Primary Causes of Death in Reported Cases of Fatal West Nile Fever,United States,2002–2006

Introduction

F atal human cases of West Nile virus (WNV) disease without neurologic involvement have rarely been reported (Klingberg et al. 1959, McIntosh et al. 1976, Campbell et al. 2002). For the first 3 years after the appearance of WNV in North America (1999–2001), all 19 fatal WNV disease cases reported to the Centers for Disease Control and Prevention (CDC) were cases of neurologic illness (West Nile neuroinvasive disease [WNND]) (Marfin et al. 2001, Nash et al. 2001, O'Leary et al. 2004). Beginning in 2002, however, the explosive expansion of the geographic range of WNV, the annual occurrence of regional epidemics, and the increased availability of commercial serologic tests for WNV infection resulted in more frequent identification of West Nile fever (WNF) cases. This suggested that WNF was a significant public health problem in the United States. As the number of WNV disease cases increased, small numbers of fatal WNF cases began to be reported, especially among persons of advanced age. Many early and some more recent descriptions of human illness from infection with WNV have used the term “West Nile fever” to refer to any clinical illness associated with infection (Nisenbaum and Wallis 1965, Schmidt 1965, Berner et al. 2002); however, this article will use the term WNF only to refer to febrile illness without apparent neurologic disease. The mechanisms by which this normally uncomplicated and self-limited illness results in death are unknown, but a better understanding of the relationship between WNF and death could identify risk factors, guide secondary prevention measures, and potentially decrease mortality. We report findings of a retrospective study of primary causes of death in fatal WNF cases reported to CDC from 2002 through 2006.

Materials and Methods

Human WNV disease cases are reported to CDC through ArboNET, an Internet-based passive surveillance system managed collaboratively by state health departments and CDC. Cases are reported by clinical syndrome categories, which include “encephalitis,” “meningoencephalitis,” “meningitis,” “uncomplicated fever,” “other clinical illness,” “unspecified illness,” and/or “acute flaccid paralysis.” All fatal cases initially reported to ArboNET as “uncomplicated fever” from 2002 through 2006 were included in this study. All cases met reporting criteria to ArboNET, which includes laboratory evidence of acute WNV infection. We requested state and local health departments to provide available medical records, death certificates, hospital discharge summaries, and other relevant clinical information for each WNF-associated death reported from their jurisdiction. We first reviewed death certificates and hospital discharge summaries. Cases for which the death certificate or hospital discharge summary indicated a diagnosis of WNND (“encephalitis,” “meningoencephalitis,” “meningitis,” or “acute flaccid paralysis”) were presumed to be misclassified in the ArboNET system and were not evaluated further. In remaining cases, the primary or proximate cause of death, as well as demographic information and underlying or comorbid medical conditions, was ascertained by review of medical records, death certificate information, autopsy reports, or other available clinical information.

Results

Between 2002 and 2006, a total of 13,482 cases of WNF were reported to ArboNET nationally; the median age among all cases was 47 (range: 1 month–96 year) (O'Leary et al. 2004, CDC 2006). A total of 43 of these cases (0.3%) were reported as fatal. Medical data were available for 35 (81%) cases. Fifteen (43%) of the 35 reviewed fatal cases were in fact WNND cases that were misclassified as WNF and should not have been reported as “uncomplicated fever” in ArboNET, and 1 death was thought to be medically unrelated to WNV infection (a fatal bicycle accident); 23 were considered to be potential WNF-associated fatalities. The median age of the 23 case-patients was 78 years (range: 54–92 years) and 18 (78%) were ≥70 years old; 14 (64%) were men. Twenty case-patients were white, 1 was Hispanic, and 1 was black; race/ethnicity was unknown in 1 case. The median interval from WNV illness onset to death was 15 days (range: 2–131 days). In 12 cases, the interval between WNV illness onset and death exceeded 14 days.

All 23 of the WNF fatalities were hospitalized. Among the 23 WNF fatalities, the primary cause of death was a cardiac condition (8 cases; 35%), including myocardial infarction, sudden cardiac death, ischemic cardiomyopathy, and cardiopulmonary arrest. Pulmonary conditions (i.e., respiratory failure and pneumonia) were also significant primary causes of death (6 cases; 25%). Other primary causes of death in this group were malignancy (3 cases), multisystem organ failure (2 cases), and stroke (1 case). In three cases, no other primary cause of death other than WNF was identified; in two of these cases, underlying medical conditions were unknown (Table 1). Overall, information on underlying or comorbid medical conditions was available for 17 (74%) of 23 WNF-associated fatalities; the most frequent comorbid conditions included cardiovascular disease (13; 76%), hypertension (8; 47%), and diabetes mellitus (6; 35%). Autopsies were performed in 5 (22%) of the 23 fatal WNF cases, substantiating the cause of death; in none of these were WNV-specific tests, such as serology or polymerase chain reaction testing for viral-specific nucleic acid, performed on autopsy material, blood, or cerebrospinal fluid.

Discussion

WNF has generally been associated with a benign course and favorable outcome. However, there is increasing data to suggest that WNF may represent a more considerable source of morbidity than initially thought. An assessment of patients hospitalized with WNV illness in Chicago found that 30% were found to have WNF (Watson et al. 2004) with 63% continuing to experience persistent symptoms such as headache and fatigue at least 30 days after illness onset. Similarly, nearly one-fourth of 221 patients hospitalized in a four-county area of Colorado in 2003 had WNF (Bode et al. 2006). An assessment of 32 patients aged 65 and older hospitalized with WNF during an outbreak in Israel suggested a fatality rate of 22% (Berner et al. 2002); however, the frequent presence of altered mental status and other neurologic findings among this cohort suggest that many of these cases were actually WNND.

Our assessment of WNF-associated fatalities reported to CDC during 2002–2006 suggests that in some individuals, especially persons of advanced age and those with underlying medical conditions, WNF may precipitate death. A substantial proportion of our reviewed cases (43%) were misclassified WNND cases, which is fatal in up to 20% of encephalitis cases (Bode et al. 2006). Among those cases that did not appear to have WNND, the most common primary causes of death were cardiac and pulmonary complications. Both respiratory failure and cardiac arrhythmias have been recognized to be frequent contributors to death in WNV encephalitis (Bode et al. 2006). Although the pathophysiology and the specific role of WNV infection in these WNF-related deaths are not clear, it is probable that the physiological stress of WNV infection precipitated or exacerbated underlying medical conditions resulting in death, such as acute myocardial infarction, sudden death from cardiac arrhythmias, and respiratory failure, in predisposed individuals. In 12 cases, an extensive time interval (exceeding 2 weeks) between onset of WNV-associated illness and death suggests that WNV infection and death were unrelated. In three cases, no other cause of death except for WNF itself was identified. In one of these, significant underlying medical conditions suggest that WNF may have precipitated death from a cardiac or pulmonary etiology; in two cases, however, underlying medical conditions were unknown, making it impossible to determine what role WNV infection had in death. Further postmortem diagnostic and virologic studies to more clearly define involvement of WNV in organs other than the central nervous system, including heart and lung, may provide additional information on the direct role, if any, of WNV in deaths from nonneuroinvasive disease.

To our knowledge, this study represents the first systematic assessment of primary causes of death in a series of patients with WNF. Despite the study's limitations (e.g., identification of cases through passive surveillance and absence of autopsy information in all but a few cases), our results suggest that fatalities after WNF are relatively uncommon. In those fatalities that appear to be temporally associated with WNF, death is frequently attributable to cardiac and pulmonary complications, especially in the elderly. In this respect, WNF resembles influenza and other influenza-like illnesses predominantly associated with febrile illness (Hui et al. 2008). While death among persons with WNND has been well recognized for decades, the observation that WNF may directly or indirectly lead to death is a relatively recent observation. The fact that the elderly are at increased risk of death from both WNND and WNF emphasizes the importance of primary prevention of WNV infection (through personal protection from mosquito bites and mosquito control programs), as well as close monitoring for cardiac and pulmonary complications in elderly patients hospitalized for WNV disease.

Disclaimer

The findings and conclusions in this report are those of the authors and do not necessarily represent the views of the Centers for Disease Control and Prevention.