The re-introduction of DENV in the continental US has made it an important infection for VA providers to understand, especially since approximately two-thirds of confirmed/probable cases and almost half of the suspected cases were hospitalized. The majority of endemic VA dengue cases during 2007–2010 were identified in Puerto Rico and Florida. In VA medical facilities in Florida, locally acquired dengue was limited to Key West, Florida, while all other VA cases detected in Florida during the epidemic had an exposure outside of the state. While the number of VA dengue cases in Florida decreased in 2011 to 3 imported cases and is no longer showing sustained local transmission, the potential for further spread of DENV infection in Florida and other parts of the US is possible. Therefore, it is important to understand the clinical presentation, diagnostic testing patterns and epidemiology of dengue within the VA system.
Overall, our dengue cases presented with expected signs and symptoms of fever, arthralgias/myalgias, headache, skin manifestations/rashes, headache, thrombocytopenia, and leukopenia. Clinically, our Florida cases appeared to be no different than those found in Puerto Rico. The similarity of clinical symptoms was not surprising since a DENV-1 strain, related to other Central American strains of dengue, was the predominant serotype identified in the Florida epidemic 
. Based on the 2010 CSTE dengue case definition, confirmed/probable patients had similar characteristics as suspected cases. However, as expected based on the definition of a confirmed/probable case, all of these patients had dengue diagnostic testing, while only 42% had testing in the suspected group (no lab test was needed to meet this definition and some suspected patients had initial dengue diagnostic testing performed but did not have convalescent serologic testing to confirm a diagnosis). Of note, less than two-thirds of suspected dengue cases in Puerto Rico received any dengue diagnostic testing, even though free diagnostic testing is available in Puerto Rico through CDC. In some cases, testing was never ordered and in others, samples were rejected because the CDC-required paperwork was either incomplete or not filled out properly. This finding highlights the need for additional education among VA providers regarding availability of testing and collection of required patient information. We also found that hospitalization, thrombocytopenia, and leukopenia rates were higher in the confirmed/probable group. Interestingly, URI symptoms were less likely in confirmed/probable cases; implying patients with URI symptoms were less likely to have dengue 
. However, 20% of confirmed/probable cases reported URI symptom(s) so their presence does not preclude the diagnosis of dengue and testing patients that have URI symptoms and are suspected of having dengue remains important.
The cases identified in Florida during the Key West epidemic were similar in nature and severity to those seen during epidemic years in Puerto Rico except for the significantly increased amount of testing and increase in cases receiving ICU care in Florida. Interestingly, there was a lower incidence of overall hospitalization for dengue in Florida during the epidemic years. This suggests that even though fewer patients were hospitalized in Florida, when they were hospitalized they received ICU level care. However, although the numbers are small, the number of hospitalized cases in Florida was low and when chart reviews were performed on the patients receiving ICU level care it revealed either atypical presentations or a provider's decision for closer monitoring in an ICU setting, suggesting a relative discomfort in treating dengue in Florida facilities compared to Puerto Rico. There were a lower percentage of cases presenting with headache and eye pain in Florida, however, this may be related to fewer providers asking patients and documenting these symptoms in the EMR. In addition, the cases of dengue during epidemic and non-epidemic years had similar characteristics.
Limitations of our study included a small cohort number that was predominantly older males. In addition, some Veteran cases could have been missed if they were treated at outside facilities and not reported or if they were not coded as having dengue. Dengue PCR was not as commonly used in the Florida cases so we are unable to determine the most common serotype involved with the VA dengue cases in the continental US, or to have samples to do further molecular typing and sequencing. Since we combined confirmed and probable dengue cases it is possible that we slightly overestimated the number of actual dengue cases since elevated anti-DENV IgM may be due to a cross reactivity with other flaviviruses (West Nile virus, St. Louis encephalitis virus, Japanese encephalitis virus, and yellow fever virus), although this is unlikely as there are few places where these viruses co-circulate or where these conditions cannot be differentiated clinically 
. In addition, many Veterans are vaccinated against yellow fever which can have cross reactivity with dengue serologic testing 
. However, the number of probable cases was small therefore it is unlikely to have greatly affected our analysis. Unfortunately, primary/secondary dengue infection status is not reliably documented in our EMR making it impossible to compare prior dengue exposure to symptom severity.
The primary goal of our study was not to compare different surveillance system performance for DENV detection. However, accurate DENV case finding required the combination of two electronic biosurveillance systems (ESSENCE and BioSense), as well as infection preventionist manual surveillance efforts at VA facilities. These electronic biosurveillance systems currently rely on outpatient diagnostic encounter codes, ICD-9, which can be searched by syndrome or individual codes. ICD-9 coding for outpatient visits in VA may not be completely accurate, and likely underestimates the true number of cases, particularly in those cases where confirmatory laboratory testing was not obtained, or was obtained and results were not available at the time of encounter close-out. Syndromic surveillance includes additional, non-specific ICD-9 codes (i.e., fever and rash), which can further reduce the specificity of the diagnosis. In addition, DENV or syndrome ICD-9 codes could reflect prior outpatient encounters for DENV disease, and contribute to an overestimation of the number of cases. Infection preventionists can access multiple data domains in the EMR (including history, laboratory data, and treatment), which help refine whether a potential DENV case is likely to be a confirmed or probable case. In addition, infection preventionists can help facilitate obtaining convalescent blood samples to further help confirm diagnoses. As demonstrated in no system of identifying cases was perfect, both infection preventionists and BioSense were able to identify 9 out of the 12 confirmed/probable cases. ESSENCE was able to capture all 12 cases, however, misidentified 3 cases. Because of the reduced specificity of electronic biosurveillance systems, VA is enhancing VA's ESSENCE system by including vital signs (temperature), laboratory orders and results, inpatient admission data, outpatient encounter severity codes, telephone care encounter data, and pharmacy prescription data, in addition to ICD-9 encounter codes, which will improve specificity and automate much of what infection preventionists currently must review by hand. Until our enhanced system is available, utilization of an electronic surveillance system in addition to manual surveillance by infection preventionists will remain important.
Although indigenous cases of DENV infection are rare in the continental US, after the epidemic of dengue in Key West, Florida, greater attention was placed on dengue surveillance, education and public health reporting. The VA Office of Public Health, CDC and Florida Department of Health collaborated on providing educational materials including a VA Dengue Health Alert that was produced in July 2010 to help educate VA providers on the presence of dengue in Florida. The alert advised providers to be vigilant for symptoms of dengue, to report suspected cases to local and state health departments, and to obtain appropriate laboratory testing for confirmation. Laboratory testing will hopefully become more widely available now that a DENV reverse transcription polymerase chain reaction (RT-PCR) assay developed by the CDC has been approved by the US Food and Drug Administration 
. Additional details on laboratory testing algorithms and clinical guidance are available on the CDC website 
. Increased efforts are necessary to improve dengue awareness in VA through patient and clinician education, and which emphasizes the need for testing, accurate coding of potential dengue cases and appropriate reporting to county and state health officials.