Maternal mortality in the United States: predictability and the impact of protocols on fatal postcesarean pulmonary embolism and hypertension-related intracranial hemorrhage




Objective


The purpose of this study was to examine the efficacy of specific protocols that have been developed in response to a previous analysis of maternal deaths in a large hospital system. We also analyzed the theoretic impact of an ideal system of maternal triage and transport on maternal deaths and the relative performance of cause of death determination from chart review compared with a review of discharge coding data.


Study Design


We conducted a retrospective evaluation of maternal deaths from 2007-2012 after the introduction of disease-specific protocols that were based on 2000-2006 data.


Results


Our maternal mortality rate was 6.4 of 100,000 births in just >1.2 million deliveries. A policy of universal use of pneumatic compression devices for all women who underwent cesarean delivery resulted in a decrease in postoperative pulmonary embolism deaths from 7 of 458,097 cesarean births to 1 of 465,880 births ( P = .038). A policy that involved automatic and rapid antihypertensive therapy for defined blood pressure thresholds eliminated deaths from in-hospital intracranial hemorrhage and reduced overall deaths from preeclampsia from 15-3 ( P = .02.) From 1-3 deaths were related causally to cesarean delivery. Only 7% of deaths were potentially preventable with an ideal system of admission triage and transport. Cause of death analysis with the use of discharge coding data was correct in 52% of cases.


Conclusion


Disease-specific protocols are beneficial in the reduction of maternal death because of hypertensive disease and postoperative pulmonary embolism. From 2-6 women die annually in the United States because of cesarean delivery itself. A reduction in deaths from postpartum hemorrhage should be the priority for maternal death prevention efforts in coming years in the United States.





See related editorial, page 1



Maternal mortality rates in the United States have remained unchanged for several decades, with some data suggesting a recent increase in such deaths. In 2008, we published a review of maternal deaths in approximately 1.5 million births from 2000-2006 and drew a number of conclusions regarding the potential value of several specific steps to address this issue effectively in our system and in the nation as a whole. These steps included protocols that were directed at prevention of death from postcesarean pulmonary embolism, intracranial hemorrhage in women with hypertensive crisis, and postpartum hemorrhage. We present a 6-year observation, detailing clinical results of these efforts. In the current study, we also sought to examine the issue of predictability of maternal death and the potential impact of an ideal system of risk identification and transport on maternal mortality rates. Finally, we examined the correlation between diagnostic accuracy regarding the cause of death that was obtained from coding data, and the data that were obtained from medical records review.


Because maternal death commonly is defined as deaths per 100,000 live births and may include deaths that are associated with early pregnancy loss or stillbirth, the use of the term maternal mortality ratio to describe these deaths statistically would be correct. However, because the term maternal mortality rate is used almost exclusively throughout the literature to describe maternal deaths during pregnancy, we have observed this convention throughout this article.


Methods


The Hospital Corporation of America encompasses 110 maternal/newborn facilities in 21 states. Our annual delivery volume is approximately 210,000 or roughly 5-6% of all births in the United States. Our present study consisted of 3 parts.


Part 1


After a review of maternal deaths that occurred in our system from 2000-2006, we developed 3 specific patient safety programs that were aimed at reduction in maternal deaths. (1) All affiliated hospitals instituted the universal use of intra- and postoperative pneumatic compression devices in women who undergo cesarean delivery. (2) We introduced specific checklist-based protocols that were directed at prompt recognition and treatment of hypertensive crisis, with either labetalol or hydralazine, and emphasized the importance of aggressive recognition and management of preeclampsia-related pulmonary edema ( Figures 1 and 2 ). (3) We developed and introduced a checklist-based protocol that is directed at summoning assistance and timely fluid, blood, and component replacement in cases of postpartum hemorrhage ( Figure 3 ).




Figure 1


Blood pressure management of severe intrapartum or postpartum hypertension with hydralazine

Checklist.

IV, intravenously; q, every.

Clark. Maternal mortality in the United States. Am J Obstet Gynecol 2014 .



Figure 2


Blood pressure management of severe intrapartum or postpartum hypertension with labetalol

Checklist.

IV, intravenously; q, every.

Clark. Maternal mortality in the United States. Am J Obstet Gynecol 2014 .





Figure 3


Recommended protocol for patients with postpartum hemorrhage

Checklist.

CBC , complete blood cell count; EBL , estimated blood loss; NS , normal saline solution; OB , obstetrician; PTT , partial thromboplastin time; RN , registered nurse; SpO2 , oxygen saturation; STAT , immediately; T&C , type and crossmatch.

Clark. Maternal mortality in the United States. Am J Obstet Gynecol 2014 .


We then conducted a similar review of deaths that occurred between Jan.1, 2007, and Dec. 31, 2012, after the introduction of these protocols. Maternal deaths were identified initially from discharge coding data for the years of interest. Pertinent medical records that pertain to the admission ending in death were then examined for extraction of data used in this analysis. For purposes of determining a causal relationship between cesarean delivery and maternal death, we determined whether delivery of the patient whose death was associated with cesarean delivery would likely have been avoided had the patient been delivered vaginally instead of by cesarean section.


Part 2


Causes of death from 2000-2012 were determined from medical record review without regard to coded discharge diagnoses. Then, deidentified discharge diagnostic and procedure codes that were associated with women who died during this period were provided to an author (J.T.C.) who is a maternal-fetal medicine specialist and who, for more than a decade, has served on the Coding Committees of the Society for Maternal Fetal Medicine and the American College of Obstetricians and Gynecologists. This investigator assigned a cause of death that was based exclusively on available discharge diagnostic and procedure codes and was blinded to any medical records. The assignments of cause of death from these 2 different types of reviews were then compared.


Part 3


Finally, we examined both cause of death and status on arrival to our facility for each patient in the context of a hypothetical ideal medical system with the following features:




  • There is a specialized high-risk maternal center with full-time maternal-fetal medicine, obstetric anesthesiology, and a medical intensivist available within 30 minutes of the facility of presentation.



  • A preexisting transport agreement exists between the facility of admission and such a center.



  • Transport services are available for all patients without delay.



  • All patients consent to any recommendation for transport.



  • There are no insurance or payment barriers to transport to any area facility.



We then assigned each patient to one of the following categories, assuming the availability of the system described earlier: (1) The patient had no identifiable risk factors for death when she was admitted that would have prompted transport to such a specialized maternal care center. (2) The patient did have identifiable risk factors that could justify transport to a specialized maternal care center. (3) The patient had known end-stage disease for end-of-life care or already had experienced cardiac arrest before evaluation.


From these data, we sought to determine the maximum possible effect of an ideal system of maternal transport and availability of specialized referral centers on maternal mortality rates.


Statistical analysis


Statistical analysis was performed with the 2-tailed Fisher exact test with a probability value of < .05 cutoff for statistical significance.


Because this project involved examination of deidentified data for quality improvement purposes, it was exempt from institutional review board approval based on 45CFR46.101(b) and 46.102(f) and 45CFR164.514(a)-(c) of the Health Insurance Portability and Accountability Act.




Results


Between Jan. 1, 2007, and Dec. 31, 2012, there were 81 maternal deaths in 1,256,020 deliveries for a rate of 6.4 per 100,000 births. The mean maternal age at time of death was 30.5 years (range, 17–44 years.) Gestational age distribution at the time of death is outlined in Table 1 . For women who delivered in our affiliated institutions, 52% had private insurance; 44% had Medicaid, and 4% were uninsured during the time period of this study.



Table 1

Gestational age at the time of maternal death






















Gestational age, wk Deaths
0-12 0
13-24 4
25-36 30
37-41 57
>41 0

Clark. Maternal mortality in the United States. Am J Obstet Gynecol 2014 .


Causes of death, as determined from medical records review and a comparison of causes of death from 2007-2012 with those from 2000-2006 are detailed in Table 2 . There was a significant decline in the rate of fatal postcesarean pulmonary embolism and a significant decline in the rate of deaths from preeclampsia during the study period, as compared with the earlier control period. During the study period, no hypertensive women died because of untreated in-hospital pulmonary edema or hypertensive crisis, in contrast to our preprotocol experience. Two women with severe preeclampsia came to our facilities with intracranial hemorrhage that had occurred before presentation to the hospital. Rates of death from other causes did not change.



Table 2

Cause of death















































































Category of death 2000-2006 (n = 1,461,270) 2007-2012 (n = 1,256,020) P value
Hemorrhage 11 19 NS .07
Amniotic fluid embolism 13 11 NS 1.0
Nonobstetric ID 7 10 NS .81
Other 11 8 NS .72
Postcesarean pulmonary embolism 7 1 .038
Other pulmonary embolism 2 7 NS .09
Cardiovascular 10 8 NS .48
End-stage medical disease 1 5 NS .10
Obstetric ID 7 3 NS .36
Trauma/overdose 6 3 NS .52
Hypertension 15 3 .02
Asthma 0 2 NS .21
Medication error/reaction 5 1 NS .23
Total 95 81 NS 1.0

ID , infectious disease; NS, not significant.

Clark. Maternal mortality in the United States. Am J Obstet Gynecol 2014 .


In the entire 12-year cohort of 2,717,290 women, there were 176 maternal deaths. Ninety-nine patients (56%) had no significant risk factors for the ultimate cause of death at the initial examination; 32 women (18%) had known, terminal, or end-stage disease or cardiac arrest. Forty-five patients (26%) had known risk factors on admission for their ultimate cause of death ( Table 3 ). However, 76% of these women were admitted to a specialized center that already met the specialist criteria outlined in the “Methods” section. Thus, only 12 of 176 maternal deaths (7%) would have been preventable potentially with an ideal system of risk identification and transport to universally available specialized centers, even assuming all women with common conditions that included preeclampsia or asthma were transported to specialized centers.


May 10, 2017 | Posted by in GYNECOLOGY | Comments Off on Maternal mortality in the United States: predictability and the impact of protocols on fatal postcesarean pulmonary embolism and hypertension-related intracranial hemorrhage

Full access? Get Clinical Tree

Get Clinical Tree app for offline access