Authors: Yon K. Sung, Jeffrey A. Kline
Categories: Editorials
Source: Annals of the American Thoracic Society
Compared with other acute thrombotic emergencies, such as stroke and myocardial infarction, there have been only modest changes in the diagnosis and treatment of pulmonary embolism over the past 50 years. Regarding diagnosis, public awareness of the symptoms of pulmonary embolism remains low; at the point of care, the decision to order diagnostic testing still relies heavily on physician acumen and intuition. Regarding treatment, although the anticoagulants of choice have shifted, and a small percentage of patients with low-risk pulmonary embolism are treated at home, anticoagulation with hospitalization has remained the mainstay of treatment for pulmonary embolism since 1960. Fifty years ago, the multicenter UPET (Urokinase Pulmonary Embolism Trial) randomized controlled trial showed that thrombolysis could produce more rapid resolution of clot burden with pulmonary embolism but without any change in mortality. Since then, despite the advent of multiple advanced therapies—recombinant plasminogen activators for systemic fibrinolysis, catheter-directed fibrinolysis and catheter-based mechanical clot extraction, extracorporeal membrane oxygenation support, and surgical embolectomy—the actual impact on pulmonary embolism mortality remains controversial. At some large academic hospitals, pulmonary embolism response teams (PERTs) have changed the initial management of pulmonary embolism from an individual craft to a team effort. However, it has been argued that over the past several decades the mortality rate has not improved, even for severe pulmonary embolism treated in prestigious academic medical centers (1). With this background, the national, database-derived work from Zghouzi and colleagues reported in this issue of AnnalsATS (pp. 1571–1577) provides provocative and concerning inferences about stagnant outcomes for pulmonary embolism (2). This study does have a number of limitations, however, and so although the findings are significant, it is not clear what needs to be done to improve pulmonary embolism–related mortality going forward.
Zghouzi and colleagues use the publicly available US Centers for Disease Control's Wide-ranging ONline Data for Epidemiologic Research (WONDER) database, maintained by the Centers for Disease Control and Prevention, to assess mortality rates for pulmonary embolism in the United States from 2006 to 2019. Several aspects of this database warrant critical consideration. To the best of our knowledge, no protocol for methods behind the WONDER database has been published in peer-reviewed literature. Herein lie several questions about potential biases that might affect the results. The frequency of errors in the accuracy of the cause of death written on death certificates has been estimated to range from 30% to 80%. This error rate appears to be particularly high with pulmonary embolism and other cardiovascular diseases (3, 4). Errors in death certificates can arise from multiple sources, including missing data for immigrants, illegible handwriting, or, more commonly, errors in identification of the cause of death (5). Moreover, death certificates do not have International Classification of Diseases (ICD) codes, so presumably these were entered by humans who read the cause of death and designated an ICD code. The error rate of transposing the cause of death into ICD codes remains unstudied, although >90% sensitivity and specificity has been reported for the accuracy of discharge diagnosis ICD coding for pulmonary embolism in hospital medical records compared with a gold standard of chart review by members of a research team (6). To our knowledge, no such decedent-level review has been published to address the accuracy of WONDER for cause of death. In addition, no quantitative analysis was done to examine for the effect of clustering by regional differences and possible reporting biases. For example, if the frequency of death from pulmonary embolism were 0.01% in one city, but 1% in another city, this would immediately raise questions about reporting bias. In addition, in large cities, pulmonary embolism as cause of death may more likely be based on imaging or autopsy evidence, whereas in small towns, this may be determined by an elected coroner, who may not have any medical training. Thus, in the interpretation that follows, we must assume with caution that the magnitude and effect of reporting biases are evenly distributed by region, race or ethnicity, sex, and age.
With that caveat, the study reports age-adjusted mortality data from 2006 to 2019 in Figures 1–5 (2). First, it demonstrates that overall age-adjusted mortality for pulmonary embolism has not changed at all during this time period. This finding agrees with some, but not all, precedent literature in the United States, perhaps reflecting the biases of different data sources (7, 8). Notably, this finding contrasts with data from Europe, which have shown a steady decrease in mortality over the past decade (9). Although the rate of pulmonary vascular imaging, the use of catheter-based interventions, and dissemination of PERTs has clearly increased during this time period, the overall lack of improvement in age-adjusted mortality underscores the question of whether these perceived advancements have made any impact on mortality.
Equal in both impact and concern, this study shows a higher mortality for Black Americans compared with White Americans, reinforcing prior findings (8). Prior work using the National Inpatient Sample as well as Medicare databases has suggested that persons of color and socioeconomically disadvantaged or uninsured Americans are less likely to be treated with advanced therapies for pulmonary embolism (10–12). Although racial disparities are generally well recognized in interhospital transfer, to our knowledge, no specific literature has described whether Black Americans lack access or are less likely to be transferred for more of these advanced therapies for pulmonary embolism.
In addition, the data also demonstrate a wide and concerning gap in care for persons who live in rural America. The mortality rate in rural areas was found to be almost twice as high as metropolitan areas (age-adjusted mortality of 4.07 [4.02–4.12] vs. 2.32 [2.30–2.34]). We suspect this is due to delays in diagnosis related to longer distances patients need to travel for medical care and likely fewer resources for diagnostic and therapeutic interventions in rural center. Sixty million Americans (one in five) live in federally designated rural areas (13). The Federal requirement for this rural designation is remarkably austere. An exceedingly small percentage of these 60 million Americans has a significant chance of receiving care from a center with a PERT or access to other advanced therapies.
Last, this study demonstrated significant differences in mortality based on geographical region. Age-adjusted mortality was higher in the South at 3.53 (3.5–3.57) and Midwest at 3.28 (3.25–3.32) compared with the Northeast at 2.29 (2.26–2.33) and the West at 1.86 (1.83–1.89). Mortality being the highest in the South aligns with literature prior to 2006 (14).
However, although each of these findings alone is significant, because WONDER does not have patient-level data, multivariable analysis is not able to be performed, which precludes any inferences about what might be driving mortality rates (15). For example, according to the Pew Research Center, the majority of the U.S. Black population lives in the South (56%), followed by the Northeast and Midwest (both 17%), and lastly the West (10%) (16). Thus, is the higher mortality in the Black population driven by differences in practice patterns in the management of pulmonary embolism in the South versus other regions in the country? Or are there factors related to the racial disparities faced by the Black population, including socioeconomic factors or higher burdens of comorbid conditions, driving this mortality? A similar question can be raised about the higher mortality in the Midwest, which has a predominance of rural counties compared with other regions (17).
Therefore, although this study brings to light the wide variability in pulmonary embolism–related mortality across the United States, based on sex, race/ethnicity, type of community, or region, the lack of patient-level data precludes firm conclusions on what steps need to be taken to improve pulmonary embolism–related mortality. Thus, it highlights the need for committed Federal funding for a comprehensive pulmonary embolism surveillance network in the United States.