Delayed Mortality in a Cohort of Persons Hospitalized with West Nile Virus Disease in Colorado in 2003

Abstract

Most mortality associated with West Nile virus (WNV) disease occurs during the acute or early convalescent phases of illness. However, some reports suggest mortality may be elevated for months or longer after acute illness. The objective of this study was to assess the survival of a cohort of patients hospitalized with WNV disease in Colorado in 2003 up to 4 years after illness onset. We calculated age-adjusted standardized mortality ratios (SMRs) to evaluate excess mortality, evaluated reported causes of death in those who died, and analyzed potential covariates of delayed mortality. By 1 year after illness onset, 4% of the 201 patients had died (SMR, 2.7; 95% confidence interval [CI], 1.3–5.2), and 12% had died by 4 years after onset (SMR, 2.0; 95% CI, 1.3–3.0). Among those who had died, the most common immediate and contributory causes of death included pulmonary disease and cardiovascular disease; cancer, hepatic disease, and renal disease were mentioned less frequently. In multivariate analysis, age (hazard ratio [HR], 2.0 per 10-year increase; 95% CI, 1.4–2.7), autoimmune disease (HR, 3.0; 95% CI, 1.1–7.9), ever-use of tobacco (HR, 3.0; 95% CI, 1.3–7.0), encephalitis during acute WNV illness (HR, 2.6; 95% CI, 1.1–6.4), and endotracheal intubation during acute illness (HR 4.8; 95% CI, 1.9–12.1) were found to be independently associated with mortality. Our finding of an approximate twofold increase in mortality for up to 3 years after acute illness reinforces the need for prevention measures against WNV infection among at-risk groups to reduce acute as well as longer-term adverse outcomes.

Introduction

W est Nile virus (WNV), a mosquito-borne flavivirus, was first detected in the United States in 1999 and has since become the leading cause of arboviral encephalitis in the United States (Nash et al. 2001, Campbell et al. 2002, Reimann et al. 2008). Mortality from WNV disease is mainly limited to patients with West Nile neuroinvasive disease (WNND), which includes encephalitis, aseptic meningitis, poliomyelitis, and other less common neurologic syndromes (Nash et al. 2001, O'Leary et al. 2004, Lindsey et al. 2010). Among persons hospitalized for WNV disease, case-fatality rates (CFR) are typically 4%–16% overall, 12%–24% in encephalitis cases, and increase with age (Tsai et al. 1998, Chowers et al. 2001, Nash et al. 2001, Pepperell et al. 2003, Bode et al. 2006, Murray et al. 2006). Most reported WNV-associated mortality has occurred during the acute or early convalescence phases of illness. However, a few reports suggest that mortality after WNV disease may continue to be elevated for at least a year after acute illness. A study of 56 WNND cases acquired in Tennessee in 2002 reported a CFR of 19% in early illness and 25% at 1 year postonset (Gottfried et al. 2005). Among 246 adult Israeli patients with WNV disease who survived hospitalization and at least 1 week postdischarge, the overall age-standardized mortality ratio (SMR) was 2.5 (95% confidence interval [CI], 1.5–3.9) after 1 year postonset and 1.7 (not statistically significant) after 2 years (Green et al. 2005). Men had even higher mortality (SMR, 3.2; 95% CI, 1.6–5.7). Other patient characteristics associated with mortality included advancing age, diabetes mellitus, and history of dementia. No studies have examined mortality among patients with WNV beyond 2 years after illness onset.

The Centers for Disease Control and Prevention (CDC) and the Colorado Department of Public Health and Environment (CDPHE) conducted a retrospective study of 221 patients with WNV disease hospitalized in northern Colorado in 2003 (Bode et al. 2006). Twelve patients (all patients with WNV encephalitis aged >50 years) died during the acute phase of illness, for an overall CFR of 5% and a WNV encephalitis-specific CFR of 18%. In multivariate analysis, advancing age, endotracheal intubation during hospitalization for WNV disease, history of stroke, and immune suppression were independently associated with death. The objectives of the current study were to assess the survival of the remaining members of this cohort up to 4 years after illness onset, calculate SMRs, evaluate reported causes of death in those who died, and analyze potential covariates of delayed mortality.

Materials and Methods

The study population was limited to the 221 patients who had been hospitalized and who were included in the original cohort study (Bode et al. 2006). We excluded the 12 patients who had died during hospitalization and the one patient who had died within a month after illness onset; we also excluded seven children aged <18 years, for comparability with previous reports. Two hundred one survivors of acute-phase WNV illness were eligible for follow-up. Demographic information, medical history, date of illness onset, and signs and symptoms during acute WNV illness data were obtained from CDPHE records. We used patient names and dates of birth to search the CDPHE Vital Records database and the National Death Index to determine how many of the 201 initial survivors were issued death certificates as of December 31, 2007. In addition, to confirm vital status of those persons not identified in the death certificate searches, we attempted to contact by mail the physician of record. Although every patient in the study had a physician listed in CDPHE records, they were often hospital or emergency room physicians (not the patient's primary care physician) or were no longer the patient's primary care physician. Therefore, the physicians of record were only able to provide information about current vital status for 72 of the 201 patients in the study. Persons not identified in death certificate searches or identified as deceased through other means were, for the purposes of the analysis, assumed to be alive. For deceased individuals, cause-of-death and other death-related information were gathered from available medical records and death certificates. This study was reviewed and approved by the human subjects review board of the CDPHE.

Survival time was measured from the date of acute WNV disease onset to the date of death, regardless of cause, censored at 4 years (48 months). A Kaplan–Meier survival curve was constructed for all patients, and age-adjusted SMRs were computed by using the person-time method. Person-months of risk were calculated from the reported onset of the acute illness to either death or 48 months (4 years) after onset. The expected number of deaths in the cohort was calculated by multiplying person-months at risk by the national death rates in 2005 (the last year for which published data were available) for the same age and sex categories.

Potential risk factors were assessed by using univariate and multivariate Cox proportional hazards models (SAS version 9.2, SAS Institute, Cary, NC). Twenty-one independent variables were considered, including demographic characteristics, underlying medical conditions, and signs/symptoms of acute WNV disease. All variables were evaluated as measured at the time of the initial hospitalization. Underlying medical conditions were defined as described in the original cohort study (Bode et al. 2006). Age was evaluated as a continuous variable; all other variables were dichotomous. Some variables that could potentially be important risk factors were not included in the multivariate modeling process, because ≤5 persons had them, including history of intravenous drug use, history of solid organ transplant, renal dialysis during acute WNV illness, and coma during acute WNV illness. The multivariate model was built by using a manual backward selection approach starting with all variables significant at p<0.2 in the univariate analysis. Variables significant at p<0.05 were retained in the multivariate model. Although not significant at p<0.2 in univariate analysis, sex and history of cancer were evaluated in the multivariate model, because they had been found to be important in previous studies of WNV outcomes; since none of these variables were significant in the multivariate model or confounded any of the associations between any of the other independent variables and survival time, they were not retained. Interactions between age and other variables included in the final main-effects model were not significant and, therefore, were not retained in the model. To test the validity of the proportional hazards assumption, the Kolmogorov-type supremum test as implemented in the PROC PHREG procedure of the SAS package was performed; none of the terms included in the final model violated the proportional hazards assumption.

Results

Among the 201 adult survivors of the acute-phase WNV illness, 24 (12%) patients died between 1 month and 4 years after illness onset, including 8 (4%) within the first year and 16 (8%) in the subsequent 3 years. All deaths were identified through death record searches. On average, the 24 deceased patients were older (mean 67 years) than those who survived (mean 51 years; Table 1). Although the proportion of patients classified as having neuroinvasive disease were similar among deceased and surviving patients, encephalitis was more common among deceased patients (63%) than among surviving patients (21%). A larger proportion of deceased patients (38%) required endotracheal intubation (i.e., ventilation) during their hospitalization for acute WNV disease than patients who survived (6%).

Table 1.

Characteristics of 201 Patients Aged ≥18 Years Surviving Initial Hospitalization with West Nile Virus Disease in 2003 by Vital Status by 4 Years After Illness Onset

Patient characteristic	Alive (n=177) No. (%)	Deceased (n=24) No. (%)	p-value
Age, mean years (range)	51 (19–86)	67 (44–87)	<0.01^a
Male sex	105 (59)	13 (54)	0.63^b
Neuroinvasive disease^c	136 (77)	18 (75)	0.84^b
Encephalitis	37 (21)	15 (63)	<0.01^b
Endotracheal intubation during acute hospitalization	11 (6)	9 (38)	<0.01^d

t-test.

Chi-square.

Includes patients with encephalitis, meningitis, and/or poliomyelitis.

Fisher's exact test.

The 24 deceased patients included 15 patients with (63%) encephalitis, 1 (4%) with meningitis, 2 (8%) with poliomyelitis, and 6 (33%) with non-neuroinvasive disease. Six patients classified as having had encephalitis and the one classified as having had meningitis were also classified as having had poliomyelitis (Table 2). The median interval from WNV illness onset to death was 732 days (range 38–1443 days). The most common immediate and contributory causes of death included pulmonary disease (pneumonia, chronic airway disease, and other causes of respiratory failure) and cardiovascular disease (congestive heart failure, coronary artery disease, and hypertension). Cancer, hepatic disease, and renal disease were mentioned less frequently (Table 2). Among the eight deaths that occurred within 1 year of illness onset, seven had WNV identified as an immediate or contributing cause of death on the death certificate. None of the deaths occurring more than a year after illness onset had WNV specifically identified as a cause of death.

Table 2.

Immediate and Contributory Causes of Death Among 24 Patients Surviving Initial Hospitalization with West Nile Virus Disease in 2003

						Cause(s) of death (as recorded on death certificate)
Patient No.	Acute WNV disease syndrome	Mechanical ventilation during acute hospitalization for WNV disease	Interval from onset to death (days)	Sex	Age (years)	Immediate	Contributory	Documented underlying conditions
1	WNE	Yes	38	M	75	Respiratory failure	WNE	Graves' disease, rheumatoid arthritis
2	WNE (+WNP)	No	41	F	87	Aspiration pneumonia	WNE
3	WNE	No	52	M	77	Respiratory failure	WNE	Nephrolithiasis with stent
4	WNE	Yes	88	F	63	Pneumonia	Chronic renal failure	Diabetes
5	WNE (+WNP)	Yes	137	F	76	Pneumonia	WNE	COPD
6	WNE (+WNP)	Yes	175	M	59	Ischemic encephalopathy	WNE, aspiration, COPD	Tracheostomy, alcoholism, current smoker
7	WNE	Yes	203	M	61	Respiratory failure	Quadriplegia, WNE	Gastrocolic-cutaneous fistula
8	WNM (+WNP)	Yes	207	M	52	West Nile virus disease	Respiratory failure, pneumothorax	Liver transplant for hepatitis C (mid-1990s); renal insufficiency
9	WNF	No	505	M	45	Encephalopathy	Cirrhosis, hepatitis C
10	WNF	No	542	F	89	Aspiration pneumonia	Cerebrovascular accident, atrial fibrillation
11	WNF	No	562	F	45	Multi-organ-system failure	End-stage liver disease	Alcoholism, celiac sprue, current smoker
12	WNE (+WNP)	No	692	M	46	Hepatitis	Esophageal cancer, renal failure	Alcoholism, hepatitis C
13	WNP	Yes	772	F	87	Exacerbation of ulcerative colitis		Ulcerative colitis with colostomy
14	WNF	No	864	F	64	Congestive heart failure	Renal failure	Diabetes, current smoker
15	WNE	Yes	902	M	58	Sudden cardiac death	Coronary artery disease, pneumonia	Alcoholism, rheumatoid arthritis, current smoker
16	WNE	No	910	M	71	COPD	Ischemic cardiomyopathy	Diabetes, hypertension, asthma
17	WNE (+WNP)	Yes	911	F	70	Chronic renal failure	Diffuse atherosclerosis	Diabetes, cerebrovascular disease
18	WNE (+WNP)	No	917	M	89	Community-acquired pneumonia		Diabetes, hypertension, previous CVA
19	WNP	No	972	M	85	End-stage dementia with dysphagia	Coronary artery disease, pneumonia, hypertension	Alcoholism, diabetes, atrial fibrillation
20	WNF	No	1,166	M	70	Lung cancer	Hypertension	Current smoker, past history of lymphoma
21	WNE	No	1,277	M	68	Pneumonia	COPD, diabetes, hypertension	Alcoholism
22	WNE	No	1,298	F	90	Adenocarcinoma	Pulmonary hypertension; chronic lymphocytic leukemia	Atrial fibrillation
23	WNE	No	1,387	F	63	COPD		Chronic prednisone therapy, smoker
24	WNF	No	1,443	F	72	Congestive heart failure	Rheumatoid arthritis, hypothyroidism

WNE, West Nile encephalitis; WNP, West Nile poliomyelitis; COPD, chronic obstructive pulmonary disease; WNM, West Nile meningitis; WNF, West Nile fever; WNV, West Nile virus.

Including the 13 deaths that had occurred during the acute phase of illness, 37 deaths occurred in the complete original cohort of 221 persons within 4 years of illness onset, for a CFR of 10% at 1 year and 17% at 4 years. Among the 65 patients with encephalitis in the original cohort of 221, the CFR was 31% at 1 year and 43% at 4 years.

The Kaplan–Meier survival curve for the 201 acute-phase survivors is shown in the Figure. Age- and sex-adjusted SMRs are shown in Table 3. Overall, 4% of the cohort had died by 1 year after illness onset (SMR, 2.7; 95% CI, 1.2–5.1). The excess risk by 2 years after onset was lower and was not statistically significant. The SMR by 3 years after onset was nearly 2.5 times higher than expected (SMR, 2.4; 95% CI, 1.1–4.8); by 4 years after onset, the excess risk was again not significantly higher than expected. Cumulatively, within the 4-year period, 12% of the cohort died (SMR, 2.0; 95% CI, 1.3–3.0). By 1 year after onset, SMRs for men (2.8; 95% CI, 1.0–6.2) and women (2.7; 95% CI, 0.7–7.4) were similar; by 4 years after onset, it was 1.7 for men (95% CI, 1.0–2.9) and 2.6 for women (95% CI, 1.3–4.4).

FIG. 1.

Kaplan–Meier survival curve for 4-year follow-up of 201 patients surviving initial hospitalization with West Nile virus disease.

Table 3.

Sex- and Age-Adjusted Standardized Mortality Ratios for 201 Patients Aged ≥18 Years Surviving Initial Hospitalization with West Nile Virus Disease

Follow-up period	No.	Observed deaths	Expected deaths	SMR	95% CI
First year	201	8	3.0	2.7	1.2–5.1
Second year	193	4	3.1	1.3	0.4–3.1
Third year	189	7	2.9	2.4	1.1–4.8
Fourth year	182	5	3.0	1.7	0.6–3.7
All years combined	201	24	11.9	2.0	1.3–3.0

SMR, standardized mortality ratio; CI, confidence interval.

In a univariate analysis, age, cardiovascular disease, hepatic disease, immunodeficiency, autoimmune disease, solid organ transplant, tobacco use, alcoholism, intravenous drug use, encephalitis during acute WNV illness, weakness during acute WNV illness, and intubation during acute WNV illness were associated with mortality (Table 4). In multivariate analysis, age, autoimmune disease, ever use of tobacco, encephalitis during acute WNV illness, and endotracheal intubation during acute WNV illness were associated with mortality.

Table 4.

Cox Regression Analysis of Factors Associated with Time to Death Among 201 Patients ≥18 Years Surviving Initial Hospitalization with West Nile Virus Disease

Variable ^a	Unadjusted hazard ratio (95% CI)	p-value	Adjusted hazard ratio (95% CI)	p-value
Age (per 10-year increase)	1.8 (1.4–2.4)	<0.01	2.0 (1.4–2.7)	<0.001
Cardiovascular disease	3.2 (1.4–7.4)	<0.01	—	—
Diabetes	2.0 (0.8–5.1)	0.13	—	—
Hepatic disease	4.0 (1.2–13.5)	0.02	—	—
Immunodeficiency	4.3 (1.6–11.5)	<0.01	—	—
Solid organ transplant^b	6.2 (0.8–46.3)	0.07	—	—
Autoimmune disease	2.9 (1.2–7.3)	0.02	3.0 (1.1–7.9)	0.03
Ever-use of tobacco	2.8 (1.2–6.4)	0.02	3.0 (1.3–7.0)	0.01
Alcoholism	3.3 (1.3–8.3)	0.01	—	—
Intravenous drug use^b	4.1 (1.0–17.6)	0.06	—	—
Encephalitis during acute WNV illness	5.5 (2.4–12.5)	<0.01	2.6 (1.1–6.4)	0.03
Weakness during acute WNV illness	2.0 (0.9–4.7)	0.11	—	—
Intubation during acute WNV illness	7.4 (3.2–17.0)	<0.01	4.8 (1.9–12.1)	<0.001

Only variables with a p-value<0.2 in the univariate analysis are listed in this table.

Not included in multivariate modeling process.

Discussion

The results of this 4-year study of 201 adult patients who survived hospitalization for acute WNV disease in Colorado confirm those of the 2-year study of 246 patients in Israel by finding a two- to threefold increase in age-adjusted, delayed mortality at 1 year post-illness onset and a small, but not statistically significant, increase in cumulative mortality by 2 years after illness onset (Green et al. 2005). Our findings are also comparable to those of the 1-year follow-up study of 39 patients with WNND surviving acute WNV hospitalization in Tennessee that reported a CFR of 8% by 1 year after onset; SMRs were not reported (Gottfried et al. 2005). Our results, however, extend the observations of previous studies to suggest that higher-than-expected mortality might persist for at least 3 years post-illness onset. In our study, mortality was about 2.5 times higher than expected among both men and women by 1 year after illness onset. However, by 4 years after onset, the SMR was higher for women (2.6) than for men (1.7). In the previous study of Israeli patients, mortality by 1 year after illness onset was more than thrice higher than expected among men, but was less than twice as high among women and not significant (Green et al. 2005). The potential reasons for the differences between the findings of the two studies are unknown. Of the 24 deceased patients, WNV was specifically listed as an immediate or contributory cause of death on 7 of the death certificates, all of which were deaths occurring within 1 year of acute WNV illness onset. Based on the data available, it is unclear how many (if any) of the other 17 deaths may have been related to the patient's previous WNV disease. These findings should be interpreted in light of the inherent limitations of information recorded on death certificates (e.g., coding errors, inadvertent omissions, unavailability of medical records to the certifying physician). The most commonly listed immediate and contributory causes of death, pulmonary and cardiovascular disease are among the most common causes of death in the general population.

In multivariate analysis, our study identified the following five covariates of delayed mortality: increasing age, encephalitis, endotracheal intubation (i.e., assisted ventilation) during hospitalization for acute WNV disease, autoimmune disease (mainly, rheumatoid arthritis), and ever-use of tobacco. Using similar methods, Green and colleagues (2005) identified age, diabetes, and dementia as covariates of delayed mortality. Although none of these observations is unexpected, particularly that increasing age, having had encephalitis (a more severe manifestation of WNND), or having required assisted ventilation during the acute illness is associated with mortality, additional studies would be useful to better understand the role of pre-existing medical conditions in long-term outcomes of WNV disease. Additionally, the impact of longer-term convalescent issues that may accompany WNND, such as prolonged mechanical ventilation in WNV poliomyelitis, should be more fully explored.

Our study is limited in several ways. It is possible that the patients in our cohort differ from the age-adjusted general population in the prevalence of underlying medical conditions or other factors that were not controlled for in the analysis. Since people with underlying illnesses are more likely to have severe WNV disease and require hospitalization at the time of acute infection, we could expect that the patients in our study have a higher prevalence of underlying medical conditions than the general population. If this were the case, then the calculated SMRs (which were based on age-adjusted death rates in the general population) would be overestimated. It is possible that if the patients included in this study were compared with a group with similar medical baselines, then WNV mortality would not be significantly elevated, particularly after the first year. The small numbers of patients in our cohort with certain potentially important underlying conditions (e.g., solid organ transplants) limits the power of our study to show an association between those conditions and delayed mortality after acute WNV disease. Although we used several methods to identify decedents among this cohort, we cannot exclude the possibility of incomplete ascertainment, which may have resulted in an underestimation of SMRs.

In summary, this study substantiates earlier findings of increased cumulative mortality in persons developing WNV disease in the years after acute illness. A two- to threefold increase in mortality may be observed up to 3 years after acute illness, and possibly longer. These findings reinforce the need for prevention measures against WNV infection among at-risk groups, in an effort to diminish acute as well as longer-term adverse outcomes.

Footnotes

Acknowledgments

The authors thank the following for their contributions to this study: Melina Evdemon-Hogan, CDPHE for surveying providers of record for cohort members; Janell Bezdek and Claire Reisenauer, CDPHE, for helping abstract information from medical records for identified fatalities; and Mark Delorey, Division of Vector-Borne Diseases, CDC, for statistical advice.

Disclaimer

The findings and conclusions in this report are those of the authors and do not necessarily represent the views of the funding agency.

Disclosure Statement

No competing financial interests exist.

References

Bode

, Sejvar

, Pape

, Campbell

, Marfin

. West Nile virus disease: a descriptive study of 228 patients hospitalized in a 4-county region of Colorado in 2003. Clin Infect Dis, 2006; 42:1234–1240.

Campbell

, Marfin

, Lanciotti

, Gubler

. West Nile virus. Lancet Infect Dis, 2002; 2:519–529.

Chowers

, Lang

, Nassar

et al. Clinical characteristics of the West Nile fever outbreak, Israel, 2000. Emerg Infect Dis, 2001; 7:675–678.

Gottfried

, Quinn

, Jones

. Clinical description and follow-up investigation of human West Nile virus cases. South Med J, 2005; 98:603–606.

Green

, Weinberger

, Ben-Ezer

, Bin

et al. Long-term death rates, West Nile virus epidemic, Israel, 2000. Emerg Infect Dis, 2005; 11:1754–1757.

Lindsey

, Staples

, Lehman

, Fischer

. Surveillance for West Nile virus disease—United States 1999–2008. MMWR Surveill Summ, 2010; 59:1–17.

Murray

, Baraniuk

, Resnick

, Arafat

et al. Risk factors for encephalitis and death from West Nile virus infection. Epidemiol Infect, 2006; 134:1325–1332.

Nash

, Mostashari

, Fine

, Miller

et al. The outbreak of West Nile virus infection in the New York City area in 1999. N Engl J Med, 2001; 344:1807–1814.

O'Leary

, Marfin

, Montgomery

, Kipp

et al. The epidemic of West Nile virus in the United States, 2002. Vector Borne Zoonot Dis, 2004; 4:61–70.

10.

Pepperell

, Rau

, Krajden

, Kern

et al. West Nile virus infection in 2002: morbidity and mortality among patients admitted to hospital in southcentral Ontario. CMAJ, 2003; 168:1399–1405.

11.

Reimann

, Hayes

, DiGuiseppi

, Hoffman

et al. Epidemiology of neuroinvasive arboviral disease—United States, 1999–2007. Am J Trop Med Hyg, 2008; 79:974–979.

12.

Tsai

, Popovici

, Cernescu

, Campbell

, Nedelcu

. for the Investigative Team. West Nile encephalitis epidemic in southeastern Romania. Lancet, 1998; 352:767–771.