| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
 |
ABSTRACT |
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES |
|---|
|
INTRODUCTION |
|
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES |
|---|
Rocky Mountain spotted fever is a nationally notifiable disease, and surveillance for RMSF is primarily conducted to help track incidence over time and changes in geographic distribution of reported cases. National surveillance records suggest that the mean incidence rate for RMSF is increasing, rising from 2.2 cases per million persons in 1997–2002 to 4.7 cases per million persons in 2001–2005 (J. Openshaw and others, unpublished data).3
Nationally, there are clear differences in the geographic distribution of reported RMSF cases. Most RMSF cases are reported from the South Atlantic and West South Central states,1,3 with North Carolina, followed by Arkansas, Oklahoma, and Tennessee reporting the highest recent incidence rates in the United States at 52.8, 40.4, 39.9, and 16.6 cases per million persons, respectively (J. Openshaw and others, unpublished data). Yet, little is known about how associated hospitalization, complication, and fatality rates are spread throughout the country. The objective of this study is to characterize the U.S. geographic distribution of severe cases with reported RMSF to determine if hospitalizations, complications, and fatalities are clustered in certain geographic areas. Specifically, we aim to identify any geographic foci within the United States with significantly different rates of reported severe RMSF-associated patient outcomes relative to all cases reported in the nation by performing a spatial cluster analysis on national RMSF surveillance data. Identifying if and where significant geographic clusters of both severe and non-severe reported RMSF cases occur is critical for understanding why these differences exist and how to minimize severe outcomes in people.
|
METHODS |
|
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES |
|---|
a confirmed case was defined as a person with a clinically compatible illness (acute onset of fever, headache, malaise, myalgia, nausea/vomiting, rash and/or neurologic signs) that was also laboratory confirmed by either a 4-fold change in serum antibody titer against R. rickettsii antigens between paired serum samples as determined by immunofluorescence antibody assay (IFA) or enzyme-linked immunosorbent assay (ELISA), a positive polymerase chain reaction (PCR) test, immunohistochemical (IHC) stain, or by isolation during culture. A probable case was defined as a person with a clinically compatible illness and serologic evidence of antibodies reactive to R. rickettsii in a single sample at a titer considered indicative of current or previous infection by the diagnostic laboratory conducting the test.5 Cases were geocoded into latitude and longitude coordinates by county centroid and classified by their level of disease severity as reported in the CRFs. Three levels of disease severity were defined: Level 1) including cases reporting any hospitalizations, complications and/or fatalities; Level 2) including only cases reporting complications and/or fatalities); and Level 3) including only cases reporting fatalities. Oklahoma did not report case complications on CRFs, and were excluded from analysis for Level 2 severity.
Data were uploaded into the statistical software program SaTScan, software for spatial and space-time scan statistics (M. Kulldorff and Information Management Services, Inc., SaTScan version 7.0, www.satscan.org, 2007), and a Poisson model was used to detect significant spatial clusters (P 
0.05) of cases for each level of disease severity examined. Under this model, all cases reported represent the total RMSF population, and those individuals meeting the criteria for each specific level of disease severity represent the "cases." The model has an expected Poisson distribution because of the rare nature of these outcome events.
A spatial scan test was used to compare the composition of these defined cases (based on their geo-coordinates) inside a circular window to the entire RMSF population. The window moved over the area, and varied in size from zero to a maximum radius of 300 km; the window was also set to never include more than 50% of the total population to allow both small and large clusters to be detected.6 For each window of varying position and size, the software tested the risk of being a defined case both within and outside the window, with the null hypothesis stating that both areas were of equal risk.7 For each index of severity, clusters of RMSF cases that were both greater than and less than expected, relative to the respective mean national rates observed among all RMSF cases, were examined. Maps of all RMSF cases reported by severity level and of all significant clusters detected were generated using ArcView GIS 9.2 (Environmental Systems Research Institute, Inc., Redlands, CA).
|
RESULTS |
|
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES |
|---|
Over 46.2% of all cases in the CRF cohort were reported from North Carolina. This state accounted for 36.1%, 37.8%, and 30.4% of cases classified as severity Level 1, 2, and 3, respectively. The next highest number of cases came from Oklahoma, which represented 15.2% cases reported in the CRF cohort. Of these, 20.4% were classified as Level 1 and 13.0% were classified as a Level 3 severity. Although Tennessee only accounted for 2.4% of all reported cases in the CRF cohort, these individuals represented 17.4% of all deaths reported. Similarly, Arizona accounted for less than 0.4% of the total number of cases reported in the CRF cohort, yet this state accounted for 8.7% of all deaths. Other states with reported deaths (and the percentage of total deaths they represented) include Arkansas (8.7%), Georgia (4.4%), Missouri (4.4%), Ohio (4.4%), and Virginia (8.7%). Twenty states reported 10 or fewer cases. Table 1
shows the breakdown of RMSF cases by state and disease severity level.
|
Two significant clusters of RMSF cases with disease less severe than expected when compared with the overall U.S. means for Level 1 (P = 0.001, RR = 0.62) and Level 2 (P = 0.001, RR = 0.45) severity were identified in south-central North Carolina (35.258420, –80.804130; radius = 277.9 km) and in central North Carolina (35.6827, –80.47712; radius = 240.6 km).
Maps demonstrating all reported RMSF cases by disease severity level (Figure 1) and of all significant clusters identified (Figure 2) were generated.
|
|
|
DISCUSSION |
|
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES |
|---|
A significant cluster of six fatal cases was detected centering in southwest Tennessee. This geographic focus was located ~235 km northeast of Memphis, Tennessee, but also incorporated some portions of Alabama, Arkansas, Illinois, Kentucky, Mississippi, and Missouri within its 250 km radius. The fatal cases included in this cluster were all adults ranging from 44–75 years of age (mean age of 62 years), four of whom were male. The risk of a case dying from reported RMSF in this implicated region appears to be over eleven times greater than the relative risk of dying for other U.S. RMSF cases. However, the validity of this method in assessing risk for severe disease may be influenced by how cases are reported via CRFs in this area. For example, it is possible that severe or fatal RMSF cases reported in this region may garner greater medical attention, and thus be more likely to be reported directly to CDC using CRFs than occurs in other states. Regardless, this cluster represents 26% of all reported deaths within the United States that occurred during the five year study period, suggesting that this is likely an area of intense RMSF activity. Active surveillance in this region is needed to determine the reasons behind this observed cluster of deaths.
There are several possible explanations that could account for observations of increased disease severity among RMSF cases reported within a specific geographic location. In 2001, a study looking at risk factors for fatal RMSF in the United States identified older patient age, delayed treatment, treatment with only chloramphenicol, and treatment that did not include tetracycline antibiotics as the primary therapy as significant causes of increased risk.8 It is possible that these factors could influence the occurrence of clusters of severe disease. The cluster of RMSF deaths detected in southwest Tennessee geographically coincided with a significant cluster of RMSF-associated hospitalizations with complications. Although these severe outcomes may be related to the anthropogenic risk factors listed by Holman and others,8 as well as others including differences in available medical care, biologic factors, such as vector characteristics or differences in virulence among R. rickettsii strains, must also be considered as potential causes of more severe disease.
Strains of R. rickettsii have been shown to vary dramatically in their virulence in animal model systems and severity of human disease.9 A phylogenetic analysis of a limited number of R. rickettsii isolates showed that multiple genotypes exist in the eastern United States, 10 though it is not known precisely how these genetically distinct strains are distributed. It is possible that the R. rickettsii strains found circulating within the implicated area may exhibit an increased level of virulence relative to those found elsewhere, and that these clusters represent a unique ecologic niche. Previous disease reporting data show that the regions encompassing these clusters are considered to be hyperendemic for RMSF, with multiple cases identified within single groups and families located there. 11 Further studies on isolates collected from tick vectors and cases from these geographic foci are needed to assess possible biologic influences on severe case clusters.
In our study, a cluster of deaths concentrated in eastern Arizona was also identified, although the findings were not statistically significant. In this area, 14 RMSF cases were reported by CRFs, including 2 deaths in children (aged 5 and 9 years, respectively). This represents a case fatality rate for this region of 14%, which is between one and three times greater than what is typically reported as the national average in recent decades (J. Openshaw and others, unpublished data). 3,8 The lack of statistical significance in this cluster is likely a result of the relatively small overall number of cases reported within this area, combined with the fact that our analysis was limited to the county level (forcing all individuals within the county to share the same geo-coordinate).7 The number of cases reported via CRFs from this area may have been influenced by an outbreak of RMSF that was investigated during 2004–2005, during which time active surveillance for cases was initiated. 12 In this region, RMSF transmission has been linked to a different tick vector (Rhipicephalus sanguineus, the brown dog tick) 13 than is responsible for most U.S. RMSF infections, and thus anthropogenic and biologic factors affecting disease severity may be quite different than in other areas. A genetic analysis on R. rickettsii isolates obtained from cases and tick vectors from this region has revealed a common genotype that appears somewhat distinct from isolates obtained in other geographic locations. 10 To better understand RMSF in this area, including whether disease is more severe than expected, the ecology of a novel tick vector, the virulence of the strain of R. rickettsii , and their effects on the epidemiology of human RMSF within the community must be more fully evaluated.
Our study also identified a significant cluster of moderately severe cases hospitalized in central Oklahoma; RMSF cases from this area were 43% more likely to be hospitalized than elsewhere in the United States. Unfortunately, because complication data were not reported by Oklahoma on CRFs, more detailed information regarding additional indices of severity was not available for analysis. The large number of RMSF-associated hospitalizations that occurred in this cluster within Oklahoma suggests closer examination of clinical data, including possible complications, should be considered.
In contrast to the findings of severe disease clusters, two significant clusters of less than expected rates of hospitalizations and complications were detected in central and south-central North Carolina, respectively. Within each of these clusters, which center on or near Charlotte, North Carolina and include parts of South Carolina within their boundaries, cases were nearly two times less likely to be hospitalized or develop complications during hospitalization relative to all U.S. RMSF cases. These clusters of seemingly milder disease outcomes are particularly interesting considering that North Carolina produces nearly 50% of all RMSF case reports in the United States. It is possible that because RMSF is so common in this area relative to the rest of the country, local clinicians and healthcare providers are more aware of it as a potential source of illness in patients, and therefore may be better at detecting, diagnosing, and appropriately treating RMSF. It may also be possible that other less pathogenic or clinically milder strains of rickettsiae may be circulating in this region. 10 A proportion of the RMSF cases reported may in fact be infected with antigenically related rickettsiae that are not presently thought to be pathogenic, but do cross-react during diagnostic tests for R. rickettsii, giving false-positive serologic results. 14 These possible spotted-fever group rickettsiae include Rickettsia montanensis, R. peacockii, R. rhipicephali, and R. amblyommii.15 Previous studies have detected evidence of R. rhipicephali and R. montanensis in dogs in North Carolina and of R. amblyommii in high numbers of ticks collected in the south, central, and mid-Atlantic states. 16,17 Strong associations have also been reported between R. parkeri and the Gulf Coast tick, Amblyomma maculatum,18 whose distribution includes parts of North Carolina. 19 More research, including active surveillance for disease focusing on milder forms of infection, will be important to help elucidate explanations for the perceived concentration of mild illness in this area, and further studies are needed to tease out the roles that other spotted-fever group rickettsiae play.
This study is limited to information reported by the states to CDC using CRFs, which is the only national surveillance system for RMSF that attempts to capture details on associated complications and other key patient characteristics. Although participation in the CRF surveillance system is not required by national reporting guidelines, most states do participate. States more frequently use another nationally-mandated reporting system (National Electronic Telecommunication System for Surveillance [NETSS]) to provide case counts, but information on disease severity is not available through that system. Comparisons with data collected during the same time period through NETSS suggest that CRFs capture on average 67% of all RMSF cases reported through NETSS (J. Openshaw and others, unpublished data). However, the frequency of reporting through both systems varies greatly by state; thus, use of CRF data may not be sufficient to adequately judge areas of severe disease risk throughout the United States. Some states that have traditionally reported high incidence rates of RMSF through NETSS, such as Wyoming and Texas (J. Openshaw and others, unpublished data), 20,21 could not be adequately assessed in this analysis because they did not participate in the CRF Surveillance System. Although it has been suggested that the submission of CRFs may be biased toward reporting of more severe disease because of the nature of the information collected, a study by Paddock and others in 2002 estimated that fatal cases reported via CRFs represented only a little over one-third of all RMSF-associated deaths that occur in the United States.2
Because many individuals infected with RMSF present with a non-specific clinical presentation,4 it is difficult to assess its actual incidence and prevalence in the population. This study attempts to assess severe disease in a given geographic area relative to overall disease severity within the United States, but should not be interpreted as an attempt to assess incidence or overall disease burden. More prospective epidemiologic studies are needed to determine how the rates of RMSF and case outcomes reported here compare, and to identify what impact this may have on the true magnitude of significance associated with the clusters observed here. Clusters of both increased and decreased rates of severe outcomes in RMSF cases reported here are the product of complex interactions among biologic, ecologic, and anthropogenic factors that vary by geographic location. Epidemiologic investigations into these spatial clusters, at a finer spatial scale, are needed to determine why these differences exist and to tease out the complicated components involved in the dynamics of RMSF.
Received July 8, 2008. Accepted for publication October 6, 2008.
Acknowledgments: This study would not be possible without the continued support from the states for the RMSF CRF surveillance system, which provides unique, additional data that are vital for better understanding RMSF epidemiology in the United States. In particular, we thank John Dunn, Kristy Bradley, Carl Williams, and Craig Levy for reviewing this manuscript, and the helpful comments received from Robert Massung.
* Address correspondence to Jennifer Zipser Adjemian, Centers for Disease Control and Prevention, 1600 Clifton Road, NE, MS-G44, Atlanta, GA 30030. E-mail: gdn5{at}cdc.gov 
Authors’ addresses: Jennifer Zipser Adjemian, John Krebs, Eric Mandel, and Jennifer McQuiston, Centers for Disease Control and Prevention, 1600 Clifton Road, NE, MS-G44, Atlanta, GA 30030.
The definition of a laboratory confirmed and probable case was changed in 2008 as follows: 1) confirmed case—has serologic evidence of a 4-fold change in immunoglobulin G (IgG)–specific antibody titer reactive with R. rickettsii antigen by IFA between paired serum specimens (one taken in the first week of illness and a second 2–4 weeks later), or detection of R. rickettsii DNA in a clinical specimen via amplification of a specific target by PCR assay, or demonstration of spotted fever group antigen in a biopsy or autopsy specimen by IHC, or isolation of R. rickettsii from a clinical specimen in cell culture; and 2) probable case—has serologic evidence of elevated IgG or IgM antibody reactive with R. rickettsii antigen by IFA, ELISA, dot-ELISA, or latex agglutination.
|
REFERENCES |
|
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES |
|---|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
