The purpose of this study was to evaluate the possibility of establishing predictors of mortality in women with amniotic fluid embolism.
The purpose of this study was to evaluate the possibility of establishing predictors of mortality in women with amniotic fluid embolism.
Our previous report identified eight factors associated with amniotic fluid embolism (AFE) fatality: dyspnea, cardiac arrest, loss of consciousness, serum sialyl Tn greater than 47 U/mL, serum interleukin-8 greater than 100 pg/mL, vaginal delivery, multiparity and term delivery. The ratio of the number of positive fatal factors to the number of possible fatal factors in the same case was calculated as the abundance ratio, which was used because information regarding all eight factors was not retrievable for all the patients at the time of registration. The patient group was divided into four quartiles based on this abundance ratio, and the mortality rate in each quartile was compared with the overall mortality rate among the 130 patients with AFE enrolled between 1992 and 2006. The validity of this approach was confirmed in another dataset from a cohort of 38 patients with AFE in 2007.
A statistically significant positive correlation was observed between the abundance ratio and the mortality in each quartile (P < 0.01) for the patients with AFE enrolled between 1992 and 2006. This result was also found in the AFE patients enrolled in 2007 (P < 0.05). Thus, an increased in the abundance ratio of the eight fatal factors resulted in an increased case fatality rate.
These data suggested that the abundance ratio of fatal factors may be a useful predictor of mortality and therefore may be expected to improve prognostic accuracy in the future.
Amniotic fluid embolism (AFE) syndrome is a devastating complication of pregnancy with an abrupt and fulminant onset and is one of the main causes of maternal mortality. Autopsy studies have demonstrated that AFE occurs following the contamination of the maternal circulation by fetal materials such as amniotic fluid and meconium.[1, 2] Although this condition has been suggested to be caused by an uncharacterized immune reaction rather than an embolic phenomenon,[3, 4] the causes and mechanisms responsible for AFE remain enigmatic. This syndrome is particularly alarming due to its unpredictability: it is likely unpreventable, and patients deteriorate quickly with high mortality. Accordingly, mortality due to AFE is often difficult to diagnose correctly. Nonetheless, risk factors for AFE have been reported. Three large population-based retrospective cohort studies have identified age over 35 years, cesarean section, forceps- or vacuum-assisted vaginal delivery, placenta previa, placental abruption and eclampsia as possible risk factors.[5-7] It may be possible to reduce the incidence of AFE through the management of these risk factors but not the case fatality rate. Although recent population-based studies indicate a decrease in case fatality rates for AFE, data regarding the factors leading to the reduction in mortality remain scarce. Eight factors associated with mortality in AFE were previously identified by our group; to the best of our knowledge, no other studies have reported factors associated with mortality due to AFE. The aim of this study was to investigate whether mortality in parturient women with AFE could be predicted using these fatal factors. Such information may establish useful predictors for mortality in patients with AFE and therefore may be expected to improve prognostic accuracy in the future.
The diagnosis of AFE was based on the clinical features listed in the Japanese Consensus Criteria for the Diagnosis of AFE. The enrollment criteria were as follows: (i) at least one of the following symptoms: cardiac arrest (acute hypoxia and hypotension), respiratory arrest (dyspnea) or consumptive coagulopathy (severe obstetric hemorrhage); (ii) the onset of all signs and symptoms occurring during pregnancy, labor, cesarean section or within 12 h post-partum; and (iii) the absence of other illnesses that could explain the observed signs and symptoms.
The study subjects were recruited from the Japan AFE Registration Center in Hamamatsu University School of Medicine, Shizuoka, which is closely linked to the AFE Association of Japan, Nara Medical University. Patients who met the criteria for AFE were enrolled directly and voluntarily to this center by the chief physician and provided informed consent. The identification of AFE at the center was based on the clinical diagnosis as recorded in the medical report, without other verifying evidence. Our study was limited by its dependence on voluntary self-reporting. The study was reviewed and approved by the institutional review board of Hamamatsu University School of Medicine, which also allowed us to contact the patients or their families.
Eight factors were identified as correlated with fatal AFE in our previous report and included dyspnea, cardiac arrest, loss of consciousness, serum sialyl Tn (STN) greater than 47 U/mL, serum interleukin (IL)-8 greater than 100 pg/mL, vaginal delivery, multiparity and term delivery.
Information regarding all eight of the fatal AFE factors was not always retrievable at the time of registration; therefore, the abundance ratio of the factors that were available was analyzed. The abundance ratio is defined as the ratio of the numbers of positive fatal factors to the number of possible fatal factors in the same case (Table 1). Because this analysis focused on increasing prognostic accuracy in patients presenting fewer than half of these factors, cases presenting between four and seven of the fatal factors were excluded. This exclusion applied to the data for both the 2007 and 1992–2006 cohorts.
|STN ≥ 47 U/mL||1||Yes (78 U/mL)||1|
|Vaginal delivery||1||No (C/S)||0|
|Term delivery||1||Yes (38 weeks)||1|
|Absent||Loss of consciousness|
|IL-8 ≥100 pg/mL|
A total of 135 patients met the inclusion criteria, comprising both fatal AFE (n = 65) and non-fatal AFE (n = 70). Of the 135 patients, 114 (84.4%) were registered voluntarily in Japan and the other 21 (15.6%) were from other countries. This cohort was the same as that in our previous report. Five (four fatal, one non-fatal) of the 135 cases were excluded based on the definition of the abundance ratio (Table 2).
|No. of missing items||Fatal||Non-fatal|
In 2007, 38 Japanese patients met the inclusion criteria, comprising both fatal AFE (n = 29) and non-fatal AFE (n = 9). All 38 patients were registered voluntarily through the same system used in 1992–2006. No patient was excluded based on the definition of the abundance ratio.
Information regarding certain items (i.e. fatal factors) was missing from some of the registration documents of the patients in the 1992–2006 cohort, and if there was a statistical bias in the missing data between the fatal and non-fatal cases, it would be inappropriate to compare these two groups. Thus, the frequency of missing data in both groups was compared using the χ2-test.
The patient group was divided into four quartiles according to the abundance ratio results: 0–24%, 25–49%, 50–74% and 75–100%. Differences in the case fatality rates of each of the four groups were investigated for both the 1992–2006 and 2007 cohorts using Fisher's post-hoc test to confirm the relationship between the case fatality rate and the abundance ratio.
The statistical analyses were performed using SPSS version 16.0 and Statview4.1.
The P-value of the χ2-test was 0.1333. Therefore, we observed no statistical bias in the amount of information provided in the registration documents between the fatal and non-fatal cases (Table 2).
A correlation was observed between the abundance ratio and the case fatality rate. Statistically significant differences in mortality between each abundance ratio group were demonstrated for the 1992–2006 cohort (P < 0.01), and significant differences in the case fatality rate between each quartile were also demonstrated for the 2007 cohort (P < 0.05) (Fig. 1). Thus, an increase in the abundance ratio of the eight fatal factors resulted in an increased case fatality rate. These data suggest that the abundance ratio of fatal factors may be a useful predictor of mortality.
Amniotic fluid embolism is a perinatal disease with a high case fatality rate. This high mortality is a result of the difficulty that the patient's body has in responding to a sudden onset of severe shock and hemorrhage. Moreover, it is not currently possible to predict AFE or treat it prophylactically, and simply extracting the risk factors reported elsewhere[5-7] may not solve the problem of high mortality. Conversely, understanding the risk factors associated with AFE may contribute to the prevention. Therefore, we considered the abundance ratio, whereby a high abundance ratio corresponds to a high case fatality rate, as a tool to reduce this rate. In Japan, approximately 50% of pregnant women deliver at private clinics, and patients with such conditions as AFE, which exhibits sudden, unpredictable and severe onset, would be difficult to resuscitate in these settings because of the lack of medical equipment and staff. If parturient women who have AFE or who may be at risk of mortality due to AFE can be identified, then the medical response may become sufficiently rapid enough to ensure the survival of the mother. Indeed, the early recognition of AFE with prompt intervention is paramount to a successful outcome and to decreasing the associated mortality. Therefore, it is necessary to transport a patient from a private clinic to a higher level medical facility as soon as possible. In addition, the abundance ratio may be a useful index for identifying the appropriate emergency level for the requested transport, and reduced maternal mortality in AFE cases may be expected when predictors of mortality have been identified using this ratio. From an alternative point of view, this ratio offers two merits: (i) a more accurate prognosis may be achieved by analyzing the survival of patients with high abundance ratios; and (ii) this ratio becomes a useful piece of information to explain the patient's condition, which can lead to an improved relationship with the patient's family.
However, this ratio is not consistent with current medical information because the two serum markers included as fatal factors, STN[10, 11] and IL-8, cannot be detected immediately. These two serum markers are crucial predictors of mortality when compared to other clinical manifestations for two reasons. First, STN is recognized as NeuAc-α-2,6-GalNAc and is present in high concentrations in meconium. The correlation between the fatality and turbid amniotic fluid has been described previously. In patients with AFE who presented AF containing thick meconium, there was a shorter time from the initial presentation to cardiac arrest and an increased risk of neurological damage or death. As mentioned above, the detection of meconium passage into the maternal circulation may be a crucial factor for accurate prognosis. Second, IL-8 in the bronchoalveolar lavage fluid is the most significant predictor of mortality in patients with acute respiratory distress syndrome. IL-8, a major chemoattractant for neutrophils, promotes the secretion of the proteolytic enzyme neutrophil elastase, which results in poor patient outcomes due to multiple organ failure. Patients with AFE exhibit a high frequency of adult respiratory distress syndrome (ARDS) suggesting that high serum IL-8 concentrations have a critical impact on the outcomes of patients with AFE complicated by ARDS. Unfortunately, in our study, no registration data indicated a correlation between ARDS and fatal AFE. Furthermore, measurements of these two serum markers will not be available to most clinicians around the world. However, these fatal factors for AFE patients could not be substituted by other clinical symptoms associated with serum STN and serum IL-8, such as turbid amniotic fluid and elevated fever, respectively, because turbid amniotic fluid (P = 0.289) and elevated fever (P = 0.514, data not shown) were not correlated with fatal AFE in our enrolled patients. We suggest that an easy-to-use kit to detect these two serum markers using monoclonal antibodies should be considered to avoid the limiting value of the abundance ratio or diagnostic criteria based on these markers.
In our study, an increase in the abundance ratio of the eight fatal factors was associated with a high case fatality rate. Because the analysis used the same population as our previous report, it was expected that the abundant ratio and the case fatality rate would vary simultaneously. To validate our result, 38 patients with AFE were enrolled in 2007 and analyzed using the same abundance ratio. A significant difference was also confirmed in the 2007 cohort, though the correlation was not as strong as in the 1992–2006 cohort, likely because of the small study population and the case fatality rate. Furthermore, the data suggest that the abundance ratio is an effective predictor of mortality because the same tendency was indicated in both cohorts.
It is possible that different results could be obtained using other databases. For example, the reported incidence of AFE varies among countries. The cause of this difference has been ascribed to methodological differences in the collection and analysis of the data. Our data were not population-based or national data but rather voluntarily provided data, and the population was both highly homogeneous and limited in number. Moreover, the AFE entry criteria are slightly different in Japan compared to the USA and the UK, though it is assumed that the interpretations of the Japanese criteria are similar to those of the USA and UK criteria. Our entire population exhibited a case fatality rate of 56%. Because recent population-based studies have consistently reported case fatality rates ranging 11–43% (six out of eight studies reported a mortality rate of under 21%), it appears that a substantial ascertainment bias toward more serious and fatal cases was present in our study. Indeed, differing results may be observed in different AFE populations. Thus, our study had several limitations. It would be interesting if other investigators evaluated the predictors of mortality using other databases and compared the results with those of our study. Any disparities in the results would likely be due to the enrollment system or differences in racial background. The evaluation of this ratio worldwide would, at the very least, show whether the AFE enrollment system should be standardized internationally, as it is unlikely that the patient AFE databases in each area or country are accurate.
The authors are grateful to the Japan Association of Obstetricians and Gynecologists for their support of this study. This study was supported by a grant (Grant-in-Aid for Scientific Research) from the Ministry of Education.