The epidemiology and early clinical features of West Nile virus infection

Jacek M. Mazurek MD, PhD^a,b,*,1, Kim Winpisinger MS^b, Barbara J. Mattson MAEd, MSTE^b, Rosemary Duffy DDS, MPH^c,

Ronald L. Moolenaar MD, MPH^d

^aEpidemic Intelligence Service, Epidemiology Program Office, Centers for Disease Control and Prevention, Atlanta,

GA 30333, USA

^bOhio Department of Health, Colombus, OH 43215, USA

^cNational Center for Chronic Diseases Prevention and Health Promotion, Centers for Disease Control and Prevention,

Atlanta, GA 30333, USA

^dEpidemiology Program Office, Centers for Disease Control and Prevention, Atlanta, GA 30333, USA

Received 9 November 2004; accepted 28 November 2004

Abstract We studied early clinical features of the West Nile virus (WNV) infection. Case patients were Ohio residents who reported to the Ohio Department of Health from August 14 to December 31, 2002, with a positive serum or cerebrospinal fluid for anti-WNV IgM. Of 441 WNV cases, medical records of 224 (85.5%) hospitalized patients were available for review. Most frequent symptoms were fever at a temperature of 38.08C or higher (n = 155; 69.2%), headache (n = 114; 50.9%), and Mental status changes (n = 113; 50.4%). At least one neurological symptom, one gastrointestinal symptom, and one respiratory symptom was present in 186 (83.0%), 119 (53.1%), and 46 (20.5%) patients, respectively. Using multivariate logistic regression and controlling for age, we found that the Initial diagnosis of encephalitis ( P = .001) or reporting abdominal pain ( P b .001) was associated with death. Because initial symptoms of WNV infection are not specific, physicians should maintain a high index of suspicion during the epidemic season, particularly in elderly patients with compatible symptoms.

This is a US government work. There are no restrictions on its use.

T Corresponding author. NIOSH, DRDS, Surveillance Branch Mailstop HG 900.2 Morgantown, WV 26505, USA. Tel.: +1 304 285 5983; fax: +1

304 285 6111.

E-mail address: [email protected] (J.M. Mazurek).

¹ Current address: National Institute for Occupational Safety and Health, Centers for Disease Control and Prevention, Morgantown, WV 26505, USA.

Introduction

The West Nile virus (WNV) was first identified in 1937 in the West Nile district of Uganda [1]. Initially, WNV infections were characterized by mild disease with low- grade fever, malaise, anorexia, nausea and vomiting, eye pain, headache, myalgia, rash, and lymphadenopathy [2- 4]. In recent years, an increase in the frequency and clinical severity of WNV infection has been observed [5-9].

0735-6757/$ – see front matter. doi:10.1016/j.ajem.2004.11.005

case definition“>The first case of WNV in North America occurred in the United States in New York City in 1999 [7]. From that year to the present, WNV has spread rapidly westward to most states [7-9]. In Ohio (2000 population, 11.3 million), the first WNV activity was observed in 2001 in crows and blue jays. In 2002, all of Ohio’s 88 counties reported WNV activity in birds, mosquitoes, horses, or human beings. The first bird tested positive on May 19, the first mosquito pool on May 28, and the first horse on June 6. The first infected human being in Ohio became ill on July 27 and was reported on August 14. A total of 441 WNV human cases was reported to the Ohio Department of Health (ODH) for 2002.

In Ohio, WNV encephalitis is a reportable disease [10]. Local health departments report cases to the ODH. In 2002, serum or cerebrospinal fluid (CSF) samples taken from hospitalized patients with suspected WNV encephalitis or meningitis were sent to the ODH laboratory for anti-WNV IgM testing by enzyme-linked immunosorbent assay (ELISA). The ODH laboratory provided free arboviral testing for WNV, St Louis encephalitis (SLE), and La Crosse encephalitis for patients hospitalized with a diagno- sis of viral encephalitis, viral meningoencephalitis, or viral (aseptic) meningitis. Patients with mild illness or those who were exposed but remained asymptomatic were referred to private laboratories for testing. Laboratories that identify WNV infection are required to report it to the ODH laboratory and to the local health department within whose jurisdiction an infected patient resides. Human WNV infections were deemed confirmed at the ODH laboratory by ELISA testing on CSF when the anti-WNV IgM result was positive and the anti-SLE IgM result was negative [11]. In addition, some cases were confirmed at the Centers for Disease Control and Prevention (CDC) by plaque reduction neutralization test (PRNT) [12]. The ODH requested this additional testing in the initial phase of the WNV outbreak, when it was thought that the patient might have been exposed to other closely related viruses (eg, SLE virus), which might result in a false-positive laboratory test for WNV. The CDC reported PRNT test results to the ODH.

The 1999 and 2000 outbreaks in New York and New Jersey and the WNV surveillance system implemented in conjunction with the CDC provided an opportunity to describe in detail the clinical characteristics of hospitalized human cases [13-15]. However, the frequency of various signs and symptoms associated with early WNV infection has not been specifically addressed. Because another WNV outbreak is possible, knowledge of early symptoms could enable physicians to more rapidly diagnose patients with WNV infection. This paper describes the epidemiology of WNV in Ohio, the initial signs and symptoms of the WNV infection of patients who presented to Ohio hospital emergency departments (EDs) and were subsequently hospitalized in 2002, initial diagnoses of cases noted within 24 hours after hospital admission, and the final diagnoses. In addition, we evaluated the association of individual risk factors with death from WNV infection.

Methods

Case definition

We defined a probable human WNV case patient as an Ohio resident with a positive serum or CSF anti-WNV IgM test result who was reported to the ODH between January 1 and December 31, 2002. A confirmed case patient was a probable case patient who tested positive for WNV infection at the CDC by PRNT (with or without serum- or CSF- positive anti-WNV IgM test results) or who had CSF test results that were positive for anti-WNV IgM and negative for anti-SLE IgM by ELISA. Probable and confirmed case patients who presented to Ohio EDs and were subsequently admitted to a hospital were included in the analysis.

Case ascertainment

We reviewed the serological WNV test results and information from WNV surveillance of human beings as reported to the ODH for 2002 and created a list of patients who tested positive for WNV at the ODH, a private, or the CDC laboratory. Sublists of WNV patients were provided to the local health departments in whose jurisdiction the patients resided and to the Infection control practitioners at the hospital where the patients were evaluated. Local health departments and hospital staff ensured that case patient medical charts were available for review.

Twenty-eight abstractors from the ODH, local health departments, and hospital infection control departments reviewed ED medical charts by using a standard data collection form and collected demographic data, clinical data, CSF laboratory results, information on initial diagno- ses, potential risk factors, disposition, and final diagnosis. Abstractors received detailed instruction on how to collect the data. If the desired information was not available, we reviewed the physicians’ and nurses’ notes. Test results other than WNV serology performed on CSF were obtained from the charts or directly from the hospitals’ laboratories.

Statistical analysis

We used Epi Info (CDC, Atlanta, GA) and SAS statistical software (SAS, Cary, NC) to analyze collected data [16,17]. Simple proportions and medians were generated to describe the demographic characteristics and the frequencies of initial diagnoses, signs or symptoms reported on admission, or underlying conditions. We used the Wilcoxon test to compare means of continuous data and the Fisher exact test to determine if any initial diagnosis, sign or symptom reported on admission, or underlying condition was associated with death; a 2-tailed P value less than .05 was accepted as significant. We calculated 95% confidence intervals (CIs) of odds ratios (ORs) using exact method.

A multiple logistic regression analysis was used to evaluate the association of initial diagnosis, signs and symptoms, and underlying conditions reported on admission with death while controlling for age. Variables were

Fig. 1 Human West Nile virus infections reported from Ohio by onset date (n = 441).

considered for inclusion in the multivariate model if they were associated with death at a statistical significance level of P b .25 in the univariate analysis and were reported by more than 5% of the study population. Using all-subsets regression, we then identified several models all of whose variables remained significantly associated with death at a significance level of P = .05.

Results

During the period of January 1 to December 31, 2002, a total of 441 probable cases with positive test results for WNV in CSF or serum (IgM ELISA) was reported to the ODH. Of these, 140 (31.7%) cases were tested and confirmed at the CDC by PRNT or at the ODH laboratory by ELISA CSF testing that was positive for anti-WNV IgM

Table 1 Number and percentage of hospitalized WNV-infected patients by selected demographic characteristics–Ohio, 2002 Characteristics Nonfatalities [n (%)] Fatalities [n (%)] Total [n (%)] Crude OR 95% Confidence limits P

and negative for anti-SLE IgM. Thirty-one (7.0%) cases were fatal. The 441 patients had a median age of 61 years (range, 2-98 years); 227 (51.5%) were women. The median age of the 31 decedents was 75 years (range, 32-98 years);

9 (29.0%) were women. Onset of symptoms of WNV infection among reported patients occurred during the period of July 27 to October 19, 2002 (Fig. 1). Cases occurred among residents of 56 (63.6%) of Ohio’s 88 counties, with 219 (49.4%) cases reported from the most populated county in Ohio, Cuyahoga County (2000 population, 1.3 million). The attack rate during the period of January 1 to December 31, 2002, was 3.9 per 100000 population for the state, compared with 15.6 per 100000 for Cuyahoga County.

Of the 441 WNV patients reported in Ohio, 262 (59.4%) had medical charts available for immediate review. Of these, 224 (85.5%) were hospitalized, 28 (10.7%) were ED-only

patients, and 10 (3.8%) were outpatients.

Of the 224 hospitalized case patients, 113 (50.4%) were women and 179 (79.9%) were white (Table 1). Their ages ranged from 11 to 98 years (median, 67 years). Over half of the patients were aged 65 years or older (n = 129; 57.6%). Fourteen (6.3%) case patients died. The median age of these 14 decedents was 75 years (range, 62-98 years); 7 (50.0%) were women. The highest proportion of deaths (1/3 [33.3%]) among hospitalized patients occurred among persons aged 90 to 99 years (Fig. 2). The median time from onset to death was 13 days (range, 2- 44 days). Case patients were hospitalized in 47 hospitals in 27 counties.

The admission diagnoses of Aseptic meningitis, enceph- alitis, and meningoencephalitis were made in 68 (30.4%),

21 (9.4%), and 9 (4.0%) of the hospitalized patients, respectively (Table 2). Frequent initial diagnoses included bfever of unknown originQ (n = 39; 17.4%), viral infections (n = 39; 17.4%), urinary tract infection (n = 33; 14.7%), and

	n = 210 (93.7%)		n = 14 (6.3%)		n = 224 (100%)		Lower	Upper
Age group (y)
10-19	5 (2.4)		0 (0.0)		5 (2.2)	0.00	0.00	26.67	1.000
20-29	7 (3.3)		0 (0.0)		7 (3.1)	0.00	0.00	18.17	1.000
30-39	14 (6.7)		0 (0.0)		14 (6.3)	0.00	0.00	8.60	1.000
40-49	31 (14.8)		0 (0.0)		31 (13.8)	0.00	0.00	3.78	.511
50-59	25 (11.9)		0 (0.0)		25 (11.2)	0.00	0.00	4.71	.538
60-69	41 (19.5)		2 (14.3)		43 (19.2)	1.00	–	–	Ref.
70-79	58 (27.6)		5 (35.7)		63 (28.1)	1.39	0.19	16.06	1.000
80-89	27 (12.9)		6 (42.9)		33 (14.7)	4.06	0.59	44.92	.118
90-99	2 (1.0)		1 (7.1)		3 (1.3)	9.75	0.11	256.84	.195
Sex
Women	106 (50.5)	7 (50.0)		113 (50.4)		0.98	0.28	3.40	1.000
Men	104 (49.5)	7 (50.0)		111 (49.6)		1.00	–	–	Ref.
Race
White	166 (79.0)	13 (92.9)		179 (79.9)		1.00	–	–	Ref.
Black	30 (14.3)	1 (7.1)		31 (13.8)		0.43	0.01	3.05	.698
Other/Unknown	14 (6.7)	0 (0.0)		14 (6.3)		0.00	0.00	3.41	.604

Fig. 2 Rate of hospitalizations and percentage of deaths among hospitalized patients by age group (n = 224).

dehydration (n = 32; 14.3%). In 12 (5.4%) cases, the admission diagnosis was WNV infection.

The signs and symptoms most frequently recorded at admission are reported in Table 3. At least one neurological symptom (headache, stiff neck, photophobia, muscle weak- ness, paresthesias, seizures, or mental status changes), one gastrointestinal symptom (nausea, vomiting, or diarrhea), and one respiratory symptom (cough or shortness of breath) was present in 186 (83.0%), 119 (53.1%), and 46 (20.5%) patients, respectively. Rarely reported symptoms included slurred speech, sore throat, Blurred vision, lymphadenopa- thy, and paresthesias.

Some of the underlying medical conditions present in case patients at the time of admission are reported in Table 3. Diabetes and hypertension were the most prominent conditions reported, and asthma and history

of cancer were the least reported. None of the patients were reported to be HIV positive or to take immunosup- pressive medications.

In 65 (29.0%) case patients, CSF was collected on or before admission. The findings reflected changes charac- teristic of viral infection: mild leukocytosis (median,

128 cells/mm³; range, 0-725 cells/mm³; normal values [18], 0-5 cells/mm³), high protein (median, 72 mg/dL; range, 23-700 mg/dL; normal values, 15- 45 mg/dL), and normal glucose (median, 61 mg/dL; range, 22-224 mg/dL; normal values, 50-80 mg/dL). Twenty-four (36.9%) patients had a predominance of neutrophils (N50%; normal values, 0%-8%) in the CSF.

The disease onset ranged between July 27 and October 2, 2002. Patients were admitted during the period of August 8 to October 17, 2002. The mean and median times from symptom onset to admission were 4.3 and 3 days, respectively (range, 0-32 days). The mean and median durations of hospitalization were 9.5 and 7 days, respec- tively (range, 1-65 days). There was no difference in time from symptom onset to admission ( P = .242) and duration of hospitalization ( P = .153) between those who subse- quently died and those who survived. Hospitalization longer than 7 days was associated only with hypertension (OR = 2.4; 95% CI, 1.1-4.8; P = .021).

The final diagnoses of the 224 hospitalized patients, as noted in the medical charts at the time of data collection, included viral encephalitis (n = 77; 34.4%), aseptic meningitis (n = 64; 28.6%), WNV infection (n = 55; 24.6%), viral meningoencephalitis (n = 21; 9.4%), others

(n = 6; 2.7%), and Guillain-Barre’ syndrome (n = 1; 0.4%). Hospitalized patients were discharged home (n = 126; 56.3%), to a skilled nursing facility (n = 66; 29.5%), or to the patient caretaker’s home (n = 5; 2.2%). Nine patients

(4.0%) were transferred to another acute care hospital.

Table 2 Admission diagnoses of hospitalized WNV-infected patients Admission diagnosis Nonfatalities [n (%)] Fatalities [n (%)] Total [n (%)] Crude OR 95% Confidence limits P
	n = 210 (93.7%)		n = 14 (6.3%)		n = 224 (100%)		Lower	Upper
Aseptic meningitis	66 (31.4)		2 (14.3)		68 (30.4)	0.36	0.04	1.71	.237
Fever	33 (15.7)		6 (42.9)		39 (17.4)	1.76	0.24	77.60	1.000
Viral infection	38 (18.1)		1 (7.1)		39 (17.4)	0.35	0.01	2.46	.473
Urinary tract infection	31 (14.8)		2 (14.3)		33 (14.7)	0.96	0.10	4.65	1.000
Dehydration	28 (13.3)		4 (28.6)		32 (14.3)	2.60	0.55	9.77	.121
Mental status changes	23 (11.0)		2 (14.3)		25 (11.2)	1.36	0.14	6.69	.660
Pneumonia	21 (10.0)		1 (7.1)		22 (9.8)	0.69	0.02	5.08	1.000
Encephalitis	16 (7.6)		5 (35.7)		21 (9.4)	6.74	1.56	25.42	.005
Sepsis	20 (9.5)		0 (0.0)		20 (8.9)	0.00	0.00	2.45	.620
WNV infection	12 (5.7)		0 (0.0)		12 (5.4)	0.00	0.00	4.44	1.000
Cerebrovascular	9 (4.3)		1 (7.1)		10 (4.5)	1.72	0.04	14.18	.483
accident/stroke
Meningoencephalitis	8 (3.8)		1 (7.1)		9 (4.0)	1.94	0.04	16.47	.447
Cephalgia	9 (4.3)		0 (0.0)		9 (4.0)	0.00	0.00	6.23	1.000
Parkinson’s disease	3 (1.4)		0 (0.0)		3 (1.4)	0.00	0.00	26.78	1.000
Rhabdomyolysis	3 (1.4)		0 (0.0)		3 (1.4)	0.00	0.00	26.78	1.000

Table 3 Clinical characteristics of nonfatal and fatal hospitalized WNV-infected patients by signs, symptoms, and underlying conditions recorded on admission

Characteristic Nonfatalities [n (%)] Fatalities [n (%)] Total [n (%)] Crude OR 95% Confidence limits P

	n = 210 (93.7%)		n = 14 (6.3%)		n = 224 (100%)		Lower	Upper
Signs and symptoms Fever z 388C (z100.48F)	145 (69.0)		10 (71.4)		155 (69.2)	1.12	0.31	5.08	1.000
Headache	113 (53.8)		1 (7.1)		114 (50.9)	0.07	0.00	0.46	.001
Mental status changes	100 (47.6)		13 (92.9)		113 (50.4)	14.30	2.06	613.11	.001
Nausea	76 (36.2)		6 (42.9)		82 (36.6)	1.32	0.36	4.53	.775
Vomiting	73 (34.8)		7 (50.0)		80 (35.7)	1.88	0.54	6.52	.262
Chills	66 (31.4)		6 (42.9)		72 (32.1)	1.64	0.45	5.61	.386
muscle weakness	63 (30.0)		6 (42.9)		69 (30.8)	1.75	0.48	6.01	.372
Confusion	55 (26.2)		3 (21.4)		58 (25.9)	0.77	0.13	3.06	1.000
Fatigue	49 (23.3)		4 (28.6)		53 (23.7)	1.31	0.29	4.81	.745
Lethargy	41 (19.5)		8 (57.1)		49 (21.9)	5.50	1.56	20.15	.003
Decreased appetite	43 (20.5)		5 (35.7)		48 (21.4)	2.16	0.54	7.59	.186
Diarrhea	35 (16.7)		2 (14.3)		37 (16.5)	0.83	0.09	4.00	1.000
Myalgia	33 (15.7)		1 (7.1)		34 (15.2)	0.41	0.01	2.94	.700
Malaise	29 (13.8)		2 (14.3)		31 (13.8)	1.04	0.11	5.05	1.000
Abdominal pain	20 (9.5)		6 (42.9)		28 (12.5)	7.13	1.82	25.85	.002
Neck stiffness	28 (13.3)		0 (0.0)		28 (12.5)	0.00	0.00	1.65	.226
skin rash	27 (12.9)		1 (7.1)		28 (12.5)	0.52	0.01	3.75	1.000
Shortness of breath	24 (11.4)		3 (21.4)		27 (12.1)	2.11	0.35	8.77	.386
Cough	24 (11.4)		2 (14.3)		26 (11.6)	1.29	0.13	6.35	.670
Dizziness	21 (10.0)		1 (7.1)		22 (9.8)	0.69	0.02	5.08	1.000
Increased sleepiness	18 (8.6)		3 (21.4)		21 (9.4)	2.91	0.47	12.40	.132
Balance problems	16 (7.6)		2 (14.3)		18 (8.0)	2.02	0.20	10.34	.312
Photophobia	15 (7.1)		0 (0.0)		15 (6.7)	0.00	0.00	3.42	.606
Back pain	10 (4.8)		2 (14.3)		12 (5.4)	3.33	0.32	18.36	.167
Join pain (arthralgia)	11 (5.2)		0 (0.0)		11 (4.9)	0.00	0.00	4.91	1.000
Tremor	9 (4.3)		1 (7.1)		10 (4.5)	1.72	0.04	14.18	.483
Weight loss	9 (4.3)		1 (7.1)		10 (4.5)	1.72	0.04	14.18	.483
Slurred speech	5 (2.4)		3 (21.4)		8 (3.6)	11.18	1.50	65.04	.009
Neck pain	7 (3.3)		0 (0.0)		7 (3.1)	0.00	0.00	8.45	1.000
Sore throat	6 (2.9)		0 (0.0)		6 (2.7)	0.00	0.00	10.25	1.000
Seizures	4 (1.9)		1 (7.1)		5 (2.2)	3.96	0.07	43.61	.274
Blurred vision	4 (1.9)		0 (0.0)		4 (1.8)	0.00	0.00	17.54	1.000
Coma	2 (1.0)		1 (7.1)		3 (1.3)	8.00	0.13	159.83	.177
Numbness	2 (1.0)		1 (7.1)		3 (1.3)	8.00	0.13	159.83	.177
Flaccid paralysis	1 (0.5)		0 (0.0)		1 (0.4)	0.00	0.00	285.00	1.000
Lympadenopathy	1 (0.5)		0 (0.0)		1 (0.4)	0.00	0.00	285.00	1.000
Paresthesias	1 (0.5)		0 (0.0)		1 (0.4)	0.00	0.00	285.00	1.000
Underlying condition
Diabetes	38 (18.1)		4 (28.6)		42 (18.8)	1.81	0.39	6.69	.305
Hypertension	33 (15.7)		5 (35.7)		38 (17.0)	2.98	0.73	10.60	.067
Chronic obstructive	11 (5.2)		1 (7.1)		12 (5.4)	1.39	0.03	11.05	.548
pulmonary disease
Dementia	12 (5.7)	0 (0.0)		12 (5.4)		0.00	0.00	4.44	1.000
Coronary artery disease	7 (3.3)	0 (0.0)		7 (3.1)		0.00	0.00	8.45	1.000
Alcoholism	5 (2.4)	0 (0.0)		5 (2.2)		0.00	0.00	12.97	1.000
Asthma	2 (1.0)	0 (0.0)		2 (0.9)		0.00	0.00	53.44	1.000
Cancer	2 (1.0)	0 (0.0)		2 (0.9)		0.00	0.00	53.44	1.000
Immunosuppression	1 (0.5)	0 (0.0)		1 (0.4)		0.00	0.00	285.00	1.000

In the univariate analysis, case patients who subsequently died were more likely than survivors to be aged 68 years or older (OR = 6.9; 95% CI, 1.5-31.4; P = .005), to be initially

diagnosed with encephalitis (Table 2), and to report the presence of the following symptoms: mental status changes (referred here to a constellation of symptoms including

altered mental status, confusion, lethargy, increased sleep- iness, slurred speech, or Memory loss), lethargy, slurred speech, and abdominal pain (Table 3). Survivors were more likely than those who died to report headache (OR = 15.1; 95% CI, 1.9-117.9). The presence of at least one neurological symptom, one gastrointestinal symptom, and one respiratory symptom at admission was not associated with death, nor was any underlying condition.

We performed multivariable logistic regression analysis and evaluated 3 models within each group–initial diagnosis, sign and symptoms, and underlying condition–with the following predictor variables included in the models, respec- tively: (1) age, diagnosis of meningitis, dehydration, and encephalitis; (2) age, presence of headache, mental status changes, lethargy, decreased appetite, abdominal pain, neck stiffness, increased sleepiness, and back pain; and (3) age and hypertension. The final models included (1) age and diagnosis of encephalitis and (2) age and presence of abdominal pain. After controlling for age, hospitalized patients who subse- quently died were more likely to be initially diagnosed with encephalitis (OR = 9.4; 95% CI, 2.4-36.7; P = .001) and reported abdominal pain (OR = 10.5; 95% CI, 2.8-39.0; P b

.001) compared with case patients who survived.

Discussion

In this paper, we described the early clinical features of 224 WNV hospitalized patients during an outbreak of WNV infection in Ohio in 2002. We found that at least one neurological symptom was present in most patients (83%), but an initial diagnosis of encephalitis, meningitis, or meningoencephalitis was made in only 90 (34.4%) case patients. Despite an ongoing outbreak, the WNV infection was initially suspected in only 12 (5.4%) case patients. The small number of cases initially diagnosed with WNV infection might reflect the nonspecific presentation of the early phase of infection and the fact that this was the first year that human WNV infection was present in Ohio.

The clinical presentation of disease in Ohio was similar to that reported from the 1996 outbreak in Romania and the 2000 outbreaks in New York and New Jersey and in Israel [6,15,19,20], with frequent reports of fever, headache, and mental status changes. The clinical findings from the New York outbreak showed that neurological and Gastrointestinal symptoms were present in about 89% and 58% of hospitalized patients, respectively. We observed similar frequencies of neurological and gastrointestinal symptoms in 83% and 53% of the hospitalized case patients, respectively (Table 2). Similarities in the frequencies of symptoms might be caused by the fact that the WNV strain circulating in the United States was genetically similar to the WNV strain that was responsible for the 2000 outbreak in Israel [21]. Although frequent, the presence of at least one neurological or one gastrointestinal symptom at admission was not associated with death.

We found that an initial diagnosis of encephalitis or reporting abdominal pain at admission placed patients at greater risk for death. West Nile virus encephalitis has been previously documented as a cause of death [22-26], but there are no previous reports discussing the significance of abdominal pain. Three categories of abdominal pain are recognized: visceral, somatoparietal, and referred pain [27]. Pain sensations are conveyed from neuroreceptors through 2 types of afferent nerve fibers in which cell bodies are located in the dorsal root ganglia of spinal afferent nerves [27]. A lesion in the dorsal root ganglion can result in acute autonomic and sensory neuropathy [28,29]. The reduced sensory nerve action caused by WNV infection has been demonstrated [30]. In addition, pathological data showed that WNV causes dorsal root ganglia lesions. The lesions might be severe enough to produce sensory deficits and correlate with the severity of the poliomyelitis [31,32].

The case fatality rate among hospitalized patients in Ohio was 6.3%, similar to that reported in Romania (4.3%) but lower than that reported in New York and New Jersey (11.0%) and in Israel (14.1%) [6,15,19,20]. The reasons for the observed differences are not apparent. Among them may have been differences in the length of the follow-up periods of hospitalized patients, the criteria for WNV-related death, completeness of the ascertainment of encephalitis cases, changes in WNV virulence [19,21], the immunological background in a population, the prevalence of comorbid conditions, age distribution of the population, and others.

This study had limitations. First, in Ohio, only hospital- ized patients presenting with neurological symptoms of infection were tested for WNV without charge. Nonhospi- talized patients with WNV infection were not studied. Consequently, the full spectrum of symptoms of WNV illness, especially mild disease, could not be discovered, and the total burden of disease might have been underestimated. Second, the information on symptoms was obtained from ED medical charts, which vary in completeness and accuracy of clinical details. We were not able to validate the data provided by each hospital. In addition, 28 recorders conducted chart reviews. Although we gave them specific instructions on how to complete the survey, interabstractor agreement was not determined and differences in complete- ness and quality of data likely occurred. Also, for analysis purposes, we assumed that signs and symptoms not recorded on the chart were not present, resulting in potential misclassification. Moreover, because of financial, time, and legal constraints, we were able to collect data on only about 60% of the reported WNV cases from 27 (48%) of 56 counties. West Nile virus cases in other counties might differ from those that we analyzed.

Our study demonstrated that WNV infection begins with nonspecific symptoms, making early clinical diagnosis challenging. Diagnosis is based mainly on serological tests for the presence of anti-WNV IgM, which can be demonstrated either in serum or in CSF. Nevertheless, the results have to be interpreted with caution because cross-

reactivity with other members of the genus Flavivirus is possible. Because no effective therapy is known, patients with WNV infection can be offered only supportive care. Physicians should maintain a high index of suspicion for WNV infection during the epidemic season, particularly when evaluating elderly patients with neurological or gastrointestinal symptoms.

Acknowledgments

We thank Virginia Abel (Akron City Hospital, Summa Health System), Marlene Becker (Massillon Community Hospital), Shari Botts (the Fort Hamilton Hospital), Jane Carmean (Clinton Memorial Hospital), Sandra Hensey (Medical College of Ohio), Joan Kimmet (Mercy Hospital of Willard), Toni Leshinski (Hillcrest Hospital), Carolyn May (Kettering Medical Center), Patricia A. Nelson (Mercy Medical Center), Joan Pugnale (Aultman Health Founda- tion), Charla Ulrich (St Vincent Hospital), Sharon L. Wells (Miami Valley Hospital), Kathy Zegarski (Southview Hospital), Vickie Davis (Coshocton City Health Depart- ment), Judy Fisher (Paulding Co Health Department), Patty Fruth (Defiance Co General Health District), Ann Glidewell (Darke Co General Health District), Jo Holt (Clark Co Health Department), Susan Irvine (Hamilton City Health Department), Cindy Jones (Auglaize Co Health Depart- ment), Amy Jones (Wood Co Health District), Joan Kowalczyk (Elyria City Department of Health), Selene Layton (Trumbull Co Health Department), Sheryl Long (Stark Co Combined General Health District), Kathy Martin (Fairfield Department of Health), Linda Mehl (Guernsey Co Health Department), Pam Pflum (Henry Co Health Depart- ment), Jacquelyn Phillips (Middletown City Health Depart- ment), Helen Rogers (Wayne Co Health Department), Cynthia Rose (Fulton Co Health Department), Sharon Schaeffer (Erie Co General Health District), Judy Louden and Peggy Scherer (Geauga Co General Health District), Heidi Scaife (Cuyahoga Co Board of Health), Janine Trottier (Lorain Co General Health District), Rose-Ann Warth (Canton City Health Department), and Barbara Wilhelm (Findlay City Health Department) for their help in accessing and abstracting medical charts.

We are grateful to Dr Amy Bode (Division of Vector- Borne Infectious Diseases, National Center for Infectious Diseases) and Dr Matthew Zack (Division of Adult and Community Health, National Center for Chronic Disease prevention and Health Promotion, Centers for Disease Control and Prevention) for their thoughtful comments.

References

1. Smithburn KC, Hughes TP, Burke AW, et al. A neurotropic virus isolated from the blood of a native of Uganda. Am J Trop Med Hyg 1940;20:471 – 92.
2. Goldblum N, Sterk VV, Paderski B. West Nile fever; the clinical features of the disease and the isolation of West Nile virus from the blood of nine human cases. Am J Hyg 1954;59:89 – 103.
3. Marberg K, Goldblum N, Sterk VV, et al. The natural history of West Nile fever: I. clinical observations during an epidemic in Israel. Am J Hyg 1956;64:259 – 69.
4. Hubalek Z, Halouzka J. West Nile fever–a reemerging mosquito- borne viral disease in Europe. Emerg Infect Dis 1999;5:643 – 50.
5. Platonov AE, Shipulin GA, Shipulina OY, et al. Outbreak of West Nile virus infection, Volgograd Region, Russia, 1999. Emerg Infect Dis 2001;7:128 – 32.
6. Tsai TF, Popovici F, Cernescu C, et al. West Nile encephalitis epidemic in southeastern Romania. Lancet 1998;352:767 – 71.
7. Centers for Disease Control and Prevention. Outbreak of West Nile- like viral encephalitis–New York, 1999. MMWR Morb Mortal Wkly Rep 1999;48:845 – 9.
8. Centers for Disease Control and Prevention. Human West Nile virus surveillance–Connecticut, New Jersey, and New York, 2000. MMWR Morb Mortal Wkly Rep 2001;50:265 – 8.
9. Centers for Disease Control and Prevention. West Nile virus activity– United States, 2001. MMWR Morb Mortal Wkly Rep 2002;51:497 – 501.
10. Ohio Administrative Code. Chapter 3701-3-04. communicable diseases and ailments, laboratory results reporting. Available at: http://www.odh.state.oh.us/Rules/Final/Chap3/FR3_lst.htm [Accessed

March 16, 2004].

1. Martin DA, Muth DA, Brown T, et al. Standardization of immunoglobulin M capture enzyme-linked immunosorbent assays for routine diagnosis of arboviral infections. J Clin Microbiol 2000;38:1823 – 6.
2. Beaty BJ, Calisher CH, Shope RE. Arboviruses. In: Schmidt NH, Emmons RW, editors. Diagnostic procedures for viral, rickettsial and chlamydial infections. 6th ed. Washington (DC)7 American Public Health Association Inc;, 1989. p. 797 – 856.
3. West Nile Virus Workgroup. Combating West Nile virus. The plan for the state of Ohio. May 2003. Available at: http://www.odh.state.oh.us/ ODHPrograms/ZOODIS/WNV/Pubs/WNVStPlan.PDF [Accessed March 16, 2004].
4. Centers for Disease Control and Prevention. Case definitions for Infectious conditions under public health surveillance. MMWR Morb Mortal Wkly Rep 1997;46:12 – 3.
5. Weiss D, Carr D, Kellachan J, et al. West Nile Virus Outbreak Response Working Group. Clinical findings of West Nile virus infection in hospitalized patients, New York and New Jersey, 2000. Emerg Infect Dis 2001;7:654 – 8.
6. Dean AG, Dean JA, Culombies D, et al. Epi Info, version 6.04: a word processing, database, and statistical program for public health on IBM compatible microcomputers. Atlanta (GA)7 Centers for Disease Control and Prevention; 1995.
7. SAS Institute Inc. The SAS system for Windows, release 8.02. Cary (NC)7 SAS Institute Inc; 1999-2001.
8. Brunzel NA. Fundamentals of urine and body fluid analysis. 2nd ed. Philadelphia (PA)7 WB Saunders Co;, 2004. p. 337.
9. Chowers MY, Lang R, Nassar F, et al. Clinical characteristics of the West Nile fever outbreak, Israel, 2000. Emerg Infect Dis 2001; 7:675 – 8.
10. Weinberger M, Pitlik SD, Gandacu D, et al. West Nile fever outbreak, Israel, 2000: epidemiologic aspects. Emerg Infect Dis 2001;7:686 – 91.
11. Lanciotti RS, Roehrig JT, Deubel V, et al. Origin of the West Nile virus responsible for an outbreak of encephalitis in the northeastern United States. Science 1999;286:2333 – 7.
12. Shieh WJ, Guarner J, Layton M, et al. The role of pathology in an investigation of an outbreak of West Nile encephalitis in New York, 1999. Emerg Infect Dis 2000;6:370 – 2.
13. Sampson BA, Ambrosi C, Charlot A, et al. The pathology of human West Nile virus infection. Hum Pathol 2000;31:527 – 31.
14. Kelley TW, Prayson RA, Ruiz AI, et al. The neuropathology of West Nile virus meningoencephalitis. A report of two cases and review of the literature. Am J Clin Pathol 2003;119:749 – 53.
15. Bouffard JP, Riudavets MA, Holman R, et al. Neuropathology of the brain and spinal cord in human West Nile virus infection. Clin Neuropathol 2004;23:59 – 61.
16. Smith RD, Konoplev S, DeCourten-Myers G, et al. West Nile virus encephalitis with myositis and orchitis. Hum Pathol 2004;35:254 – 8.
17. Glasgow RE, Mulvihill SJ. Abdominal pain, including the acute abdomen. In: Sleissenger MH, Fordtran JS, editors. gastrointestinal disease. 7th ed. Philadelphia (PA)7 WB Saunders Co;, 2002. p. 71 – 83.
18. Colan RV, Snead III OC, Oh SJ, et al. Acute autonomic and sensory neuropathy. Ann Neurol 1980;8:441 – 4.
19. Satake M, Nakagawa Y, Yamashita S, et al. Subacute autonomic and sensory ganglionopathy: a postmortem case. J Neurol Neurosurg Psychiatry 1998;64:561.
20. Doron SI, Dashe JF, Adelman LS, et al. Histopathologically proven poliomyelitis with quadriplegia and loss of brainstem function due to West Nile virus infection. Clin Infect Dis 2003;37:e74-7 [Electronic publication 2003 Aug 11].
21. Fratkin JD, Leis AA, Stokic DS, et al. Spinal cord neuropathology in human West Nile virus infection. Arch Pathol Lab Med 2004; 128:533 – 7.
22. Kelley TW, Prayson RA, Isada CM. Spinal cord disease in West Nile virus infection. N Engl J Med 2003;348:564 – 6.