Introduction Dengue virus (DENV) transmission is endemic throughout most of the tropics and sub-tropics and is estimated to result in ∼50 million symptomatic infections and ∼20,000 deaths each year [1], [2]. Infection with any DENV can result in dengue, an illness characterized by fever, headache, retro-orbital eye pain, myalgia and rash [2]. In some cases, dengue can progress to severe dengue [2], which includes dengue hemorrhagic fever (DHF) and dengue shock syndrome (DSS) [3] and is characterized by thrombocytopenia, increased vascular permeability with plasma leakage, severe organ involvement, and/or clinically significant bleeding [2]. Supportive care with appropriate intravascular volume repletion has been shown to lower mortality associated with severe dengue [2]. The four related but serotypically distinct DENV-types, DENV-1, -2, -3 and -4, are transmitted by Aedes aegypti or Ae. albopictus mosquitoes [4], [5]. Following infection, individuals develop short-lived, heterotypic immunity and long-lived, type-specific immunity [6]. Primary infection is an individual's first DENV infection, and secondary infection is any subsequent infection with a DENV-type different from the first. Severe dengue is more common upon secondary infection [2], [7] and may be affected by the order in which an individual is infected with the respective DENV-types [2], [8]. Thus, increases in the force of DENV infection can result in a decrease in the age of primary and secondary infection [2]. Both local patterns of circulation of the four DENV-types and force of infection can influence the age groups most affected by dengue and severe dengue. The unincorporated United States territory of Puerto Rico is composed of 78 municipalities, an area of 3,515 square miles, and a population of 3,725,789 [9]. The demographics of Puerto Rico are similar to the United States as median age is 36 years and 78.6% are white, although 99% are self-described Hispanic [9]. Since the mid-1990's the health care system in Puerto Rico has included both public and private health care services, and dengue has been a reportable condition for several decades. Ae. aegypti is the predominant DENV vector on the island. Dengue was first described in Puerto Rico in 1915 [10] and outbreaks have been recognized since 1963 [11], [12]. DHF was first reported in 1975 [13], [14], all four DENV-types have been identified on the island since 1982 [15], [16], and the first confirmed dengue-related death was reported in 1986 [17]. Recent epidemics were detected in 1994–1995, 1998 and 2007, with 24,700 [18], 17,000 [19] and 10,508 [20] reported suspect cases, respectively (Table S1). During both epidemic and non-epidemic periods, 10–19 year olds have been the most affected age group for several decades. In the present investigation, we describe a dengue epidemic that occurred in 2010, including differences in the epidemiology of cases infected with different DENV-types with respect to primary versus secondary infection. Materials and Methods Investigation design A retrospective analysis of suspected dengue cases reported to surveillance systems was performed to: 1) describe the epidemiology of the 2010 dengue epidemic, including disease severity; 2) determine the proportion of primary and secondary DENV infections, and the molecular epidemiology of the DENVs responsible for the epidemic; and 3) describe relationships between demographic variables (e.g. age, sex, municipality of residence) and characteristics of illness (e.g. infecting DENV-type, severity of illness). This investigation underwent institutional review at CDC and was determined to be public health practice and not research, including the post-hoc determinations of DENV molecular epidemiology and primary/secondary infection rates in reported cases; as such, Institutional Review Board approval was not required. Data sources Surveillance data from five sources were used to identify cases. First, Centers for Disease Control and Prevention Dengue Branch (CDC-DB) and Puerto Rico Department of Health (PRDH) jointly operate the island-wide Passive Dengue Surveillance System (PDSS) that requires an acute serum specimen and completion of a Dengue Case Investigation Form (DCIF) (cdc.gov/dengue/resources/dengueCaseReports/DCIF_English.pdf) for case reporting and diagnostic testing. Second, the Enhanced Dengue Surveillance System (EDSS) operates solely in the municipalities of Patillas and Guayama and utilizes an on-site nurse epidemiologist to encourage case reporting and patient follow-up to obtain a convalescent serum specimen [21]. Third, identification of fatal dengue cases is conducted via PDSS and EDSS [22], and enhanced fatal case surveillance was initiated in January 2010 in collaboration with the Instituto de Ciencias Forenses de Puerto Rico, which obtains blood and tissue specimens at autopsy from suspected dengue-related deaths. Fourth, PRDH operates the Notifiable Diseases Surveillance System (NDSS) wherein suspected dengue cases are reported without diagnostic testing at CDC-DB. Last, in addition to dengue diagnostic testing performed at CDC-DB for PDSS and EDSS, testing is performed by two private diagnostic laboratories outside of Puerto Rico according to their internal protocols [23]. Diagnostic test results from these laboratories and patient data, including sex, age, and date of illness onset (if unavailable, specimen collection date was used instead), were entered into an independent database. Deduplication of individuals reported to more than one data source was achieved by matching records on name and date of birth and consolidation into a single case if two or more reports from any data source had symptom onset dates within 14 days of each other. As case-patients' travel history is not well captured via the surveillance systems used in this investigation, reported cases may represent both locally-acquired as well as travel-associated cases. Dengue diagnostic testing All diagnostic testing was performed at CDC-DB for serum specimens received through PDSS or EDSS using the following algorithm: acute specimens (collected ≤5 days after symptom onset) were tested by DENV-type-specific real-time reverse-transcriptase-polymerase chain reaction (RT-PCR) [24] adapted for high throughput using MDX-10 Universal and M48 systems (Qiagen, Valencia, CA); acute specimens collected 5 days after symptom onset and all convalescent specimens (collected ≥6 days after symptom onset) were tested for the presence of anti-DENV immunoglobulin M (IgM) antibody with an antibody-capture enzyme-linked immunosorbent assay (MAC ELISA) and a cut-off value of the OD450 of the specimen versus that of the negative control (ie. P/N ratio ) ≥2.0 [25], [26]. All serum specimens from fatal cases were tested by both RT-PCR and MAC ELISA. Tissue specimens were tested at CDC Infectious Diseases Pathology Branch in Atlanta, GA by immunohistochemistry (IHC) [27] and flavivirus-specific RT-PCR [28] followed by sequencing. Definitions A suspected dengue case was a dengue-like illness in a person in Puerto Rico whose health care provider: 1) submitted a DCIF and serum or tissue specimen to CDC-DB for dengue diagnostic testing; 2) submitted a serum specimen to a private laboratory for dengue diagnostic testing; or 3) reported the case via NDSS. A laboratory-positive case was a suspected dengue case that met the following criteria for current (criteria 1 and 2) or recent (criterion 3) DENV infection: 1) detection of DENV nucleic acid in a serum or tissue specimen; 2) detection of DENV antigen in a tissue specimen; or 3) detection of anti-DENV IgM antibody in a serum specimen. A laboratory-negative case was a suspected dengue case with: 1) no anti-DENV IgM antibody detected in a convalescent specimen; or 2) no DENV nucleic acid or antigen detected in a fatal case with only a tissue specimen submitted. A laboratory-indeterminate case was a suspected dengue case with no DENV nucleic acid or anti-DENV IgM antibody detected in an acute specimen with no convalescent specimen available for testing. Dengue with warning signs and severe dengue were defined according to 2009 WHO clinical guidelines [2]; dengue, DHF and DSS were defined according to 1997 WHO clinical guidelines [3]. Primary and secondary DENV infections A representative sample of all RT-PCR-positive cases reported to PDSS or EDSS with illness onset between January 1 and December 31, 2010 was selected to determine the rates of primary and secondary DENV infection. Cases were stratified by age group with optimal allocation to allow for comparison between age groups, and further allocated to reflect the proportion of DENV-types that occurred during 2010 to allow for comparison between DENV-types and age groups. Sample size was calculated using an estimate of the proportion of secondary infections by age group based on data from the 2007 dengue epidemic [20], an error of 20%, 95% significance, and an expected 20% of specimens having insufficient specimen volume remaining for testing to be completed. Of the 1,000 selected cases, 818 had sufficient specimen volume and were tested at a dilution of 1∶100 for the presence of anti-DENV IgG antibody by ELISA using DENV-1–4 antigen and a cut-off value of OD450≥0.15 [29], [30]. A secondary DENV infection was defined by detection of anti-DENV IgG antibody in an acute specimen, and a primary DENV infection by lack of anti-DENV IgG antibody detection in an acute specimen. Sequencing and phylogenetic analysis Serum specimens with DENV-1 (n = 7), DENV-2 (n = 2) or DENV-4 (n = 4) detected by RT-PCR were randomly selected from municipalities with the highest incidence of the respective DENV-type and inoculated into cultured C6/36 cells; the presence of virus was confirmed by RT-PCR and indirect immunofluorescence [31]. Isolates were further propagated and viral RNA was extracted from culture supernatants using the M48 BioRobot System (Qiagen; Valencia, CA). The envelope glycoprotein (E) gene was amplified and sequenced; sequence data were restricted to the E gene open reading frame (1,485 basepairs). Multiple sequence alignment was performed using MUSCLE available in MEGA 5 (megasoftware.net) and GTR+Γ+I4 was selected as the best nucleotide substitution model as determined by MODELTEST v3.7. Genetic relatedness was inferred and represented with phylogenetic trees using the maximum likelihood method in MEGA 5. MCMC was run in BEAST v1.6.1 (beast.bio.ed.ac.uk) under Bayesian skyline prior, constructed in TreeAnnotator found in the same BEAST package, and visualized in FigTree v1.3. Both trees rendered almost identical tree topologies, therefore confirming genetic relatedness. Evolutionary distances were corroborated by pairwise alignment in BioEdit v7.1.3 and E gene sequences from GenBank were included in the phylogenetic tree to support tree topology by currently circulating genotype. Tree topology was supported by bootstrapping with 1,000 replicates. Genotypes were referred to by previously described nomenclature [32], [33]. Statistical analyses The frequencies of clinical, demographic and laboratory data were calculated by performing descriptive analyses of all suspected dengue cases identified in 2010. Rates of suspected dengue and laboratory-positive cases were calculated using population denominators obtained from the 2010 United States Census [9]. Statistical differences in proportions were tested by applying the Chi-squared test and Fisher's exact test when applicable. Unless otherwise noted, relative risk ratios were used to calculate all differences between effect sizes. All data analyses were conducted using SAS 9.2 (SAS Institute Inc., Cary, NC), graphs were produced in Microsoft Excel (Microsoft Corp., Redmond, WA), and maps were created using ArcView (ESRI, Redlands, CA). Results Suspected cases We identified 26,766 suspected dengue cases with illness onset between January 1 and December 31, 2010 (7.2 suspected dengue cases per 1,000 residents). Of these, 22,496 (84.0%) were reported to PDSS, 1,846 (6.9%) were identified though diagnostic testing at a private laboratory, 1,304 (4.9%) were reported to NDSS, and 1,120 (4.2%) were reported to EDSS (Fig. S1). Suspected dengue cases exceeded the PDSS epidemic threshold in the first week of 2010, increased steeply in week 20 (May 14–20), and peaked at 1,157 in week 32 (August 6–12) (Fig. 1). Suspected dengue cases slowly declined thereafter and returned to below the historic average in mid-December. 10.1371/journal.pntd.0002159.g001 Figure 1 Epidemic curve of suspected dengue cases by week of illness onset, Puerto Rico, 2010. Surveillance data from cases reported via the Passive Dengue Surveillance System, Enhanced Dengue Surveillance System, Notifiable Diseases Surveillance System, or private laboratory dengue diagnostic test results were compiled and grouped by diagnostic test result as indicated. Of all suspected dengue cases, 25,852 (96.6%) had a specimen tested for evidence of DENV infection, of which 25,246 (97.7%) were tested by CDC-DB and the remainder by a private laboratory; paired specimens were available for 1,996 (7.5%) cases. Of all cases with a specimen tested, 3,664 (14.2%) were laboratory-negative, 10,140 (39.2%) were laboratory-indeterminate, and 12,048 (46.6%) were laboratory-positive (3.2 laboratory-positive cases per 1,000 residents). The median weekly proportion of cases that tested laboratory-positive was 48.3%, and was highest (64.5%) in week 24 (June 11–17) and lowest (11.1%) in week 53 (December 31). Laboratory-positive cases Laboratory-positive case-patients resided in all 78 municipalities of Puerto Rico (Fig. 2A), and the median rate of laboratory-positive cases by municipality was 2.68 per 1,000 residents. Rates were the highest in the municipality of Patillas (16.34 cases per 1,000 residents), the southeastern municipality where the EDSS site is located [21], and lowest in Aibonito (0.12 cases per 1,000 residents) in the mountainous center of Puerto Rico. Of 7,426 RT-PCR-positive cases, DENV-1 was detected in 5,126 (69.0%) and incidence was highest in the southeast (Fig. 2B). DENV-2 was detected in 545 (7.3%) cases primarily in the west (Fig. 2C), whereas DENV-4 was detected in 1,757 (23.7%) cases and incidence was highest in south-central and northwestern Puerto Rico (Fig. 2D). DENV-3 was detected in just two ( 0.05) except for the 50–59 year-old age group, for which serum specimens were collected later (median: 6 days post-illness onset [DPO]) than all other age groups (median: 4 DPO) (Fisher's exact, p = 0.04) and thus tested less frequently by RT-PCR. Despite this, the distribution of DENV-types was not consistent among age groups (Fig. 3B). The strong majority (89.3%) of RT-PCR-positive cases in individuals 1–4 years of age were due to infection with DENV-1, whereas 8.1% and 2.6% were due to infection with DENV-4 and -2, respectively. The percent of infections due to DENV-1 decreased and those due to DENV-4 increased with age until a plateau of approximately 60% DENV-1, 30% DENV-4 and 10% DENV-2 was reached in the 20–29 year old age group. Primary and secondary DENV infections From the sample of 818 RT-PCR-positive specimens tested for primary versus secondary DENV infection, 169 (20.7%) were primary and 649 (79.3%) were secondary. The median age of individuals experiencing primary infection was 14 years, compared to 23 years for individuals experiencing secondary infection. Eighty-one percent of individuals 1–4 years of age had primary infection and were the only age group for which primary infection was significantly more common than secondary (p = 0.003) (Figure 3C). More than 89% of infections in all adult age groups (i.e. age ≥20 years) were secondary. The frequency with which anti-DENV IgG antibody was detected in specimens taken from infants was likely due to the presence of maternal antibody [2]. Whereas 28.5% of all DENV-1 infections were primary, significantly fewer DENV-2 (6.8%) and DENV-4 (7.1%) cases were primary infections (p 90% of case-patients on only one occasion and some data variables were incompletely reported (e.g. only 56% of suspected cases had a reported status of hospitalization), severity of disease and the rates of dengue with warning signs and severe dengue reported here were likely underestimated. Finally, the description of the epidemiology and molecular characteristics of dengue reported here is only representative of reported, clinically apparent DENV infections and may not be reflective of asymptomatic and sub-clinical DENV infections. The 2010 dengue epidemic in Puerto Rico demonstrated that dengue continues to be a public health concern for Puerto Rico residents and visitors, and surveillance systems and control initiatives should continue to be supported and strengthened. This epidemic also highlights the need for effective primary prevention tools such as a dengue vaccine to reduce disease morbidity and mortality. Supporting Information Table S1 Summary of epidemiologic data from previous dengue epidemics in Puerto Rico. (DOCX) Click here for additional data file. Figure S1 Flow diagram of data sources, diagnostic test results, and sub-analyses of suspected dengue cases, Puerto Rico, 2010. A: Data sources and diagnostic test results. B: Sub-analyses using RT-PCR-positive specimens. PDSS = Passive Dengue Surveillance System; PrivLab = private diagnostic laboratories; NDSS = National Disease Surveillance System; EDSS = Enhanced Dengue Surveillance System; IHC = immunohistochemistry; IgG ELISA = anti-DENV immunoglobulin G enzyme-linked immunosorbent assay; RT-PCR = real-time reverse-transcriptase polymerase chain reaction; DENV = dengue virus; * = includes two co-infections. (TIF) Click here for additional data file.