Conceived and designed the experiments: ADH AO. Performed the experiments: ADH AO. Analyzed the data: ADH AO. Wrote the paper: ADH AO.
Current address: Department of Pathology and the Center for Biodefense and Emerging Infectious Disease, The University of Texas Medical Branch, Galveston, Texas, United States of America
The authors have declared that no competing interests exist.
Although La Crosse virus (LACV) is one of the most common causes of pediatric arboviral infections in the United States, little has been done to assess its geographic distribution, identify areas of higher risk of disease, and to provide a national picture of its clinical presentation. Therefore, the objective of this study was to investigate the geographic distribution of LACV infections reported in the United States, to identify hot-spots of infection, and to present its clinical picture.
Descriptive and cluster analyses were performed on probable and confirmed cases of LACV infections reported to the Centers for Disease Control and Prevention from 2003–2007. A total of 282 patients had reported confirmed LACV infections during the study period. Of these cases the majority (81 percent) presented during the summer, occurred in children 15 years and younger (83.3 percent), and were found in male children (64.9 percent). Clinically, the infections presented as meningioencephalitis (56.3 percent), encephalitis (20.7 percent), meningitis (17.2 percent), or uncomplicated fever (5 percent). Deaths occurred in 1.9 percent of confirmed cases, and in 8.6 percent of patients suffering from encephalitis. The majority of these deaths were in patients 15 years and younger. The county-level incidence risk among counties (n = 136) reporting both probable and confirmed cases for children 15 years and younger (n = 355) ranged from 0.2 to 228.7 per 100,000 persons. The southern United States experienced a significantly higher (p<0.05) incidence risk during the months of June, July, August, and October then the northern United States. There was significant (p<0.05) clustering of high risk in several geographic regions with three deaths attributed to complications from LAC encephalitis occurring in two of these hot-spots of infections.
Both the incidence risk and case fatality rates were found to be higher than previously reported. We detected clustering in four geographic regions, a shift from the prior geographic distributions, and developed maps identifying high-risk areas. These findings are useful for raising awareness among health care providers regarding areas at a high risk of infections and for guiding targeted multifaceted interventions by public health officials.
La Crosse virus (LACV) is a member of the genus
LACV has traditionally been associated with forested areas in the upper-midwestern United States
Our study area encompassed the eastern United States, the geographic region that includes the majority of previously reported LACV infections and the range of the primary vector,
State | Reported Cases | Percentage of Reported Cases | Incidence Risk by State | Range of Incidence Risk Among Counties Reporting Cases |
West Virginia | 83 | 23.4 | 57.8 | 7.5–228.7 |
Ohio | 73 | 20.6 | 5.0 | 0.3–37.1 |
North Carolina | 66 | 18.6 | 21.0 | 0.5–206.7 |
Tennessee | 50 | 14.1 | 18.4 | 3.6–166.6 |
Wisconsin | 21 | 5.9 | 12.1 | 1.1–76.7 |
Illinois | 18 | 5.1 | 1.3 | 0.2–30.8 |
Minnesota | 9 | 2.5 | 10.6 | 7.3–29.8 |
Virginia | 8 | 2.3 | 7.8 | 1.6–35.2 |
Georgia | 7 | 2.0 | 2.6 | 0.5–129.8 |
Indiana | 6 | 1.7 | 2.0 | 0.5–20.6 |
Louisiana | 4 | 1.1 | 6.5 | 4.0–23.2 |
Iowa | 3 | 0.8 | 41.4 | 27.7–55.1 |
Kentucky | 3 | 0.8 | 13.5 | 7.7–23.6 |
Michigan | 1 | 0.3 | 1.0 | 1.0 |
Mississippi | 1 | 0.3 | 8.3 | 8.3 |
Alabama | 1 | 0.3 | 0.7 | 0.7 |
South Carolina | 1 | 0.3 | 1.7 | 1.7 |
Florida |
– | – | – | – |
Overall | 355 | 100 | 7.2 | 0.3–228.7 |
Incidence risk was calculated in counties reporting probable and confirmed cases of La Crosse virus infections and presented as the number of cases per 100,000 persons in children 15 years and younger, and are expressed here as a range in those states with two or more counties reporting cases.
No incidence risk is reported for states not reporting cases 15 years and younger.
This study was conducted for all states reporting probable and confirmed cases of LACV infections in the United States, 2003–2007, through the ArboNET surveillance system
Personal identifiers of patients were deleted before database construction. Clinical and epidemiological LACV data for 151 probable and 275 confirmed cases of LACV infections reported during this time period were provided by the Centers for Disease Control and Prevention. Cases that acquired infection outside of the county/state were excluded from spatial analyses (n = 10). This research was deemed exempt from review and certification by the University of Tennessee's Institutional Review Board following review by the Department of Entomology and Plant Pathology's Departmental Review Committee under the University of Tennessee's guidelines for research involving human subjects.
Confirmed cases of LACV infections are required to meet both the clinical and laboratory requirements set by the Centers for Disease Controls and Prevention's case definition for neuroinvasive domestic arboviral diseases
In the absence of a more likely clinical explanation as documented by a physician, confirmed cases must meet all of the following criteria:
2) Acutely altered mental status, or other acute signs of central or peripheral neurologic dysfunction, or pleocytosis associated with illness clinically compatible with meningitis, AND
The 2005, United States Census Bureau, Estimated County Population Dataset
Incidence risk was calculated and spatial analyses were performed on 123 probable and 232 confirmed cases 15 years and younger and on 151 probable and 275 confirmed cases for all ages occurring during the study period for which county level data were available (
To determine if there was a significant difference in incidence risk by month between states reporting cases in the northern (n = 7) and southern (n = 10) regions of the study area, we calculated the incidence risk by region and month using the both probable and confirmed cases 15 years and younger. The northern region was comprised of Illinois, Iowa, Indiana, Michigan, Minnesota, Ohio, and Wisconsin. The southern region was comprised of Alabama, Georgia, Kentucky, Louisiana, Mississippi, North Carolina, South Carolina, Tennessee, Virginia, and West Virginia.
All incidence risk computations and descriptive analyses were performed using STATA 10.0
Global Moran's I
A total of 282 patients had confirmed LACV infections reported to the Centers for Disease Control and Prevention (
Variable | Total Confirmed (%) | Probable and Confirmed Cases Combined (%) |
Sex | ||
Male | 183 (64.9) | 264 (60.6) |
Female | 99 (35.1) | 172 (39.4) |
Unknown | – | 1 |
Age | ||
0.1–0.9 yr | 3 (1.06) | 5 (1.15) |
1 yr | 11 (3.9) | 13 (2.98) |
2–5 yr | 59 (20.9) | 98 (22.8) |
6–10 yr | 98 (34.8) | 157 (26.4) |
11–15 yr | 64 (22.7) | 87 (19.95) |
16–20 yr | 7 (2.5) | 15 (3.44) |
≥21 yr | 40 (14.2) | 61 (13.99) |
Unknown | – | 1 |
Race | ||
White | 241 (95.3) | 368 (95.6) |
Black or African American | 9 (3.6) | 11 (2.86) |
American Indian or Alaska Native | 3 (1.2) | 4 (1.04) |
Asian | 0 (0) | 1 (0.259) |
Other | 0 (0) | 1 (0.259) |
Unknown | 29 | 52 |
Month of presentation | ||
March | 1 (0.35) | 2 (0.458) |
April | 3 (1.06) | 3 (0.686) |
May | 1 (0.35) | 2 (0.454) |
June | 23 (8.2) | 29 (6.64) |
July | 71 (24.9) | 110 (25.2) |
August | 92 (32.6) | 138 (31.6) |
September | 65 (23) | 105 (24.0) |
October | 23 (8.2) | 44 (10.1) |
November | 2 (0.71) | 3 (0.686) |
December | 1 (0.35) | 1 (0.229) |
Clinical manifestation | ||
Meningioencephalitis | 157 (56.3) | 242 (55.3) |
Encephalitis | 58 (20.7) | 78 (17.8) |
Meningitis | 48 (17.2) | 87 (19.9) |
Uncomplicated fever | 14 (5.0) | 18 (4.4) |
Other | 2 (0.7) | 3 (0.7) |
Unknown | 3 | 8 |
Death | 5 (1.86) | 6 (1.43) |
Unknown |
13 | 16 |
Unknown, represents the number of confirmed and probable cases for which the case outcome was not reported.
The sex-specific incidence risk (per 100,000 persons) for all counties reporting confirmed cases was 1.9 and 1.0 for males and females, respectively. The age-specific incidence risk was 1.2 for children under one year; 5.4 for 1–5 years; 6.8 for 6–10 years; 4.7 for 11–15 years; 0.5 for 16–20 years; and 0.2 for 21 years and older. Blacks had the lowest race-specific incidence risk (0.3 per 100,000 persons) and American Indians had the highest (4.9 per 100,000 persons). The incidence risk for all counties reporting probable and confirmed cases combined in children 15 years and younger was 7.2 per 100,000 persons (mean 30.2), and 1.6 per 100,000 persons (mean 5.8) for the total population.
The incidence risk was significantly higher in the states in the southern region than those states in the northern region for June (p = 0.0018), July (p = 0.0002), August (p = 0.0033), and October (p = 0.0053) (
Northern Region |
|
Month | Incidence Risk |
June | 2.9 |
July | 1.8 |
August | 1.8 |
September | 4.0 |
October | 3.7 |
Southern Region |
|
Month | Incidence Risk |
March | 2.4 |
April | 16.7 |
May | 20.7 |
June | 10.7 |
July | 8.9 |
August | 9.8 |
September | 12.6 |
October | 8.5 |
November | 2.0 |
Incidence risk was calculated by region and month using the both probable and confirmed cases of La Crosse virus infections and presented as the number of cases per 100,000 persons in children 15 years and younger.
Illinois, Iowa, Indiana, Michigan, Minnesota, Ohio, and Wisconsin.
Alabama, Georgia, Kentucky, Louisiana, Mississippi, North Carolina, South Carolina, Tennessee, Virginia, and West Virginia.
Clinically, the infections presented as meningioencephalitis (56.3 percent), encephalitis (20.7 percent), meningitis (17.2 percent), uncomplicated fever (5 percent), or other (0.7 percent) (
In children 15 years and younger (probable and confirmed combined), state-level incidence risk ranged from 0.7 to 56.4 per 100,000 persons whereas at the county-level incidence risk in counties reporting cases, ranged from 0.2 to 228.7 per 100,000 persons (
The map on the left represents the distribution of a) unsmoothed risk, and on the right b) smoothed risk of La Crosse virus infections at the county level for the eastern United States.
This map shows the significant clustering of La Crosse virus infections at the county level detected by the Moran's I Local Indicators of Spatial Autocorrelation (LISA) for the eastern United States. Four types of spatial autocorrelation are observed using the LISA statistic (High-High, Low-Low, High-Low, and Low-High). Positive spatial autocorrelation is represented by High-High and Low-Low, and negative spatial correlation by High-Low and Low-High. Positive spatial autocorrelation (i.e. an association of areas of similar values) were represented as either High-High (i.e. a high risk in an area surrounded by similarly high values in neighboring areas) or Low-Low (i.e. a low risk in an area surrounded by similarly low values in neighboring areas). Negative spatial autocorrelation (i.e. an association of areas of dissimilar values) was represented as either High-Low (i.e. a high rate in an area surrounded by low values in neighboring areas) or Low-High (i.e. a low rate in an area surrounded by high values in neighboring areas).
Our study provides the first risk map of LACV infections for the United States, and presents insights into the clinical picture of LACV infections. We found both a higher incidence risk than previously reported ranging up to 228.7 cases per 100,000 persons in children 15 years and younger and a case fatality rate of 1.9 percent
The majority of cases reported occurred in the summer months (June, July, and August) in agreement with previous studies
Temperature may play a role in the higher incidence risk in southern region of the United States. Given that transovarial transmission following one gonotrophic cycle is likely rare in nature
Previous studies have shown that over 90 percent of reported cases occur in children 15 years and younger
Of the 426 reported probable and confirmed LACV infections during our study period, the Appalachian region including West Virginia, North Carolina, Ohio, and Tennessee reported 22.2 percent, 20.8 percent, 18.5 percent and 22.2 percent of reported cases, respectively, and accounted for 74.5 percent of all cases during the study period. A previous study from 1964–1981, reported 1348 cases in the United States of which 88.8 percent came from Ohio, Wisconsin, Minnesota, Illinois, and Iowa
Though our finding of an mean incidence risk of 30.2 per 100,000 persons for children 15 years and younger in probable and confirmed cases combined for eastern United States was similar to previous reports of an incidence risk of 20 to 30 per 100,000 persons in the same age group
A recent study of the incidence of WNV infections in the United States by Lindsey et al.
When the results from our analysis of those counties in the study area reporting both probable and confirmed cases of LACV infections are compared with pediatric WNV infection cases reported in Cuyahoga County, Ohio, in 2002
From the results of LACV serosurveys conducted in endemic regions it is clear that the risk of asymptomatic infections is much higher than symptomatic infections, with estimates of asymptomatic infections to clinical infections in pediatric populations ranging from 2∶1 to 1500∶1
The use of probable and confirmed cases combined increased overall cluster detection from 47 to 54 counties. The significant high risk clusters detected should be targeted for future studies and for interventions by public health officials. Reporting the mean incidence risk at the state or county level may lead to a distorted picture of the spatial patterns of LACV infections and thus decrease the perception of risk. Future reporting should include the range of incidence risk occurring in those counties reporting cases. This will minimize misperceptions of risk, as the use of incidence risk continues to remain the most used tool to identify high risk areas for education, prevention, and intervention.
Though LACV infections are typically reported as LAC encephalitis there appear to be four distinct clinical syndromes, as well as asymptomatic infections
Our study found a case fatality rate of 1.9 percent in confirmed cases, with all of the deaths occurring in patients presenting encephalitis. The majority of deaths were in children 15 years and younger. This finding is much higher than the case fatality rate of 0.3 percent reported in a previous study of LACV infections in the United States
Our study has some limitations. Probable and confirmed cases were reported through a passive surveillance system, which inherently suffers under-reporting. Not withstanding this limitation we feel that the majority of cases progressing to severe illness were diagnosed and reported to public health officials. Using only confirmed cases as well as probable and confirmed cases combined, we were able to demonstrate similar high incidence risk and case fatality rates. We were also able to show similar patterns in disease clustering. Clinical data was reported from multiple state health departments to the Centers for Disease Control and Prevention, which did not allow for the verification of laboratory results and the diagnosis of the specific clinical presentation for each patient. We feel that this is a small limitation and that the majority of cases reported have been correctly separated into the four manifestations of severe LACV infections by clinicians.
One drawback to using the LISA is the issue of multiple comparisons which increases type I error rates. We didn't make an attempt to adjust for this because some authors have suggested that any adjustments made to reduce the type I errors would increase type II errors
Our findings of a high incidence risk within significantly high spatial clusters and high case-fatality rate indicate a much higher burden of disease than previously reported, and demonstrate that LACV infections are much more common than previously reported. We have demonstrated the usefulness of these spatial statistical techniques to detect hot-spots of infections, thus allowing for targeted interventions by public health officials while raising awareness among health care providers of geographic areas at the highest risk of disease.
Our results will allow focused national serological studies, form the basis for the development of predictive models of virus transmission, provide a methodology for the use spatial analyses at a national level for other infectious diseases, and demonstrate the need for the reporting of arboviral and other disease cases at smaller geographic scales.
We would like to thank the Arboviral Diseases Branch of the Centers for Disease Control and Prevention for releasing the surveillance data to our group, and their comments on the draft manuscript. We would also like to offer our sincere thanks to the staff of all the many local and state health departments who collected La Crosse virus infection case data and submitted them to the Centers for Disease Control and Prevention. Without their contributions this work would not have been possible.