Special Report on Maternal Mortality and Severe Morbidity in Canada Enhanced Surveillance: The Path to Prevention

Discussion

Current Provincial/Territorial Maternal Death Review Activities

Investigation and reporting of maternal death in Canada is highly variable across the provinces and territories. The diversity of existing systems is striking -- from comprehensive and highly detailed investigations to non-existent review mechanisms. Only four jurisdictions have province/territory-wide maternal death review committees with a mandate to investigate all maternal deaths. These committees work closely with coroners/medical examiners, vital statistics offices and hospitals to ensure complete ascertainment of maternal deaths. Once cases are ascertained, each committee conducts thorough reviews and determines the preventability of deaths, as well as likely sources of error. These committees represent an ideal model for provincial/territorial maternal death review activities.

In provinces and territories without active maternal death review committees, the coroners/medical examiners are likely notified of most, if not all, maternal deaths occurring in their jurisdiction. For most of these notified cases, a full coroner/medical examiner investigation will occur. Despite these efforts, in most provinces and territories, the results of coroner/medical examiner maternal death investigations are not collectively analyzed or categorized together as maternal deaths. Furthermore, the provinces and territories vary in defining which pregnancy-related deaths will be investigated, as well as the specific details collected for each maternal death. This variability is present even where review committees exist.

In many instances, the case details provided for this national review lacked the depth to completely assess the circumstances surrounding death. The data collection form, which was constructed from several existing maternal death review data collection tools, proved too detailed for this retrospective data collection effprt. In addition to pertinent details regarding medical management, specific information on low socioeconomic groups, Aboriginal women, recent immigrants to Canada and other vulnerable populations was not routinely available. Collection of these data, which may shed light on potential disparities in maternal death and severe illness in Canada, should be considered a priority for future reports. Finally, very limited information is disseminated regarding maternal death review activities in provinces and territories, and information is not systematically shared across jurisdictions.

Of note, efforts to highlight current provincial/territorial maternal death review activities for this report have already resulted in important steps to enhance maternal death review in Canada. First, in response to difficulties we encountered in obtaining case details of maternal deaths in Quebec, the Collège des Médecins du Québec has committed to establishing a process for maternal death reviews in Quebec and to collaborate on future Canadian maternal death review projects. Second, in response to our request for case details, one provincial death review committee identified potential gaps in its case notification mechanism. Subsequent improvement in this province-specific reporting procedure will ensure optimal ascertainment of maternal deaths in the future. Finally, requests for case details to one provincial chief coroner identified cases that were not reported to the coroner as required in the legislation; consequently, information sessions on reporting requirements were conducted for the area hospitals. These changes highlight the evolutionary process of maternal death surveillance in Canada. It is anticipated that each subsequent national report will result in further enhancements to the maternal death surveillance system at all levels, in addition to providing a valuable overview of maternal deaths in Canada.

Maternal Mortality in Canada

The total MMR of 6.1 per 100,000 live births (based on this review) is consistent with international estimates of the Canadian MMR, which accounted for probable under-reporting in the Canadian Vital Statistics System.² It is also consistent with estimates obtained from a previous record-linkage study.⁴ As highlighted in the introduction of this report, Canada's MMR is lower than MMR estimates from most other developed countries,² and also lower than MMRs from countries based on similar individual case reviews.⁶ The reason for Canada's relatively low MMR is not entirely clear. Universal health care coverage has been proposed as one possible explanation for Canada's enviable ranking. ⁴ A population-based study conducted in North Carolina in the United States reported an adjusted odds ratio of 0.2 (95% CI 0.1-0.6) for pregnancy-related death associated with the receipt of prenatal care.¹⁹

The direct and indirect MMRs of 4.2 and 1.9 per 100,000 live births, respectively, were also within the range of previous Canadian estimates.⁴ The ratio of direct to indirect maternal deaths in Canada during 1997-2000 of 2.2:1 is very similar to the 2.3:1 ratio observed in Australia during 1994-1996.⁸ In contrast, the ratio of direct to indirect maternal deaths in the U.K. during 1997-1999 was 0.78, reflecting a greater inclination to classify non-direct deaths as indirect rather than incidental (see incidental deaths below).⁶Furthermore, changes in maternal death classification guidelines between ICD-9 and ICD-10 clearly influence the distribution of deaths between the direct and indirect categories.⁴ For example, if future reports classify intracranial hemorrhage (ICH) as an indirect maternal death (in accordance with ICD-10), a marked decrease in the direct MMR and a concomitant increase in the indirect MMR is likely to be observed.

Maternal age was associated with the MMR, with a higher MMR observed with increasing maternal age after age 20. This maternal age effect has been consistently reported elsewhere.^5,6,8 As the proportion of women who are delaying childbearing to later in life has increased markedly in Canada in recent years,¹³ an age-related increase in the Canadian MMR may also be expected.

The five leading causes of direct maternal deaths were pulmonary embolism, pre-eclampsia/ pregnancy-induced hypertension, amniotic fluid embolism (AFE), ICH and ectopic pregnancy. With the exception of ICH, these were also the most frequent causes of direct maternal death in the most recent report on maternal deaths in Australia.⁸ In the most recent U.K. data, thrombosis and thromboembolism, hypertensive disease of pregnancy, genital tract sepsis, ectopic pregnancy and AFE were the five leading causes of direct maternal deaths.⁶ As highlighted above, classification of ICH deaths will likely shift in future reports, and ICH will no longer remain a leading cause of direct maternal deaths.

Prevention of pulmonary embolism, the leading cause of direct maternal death in most reviews, has received particular attention. In the U.K., a substantial decline in the number of deaths due to pulmonary embolism has been attributed mainly to a marked decline in deaths after cesarean section following the 1995 publication of thromboprophylaxis guidelines.⁶ The most recent Canadian guidelines on the prevention and treatment of venous thromboembolism (VTE) in obstetrics were published by the Society of Obstetricians and Gynaecologists of Canada (SOGC) in 2000.²⁰ The case details that were available to us did not allow a complete review of the presence of risk factors and the compliance with preventive guidelines for all pulmonary embolism cases. However, suggestions of inadequate thromboprophylaxis were observed in at least two maternal deaths attributed to pulmonary embolism. Ongoing efforts should ensure that all health care workers are aware of, and in compliance with, SOGC clinical practice guidelines on the prevention and treatment of VTE in obstetrics. Adoption of more aggressive thromboprophlyaxis guidelines may be an additional consideration.

As stated, pre-eclampsia/pregnancy-induced hypertension was a leading cause of direct maternal death in Canada during 1997-2000. This has been consistently reported in other countries.^5,6,8 The majority of deaths attributed to pre-eclampsia/pregnancy-induced hypertension in this series involved ICH. The predominance of ICH among deaths due to hypertensive disease has also been reported elsewhere and attributed to a failure of effective antihypertensive management.⁶ The available case details did not allow a complete assessment of the role of medical management in these deaths. Nonetheless, this review highlights that pre-eclampsia/pregnancy-induced hypertension remains a potentially lethal disease that requires a collaborative, intensive approach to treatment. Recent advances in the clinical management of pre-eclampsia/pregnancy-induced hypertension, for example, magnesium sulphate treatment, may result in a reduction in the number of maternal deaths attributed to this condition.²¹

Amniotic fluid embolism was a major cause of direct maternal death in Canada. Again, this finding has been reported elsewhere.^6,8 Several possible risk factors for AFE have been previously identified, including advanced maternal age and use of oxytocic drugs.⁶ In this series, none of the seven women whose death was attributed to AFE was older than 34 years of age; three of seven cases involved induction of labour with oxytocic drugs. Despite the identification of risk factors, the ability to prevent AFE remains elusive.⁶

Of particular concern is the number of women who died due to an ectopic pregnancy despite seeking medical care for their symptoms. In one instance, further investigations (i.e., ultrasound) were not performed despite abdominal cramping and a positive pregnancy test. In a second case, symptoms were primarily gastrointestinal in nature. This concerning pattern of death from a ruptured ectopic pregnancy following presentation with gastrointestinal or urinary tract symptoms has been described elsewhere.⁶ Health care provider education and training should reinforce the need for a high index of suspicion of ectopic pregnancy, particularly with atypical presentations.

The absence of a single maternal death attributed to primary post-partum hemorrhage is encouraging. Once a leading cause of maternal death in Canada and still a leading cause of maternal death worldwide,²² this finding is emblematic of the progress in reducing maternal mortality in many developed countries, including Canada. A similar reduction in maternal deaths attributed to primary post-partum hemorrhage has been reported elsewhere.^6,8 Nonetheless, as discussed in this report, hemorrhage remains an important factor in many maternal deaths ultimately attributed to other causes. Furthermore, obstetric hemorrhage remains a major cause of severe maternal morbidity, even where its importance as a cause of maternal mortality has been reduced.¹¹

Three direct maternal deaths resulted from surgical or anaesthetic misadventure - a lacerated uterine artery during assisted vaginal delivery, a perforated abdominal aorta during laparoscopic management of an ectopic pregnancy and an inadvertent extubation postoperatively. While Canada's low MMR speaks to the safety of childbirth for women in Canada, these deaths are a sobering reminder of the risks associated with surgical intervention. As the rate of cesarean deliveries has been increasing steadily in Canada in recent years,¹³ an increase in the number of maternal deaths due to surgical and anaesthetic complications may be expected. Sound education, rigorous training requirements and constant vigilance are required to avoid such tragic cases.

Cardiovascular disease was the leading cause of indirect maternal deaths, as well as the single leading cause of maternal death overall. This prominence of cardiovascular disease deaths among all maternal deaths has also been reported in the U.K. and Australia.^6,8

As noted, among indirect maternal deaths attributed to cardiovascular disease during 1997-2000, coronary artery disease was the co-leading cause. Two of the four women who died of coronary artery disease had a medical history of diabetes. A high mortality rate among pregnant diabetics with ischemic heart disease has been previously reported.²³ As maternal age continues to increase and risk factors for coronary artery disease, such as smoking and diabetes, remain prevalent, maternal deaths due to coronary artery disease may increase.⁶ Health care providers must be aware of the important role of cardiovascular disease and, increasingly, coronary artery disease in maternal deaths.

Ascertainment of incidental deaths, and deaths due to injuries in particular, was likely not complete in this study. Nevertheless, careful review of the case reports revealed the prominent role played by injuries and, in particular, motor vehicle crashes as the leading cause of incidental deaths. In two of these deaths, the woman was not wearing a seat belt. The correct use of seat belts is one clear measure to reduce morbidity and mortality among pregnant women. In one study, proper seat belt use and crash severity were the two best predictors of maternal-fetal outcomes in motor vehicle crashes.²⁴ A survey from the U.K. published in 2000 reported that only 48% of women identified the correct way to wear a seat belt in pregnancy.²⁵ Another study reported that while 86% of women used restraints while pregnant, almost half used them incorrectly.²⁶ Transport Canada's road safety material regarding seat belt use in pregnancy states:

"Pregnant women should always wear the lap and shoulder seat belt. The lap belt should be snug and low over the pelvic bones and not against the soft stomach area.The shoulder belt should be worn across the chest. Worn properly, the seat belt will not harm the baby."²⁷

Ongoing efforts in Canada must ensure that all pregnant women are aware of the importance and correct usage of seat belts in pregnancy.

As noted earlier, many deaths classified as incidental in this review might also be reasonably classified as indirect maternal deaths. For example, in U.K. enquiries and reports, deaths due to epilepsy and suicide (in the absence of a long-standing history of previous mental illness) are classified as indirect.⁶ In this review, these deaths were all classified as incidental. Clearly, these divergent approaches to the classification of death will affect both the overall and the indirect MMR. Future reports should revisit the ideal classification of various causes of death.

Severe Maternal Morbidity in Canada
Severe maternal morbidity analyses represent an important supplement to maternal mortality reviews. Mantel el al. proposed a framework to define severe maternal morbidity or "near miss."¹¹ They suggest that near miss and maternal death are events on a continuum from normal healthy pregnancy, to organ dysfunction, to organ failure, to death. The conditions and procedures analyzed in this study include many of the conditions suggested by Mantel et al. The overall rate of severe maternal morbidity in this analysis was 4.62 per 1,000 deliveries. The availability of comparative rates from population-based studies is limited. Two Nova Scotia studies reported rates of eclampsia and uterine rupture (including both complete rupture and dehiscence) of 0.27 and 0.34 per 1,000 deliveries, respectively.^28,29 These rates are lower than those observed in this analysis (0.38 per 1,000 deliveries for eclampsia and 0.74 per 1,000 deliveries for uterine rupture).

The reason for the observed variation in rates is not clear. The Nova Scotia studies were based on a database that is more comprehensive with greater clinical detail than CIHI's DAD -- the database used in this study. This additional detail may eliminate false positive diagnoses of specific conditions, resulting in lower rates. Regional variation in outcomes secondary to variation in clinical practice is another possible explanation that should be explored in future studies.

The reasons for the observed temporal trends in severe maternal morbidity are also complex. Some trends reflect changes in demographic and clinical characteristics of pregnant women in Canada. In the past decade, Canada has witnessed a substantial increase in advanced maternal age and multi-fetal pregnancy,¹³ both factors that increase the risk of maternal morbidity.^13,30,31 In addition, changes in clinical practice can result in corresponding changes in specific measures of maternal morbidity. This is particularly true for procedures that are directly related to differences in clinical management, such as hysterectomy and transfusion. In this analysis, the rate of post-partum hemorrhage requiring hysterectomy increased, while the rate of post-partum hemorrhage requiring transfusion decreased in recent years.

Trends in other conditions, such as uterine rupture, may also be related to changes in clinical practice. The increase in the rate of uterine rupture mirrors quite closely the increase in trial of labour and attempted vaginal birth after previous cesarean delivery in Canada.¹⁶ Finally, improvements in diagnosis can lead to disease reclassification, which can artificially increase the rates of some conditions, while reducing the rates for others.³² In-depth analysis of each specific condition is needed to clearly elucidate the etiology behind the observed trends. Consideration of indicators that combine severe maternal morbidity and maternal death, for example, the ratio of deaths to near misses, may provide additional useful information. Future efforts to develop comprehensive and consistent approaches to the analysis of severe maternal morbidity will ensure its central role in the surveillance of maternal health in Canada and elsewhere.

trengths and Limitations
This study of maternal death and severe illness in Canada has several strengths. It is the first national report on maternal death to go beyond vital statistics and hospitalization databases and death certificate reviews to actually examine individual case reports and the medical circumstances surrounding death. It is also the first report to describe the nation-wide occurrence of overall severe maternal morbidity, as well as trends over time.

The study was able to ascertain and collect case details for most maternal deaths. The use of both the vital statistics and hospitalization databases to construct the list of database deaths likely captured most direct and indirect maternal deaths. Subsequently, collaboration with coroners/medical examiners, as well as existing provincial and territorial death review committees, resulted in coroner/medical examiner reports or completed data collection forms for most of these maternal deaths. Reports or data collection forms were obtained for almost 80% of the cases on the database death list.

Review of the remaining cases on the list for which reports could not be obtained suggested that several of these were not maternal deaths. Of the 64 direct or indirect maternal deaths ultimately included in the case review, all but three had been ascertained through the original database search. The MMRs based on this case review were consistent with previous estimates of the Canadian MMR, which had accounted for anticipated underascertainment of maternal deaths in the Canadian Vital Statistics System. Also, the distribution of maternal death by direct vs. indirect and specific causes was fairly consistent with reports from other countries.^6,8 This further supports the relatively complete ascertainment of maternal deaths in this report for the population under study.

This study also suffers from several limitations. Quebec was not included in the maternal death review because of the limited availability of case details from that province. Manitoba, Quebec and Nova Scotia were all excluded from the analysis of severe maternal morbidity because of the limitations of the database used for that period of study.

Some maternal deaths were not included in the data presented. First, three cases from the list of database deaths were confirmed with provincial and territorial contacts, but additional case details were not available. In two instances, despite notification to the coroner, the deaths were classified as "non-coroner cases" and no formal investigation was conducted. In the third case, the coroner was able to confirm that the case existed, but the hospital had not complied with reporting procedures, and case details were not available in time for this report. (These three cases were not included in the data presented.) Second, previous reports have concluded that cerebrovascular disorders may be particularly prone to underreporting in the vital statistics system.⁴ In addition, the hospitalization database search may not have captured all cases of maternal death due to cerebrovascular disorders, particularly if the event occurred in the post-partum period. As a result, these deaths may not have been completely ascertained.

As well, the new category of "late maternal death" was likely not complete in this analysis. Only one late maternal death was ascertained during 1997-2000. This finding is not unexpected, as the ICD-10 coding system with a definition for late maternal death was implemented for year 2000 deaths only. Furthermore, the definition of maternal death used by most of the provincial/territorial death review committees does not include deaths beyond 90 days. Future reports may capture an increasing number of late maternal deaths.

As highlighted, incidental deaths were likely not well ascertained in this study. Only eight of thirty incidental deaths reviewed in this analysis were detected in the database review. The additional 22 reports of incidental deaths were received from just two of the ten provinces and three territories. It is likely that other jurisdictions had incidental deaths that were either not reported to the coroner/medical examiner or provincial/territorial review committee, or were not easily retrievable as incidental deaths by those bodies. As neither late maternal deaths nor incidental deaths are included in the MMR calculations, the poor ascertainment of these cases did not affect the reported MMRs.

As discussed above, another important limitation of this report is the variability in the detail and quality of the reports/data collection forms obtained. In many instances, the case details provided lacked the depth to completely assess the circumstances surrounding death.

It is important to note that the scope of this study did not include the broader issue of reproduction-related deaths, such as those due to sexually transmitted diseases or contraception.

Finally, the analysis of severe maternal morbidity was based solely on an administrative database. Such databases often lack pertinent clinical details and are prone to a certain degree of coding errors.³³ Conditions that are difficult to accurately and consistently diagnose, such as amniotic fluid embolism,³⁴ may be particularly prone to misclassification using administrative databases.

Fortunately, maternal death is rare in Canada. Nevertheless, review of the circumstances surrounding maternal death and severe illness provides important lessons for maternity care providers. Ongoing audit of death and severe illness must be a requirement for the maternity care system and for comprehensive perinatal health surveillance. This report marks the first step towards ensuring that these vital measures are consistently and routinely available.

Key Recommendtions to Enhance Maternal Mortality and Severe Maternal Morbidity in Canada
1	Where feasible, specific maternal death review committee should be established (or maintained) as the ideal maternal death review mechanism.*
2.	In jurisdictions without a specific maternal death review committee, the coroner/medical examiner should be a focal point for maternal death review activities.*
3.	Whether in the form of a specific maternal death review committee or in collaboration with the coroner/medical examiner, an appropriate body should be authorized to review reports of maternal death and seek additional, pertinent case information as necessary.
4.	Legislation on notification to coroners/medical examiners in all jurisdictions should specifically mention "pregnancy" to ensure complete ascertainment of maternal deaths.
5.	Coroner/medical examiner reports on deaths during pregnancy or following pregnancy should be collated so that they are easily retrievable for maternal death review activities.
6.	Consistency in the definition of maternal death and in the information collected on each maternal death should be attained across all jurisdictions, including attention to vulnerable populations.
7.	An ongoing mechanism should be established for national synthesis and reporting of provincial/territorial maternal death investigations.
8.	Maternal death review activities at the provincial/territorial, regional and national level must ensure timely feedback to health care providers and facilities active in maternity care.
9.	Future efforts should refine the coding and classification system for severe maternal morbidity in Canada's hospitalization databases, with particular attention to the change from ICD-9 to ICD-10.
10.	Future reports should explore the use of indicators that combine severe maternal morbidity and maternal mortality, for example, the ratio of maternal deaths to "near miss."
11.	Consideration should be given to reviewing individual cases of specific types of severe maternal morbidity, where feasible.

*The size of the population may necessitate a regional-level review mechanism.

[ Previous ] [ Next ]