Chikungunya Fever Cases Identified in the Veterans Health Administration System, 2014

Background During December 2013, the first locally transmitted chikungunya virus (CHIKV) infections in the Americas were reported in the Caribbean. Although CHIKV infection is rarely fatal, risk for severe disease increases with age and medical comorbidities. Herein we describe characteristics of Veterans Health Administration (VHA) patients with CHIKV infection and, among those with infections diagnosed in Puerto Rico, investigated risk factors for hospitalization. Methodology We queried VHA’s national electronic medical records to identify patients with CHIKV testing during 2014. Demographics, clinical history, laboratory results, and outcomes were abstracted. We investigated risk factors for hospitalization among patients with laboratory-confirmed CHIKV infection in Puerto Rico. Principal Findings We identified 180 laboratory-confirmed CHIKV infections; 148 (82.2%) were diagnosed in Puerto Rico, and 32 (17.8%) were diagnosed among returning travelers elsewhere in the United States. In Puerto Rico, where more patients were hospitalized (55.4% versus 20.0%) and died (4.1% versus 0%), risk for hospitalization increased with age (relative risk [RR]/each 10-year increase, 1.19; 95% confidence interval [CI], 1.06–1.32) and, adjusted for age, increased among patients with congestive heart failure (RR, 1.58; 95% CI, 1.25–1.99), chronic kidney disease (RR, 1.52; 95% CI, 1.19–1.94), diabetes mellitus (RR, 1.39; 95% CI, 1.06–1.84), or chronic lung disease (RR, 1.37; 95% CI, 1.03–1.82). Conclusions/Significance CHIKV infection is an emerging problem among Veterans residing in or visiting areas with CHIKV transmission. Although overall mortality rates are low, clinicians in affected areas should be aware that older patients and patients with comorbidities may be at increased risk for severe disease.


Introduction
Chikungunya virus (CHIKV), an alphavirus most commonly transmitted by Aedes species mosquitoes, causes chikungunya fever (CHIK), characterized by acute-onset of fever and what is frequently described as incapacitating polyarthralgia [1,2]. Since CHIKV was first identified in Tanzania in 1953 [2], epidemics have occurred in South and Southeast Asia, Africa, and Europe [1,3,4]. Three distinct genotypes have been described, including East/Central/South African, Asian, and West African [1,5]. In December 2013, locally transmitted infection in the Western Hemisphere was first reported; the predominant strain is closely related to the Asian genotype [6][7][8]. CHIKV has rapidly disseminated among this largely immunologically naïve population; the Pan American Health Organization (PAHO) reported >1.1 million suspect cases, involving the majority of Western Hemisphere countries by the end of 2014 [9]. During 2014, over 31,000 cases (14% laboratory-confirmed) were reported in Puerto Rico [10] and over 1,500 cases (17% laboratory-confirmed) were reported in the U.S. Virgin Islands [11]; these figures are thought to underestimate disease burden because they do not include patients who did not present for care nor those who presented, but were not reported or for whom diagnostic testing was not completed [12]. During 2014, 2,811 laboratory-confirmed infections in the United States were reported to the Centers for Disease Control and Prevention (CDC) through the ArboNET surveillance system; the majority were among returning travelers, except for 12 persons in Florida with locally transmitted infection [13].
CHIK is usually self-limited, with the majority of symptoms typically resolving in 7-10 days [1]; however, patients can have prolonged rheumatologic symptoms [14,15]. CHIKV infection can also be associated with severe illness, involving neurologic, cardiovascular, respiratory, renal, and ocular manifestations [16]. Although overall mortality is low, estimated at 0.3/1,000 population per year on Réunion Island [17], risk for severe disease and death increases with age and is higher among patients with certain comorbidities [17][18][19].
The Veterans Health Administration (VHA) has health care facilities throughout the United States and U.S. territories. Because 45% of all U.S. Veterans and 62% of Veterans in Puerto Rico are aged 65 years [20], and VHA patients have more comorbidities than Veterans who receive care at non-VHA facilities or non-Veterans [21], VHA patients might be at higher risk for severe CHIK than those in the general U.S. population exposed to the virus (i.e. returning travelers and those living in areas with CHIKV transmission). VHA's Public Health Surveillance and Research Group (PHSR) performs surveillance for emerging infections among VHA patients. After the CHIK epidemic involved U.S. territories, PHSR began performing CHIK epidemiologic surveillance in July 2014 of all patients with laboratory-confirmed CHIKV infection diagnosed at VHA facilities during 2014. Herein, we describe characteristics of these patients, compare clinical findings with patients who tested negative, investigate risk factors for hospitalization, and report phylogenetic analysis of CHIKV strains detected.

Methods
We identified patients from all VHA facilities with CHIKV test results for specimens collected during January 1, 2014-December 31, 2014, utilizing VHA's Corporate Data Warehouse, a national data repository from VHA's VistA electronic medical record system. Because VHA has no standardized naming convention for laboratory tests and inconsistent application of universal codes that can be utilized for test identification, we queried this data warehouse for any test name containing the term "chik." After distributing CHIKV surveillance reports to VHA facilities beginning in July 2014, PHSR was contacted by VHA facilities that were not listed in the reports but had diagnosed infections. Consequently, we expanded the query to search for "chik" in the comments field to capture additional CHIK diagnostic test results associated with other test names (e.g., "dengue", "miscellaneous", and "reference laboratory"). The last query was performed on February 26, 2015.
Apart from VHA's Public Health Reference Laboratory (PHRL), VHA clinical laboratories do not have CHIK diagnostic testing capability, and thus specimens are sent to non-VHA laboratories for testing. When capacity of non-VHA laboratories in Puerto Rico became overwhelmed by demand [12], not all specimens submitted for CHIKV testing by Veterans Affairs Caribbean Healthcare System (VACHS) in Puerto Rico (1 medical center and 8 outpatient clinics) were processed, including some specimens from patients who subsequently died. To further investigate, we queried the data warehouse for CHIKV testing orders. For patients without results, medical records were reviewed to determine whether tests remained pending, or if cancelled, the reason for cancellation. During December 2014, VHA PHSR provided a list of patients without results to laboratories in Puerto Rico to recover any remaining specimens for testing by PHRL. PHRL performs internally and externally validated CHIKV reverse transcriptase PCR (RT-PCR) (CDC, CHIKV RT-PCR assay protocol, Fort Collins, Colorado) [22] and IgM enzyme-linked immunosorbent assay (ELISA) (Abcam, Inc., Anti-Chikungunya Virus IgM Human ELISA kit, Cambridge, Massachusetts) according to manufacturer's recommendations. For patients who had specimens collected for CHIKV testing in Puerto Rico or elsewhere in the United States and died before the last query was performed, a physician-epidemiologist reviewed medical records to confirm presenting symptoms were consistent with CHIK (i.e. fever and either oligoarthralgia, polyarthralgia, or myalgia) and the patient resided in or had recently visited an area with CHIKV transmission, and determined whether CHIKV infection might have contributed to death (i.e., the patient had died before recovery from acute illness and death was not clearly from unrelated causes). Although PAHO defines a suspect case as a patient with acute onset of objective fever (>38.5°C or 101.3°F [23,24]; >38.3°C or 101.0°F [9]; >38.0°C or 100.4°F [9]) and severe arthralgia, not otherwise explained, who resides in or has visited epidemic or endemic areas 2 weeks before symptom onset [9,23,24], we did not utilize this definition because the majority of our laboratory-confirmed cases did not meet these criteria.
For all patients with available CHIK diagnostic test results, we abstracted demographics, clinical history, including symptoms during the acute illness, laboratory results, and outcomes from medical records. Comorbidities were identified by review of problem lists and provider notes. A laboratory-positive case was defined as a patient with detectable CHIKV RNA by RT-PCR or anti-CHIKV IgM antibody. Patients with positive serology, but without travel to an area with known CHIKV transmission, consistent with a false-positive result, as well as patients with inadequate testing to rule out CHIKV infection (negative RT-PCR >8 days after symptom onset and no serology or negative IgM <4 days after symptom onset and no RT-PCR or convalescent serology) were excluded from statistical analysis [24]. Demographics and clinical characteristics of patients with laboratory-confirmed infection diagnosed in Puerto Rico versus elsewhere in the United States were compared, and clinical findings of CHIKVpositive versus CHIKV-negative patients (all remaining patients with negative CHIKV tests) were compared by using chi-square test for categorical (Fisher's exact test when n <5 in any cell) and t-test for continuous variables.
To determine whether the association between age and hospitalization was modified by whether CHIK was diagnosed in Puerto Rico versus elsewhere in the United States, analysis was stratified by patient location. Poisson regression with robust error variance [25,26] was utilized to investigate risk factors for hospitalization among patients with laboratory-confirmed infection in separate age-adjusted models. Potential risk factors investigated included comorbidities, vital signs on presentation, and laboratory abnormalities at presentation and during clinical course. Analyses were performed by using SAS 9.2 (SAS Institute, Inc., Cary, North Carolina).
Finally, we sequenced the CHIKV E1 envelope glycoprotein gene from a convenience sample of patients from Puerto Rico with detectable CHIKV RNA. RNA was extracted from serum by using Qiagen DSP viral RNA mini kits (Qiagen, Germantown, Maryland) according to manufacturer's instructions. RT-PCR was performed by using Superscript One-Step Platinum Taq HiFi (Thermo Fisher Scientific, Inc., Waltham, Massachusetts) and primers CV1F and CV1R [27]. Resulting PCR products were purified by using Qiagen PCR purification kits and subjected to population-based sequencing by using standard dideoxyterminator techniques. Sequences were assembled, edited, and compared with sequences in GenBank from the Western Hemisphere and previous outbreaks using Geneious software (Biomatters, Inc., San Francisco, California).

Ethics Statement
This study was reviewed by CDC for human subjects protection and was deemed to be nonresearch. It was also approved by the Stanford University Institutional Review Board and fulfilled the requirements of regulation OHRP 45 CFR 46.116 (d) for waiver of informed consent. Patient data was anonymized after data abstraction.

Results
We identified 264 patients with CHIKV results during 2014 (Fig 1). Results for 184 patients from VACHS in Puerto Rico were available (22.8% of 806 patients from whom specimens were collected for CHIK diagnostic testing; testing was not completed for the remainder) ( Overall, 180 patients had laboratory-confirmed CHIKV infection (Table 1). Specimens for CHIKV testing from Puerto Rico were collected a mean of 4.5 d ± 5.4 d from symptom onset. Specimens from the rest of the United States were collected a mean of 28.7 d ± 40.3 d from symptom onset (p = 0.003). In Puerto Rico, 148 (82.2%) of 180 patients tested were CHIKVpositive; 145 (98.0%) were confirmed by RT-PCR and 3 (2.0%) by serology alone. Outside Puerto Rico, 32 (44.4%) of 72 patients tested were CHIKV-positive; 4 (12.5%) were confirmed by RT-PCR and 28 (87.5%) by serology alone. Fourteen (19%) of 72 patients tested for CHIKV infection outside Puerto Rico had no recorded travel to an area with CHIKV transmission prior to symptom onset, all of whom were CHIKV-negative; none of the patients tested in Puerto Rico had symptom-onset prior to possible exposure to circulating CHIKV. Patients with CHIKV infection in Puerto Rico were older (mean, aged 69 versus 55 years; p < .0001) and had more comorbidities than patients outside Puerto Rico.

Fatalities Associated with Confirmed and Suspected CHIKV Infections in Puerto Rico
Among the 148 patients with laboratory-confirmed CHIKV infection, 6 (4.1%) died. All had viremia demonstrated by RT-PCR (Table 5). In addition to these six patients, there were 15 patients who died after presenting with an illness compatible with CHIK that may have contributed to death; specimens for these patients were submitted for CHIKV testing but were not processed. Although attempts were made to obtain these specimens submitted to non-VHA laboratories, they were not recovered for testing. These 15 patients are thus considered suspect cases (data available upon request). The mean age of patients with laboratory-confirmed CHIKV infection in Puerto Rico who died was 78 years (range, 66-99 years), compared with 68 years (range, 23-94 years) for those in Puerto Rico who survived (p = 0.12). All 6 patients who died were afebrile on presentation (only one had a recorded temperature >100.4°F or >38.0°C during hospitalization); four were tachycardic on presentation. Of 5 with known medical history, all had multiple comorbidities.

Phylogenetic Analysis
CHIKV E1 envelope glycoprotein gene sequences (GenBank accession numbers: KU724228-KU724266) for 39 patients in Puerto Rico were closely related (<0.5% nucleotide difference) and nearly identical to strains in GenBank from St. Martin and the British Virgin Islands (Fig 3) [6,7].

Discussion
Our study is the first to characterize U.S. Veterans with laboratory-confirmed CHIKV infection. The majority received a diagnosis in Puerto Rico, although 32 were returning travelers who received a diagnosis elsewhere in the United States. Although the majority reported subjective fever, only one-third had objective fever, indicating that use of recommended CHIK case definitions, requiring objective fever, might underestimate disease burden among VHA patients [24]. Among returning travelers, approximately one-fifth were hospitalized, whereas in Puerto Rico, where patients were older and had more comorbidities, approximately half of laboratory-confirmed patients were hospitalized, 6.8% required intensive care, and 4.1% died. Missing data, findings on presentation: tachycardia (n = 2), fever (n = 2), leukocytosis (n = 4), leukopenia (n = 4), lymphopenia (n = 4), thrombocytopenia (n = 5), acute kidney injury (n = 14), hepatic transaminitis (n = 39); findings during clinical course: thrombocytopenia (n = 3), acute kidney injury (n = 13), and hepatic transaminitis (n = 36). RR, relative risk. a Reference groups for all categorical variables were patients with laboratory-confirmed CHIKV infection without the condition indicated.
doi:10.1371/journal.pntd.0004630.t004 The fatality rate was higher among VHA patients compared with a preliminary report from Puerto Rico in December 2014 that described only 4 deaths identified by passive surveillance on the island [12]. Although we performed population sequencing of only a portion of the E1 envelope glycoprotein gene, we did not find substantial difference among sequences from Puerto Rico compared with other strains circulating in the Western Hemisphere [6,7]. As in previous studies [18,19], we report that age was associated with increased risk for hospitalization. After adjusting for age, CHF (but not coronary heart disease), CKD, diabetes, and chronic lung disease were associated with increased risk for hospitalization. While we cannot be certain of the individual physicians' criteria for hospitalization in many cases, we do know that patients were more likely to be hospitalized if they had unstable vital signs (e.g. tachycardia) or had abnormal laboratory results (e.g. leukocytosis, acute kidney injury, or hepatic transaminitis) at presentation. Although literature review demonstrates low overall mortality, among persons with CHIKV infection presenting for care, the case fatality rate is not insignificant. During the CHIKV outbreak on Réunion Island (East/Central/South African genotype) [29], among 157 patients with laboratory-confirmed infection who presented to a medical center, 61.8% were hospitalized and 3.2% died [18]. Although patients in that study were >10 years younger and fewer had comorbidities, hospitalization and case fatality rates were similar to VACHS. They reported that diabetes and ischemic heart disease, unadjusted for age, was associated with increased odds of hospitalization [18]; only 11 patients in that study had CKD. Economopoulou et al. reported that among a subset of hospitalized patients with CHIK, cardiac disease, respiratory disease, as well as hypertension were associated with increased risk for severe disease [19].
Because our study included patients with laboratory-confirmed CHIKV infection, we cannot assess the overall burden of CHIKV infection among Veterans, only those who presented to VHA facilities and had appropriate diagnostic testing completed. During 2014, investigation of households surrounding laboratory-positive cases demonstrated that 28% of participants were laboratory-positive for current or recent CHIKV infection, and only 63% of symptomatic persons had sought care [12]. Although our query was robust, any patient with results not entered into the laboratory component of the electronic medical record (e.g., scanned or recorded in a progress note) would not have been captured. Sample size limited analysis of returning travelers, and in Puerto Rico, prevented inclusion of >1 comorbidity in age-adjusted models or assessment of risk factors for intensive care or death among patients with laboratory-confirmed infection. This study identified the lack of testing availability for VACHS, as well as deficiencies in CHIKV diagnostic testing across VHA. We identified 11 patients (4 in Puerto Rico and 7 elsewhere in the United States) with inadequate testing to diagnose CHIKV infection, which may have contributed to underdiagnosis of CHIKV-infection. In Puerto Rico this was because of underuse of serology for patients who presented >8 days after symptom onset. Outside Puerto Rico this was because of underuse of CHIKV RT-PCR or convalescent serology for patients who presented during the first week of symptom onset. Among returning travelers, many of whom presented for care in the U.S. during the convalescent period, when diagnosis is dependent upon serology, some diagnoses could have been missed as CHIKV IgM typically declines after several weeks to months [1]. Only 4 of the returning travelers had CHIKV IgM performed without simultaneous CHIKV IgG, and no patients had a negative CHIKV IgM and positive CHIKV IgG, however, suggesting that few cases may have been missed for this reason. Outside Puerto Rico, only 44% of patients tested for CHIKV had laboratory-confirmed infection; this not only reflects the lower prevalence of CHIKV infection outside Puerto Rico, but also improper testing of patients without symptom-onset after travel to an area with CHIKV transmission. Our surveillance activities determined that CHIKV testing availability for VACHS was lacking and resulted in VHA PHRL offering CHIKV (and dengue virus) testing. Further education of VHA providers regarding CHIKV infection, correct diagnostic testing (RT-PCR versus serology) on the basis of time from symptom onset, and available testing through VHA is needed.
Lessons learned from Puerto Rico are important for areas in the Western Hemisphere with ongoing CHIKV transmission as well as countries, including the United States, with similarly immunologically naïve populations and Aedes aegypti or Aedes albopictus vectors [6]. For clinical management, newly required CHIK public health reporting, and surveillance, having adequate laboratory testing capacity for timely results is helpful. Although testing all symptomatic persons might be infeasible, sufficient capacity to test those with severe (e.g. hospitalized patients) or atypical illness is needed. Clinicians practicing in areas with CHIKV transmission should be aware that CHIKV infection among elderly patients and patients with comorbidities, including CHF, CKD, diabetes, and chronic lung disease may be associated with more severe disease. To determine whether the risk of atypical complications is greater for CHIKV infection compared with other viral infections, a larger cohort of patients presenting with a viral syndrome would need to be studied. Further work to examine risk factors for intensive care and death among a larger sample of patients with laboratory-confirmed infection is needed to provide closer monitoring for those at greatest risk and to investigate the effect of prevention strategies [30] targeted to populations at greatest risk should they acquire CHIKV infection.