000 live births) is higher than that reported based on national pregnancy mortality surveillance (1.6–4.5 suicides/100,000 live births)11 and data from the National Violent Death Reporting System (2.0 suicides/100,000 live births).12 Colorado's higher suicide ratio may be the result of the comprehensive case identification approach used by this study. Alternatively, our rates of suicide in pregnancy and postpartum may reflect the high rates of suicide in the general Colorado population. Suicide was the most common cause of death in Coloradans aged 10–44 years with a rate of 19.7/100,000 population in 2012.13
The lack of ongoing treatment for psychiatric disorders observed in our case study is consistent with a recently published study from the United Kingdom demonstrating that perinatal women were more likely to carry a diagnosis of depression, yet were half as likely to receive any active treatment than nonperinatal women.14 These findings contrast with the accumulating body of evidence supporting the safety of antidepressants such as selective serotonin reuptake inhibitors in pregnancy.15,16
Maternal deaths were infrequent during pregnancy, consistent with existing literature.17,18 Less than 50% of women with maternal deaths from self-harm attended a postpartum visit; thus, targeting postpartum visits alone for depression screening and management will be inadequate to reach women at risk. Rather, each point of contact with women at risk for self-harm should be considered an opportunity for intervention including preconception visits, antenatal care, hospitalization for delivery, and postpartum visits for both the mother and the neonate. During pregnancy and postpartum, women at risk for self-harm may encounter various types of health care providers including social workers, nurses, and physicians (psychiatrists, addiction specialists, obstetricians, adult primary care providers, pediatricians, and emergency medicine physicians). Resources for health care providers for screening and referring women at risk for self-harm are readily available from multiple organizations such as the National Institute of Child Health and Human Development (https://www.nichd.nih.gov/ncmhep/MMHM/Pages/index.aspx), the Substance Abuse and Mental Health Services Administration (http://store.samhsa.gov/shin/content/SMA14-4878/SMA14-4878.pdf), and the Council on Patient Safety in Women's Health (http://www.safehealthcareforeverywoman.org).
Successful interventions for prevention of maternal self-harm will likely require a multitiered approach with interventions based at the individual patient and health care provider level as well as broader changes at the health care delivery system and community level. We provide examples of potential interventions at each of these four levels in Table 3. Our group has integrated perinatal mental health screening, assessment, and treatment into prenatal clinics, similar to a recently published study by Avalos et al.19 Universal screening will be necessary to identify the 22% of women in our series who did not have any identifiable risk factors for self-harm. To this end, we hope to establish virtual and electronic access to mental health resources for physicians in rural communities without local psychiatric services so that health care providers can screen women for mental health disorders and then promptly refer for further assessment and treatment. We also now include substance use and psychiatric experts on our Maternal Mortality Review Committee.
Strengths of this study include 9 years of data from multiple sources that were reviewed by the Colorado Maternal Mortality Review Committee. Detailed chart abstraction allowed for ascertainment of data that are not available by collecting diagnostic codes alone. Limitations of this study include a possible underestimation of psychiatric diagnoses, treatments, and medications as well as substance use treatment, because psychiatric records may not always have been available through the hospital or prenatal care providers. Furthermore, we could not make comparisons to other women with live births in Colorado over the same time period for prevalence of risk factors given the limited demographic data available on birth certificates.
Our findings suggest that initiatives to increase community awareness, screening, and referral to treatment for mental health and substance use disorders in pregnancy along with recognition of the need for ongoing care beyond the early postpartum period will be critical in reducing pregnancy-associated deaths from self-harm.
1. Main EK, Menard MK. Maternal mortality: time for national action. Obstet Gynecol 2013;122:735–6.
2. D'Alton ME, Main EK, Menard MK, Levy BS. The National Partnership for Maternal Safety. Obstet Gynecol 2014;123:973–7.
4. Kozhimannil KB, Adams AS, Soumerai SB, Busch AB, Huskamp HA. New Jersey's efforts to improve postpartum depression care did not change treatment patterns for women on Medicaid. Health Aff (Millwood) 2011;30:293–301.
5. Flanagan T, Avalos LA. Perinatal obstetric office depression screening and treatment: implementation in a health care system. Obstet Gynecol 2016;127:911–5.
7. Rockett IR, Smith GS, Caine ED, Kapusta ND, Hanzlick RL, Larkin GL, et al.. Confronting death from drug self-intoxication (DDSI): prevention through a better definition. Am J Public Health 2014;104:e49–55.
8. Stanistreet D, Taylor S, Jeffrey V, Gabbay M. Accident or suicide? Predictors of coroners' decisions in suicide and accident verdicts. Med Sci Law 2001;41:111–5.
9. Berg CDI, Atrash H, Zane S, Bartlett L. Strategies to reduce pregnancy-related deaths: from identification and review to action. Atlanta (GA): Centers for Disease Control and Prevention; 2001.
10. Harris PA, Taylor R, Thielke R, Payne J, Gonzalez N, Conde JG. Research electronic data capture (REDCap)—a metadata-driven methodology and workflow process for providing translational research informatics support. J Biomed Inform 2009;42:377–81.
11. Wallace ME, Hoyert D, Williams S, Mendola P. Pregnancy-associated homicide and suicide in 37 US states with enhanced pregnancy surveillance. Am J Obstet Gynecol 2016 Mar 26 [Epub ahead of print].
12. Palladino CL, Singh V, Campbell J, Flynn H, Gold KJ. Homicide and suicide during the perinatal period: findings from the National Violent Death Reporting System. Obstet Gynecol 2011;118:1056–63.
14. Khalifeh H, Hunt IM, Appleby L, Howard LM. Suicide in perinatal and non-perinatal women in contact with psychiatric services: 15 year findings from a UK national inquiry. Lancet Psychiatry 2016;3:233–42.
15. Huybrechts KF, Hernández-Díaz S, Avorn J. Antidepressant use in pregnancy and the risk of cardiac defects. N Engl J Med 2014;371:1168–9.
16. Huybrechts KF, Bateman BT, Palmsten K, Desai RJ, Patorno E, Gopalakrishnan C, et al.. Antidepressant use late in pregnancy and risk of persistent pulmonary hypertension of the newborn. JAMA 2015;313:2142–51.
17. Marzuk PM, Tardiff K, Leon AC, Hirsch CS, Portera L, Hartwell N, et al.. Lower risk of suicide during pregnancy. Am J Psychiatry 1997;154:122–3.
18. Appleby L. Suicide during pregnancy and in the first postnatal year. BMJ 1991;302:137–40.
19. Avalos LA, Raine-Bennett T, Chen H, Adams AS, Flanagan T. Improved perinatal depression screening, treatment, and outcomes with a universal obstetric program. Obstet Gynecol 2016;127:917–25.