Ethnic Differences in Out-of-Hospital Fatal Pulmonary EmbolismClinical Perspective
Background—In-hospital pulmonary embolism (PE) has been extensively studied in large populations; however, out-of-hospital fatal PE studies are rare. Here, we systematically evaluated a large number of decedents who suffered fatal PE outside of hospitals and were subsequently investigated by the New York City Office of Chief Medical Examiner.
Methods and Results—A total of 578 consecutive out-of-hospital fatal PE cases were analyzed. All underwent autopsy, toxicology, microbiology, and genetic testing. Incidence rates and baseline characteristics were analyzed. Race-adjusted incidence rates of out-of-hospital fatal PE (per 100 000 people per year) were as follows: blacks, 3.73 (95% confidence interval, 3.31 to 4.11); whites, 1.15 (95% confidence interval, 0.96 to 1.33); and Hispanics, 0.93 (95% confidence interval, 0.72 to 1.10). Overall, obesity (body mass index ≥30 kg/m2) was 2.5- to 3-fold higher in fatal PE cases than in the New York City population as a whole. Carrier frequencies for prothrombin G20210A in fatal PE were 2- to 10-fold higher than reported frequencies in ethnically matched controls. Cumulative distribution curves showed that compared with whites, blacks and Hispanics died at significantly younger ages (P<0.001). Univariate and multiple linear regression analyses showed that in addition to nonwhite ethnicity, heterozygous carriers for factor V Leiden (P=0.001) and obesity (P=0.002) are significantly associated with younger age at death.
Conclusion—There are unique epidemiological differences in out-of-hospital fatal PE between ethnic groups in New York City.
Fatal pulmonary embolism (PE) is a common immediate cause of death, usually presenting as a complication of deep venous thrombosis (DVT). Each year, >100 000 Americans die of DVT.1,2 Most current knowledge of PE was obtained from studies on in-hospital DVT patients in geographic regions consisting of mainly white American or European populations.3,–,5 Ethnic differences in epidemiology of DVT/PE have gained increased awareness, largely as a result of several recent studies conducted in the United States.6,–,9 It has become clear that the incidence rate of DVT/PE is higher in blacks than in whites and Hispanics, whereas known genetic risk factors that are common in the white population, such as factor V (FV) Leiden and prothrombin (factor II [FII]) G20210A, are relatively infrequent in blacks. Acquired risk factors also seem to contribute differently among ethnic groups with DVT and/or PE.6,9
Editorial see p 2189
Clinical Perspective on p 2225
Out-of-hospital fatal PE may represent a different subpopulation of PE than in-hospital PE because its clinical presentation is often acute and fatal. In New York City (NYC), all sudden, out-of-hospital deaths are investigated by the Office of Chief Medical Examiner (OCME). As a consequence, nearly all cases of fatal, out-of-hospital PE are identified by the NYC OCME. The standard investigation protocol involves a complete autopsy, laboratory tests, and a medicolegal investigation. This is in contrast to fatal, in-hospital PE deaths for which not all clinical diagnoses of fatal PE are confirmed by autopsy.10 Conversely, the percentage of clinically unsuspected PE causing sudden death has been reported to be as high as 50%11 to 80%.12
Here, we report the first autopsy-based study involving a large number (578) of consecutive out-of-hospital fatal PE cases in NYC. This study compared the baseline, genetic, and acquired risk factors for PE among white, black, and Hispanic populations.
The NYC OCME investigates all deaths within the city's 5 boroughs (8 million people) when caused by criminal violence, accident, or suicide, or if the decedent died suddenly when in apparent good health, was unattended by a physician, was in a correctional facility, or died under suspicious or unusual circumstances when application is made for cremation. Thus, nearly all out-of-hospital fatal PE cases are identified by the OCME. For this retrospective study, we examined 578 consecutive out-of-hospital PE cases investigated between June 2005 and December 2009 by the NYC OCME. Fatal PE was determined as the immediate cause of death for all cases. Determination of PE lesions as the immediate cause of death was based on the size, location, and distribution of embolisms, and was consistent with the clinical history of how a person died. This study included only cases in which PE lesions were fresh, recent, and fatal, and were determined to be the immediate cause of death. If a PE lesion was in a small pulmonary artery branch and was not considered the immediate cause of death, the case was excluded from in this study. All cases were subjected to thorough cardiac investigations. None of the cases included in our study had acute myocardial infarction or other autopsy findings that could be considered the immediate cause of death. A forensic examination includes autopsy, toxicology, scene investigation, review of clinical history (when possible), and molecular testing for common prothrombotic variants: FV Leiden, FII G20210A, and methylenetetrahydrofolate reductase (MTHFR) C677T. Postmortem blood, urine, bile, gastric content, liver, and brain were used for forensic toxicology testing. Postmortem tissue samples (spleen, liver, or heart) collected and preserved in RNAlater (Qiagen, Valencia, CA) or dry blood card samples were used for molecular testing. All testing was deemed necessary for diagnostic purposes and conducted in OCME facilities.
Genomic DNA was extracted and purified from postmortem tissues or dried blood cards with the Qiagen BioRobot M48 with MagAttract DNA Mini M48 Automated Purification Kit following the manufacturer's instructions. Multiplex polymerase chain reaction and multiplex SNaPshot methods were used for single-nucleotide polymorphism analysis for FV G1691A, FII G20210A, and MTHFR C677T. Primer sequences and thermal cycler conditions are shown in Table I in the online-only Data Supplement. All reagents and supplies were from Applied Biosystems Life Technologies (Carlsbad, CA), unless otherwise specified. Genomic DNA 1 to 30 ng was used for the multiplex amplification in a total volume of 25 μL with the following reagents (final concentration): 1× PCR Buffer II, 3 mmol/L MgCl2, 125 μmol/L deoxyribonucleotide triphosphates, and 0.1 U Taq gold polymerase. Total polymerase chain reaction products (0.01 to 0.4 pmol) were treated with ExoSAP-IT (USB, Cleveland, OH) to remove primers and deoxyribonucleotide triphosphates before continuing to the multiplex SNaPshot reaction following the manufacturer's instructions. The SNaPshot reaction products were detected by an ABI 3130xl genetic analyzer. GeneMapper data analysis software was used for allele calling.
Demographic, Genetic, and Risk Factor Variables
Age, sex, ethnicity, height, weight, complete autopsy report, and laboratory test results were available for all cases. Age and ethnicity were provided by the family. Height and weight of decedents were measured at autopsy. Body mass index (BMI) was calculated as kilograms per meter squared. Genotype frequency was calculated as the proportion of a specific genotype to the total number of individuals tested. When available, acquired risk factors that were considered by the medical examiner to contribute to death were noted. For this study, immobility was categorized as either acute or chronic. Acute immobility was associated with a recent decrease of physical activity owing to injury, surgery, hospitalization, or a prolonged flight or car ride. Chronic immobility was associated with long-term physical inactivity resulting from either chronic diseases, such as multiple sclerosis, or choice of sedentary lifestyle. Idiopathic PE was defined as fatal PE with no identifiable risk factors. Information on acquired risk factors, eg, immobility and oral contraception, was not available for all cases because of limitations in obtaining information from families or physicians.
Because the NYC OCME investigates all out-of-hospital deaths in the city, overall and race-adjusted incidence rates of out-of-hospital PE fatal can be calculated from the racial composition of NYC in which whites (non-Hispanic) represent a plurality of the population (35.1%), followed by Hispanics (27.5%), blacks (25.1%), and Asians (11.8%).13 The different incidence rates were calculated with the assumption of a Poisson distribution. Baseline characteristics of fatal PE were compared for whites, blacks, and Hispanics. Although the numbers of Asians in each category are identified throughout this study, statistical analyses were not performed, because the total number (7) was too small. The percentage of obesity (BMI ≥30.0 kg/m2) in fatal PE cases was compared with that reported in ethnically matched NYC residents.14 The genotype frequencies in PE cases were compared with the published average frequencies in ethnically matched controls.15,–,18
For categorical variables, frequencies among groups were compared by use of the Pearson χ2 test or Fisher exact test in the cases of small cell sizes (≤5). For continuous variables, means and SDs were calculated, and differences were tested by use of the Student t test. Sex, BMI, genotypes, and acquired risk factors were compared between ethnic groups. Bonferroni correction was used to adjust for multiple testing. Cumulative distribution function was used to compare the age at death in different ethnic groups. The significance of the age at death distribution between ethnic groups was tested by the Kolmogorov-Smirnov test. The normality of age at death was examined by the Kolmogorov-Smirnov test. Univariate and multiple linear regression models were used to estimate the mean differences in age at death among the ethnic groups after adjustment for sex, BMI, and genotypes. All P values were tested by the 2-tailed test in which values of P<0.05/3 indicated statistical significance. SAS 9.1 was used for all analyses.
Incidence Rate and Baseline Characteristics of Fatal Pulmonary Embolism by Ethnicity
The baseline characteristics of 578 consecutive out-of-hospital fatal PE cases from the NYC OCME were categorized and compared among the 3 different ethnic groups (Table 1). Blacks represented a majority of deaths (58.2%), followed by whites (25.1%), Hispanics (15.6%), and Asians (1.2%). This ethnic breakdown is in contrast to the general population of NYC (8 million people) in which non-Hispanic whites represent 35.1% of the population, followed by Hispanics (27.5%), blacks (25.1%) and Asians (11.8%) (2006 to 2008 data, US Census Bureau).13 When standardized to the NYC population, the overall incidence rate of out-of-hospital fatal PE per 100 000 people per year is 1.6 (95% confidence interval, 1.47 to 1.74). The race-adjusted incident rate of out-of-hospital fatal PE for blacks is >3 times higher (3.73; 95% confidence interval, 3.31 to 4.11; P<0.001) than for whites (1.15; 95% confidence interval, 0.96 to 1.33) and Hispanics (0.93; 95% confidence interval, 0.72 to 1.10). There was no statistically significant sex difference within ethnic groups, and the overall sex ratio of fatal PE cases is consistent with the sex ratio of the general population of NYC (male-to-female ratio=0.9).13
Obesity (BMI ≥30.0 kg/m2) was present in a majority of cases in all ethnic groups (51.1% of whites, 56.9% of blacks, and 62.3% of Hispanics). This is in contrast to obesity rates for NYC residents (17.1% of whites, 29% of blacks, and 26.2% of Hispanics).14 Overall, obesity is 2.5- to 3-fold greater in fatal PE across all ethnicities compared with NYC residents. Although numbers were too small for statistical analysis, it was observed that >10% of decedents were morbidly obese (BMI ≥45 kg/m2; Table II in the online-only Data Supplement).
Genetic analysis for FV Leiden, FII G20210A, and MTHFR C677T was performed on all 578 fatal PE cases. Cases were considered positive when decedents were heterozygous for FV 1691A or FII 20210A (no homozygote for either FV 1691A or FII 20210A was identified) or homozygous for MTHFR 677T. A few individuals carried compound variants; ie, >1 variant was identified (Table III in the online-only Data Supplement). Overall, the 3 variants were higher in whites than in blacks (P<0.001). To evaluate the significance of genotype frequencies in fatal PE, data from this study were compared with published data in healthy control subjects for each ethnic group.15,–,18Table 1 shows that genotype frequency differences between decedents and previously published ethnically matched control subjects for FII G20210A heterozygotes were ≈4-fold greater in whites (9.7% versus 2.5%), 2-fold greater in blacks (1.2% versus 0.5%), and 18-fold greater in Hispanic (8.9% versus 0.5%). On the other hand, genotype frequencies for FV G1691A heterozygotes and MTHFR C677T homozygotes in decedents were similar to those published in ethnically matched control populations.
Data for known acquired risk factors for DVT fell into 2 categories: those that could be discerned at autopsy (eg, pelvic vein compression) and those that typically require information from the decedent's family or physician (eg, oral contraceptive hormone use). One commonly identified risk factor in the second category was immobility (Table 1). Nearly half of all decedents suffered from acute or chronic immobility. Thus, even in the absence of 100% recording of acquired factors, a large proportion of decedents were immobile in all ethnic groups. Autopsy-determined atherosclerotic and hypertensive cardiovascular disease was approximately 6% in whites and Hispanics, but nearly 10% in blacks. In addition, a higher number of idiopathic PE deaths was found in blacks (17.9%) than in Hispanics (7.8%) and whites (10.4%).
Age at Death by Ethnicity
Age at death for blacks and Hispanics was compared with that of whites by cumulative distribution function analyses (Figure). The Kolmogorov-Smirnov test showed that the distribution of age at death for whites was significantly different from that for blacks (P<0.001) and Hispanics (P<0.001). Blacks and Hispanics died at a younger age compared with whites. The difference in age at death between blacks and Hispanics was not significant (P<0.78).
To determine whether sex, BMI, or genetic variables contributed to younger age at death, univariate and multiple linear regression analyses were performed (Table 2). Both blacks (P<0.001) and Hispanics (P<0.001) remained significantly associated with younger age at death compared with whites after adjustment for BMI, sex, and genetic variants. In addition, younger age at death was significantly associated with obesity (P=0.002) and being a carrier for heterozygous FV 1691 G>A variant (P=0.001).
This study presents data from 578 consecutive out-of-hospital fatal PE cases investigated between June 2005 and December 2009 in NYC. Principal findings are the following: (1) There is a 3-fold higher race-adjusted incidence rate of out-of-hospital fatal PE for blacks than for whites and Hispanics in NYC; (2) black and Hispanic decedents with fatal PE died at a significantly younger age compared with whites; (3) obesity may be associated with an increased risk of fatal PE and young age at death; (4) carriers of the FII G20210A variant may be at greater risk than the general population for developing fatal PE, but this variant is not associated with an younger age at death; and (5) carriers of FV Leiden are not at higher risk than general population for developing fatal PE but are associated with a younger age at death resulting from fatal PE.
Previous Studies Comparing Ethnic Differences of Deep Venous Thrombosis/Pulmonary Embolism
To the best of our knowledge, this is the first large-scale, epidemiological study of out-of-hospital fatal PE in a metropolitan area with a diverse ethnic background. Consequently, there are no national, state, or municipal data with which to compare our results. For comparison, several previous studies involving large numbers of in-hospital DVT/PE patients with different ethnic backgrounds are discussed. In 1 study, Schneider and colleagues7 used New Jersey hospital discharge data to determine death rates from PE and found that the rate for blacks was almost double that of whites. In a series of studies conducted in California by White and colleagues,8,19,–,21 blacks were shown to have twice the rate of whites and Hispanics for the first incidence of DVT, recurrent DVT, idiopathic DVT, and proportion of PE after the initial incidence of DVT. This study showed that blacks had a 3-fold higher incidence rate of out-of-hospital fatal PE than whites and Hispanics.
Two large population studies examined baseline characteristics, genetics, and acquired risk factors of hospital DVT/PE in a comprehensive manner. One study involved large numbers of enrolled DVT patients in the 7 centers within the Centers for Disease Control and Prevention Thrombosis and Hemostasis Centers Research and Prevention Network.9 Results from that study showed that compared with white patients with venous thromboembolism, blacks with venous thromboembolism had a lower proportion with transient venous thromboembolism risk factors (ie, less occurrence of recent surgery, trauma, and infection; among women, less recent oral contraceptive use and hormone therapy) and a significantly higher proportion with PE. Blacks had a significantly higher mean BMI, and a significantly lower proportion had recent surgery, trauma, or infection; family history of venous thromboembolism; and documented thrombophilia. These results are consistent with those of the Genetic Attributes and Thrombosis Epidemiology (GATE) study.6 The association of obesity (BMI ≥30.0 kg/m2) with an increased likelihood of fatal PE observed in our study is consistent with both these studies, as are our data showing reduced frequencies of genetic variations for FV, FII, and MTHFR in the black population.
Unique Findings in This Study
A significant finding in our study is the younger age at death resulting from fatal PE for Hispanics and blacks compared with whites. These results are in contrast to reported findings of in-hospital DVT and PE, in which the incidents of in-hospital fatal PE increased with age for both blacks and whites.22 Reasons for the younger age at death are not clear. It is possible that there are interactions among multiple risk factors, such as obesity, immobility, and hypertensive and atherosclerotic cardiovascular diseases, as well as socioeconomics-related healthcare issues, that were not possible to examine in this study. In addition, it is possible that specific genetic factors in blacks and Hispanics may play important roles in determining the younger age at death. Future research in these areas is warranted.
A second significant finding in our study concerns genotyping frequencies. We found that the carrier frequency for FII was 2- to 10-fold greater than published ethnically matched controls. Our FII variant frequency is higher than that observed in several DVT studies,23,24 suggesting that FII carriers might be associated with a higher risk for developing acute fatal PE. The underlying pathophysiological mechanism may be related to FII gene expression. The FII variant, located in the 3′ untranslated region of the FII gene,23 represents a gain-of-function mutation, causing increased mRNA and protein synthesis.25 When this variant coexists with FV variants, it is associated with earlier onset of DVT.26 It is also possible that this variant interacts with other as-yet unknown variants or risk factors and results in fatal PE. Another interesting finding in our study is that although FV Leiden does not contribute to an overall increased risk for fatal PE, it is strongly associated with an younger age at death. The FV Leiden carrier frequency in out-of-hospital fatal PE cases in our study is consistent with that in several previous studies27,–,32 showing that the frequencies are higher in DVT but not increased in fatal PE. However, our results indicate that FV Leiden is associated with an younger age at death resulting from fatal PE. When found in conjunction with other risk factors for PE, such as pregnancy or the FII variant, FV Leiden has been reported to be associated with earlier onset of thrombosis.33 Therefore, it is reasonable to speculate that there are possible interactions between FV Leiden and other variants or risk factors. Although the significance of this finding is at the moment unclear, it will be interesting to see whether our results are confirmed in future studies in which age at death or onset of DVT and PE is taken into account compared with genotyping.
Strengths and Limitations
This investigation is the first demographic and baseline comparison of out-of-hospital fatal PE for 3 major ethnic groups from a large metropolitan area in United States, and the results reveal unique and clinically relevant findings. However, this study has several limitations. First, there are no matched, healthy control subjects from the same geographic area for comparing genotypes. Consequently, we used genotype frequencies from previously published studies for each ethnic group. However, the clinical criteria for healthy control subjects in those studies, the accuracy of ethnic identification, and the differences in black and Hispanic populations in different geographic regions of the United States are not clear. Second, the ethnic information for decedents in this study relied on family reports, which can be culturally or socially influenced and do not have molecular confirmation. Therefore, it is not possible to assign, with complete accuracy, the ethnic identity of every decedent. Inaccurate ethnic assignments directly affect genotype frequency results, especially for ethnic-specific variants. For instance, if a person were a mixture of white and black and identified as black by the family, the genotype frequency of a variant would be counted toward black, not white. Finally, we were limited in our ability to find and interview families, friends, and treating physicians for all decedents. Consequently, medical histories in some cases of fatal PE are incomplete, and acquired risk factors were therefore not always identified. However, because some risk factors can be reliably identified at autopsy and through laboratory tests, we are confident that major contributing factors identified in our study play significant roles in out-of-hospital fatal PE. Furthermore, it is unlikely that there is a bias in the reporting of the acquired risk factors by medical examiners for different ethnic groups; therefore, a cross-ethnic comparison remains valid.
This study provides unique characteristics of incidence rates, age at death, BMI, and genotype frequencies of FII (G20210A) and FV Leiden in out-of-hospital fatal PE from 3 major ethnic groups in NYC. These data should help inform clinicians, medical examiners, and public health officials and policy makers in monitoring PE, especially in nonwhite ethnic groups, and addressing programs and research to identifying genetic and other risk factors that lead to PE.
The online-only Data Supplement is available with this article at http://circ.ahajournals.org/cgi/content/full/CIRCULATIONAHA.109.976134/DC1.
- Received July 8, 2010.
- Accepted March 14, 2011.
- © 2011 American Heart Association, Inc.
- Rathbun S
- Anderson FA Jr.,
- Wheeler HB,
- Goldberg RJ,
- Hosmer DW,
- Patwardhan NA,
- Jovanovic B,
- Forcier A,
- Dalen JE
- Holst AG,
- Jensen G,
- Prescott E
- Sandler DA,
- Martin JF
US Census Bureau. 2006–2008 American Community Survey 3-Year Estimates. Available at: http://factfinder.census.gov/servlet/DatasetMainPageServlet?_program=ACS. Accessed July 8, 2010.
- Van Wye G,
- Kerker BD,
- Matte T,
- Chamany S,
- Eisenhower D,
- Frieden TR,
- Thorpe L
- Poort SR,
- Rosendaal FR,
- Reitsma PH,
- Bertina RM
This report presents the first epidemiological study involving a large number of autopsy-based out-of-hospital fatal pulmonary embolism investigations in New York City, with its diverse ethnic population. New data presented should alert and aid clinicians in evaluating patients at risk for acute and fatal pulmonary embolism on the basis of different ethnic backgrounds. Because blacks and Hispanics suffer fatal pulmonary embolism at a significantly younger age than whites, physicians should closely monitor these populations for known risk factors (such as body mass index) and counsel healthy lifestyles, especially at younger ages. Because of the large number of prothrombin G2010A carriers observed in white and Hispanic out-of-hospital pulmonary embolism decedents, testing for prothrombin G20210A in high-risk patients is indicated. Finally, our results clearly point to the need for additional research to identify other genetic causes of fatal pulmonary embolism, especially in blacks, who have low frequencies of known genetic risk factors.