In this study, a longitudinal surveillance dataset spanning 25 years in China was used to investigate changes in the epidemiological characteristics of imported and indigenous dengue during the period of dramatic social-economic changes that has occurred in China over the last three decades. We found that the geographic distribution of provinces affected by imported and indigenous dengue has expanded, especially since 2000, and recently the incidence rate of indigenous dengue has increased dramatically with a peak in the most recent recorded year.
Magnitude and geographic extent of indigenous dengue
The incidence of dengue in China during the period 1990-2014 was lower than the epidemics in 1980 and 1985-1986, which resulted in more than 600,000 cases with 475 deaths overall in Hainan [17,25]. However, since 1990, indigenous dengue has not been limited to Hainan and Guangdong provinces, but has spread gradually from southern coastal tropical or subtropical regions (Guangdong, Guangxi, Hainan) to the neighboring northern and western regions (Fujian, Zhejiang, and Yunnan), and even to the central part of China - Henan province with a generally warm climate (Figure 5) [15,16,26]. Compared to the major epidemic in the 1980s, Hainan showed a dramatically decreased incidence of dengue with a few indigenous cases reported only in 1991 and 2014. Guangdong had the highest incidence of indigenous dengue over the last 25 years, with cases reported each year since 1997. Dengue transmission has also become evident in some previously unaffected areas, such as Ningbo city in the north of Zhejiang in 2004, Yiwu city in the inland part of Zhejiang in 2009, the central region of Henan province in 2013, and Nanping city in the central region of Fujian in 2014 [27-29]. This highlights the fact that the geographic range of dengue has apparently expanded in China, which is valuable information for consideration in national planning on dengue prevention and outbreak response. If dengue does continue to expand in China, this will need to be acknowledged in control planning, which currently focuses on Guangdong, Hainan, and Yunnan provinces in south China.
However, the number of reported dengue cases might be influenced by the change of diagnosis criteria and case definitions, especially through the introduction of more sensitive and rapid laboratory tests between 1990 and 2014, which could result in an increased number of reported cases without increased transmission. Compared to the 1988 criteria, the 2001 edition introduced the enzyme-linked immunosorbent assay (ELISA), immunofluorescence method, and dengue blot for serologic testing, RT-PCR for nucleic acid detection, and monoclonal antibody immunofluorescence for antigen detection. Then, the 2008 edition included MAC-ELISA for serologic testing and real-time fluorescence quantitative PCR for detecting nucleic acids, and classified a positive DENV-IgM result from a confirmed to a probable dengue case (see Additional file 2: Table S2) [19-21].
Demographic characteristics of imported and indigenous cases
The age and gender distributions of imported and indigenous cases in China differ in a number of ways. Imported cases were younger than indigenous cases, and were more likely to be male. This may reflect a population of younger working male adults who tend to travel more domestically and regionally and thereby have more exposure risk to dengue. In addition, the indigenous cases occurred across all age groups, including the elderly, which is different from other countries in Southeast Asia where dengue is endemic and where most dengue cases occur in children or younger adults . This pattern most likely is due to the fact that the population in China has very low seroprevalence of dengue antibodies, and is therefore broadly susceptible to dengue infection, whereas the population in dengue endemic countries has higher rates of immunity, especially in adults and the elderly [27,29,31]. However, the history of “mosquito bites” as part of the definition of imported case was impractical and likely introduced recall bias, which therefore probably underestimated the numbers and proportions of imported cases. A new guideline was issued in October 2014, which excludes “mosquitoes bite” in the definition of an imported case [23,32].
Dengue and Aedes mosquitoes
Aedes albopictus has been found in nearly one third of China and is the most predominant species in south China except in Hainan province, which has both types of Aedes mosquitoes . Aedes aegypti was implicated in outbreaks in Hainan in 1980 and 1985-1986 . However, Ae. albopictus was the only vector species present in the outbreaks reported in Guangdong, Fujiang, and Zhejiang from 2004 to 2010 [27,33]. The importance of Ae. albopictus in dengue outbreaks appears to be increasing in China, which is worrisome because Ae. albopictus seems to adapt easily to new environments, even in a temperate climate, and to be associated with the huge population migration and urbanization in China and climatic change [16,34]. However, under a national sentinel vector surveillance project for dengue, only 16 counties out of 483 counties in the five provinces in south China conducted Aedes mosquito surveillance between June and October since 2005, and China did not have a national vector control program for dengue . Therefore, it may be prudent for China to put more effort into mosquito surveillance and control for Ae. albopictus.
Dengue virus serotypes
In this study, we found all four serotypes of dengue virus in dengue patients in China, all of which are capable of causing dengue of any clinical severity [16,30]. DENV-3 was the first serotype documented in Guangdong in 1978  and in Hainan in 1980 . Then, in 2009 and 2010, DENV-3 was isolated again in Guangdong from imported cases, but the 2010 outbreak was not a reemergence of the 2009 strain . DENV-3 was also isolated during the outbreak in Zhejiang in 2009, in Yunnan in 2013, including from severe cases, and in the first outbreak in central China in 2013 [28,29,36]. DENV-1 has become the predominant serotype since the 1990s [27,37]. During 2005-2011, DENV-1 was the predominant serotype in circulation in Guangdong, while all four serotypes have been identified in indigenous patients from different outbreak localities since 2009 [31,38]. In addition, after an absence of 20 years since the DENV-4 outbreak in 1990, DENV-4 was detected during the outbreak in Guangzhou in 2010, in a Guangzhou resident who traveled back from Thailand . DENV-2 was confirmed in Hainan in 1985-1986 , and a few cases were reported in 2013 and 2014. The increasing diversity in DENV strains imported to China, especially in 2013 and 2014, might increase the risk of DENV outbreaks and their severity in the near future, as well as the difficulty of dengue control. Therefore, monitoring this viral diversity should be considered in the design of surveillance and control strategies for China.
Is dengue an endemic disease in China? Seasonality and virus source
Because of the geographic and seasonal restriction of cases, dengue in mainland China is still characterized as an imported disease and is not recognized as endemic . This characterization rests on the assumption that imported cases play a key role in initiating outbreaks in China [27,41]. From this study, we have shown that imported cases were reported in nearly every month during 2005 to 2014. However, indigenous cases were mainly reported from July to November, which indicates a strong seasonality to dengue transmission in China, with peak transmission occurring mostly in the hot and humid seasons. Two factors are likely to contribute to this pattern. Firstly, the large amount of rainfall from July to October increases the availability of breeding habitats of mosquitoes, thereby causing increases in mosquito population densities and the potential for dengue transmission . Secondly, transmission intensity can also fluctuate with temperature due to concomitant fluctuations in the length of the incubation period in the mosquito or mosquito mortality or blood feeding rates [43-45].
The dengue case data presented here represent only the clinically apparent infections which presented to health care facilities. Previous studies have shown that a large and variable proportion of DENV infections are clinically inapparent or mildly symptomatic [46,47], though adults are more likely to experience symptomatic illness than children . This suggests that there is likely a larger pool of DENV infections and cases in China than is represented in this dataset. However, the overall incidence likely remains low compared to that in neighboring endemic countries [49,50].
In addition, most of the first local dengue outbreaks in each city and year can be traced back to imported cases that sparked the outbreaks [27,33,39]. Although for some outbreaks initial imported cases cannot be identified [28,37], the molecular fingerprints of strains often suggest that the outbreak is likely due to viruses imported from other countries . Molecular epidemiological analysis in the last three decades also did not identify any new variants of viruses that are unique to mainland China . Although DENV-1 was predominant in most years in Guangzhou city during 2001-2010, the strains from each year belonged to different genotypes and none of them was found to be predominant, though Southeast Asian countries were generally found to be the most likely source . This suggests that dengue in China is due to localized transmission sparked by regular virus importations from returned travelers or visitors, rather than endemic transmission. Therefore, more attention should be directed toward the early identification of imported cases from other countries, especially from Southeast Asia.
There are some limitations in this study. Firstly, the data used were collected from passive public health surveillance. The data quality may be influenced by the key steps in surveillance including changing case definitions, reporting methods, availability of health facilities and laboratory diagnostics, under reporting, and completeness and accuracy of data over the years. Secondly, the individual case data were not reported before 2005, so demographic characteristics, laboratory confirmation, and the distribution of indigenous versus imported cases could only be analyzed from 2005-2014, and cases were not reported by the classification of disease severity.
Challenge of dengue control in mainland China
The expansion of global air travel and seaborne trade, and the huge population movements in China overcome geographic barriers for both disease vectors and pathogens, enabling them to move great distances in short periods of time [51-53]. With the rapid growth of the economy and urbanization in China, more and more people in China have moved away from their original residences, especially from central China to coastal provinces, and from poor rural areas to urban centers . This migration changes epidemiological dynamics and environments and can promote the transmission of dengue virus, increasing the population at risk of infection, and creating major challenges for prevention and control. Further, the increasing labor movements in and out of China to dengue endemic countries all over the world are driving changes in imported dengue dynamics.
The exceptionally high number of dengue cases in 2014 - a historical record since dengue became a notifiable disease in China in 1989 - serves as a reminder that even if dengue is not yet endemic in China, the possibility exists that the receptivity and vulnerability of certain areas to outbreaks could be increasing. Exploring the role of putative drivers of this huge outbreak, modeling and mapping the risk of importation and local transmission, and extracting lessons about how it could have been averted should be pursued immediately so as to inform future outbreak prediction and mitigation.