In 2010, Puerto Rico experienced the largest and longest dengue epidemic ever documented on the island. In total, more than 12,000 individuals had laboratory-confirmed dengue, of which more than 1,300 experienced severe dengue and 40 died. The most common DENV identified was DENV-1, and 1–4 years old were the only age group more frequently experiencing a primary versus secondary DENV infection. Individuals infected with DENV-1 were four times more likely to have a primary infection than were those infected with DENV-2 or -4. A strength of this investigation was utilization of multiple surveillance systems to identify all reported suspect dengue cases. However, a minor weakness was that data obtained from each system may not be directly comparable due to different diagnostic algorithms used by CDC-DB and private laboratories, and we were not able to determine status of primary versus secondary infection or perform sequencing on specimens from private laboratories. Because private laboratories contributed <5% of all laboratory-positive dengue cases, this likely did not affect the conclusions of this investigation.
The 2010 dengue epidemic was similar in several respects to the 1998 epidemic: both began in January during El Niño events accompanied by above average temperatures, which while not a determinant of epidemics in Puerto Rico 
may contribute to increased DENV transmission 
; and both epidemics peaked in week 32 of the calendar year and were predominated by transmission of DENV-1 and -4 
. A notable difference was that DENV-3 was essentially absent in 2010, whereas it accounted for ~6% of cases during the 1998 epidemic 
. DENV-3 was re-introduced into Puerto Rico in 1998 following a 20-year absence and was the predominant virus-type in the 2007 dengue epidemic 
. Thus, susceptibility to DENV-3 infection was likely high in 1998 and low in 2010, which likely explains these observations.
The American-African and Indonesian genotypes of DENV-1 and -4 have been circulating in Puerto Rico since introduced in 1978 and 1981, respectively 
. However, the DENV-1 isolated in 2010 was distinct from the DENV-1 isolated during the 1998 epidemic ( and 
) and was more closely related to the DENV-1 isolated during the 2007 epidemic (). Similarly, the DENV-4 isolated during the 2010 epidemic was distinct from the DENV-4 isolated in 1998 and was more closely related to viruses circulating since 2004 (). These findings suggest that DENV-1 and -4 may have both experienced clade replacements at some point after 1998 but prior to 2007. After the re-introduction of DENV-3 into Puerto Rico in 1998, DENV-1 was not detected between 2001 and 2006 and DENV-4 was not detected between 2000 and 2005 
. Nonetheless, apparent re-introductions of DENV-1 in 2007 and DENV-4 in 2006 were soon followed by the disappearance of DENV-3 in 2010 (this paper and 
). In place of the convenience sample used in this investigation to describe the DENVs responsible for the epidemic, sequencing of a representative sample of specimens and longitudinal sequence analysis will be necessary to both confirm apparent clade replacements and determine if other DENV clades contributed to the 2010 epidemic.
Similar to previous epidemics in Puerto Rico (Table S1
), 10–19 year olds were most affected during the 2010 epidemic; however, unlike previous epidemics, 5–9 year olds were the next most affected age group. The median age of individuals experiencing secondary DENV infection declined from 27 years in 2007 
to 23 years in 2010, likely due to the relative proximity of the periods of high infection pressure. Taken together, these observations indicate an increase in incidence of dengue and a decrease in the age of secondary infection, suggesting that the overall force of DENV transmission may have been higher in 2010 than in previous epidemic years.
The observation that DENV-2 and -4 cause relatively infrequent clinical apparent illness upon primary DENV infection is consistent with previous studies 
. Similarly, our finding that DENV-1 was a more frequent cause of clinically apparent illness upon primary infection has also been previously reported 
, including the observation of increased disease severity during primary infection with DENV-1 compared to other DENV-types 
. Nonetheless, of 545 DENV-2 and 1,755 DENV-4 infections, roughly 7% were primary, indicating that primary infection with these DENVs can cause clinically apparent illness, contrary to previous assertions 
. The relative abundance of DENV-1 compared to DENV-2 and -4 is unlikely to be responsible for the observed differences in likelihood of causing clinically apparent illness upon primary infection, as relative risk ratios compare the proportion of exposed individuals experiencing the outcome of interest. This is supported by the findings in the 1–4 year-old age group, of which ~80% experienced a primary infection with DENV-1. Alternative explanations for these observations include potential variations in the sensitivity of detection of DENV-type-specific anti-DENV IgG antibody and differences in force of infection between the DENV-types circulating in 2010.
We also saw that DENV-1 and -2 were less frequently a cause of severe dengue than DENV-4. This is in contrast to previous studies where DENV-1 was a more frequent cause of DHF than DENV-4 
, and a study where DENV-2 was twice as likely to result in DHF as DENV-4 
. Possible explanations for these differences include: the comparatively small number of DENV-4 infections observed in previous studies; differences in clade and/or viral fitness leading to differential pathogenicity 
; and/or the DENV-type(s) and sequence to which individuals were previously exposed, which may affect the likelihood of developing severe dengue 
This investigation had several limitations. First, because individuals experiencing secondary infection may have a diminished anti-DENV IgM antibody response 
, suspected dengue cases tested solely for anti-DENV IgM antibody may have been misclassified. Second, although DENV is the sole flavivirus known to cause clinically apparent illness in humans in Puerto Rico (CDC, unpublished data), some proportion of anti-DENV IgM or IgG positive results could have been due to infection with or vaccination against another flavivirus 
, resulting in misclassification. Third, because clinical data was provided for >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.