Pulmonary Embolism Mortality in the United States, 1979-1998: An Analysis Using Multiple-Cause Mortality Data

Background  Pulmonary thromboembolism (PTE) is a common clinical problem that is associated with substantial morbidity and mortality. Estimates of PTE mortality and predictions of PTE trends have varied widely. These estimates play a role in the planning of national health strategies. The analysis of pulmonary embolism mortality trends and comorbidities may elucidate how well we treat and prevent the disease as well as identify additional risk factors.

Methods  We analyzed PTE (International Classification of Diseases, Ninth Revision code 415.1) as reported on death certificates in the Multiple-Cause Mortality Files compiled by the National Center for Health Statistics from 1979 to 1998.

Results  Of all the 42 932 973 decedents, 572 773 (1.3%) had PTE listed on their death certificates and 194 389 of these (33.9%) had PTE as the underlying cause of death. The age-adjusted rate of deaths with PTE decreased from 191 per million in 1979 to 94 per million in 1998 overall, decreasing 56% for men and 46% for women. During the study period, the age-adjusted mortality rates for blacks were consistently 50% higher than those for whites, and those for whites were 50% higher than those for people of other races (Asian, American Indian, etc). Within racial strata, mortality rates were consistently 20% to 30% higher among men than among women. Conditions that were of higher likelihood in persons who died with PTE included thrombophlebitis, fractures, trauma, postoperative complications, certain cancers, and the inflammatory bowel diseases.

Conclusions  Mortality with PTE in the United States has decreased during the 20-year period. The mortality rates between men and women and between racial groups vary substantially. These findings may be useful in better directing preventive therapy efforts.

PULMONARY THROMBOEMBOLISM (PTE) is a common clinical problem that is associated with substantial morbidity and mortality. Hospital-based studies have played a major part in the estimation of PTE mortality in the United States.¹ From these studies, the incidence of PTE in the United States is estimated at 1 case per 1000 persons per year, or 200 000 to 300 000 hospitalizations per year.^2,3 The Prospective Investigation of Pulmonary Embolism Diagnosis study found a case fatality rate of 15% within the 3 months after a PTE,⁴ but only 10% of these deaths were directly attributable to PTE.⁵ It is estimated that as many as 50 000 to 100 000 people die of PTE annually in the United States.^6,7 Some investigators state that the mortality rate from PTE is stable,^1,8 while others say it is decreasing.^4,9-12

We were prompted by discrepancies in mortality estimates to use the Multiple-Cause Mortality Files to analyze national PTE mortality trends from 1979 through 1998 when PTE appeared on the death certificate as the underlying cause of death (UCD) or as a comorbid condition. We also assessed effects of age, sex, and race and examined the prevalence of comorbid conditions to help delineate risk factors associated with PTE death.

The National Center for Health Statistics annually compiles data from all death certificates filed in the United States, using the vital records from the individual states. These data contain the International Classification of Diseases codes for the underlying cause of death and up to 20 conditions listed on the death certificate. The International Classification of Diseases, Ninth Revision (ICD-9) was implemented in 1979 and was in effect throughout the 20-year period used in this study. The conditions are recorded in 2 places: on the entry axis, which contains the conditions exactly as reported on the death certificate, and on the record axis, which is edited by a computerized algorithm known as the translation of axes. The automated classification of medical entities algorithm determines the UCD from the conditions and their positions as listed on the death certificates. Quality assurance of the data is maintained by a trained nosologist who codes conditions at the state level and, in turn, by a nosologist at the National Center for Health Statistics who periodically reviews data from a 10% sample of the submitted death certificates. The result of this process is the Multiple-Cause Mortality File.

We searched the record axis portion of the 1979 through 1998 Multiple-Cause Mortality Files for records containing ICD-9 code 415.1, pulmonary embolism (this category excludes postpartum pulmonary embolism, amniotic fluid emboli, fat emboli, and tumor emboli). Within this group of death records, we searched for the UCD and comorbid conditions including obesity (ICD-9 278.0), septicemia (ICD-9 038.9), bronchopneumonia (ICD-9 485-486), chronic obstructive pulmonary disease (ICD-9 492-496), thrombophlebitis (ICD-9 451), cerebrovascular accident (ICD-9 436), congestive heart failure (ICD-9 428.0), acute myocardial infarction (ICD-9 410), arteriosclerotic cardiovascular disease (ICD-9 429.2 and 414.0), hypertension (ICD-9 401.1 and 401.9), and neoplasm (ICD-9 140-239.9), including specific organs: stomach (ICD-9 151.9), colon (ICD-9 153.9), pancreas (ICD-9 157.9), lung (ICD-9 162.9), breast (ICD-9 174.9), prostate (ICD-9 185), brain (ICD-9 191.9), kidney (ICD-9 189.0), lymphoma (ICD-9 202.8), multiple myeloma (ICD-9 203.0), and uterus (ICD-9 179-182.0). We also included fractures (ICD-9 829) and trauma (ICD-9 810-930).

The PE group was analyzed according to age, sex, and race, including white, black, and other (other includes Asian, American Indian, etc). The US census data were used to calculate population denominators. For age-adjusted rates, the 2000 US population was used as the standard. All analyses were performed with SAS software (SAS Institute Inc, Cary, NC). We evaluated the likelihood that decedents who had PTE listed on their death certificates also had other medical conditions. We calculated proportionate mortality ratios (PMRs),¹³ adjusted for age, sex, and race (APMR), to assess whether the presence of a medical condition is more likely (APMR >1) or less likely (APMR <1) in deaths in which PTE is present than in all deaths (PMR is the proportion of deaths in which PTE is present that are associated with a particular medical condition divided by the proportion of all deaths in which that condition is present [comparison population]).

In calculating PMRs, we used SAS software. To calculate the denominator of the PMRs, we used 100% of US deaths from 1979 to 1998.

During the 20-year period, 572 773 death records listed PTE (ICD-9 415.1) either as the UCD or as a contributing cause of death among the 42 932 973 decedents. Of those 572 773 death records, 33.9% (194 389) had PE listed as the UCD.

The total number of deaths decreased from 35 750 in 1979 to 24 947 in 1998, and the age-adjusted mortality rate due to PE decreased from 191 per million in 1979 to 94 per million in 1998. Among men, the number of deaths decreased from 17 758 in 1979 to 11 270 in 1998 and age-adjusted mortality rates decreased from 232 per million in 1979 to 103 per million in 1998 (Figure 1). For women, the number of deaths decreased from 17 989 in 1979 to 13 675 in 1998 and the age-adjusted rate decreased from 163 per million in 1979 to 88 per million in 1998. The proportion of PTE deaths with trauma or cardiac disease present decreased from 1979 to 1998, whereas the proportion for cancer increased (Table 1).

The age-adjusted mortality rates for PTE were consistently 50% higher among blacks than those of whites, and rates for whites were 50% higher than those for people of other races. Within these racial strata, mortality rates were consistently 20% to 30% higher among men than among women (Figure 1).

The age-specific mortality rates due to PTE varied by age stratum (Figure 2). The largest number of deaths (163 216) occurred in the 75- to 84-year age group. For both men and women of all racial strata, age-specific mortality rates doubled for each age group older than the 15- to 24-year age group. The rate per 100 000 decedents did not vary appreciably between the 35- to 84-year age groups (Table 2). Some conditions, when associated with the mortality from PTE, were dependent on age. Age-specific patterns were different when the mortality from PTE related to trauma, fractures, and cancer was compared with the mortality from PTE related to concurrent cardiac disease, chronic obstructive pulmonary disease, or thrombophlebitis (Table 3).

Trauma, fractures, postoperative period, and obesity were overrepresented in persons who died with PTE (APMR = 1.89, 5.04, 1.85, and 4.55, respectively). Cancer, specifically ovarian, uterine, prostate, and brain (APMR = 1.03, 1.05, 1.01, and 1.01), were more likely to appear on the death certificate with PTE than were other cancers. Thrombophlebitis had the highest APMR, being present very often in association with death with PTE (APMR = 24.2). Pulmonary fibrosis, pulmonary hypertension, and the inflammatory bowel diseases (including Crohn disease and ulcerative colitis) had APMRs higher than 2.0 (APMR = 2.04, 2.48, 2.20, and 2.29, respectively).

The results of our study demonstrate that PTE is the UCD in approximately one third of death certificates that list PTE, and annual PTE mortality has decreased about 30% from 1979 to 1998. Furthermore, a consistent, steady decline has been seen during the years studied. In both females and males and in all the ethnic groups, the same consistent decline in mortality is seen. Black men have the highest mortality throughout the study period, while the mortality rates for black women are somewhat lower, followed by white men, white women, other men, and then other women. Most comorbid conditions listed on the death certificates that also list PTE are expected. Conditions that were more likely to appear along with PTE have higher APMRs.¹³ The APMR helps us ascertain comorbid conditions that are overrepresented in patients who die with PTE. The high APMRs seen with the inflammatory bowel diseases are somewhat unexpected, as are the low APMRs with colon and lung cancers, entities believed to have high rates of deep venous thrombosis (DVT) and PTE. The high APMRs with comorbid conditions such as thrombophlebitis and trauma are expected, offering a level of internal consistency.

Figure 1 and Figure 2 show the trend in decreasing mortality in the United States. The mortality from PTE decreased almost every year during the 20 years studied. Our data support previous reports of decreasing mortality rates from PTE.^9,14,15 There are several possible explanations for this reduction in reported PTE mortality: a decreased incidence of PTE, a decreased case fatality rate from PTE, or changes in diagnostic patterns. The case fatality of PTE has been studied by previous investigators and is generally reported as not changing^2,12; however, there may be differences in case fatality rate trends between races.¹⁶ A decreased incidence of PTE death could result from improved prevention of DVT,^17,18 better detection and treatment of DVT, or a decrease in risk factors for thrombosis. There are risk factors that have changed during the 20 years studied, including smoking prevalence,¹⁹ therapy for congestive heart failure,^20-26 and the increasing number of older persons in the US population.²⁷Table 1 illustrates the 20-year average association of PTE deaths with trauma, cardiac disease, and cancer. Trauma decreased slightly during the 20-year period from 11.8% to 10.8%. Cancer increased modestly, from 21.0% to 24%. Of interest is the more dramatic change that cardiac disease has shown in assocication with PTE deaths, decreasing significantly from 34.1% to 21.2%. A final possible explanation for a decreasing mortality rate is that more accurate PTE diagnosis results in less misclassification of cases as compared with 20 years ago (a lower false-positive rate).

The total number of deaths associated with PTE decreased from 35 750 in 1979 to 24 947 in 1998. These numbers are less than those cited by other investigators, which range from 50 000 to 100 000 per year.^1,7 The mortality rate within 3 months of a venous thromboembolic event is 15% to 17.5%.^{4,5,10,12,16,28} With this rate and the estimated 250 000 patients hospitalized with PTE annually,⁷ 37 500 to 43 750 would die, estimates that are consistent with our findings.

Although the mortality from PTE increases with age, our findings show that the proportion of decedents with PTE at the age groups between 25 and 84 years are similarly between 1% and 1.6% (Table 2). The mortality rate within each age group has decreased during the 20-year period (Figure 2). While the highest number of PTE deaths is in the 75- to 84-year age group, the mean age has remained relatively stable at 69 to 70 years during the entire 20 years studied. The highest rate (per 100 000 decedents) of PTE-related death is at age 55 to 64 years (see Table 2).

Our study found that males have a consistently higher rate of PTE-related deaths than females in all racial and age strata (Figure 1). This agrees with others,²⁹ including Lilienfeld,⁹ who reviewed PTE as the underlying cause of death from death certificates between 1979 and 1996. A study analyzing the patients from the Prospective Investigation of Pulmonary Embolism Diagnosis found that men had a higher frequency of PTE than women.³⁰ In agreement, elderly men and women in different ethnic strata have been studied through the Medicare system and their data show that men have a higher mortality rate than women from PTE.¹⁶ Male sex as a risk factor for PTE has conflicting data.^4,31,32 The Tecumseh Community Health Study and others have found that women had a higher frequency of DVT and PTE than men,^31,33 while more recent studies have found the opposite, or no difference beween the sexes.^9,30,34 Again, our study found that men have a higher rate of death with PTE than women.

We showed that blacks have a higher rate of PTE death than do whites, which is higher than those of other races (Figure 1). Blacks have increased amounts of comorbid conditions that may predispose them to having a fatal PTE, such as certain cancers, chronic renal and cardiac failure,^35-39 and an increased prevalence of smoking.^40,41 The mortality rate has decreased for each racial strata, including blacks. Siddique et al²⁹ found in the Medicare population from 1984 through 1991 that there was a decreasing mortality for whites, but this was not consistent for blacks, whose rates remained more stable. Unlike our US death certificate study, the Medicare data were limited to older, hospitalized patients.

Recognition of and therapy for PTE and its risk factors affect trends in mortality related to PTE. In an evaluation of the patients from the Prospective Investigation of Pulmonary Embolism Diagnosis study, factors associated with mortality were age greater than 60 years, congestive heart failure, pulmonary hypertension, chronic lung disease, cancer, and stroke.⁵ Others have added to this a history of recent surgery, myocardial infarction, pneumonia, trauma, hospital or nursing home confinement, central venous catheter or pacemaker placement, neurologic disease, chronic renal disease, tobacco use, chemotherapy, varicose veins, thrombophlebitis, oral contraceptive use, elective abortion, and others.^4,6,32,42-44 Recently, hypercoagulable states have been further characterized. Only 10% to 20% of hypercoagulable states have been implicated in PTE.^1,32

The implementation of preventive therapies and new diagnostic techniques as well as improvements in PTE therapy is likely to affect the incidence of as well as the mortality from PTE. Diagnostic methods have advanced during the past 20 years,⁴⁵ and technologies such as the helical computed tomographic scan have added to our ability to detect the presence of PTE.⁴⁶ However, these imaging studies still have their limitations,^1,47 and before a diagnostic test can be ordered, there must be clinical suspicion. The clinical diagnosis of PTE remains insensitive and nonspecific, as there are many factors that confound the signs and symptoms.⁴⁸ As a result, many fatal PTEs remain undiagnosed,⁸ and it is estimated that only 30% to 40% of clinically suspected PTEs are actually documented.^45,49,50

Hyers,⁴⁷ in a recent review, stated that major predisposing conditions to DVT are venous stasis from any cause, any type of trauma including surgery and childbirth, and increasing age.^6,47 He contends that all cancers increase risk, that adenocarcinoma of the visceral organs is the most recognized malignancy associated with DVT, and that lung cancer is the most common malignancy associated with venous thromboembolism. We found that lung cancer is the most common cancer associated with PTE death secondary to the high volume of lung cancer deaths, but proportionately fewer people die of lung cancer with PTE on their death certificate than of ovarian, uterine, brain, or prostate cancers. Lung cancer killed more than 2.8 million people during the 20-year study period. Deaths with PTE were listed along with lung cancer in approximately 28 000, yielding an APMR of only 0.66. Cancer has been found in nearly 10% of people with PTEs who did not have another predisposing factor.⁵¹

Our data show that not only are people with inflammatory bowel disease at risk for thrombosis, but they are in the group dying with PTE. Patients with inflammatory bowel disease have a known increased tendency toward thrombosis, and there are reports of DVT as well as PTE in the literature.⁵² The reason for this increase appears to be multifactorial, but of note is that there are serologic abnormalities during inactive Crohn disease and ulcerative colitis that may predispose patients to PTE, including a high incidence of abnormalities in fibrinolysis.⁵³ In the appropriate clinical setting, PTE needs to be considered in patients with inflammatory bowel disease.

Preventive therapies directed toward the risk factors for PTE have become better understood, although there are data to suggest that these interventions are underused and not affecting the incidence of PTE.⁵⁴ On the other hand, there are ample data showing an impact on at least early mortality with treatment of PTE, and reduced incidence of DVT and PTE with prophylaxis in high-risk situations.⁵⁵

The major limitation of this study is the death certificate and its reliability. The accuracy of the physician reporting on the death certificate has come into question in the past. Most recently it has been recommended that the method of completion of this certificate be changed, so that it is completed in conjunction with the postmortem examination and amended when the autopsy findings show a discrepancy.⁴⁸ This change may help the problem of improper coding of PTE on the death certificate noted by Proctor and Greenfield in 1997.⁵⁶ Unlike much of the data used to estimate PTE mortality in the past, the death certificate is filled out by a practitioner on all deaths in the United States, including those that never arrive at the hospital. To our knowledge, no one has specifically analyzed the trends in death certificate reporting in these situations. Better known is the diagnostic accuracy of the hospital clinician, who really does not even suspect PTE much of the time. Autopsy data on a sample of hospitalized patients found that 50% to 70% of PTEs were unsuspected by the clinician, even if PTE was believed to be the cause of death at autopsy.^8,48

Pulmonary thromboembolism remains a controversial topic in light of its complexities in diagnosis and treatment. However, many advances have been made that may have affected mortality, including improved prophylaxis, diagnosis, and treatment. The decreasing trends in PTE-related mortality are encouraging and are likely related to better prevention of DVT and PE and improved identification and treatment of underlying risk factors.

Corresponding author: Kenneth T. Horlander, MD, Department of Pulmonary Medicine, Clark-Holder Clinic, 303 Smith St, LaGrange, GA 30240 (e-mail: Pulm_Doc@yahoo.com).

Accepted for publication October 21, 2002.