Dengue Viral Infection And Risk Of An Epidemic Biology Essay

Published:

Dengue is an arbovirus disease caused by four dengue virus serotypes (DEN-1, DEN-2, DEN-3 and DEN-4) and transmitted to humans by mosquito as a vector.

2.1.1 History overview

It was believed that dengue viral infection has a long history. First reported cases that were believed to be dengue-like syndrome were recorded in China in 992 AD. They were followed by the sporadic outbreaks in the French West Indies in 1935 and in Panama in 1699. However, the major epidemics of illness believed to be dengue were reported in Asia, Africa and North America in 1779 and 1780 (6). At that time, the etiology was not clear until researchers enabled to isolate and characterize the dengue viruses in laboratory during World War II. So dengue transmission before 1940 was characterized by infrequent, mostly sporadic cases and uncertainty of etiology. The transmission increased significantly in Southeast Asia and Pacific during and after World War II when the ecology disruption and demographic changes occurred. The condition after World War II was suitable for the transmission and the co-circulation of multiple dengue virus serotypes. As a result, Dengue Hemorrhagic Fever (DHF) emerged in this region. The first epidemic of DHF occurred in Manila, Philippines (1953 to 1954), was followed by Bangkok, Thailand (1958) and Malaysia, Singapore and Vietnam (1960) and spread to the whole region during the 1970s. In addition, the transmission spread further to other regions in Asia: westward to India, Sri Lanka, Maldives and Pakistan and eastward to China. The number of epidemic also increased in the Pacific Islands in 1970s. Therefore, DHF became a major public-health problem in South East Asia regions and wide spread in the Asian continent.

Lady using a tablet
Lady using a tablet

Professional

Essay Writers

Lady Using Tablet

Get your grade
or your money back

using our Essay Writing Service!

Essay Writing Service

Yet, the transmission was slightly different from what happened in America and Africa. In the former continent, the epidemic of dengue was rare from 1950s until 1970s due to the successful eradication of A. aegypti, the principal vector, from most of Central and South America. However, control programs were not sustained and there were re-infestation of the mosquito. As a result, the major epidemics of dengue occurred , first in Cuba in 1981, after it had been free from this disease for over three decades (7). Annual cases have been reported almost every year since this moment. Up to now, more than thirty countries in America reported dengue cases and multiple dengue serotypes circulate in this region. In addition, the total number of DHF cases reached to 106,037 and the case fatality was 1.2% from 2001 to 2007 (1). Speaking about DF/DHF in Africa, there is little information due to poor dengue surveillance. However, epidemic DF caused by all four dengue serotypes has increased since 1980 with which the most epidemics occurred in the eastern Africa and to a smaller extent in the western Africa (1).

The emergence of the epidemic DF/DHF as a global public health problem related to some factors, such as global population growth, unplanned and uncontrolled urbanization with substandard living conditions, lack of vector control, virus evolution, international traveling, changing in public health policy, inadequate water supply, disparity and poor sanitation (7-10). Of all these factors, urbanization has probably had the most impact on the amplification of dengue within a given country, and international traveling has had the most impact for the spread of dengue from country to country and continent to continent (10).

Furthermore, the pattern of the disease observed varies across places and time. Seasonality and cyclical patterns were presence in a majority of endemic countries. In general, the epidemics occur every three to five years. It is perhaps due to the combination of demographic, immunologic and environmental changes. Besides this, climatic factors, such as the El Niňo Southern Oscillation (ENSO) and global warming may have contributions to the cyclical pattern of dengue activity (10, 11).

2.1.2 Burden of disease

In the last fifteen years, Dengue/DHF has widely accepted as the most important arthropod borne viral disease of humans. The incidence have increased thirty folds with hyper endemic transmission expanding geographically to new countries, from urban to rural setting as shown in Figure 2.1.

Source: WHO. Dengue: guidelines for diagnosis, treatment, prevention and control, 2009

Figure 2.1 Areas at risk of dengue transmission, 2008

The disease burdens are intense among countries in South-East Asia and the Western Pacific. However dengue fever and DHF/DSS were endemic as well in Africa, America and The Eastern Mediterranean, particularly in tropical and sub tropical countries. (12). Over 100 countries reported this disease and approximately 2.5 to 3 billion people live in these endemic areas. It is estimated that 50 million infections occur annually, including DHF cases and deaths which are respectively 500,000 and 21,000 (9). Moreover, the wide spread of the transmission can be observed also from the countries reporting cases to WHO. During nine years from 1970 to1979, only nine countries in the world had dengue epidemics, but twenty years later the number of affected countries had increased more than four times. The average number of DHF cases which were reported to WHO had increased dramatically from 908 in the 1950s to 925,896 in the period of 2000 - 2007, as described in Figure 2.2 (1, 12). While the case fatality rate (CFR) varies among countries, less than 1% in some but can be as high as 10-15% (2).

Lady using a tablet
Lady using a tablet

Comprehensive

Writing Services

Lady Using Tablet

Plagiarism-free
Always on Time

Marked to Standard

Order Now

Source: WHO. Dengue: guidelines for diagnosis, treatment, prevention and control, 2009

Figure 2.2 Average number of DF/DHF cases reported to WHO and countries reporting dengue 1955 - 2007

Dengue inflicts a significant health, economic and social burden in endemic areas (1). Children mostly suffer from DHF/DSS, with the average hospital stays of 5-10 days for severe cases. There are both direct and indirect costs for each dengue patient, from inconvenience with uncomplicated DF to costs for hospitalization and significant disruption of earning potential. In addition, there are costs to local municipalities for vector control activities, and often revenue lost through reduced tourism (13). Globally the estimated number of disability-adjusted-like-years (DALYs) lost to dengue in 2001 was 528 (1). Prospective studies on the cost of dengue cases in eight countries in America and Asia found that the overall mean costs were $514 for an ambulatory and $1,394 for hospitalized cases. With an annual average of 574,000 cases reported, the aggregate annual economic cost of dengue for the eight study countries in Asia and America is at least $587 million. Preliminary adjustment for under-reporting it could raise this total up to $1.8 billion, and incorporating costs of dengue surveillance and vector control would raise the amount further. Dengue imposes substantially financial burden on both the health sector and the overall economy (14).

2.1.3 Dengue in South-East Asia and Western Pacific Regions

Approximately 1.8 billion (more than 70%) of population at risk for dengue live in South-East Asia and Western Pacific Regions (1). After first being recognized as DHF in the Philippines in 1953, the mean number of annual cases of DHF has increased from below 10,000 in the period of 1950-1960 to 46,458 and 188,684 in 1986 and 2006, respectively. The trend of reported cases is rising, whereas the case fatality rate is maintained below 1% (4, 15), as shown in Figure 2.3.

Source: http://www.searo.who.int/en/Section10/Section332_1101.htm.

Figure 2.3 Trend of dengue cases and CFR as reported by countries in SEAR (1986 - 2006)

Thailand and Indonesia had significantly the highest dengue cases among all countries in these region. As can be seen in Figure 2.4, up to 2004, approximately half of cases in this region came from Thailand alone. Since then, Thailand has likely been successful to control this disease up to less than 50.000 cases per year. Whereas Indonesia shows an upward trend and become the leading countries with dengue cases in this region since this time. A note to mention, the pattern of dengue transmission in the DHF endemic countries of this region is not very different. Firstly, the cases of DHF were sporadic; then they were followed by DHF epidemics which progressively became more frequent until DHF cases are seen every year. Meanwhile, major epidemics can be observed every three to five year intervals. It is equally important that all four dengue serotypes are present in this region (13).

Source: http://www.searo.who.int/en/Section10/Section332_1101.htm.

Figure 2.4 South East Asia Region Country Report for Dengue Cases (1985-2006)

The countries of the region have been stratified into four climatic zones. First is Thailand, Myanmar, Indonesia, Srilanka and Timor Leste where dengue becomes a major public health problem mostly during the tropical monsoon and the equatorial zone. Here, A. aegypti is widespread in both urban and rural areas, and multiple virus serotypes are circulating. As a matter of fact, Dengue is a leading cause of hospitalization and death among children in these countries. Included in the second zone is Bangladesh, India and Maldives, characterized by cyclic epidemics which commonly occur during dry and wet climatic zones. By this manner, it leads to an increase in multiple virus serotypes circulating in the population. The third climatic zone is Bhutan and Nepal which have epidemic dengue activities reported over the past four years. The last zone is only The Democratic Peoples' Republic of Korea which has no reports of dengue (1).

2.1.4 Dengue in Indonesia

Dengue virus infection was firstly reported in the cities of Jakarta and Surabaya in 1968. The number of cases reported was 58 with 24 deaths (CFR= 45.3%) at that time (5). Since then, cases have increased and effected more areas. After four decades, the incidence of DHF increased almost three folds from the first outbreak and the transmission extends to new areas.

More areas are affected and the number tends to increase every year. Up to 2009, the disease has spread to 382 districts which is over 80% of the total number of districts (Ministry of Health Republic of Indonesia, 2010, unpublish data), as shown in Figure 2.5.

Lady using a tablet
Lady using a tablet

This Essay is

a Student's Work

Lady Using Tablet

This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.

Examples of our work

Source: Sub Directorate of Arboviruses, MOH RI

Figure 2.5 Number of District/Municipality with DHF (1968 - Nov 2009)

Still in the same year, the total number of DHF cases was 156,052 though CFR was maintained below 1% (16).

The thematic map shows the distribution of DHF incidence in all provinces in 2009 (Figure 2.6). Provinces with the high incidence of DHF (more than 50 per 100,000 population) can be seen in every big island. Indeed, Kalimantan was almost covered by provinces with high incidence of DHF. However, provinces in Java Island contributed more cases among other island since that is the most populous island in Indonesia.

Kalimantan Island

Papua Island

Sulawesi Island

Jawa Island

Source: Sub Directorate of Arboviruses, MOH RI

Figure 2.6 DHF incidence by province in Indonesia, 2009

The incidence of DHF in Indonesia fluctuates annually due to outbreaks every year in different areas of Indonesia, illustrated in Figure 2.7 (17).

Source: Sub Directorate of Arboviruses, MOH RI

Figure 2.7 DHF incidence and CFR in Indonesia (1968-2009)

There were five major epidemics which occurred in 1973, 1983, 1988, 1998, and 2004. However, DHF incidences have been increasing year by year since 2004 (16). The outbreak trends of dengue initially have been characterized as a cyclic pattern recently become irregular with a high endemic background (18) and are transmitted to both urban and rural areas. Even though DHF can occur at any time of year, the monthly peak incidence of DHF usually increases from December to January which coincides with the rainy season. However, in big cities such as Jakarta, Bandung and Surabaya, the incidence reaches its peak in April and May (17).

Speaking about virus serotype which circulates in Indonesia, indeed, all dengue serotypes were found. Some studies provided the evidence and marked that serotype 3 is the most frequently associated with the severe infection (18-21). In addition, A. aegypti, the most domesticated mosquito, is the principal dengue vector in Indonesia. The vector is widely spread and difficult to control.

By all means, DHF becomes the major public health problem and government declares it as health priority. For this reason, DHF control becomes one of the ten national priority health programs (22).

Among all provinces in Indonesia, Jakarta had the highest incidence. It is important to remember when the national outbreak occurred in 2004; Jakarta was the most affected area and suffered greatly from the dramatically rising cases in a short time(23).

Source: Jakarta Province Health Office ( unpublished data)

Figure 2.8 Graph of reported DHF incidence in Jakarta Province

Figure 2.8 shows temporal distributions of annual DHF incidence in Jakarta for four decades. Obviously, we can see that CFR has been decreasing significantly to less than 1% since 2000. On the other hand, the incidence was increasing dramatically in the last decade. In the setting of high population density, urbanization, and unoptimal of vector control program, the transmission of dengue still persists at a high level in this province. Under those circumstances, it is not surprising if Jakarta becomes the highest province with DHF among all provinces in Indonesia recently.

2.1.5 Transmission of Dengue

Interaction between virus, host and vector contributes to the clinical spectrum and geographic distribution of dengue infection. Each factor influences and effects the others in the ecology setting (24).

2.1.5.1 Dengue Virus

Dengue virus is a single - stranded RNA virus. It belongs to the genus Flavivirus and the family Flaviviridae. The virus consists of four distinct serotypes (DEN-1, 2, 3, 4) which can be identified by serological methods. Infection in humans by one serotype produces lifelong immunity against re-infection by that same serotype, but there is no cross-protective immunity to other serotypes (6).

All four dengue serotypes cause a wide range of clinical manifestation, from asymptomatic or mild (DF) to more severe disease such as DHF and DSS (11). However, the predominant serotype associated with the severe manifestation varies in different epidemics. A systematic review conducted by Kyle and Harris (2008) shows that DEN-2 have mostly associated with the severe disease (DHF/DSS)(24). In contrast, studies from Indonesia have found that DEN-3 was the frequently predominant serotype in the severe disease (19, 25), and this serotype is found as the most dominating serotype in Indonesia (19, 25-28). Furthermore, some studies confirmed that the secondary infection by any serotype has been associated with the severe disease (29-33).

2.1.5.2 The vector

The vector for dengue transmission is female Aedes (Stegomyia) spp. mosquito. Depending on the geographical distribution, the species include A.aegypti, A. albopictus, A. polynesiensis and A. scutellaris groups (6). However, the principal vector of dengue transmission is A. aegypti. This domesticated tropical mosquito is found widely and spread worldwide between the latitudes of 35°N and 35°S but is relatively uncommon in high altitude areas (above 1000 meters) due to lower temperatures (1, 15). The mosquito adapts well to the urban environment. It prefers to lay its eggs in an artificial container and is mostly found in and around human dwellings. It prefers to rest indoors and feed on humans during daylight hours. Importantly, female mosquitoes often feed on several persons during a single blood meal which allows the mosquitoes to transmit dengue virus to multiple persons in a short time. Thus, A. aegypti is an efficient vector to cause an epidemic (6).The secondary vector of DEN is A. albopictus which is found in Southeast Asia, in the Western Pacific and recently in Central and South America(24). This Asian tiger mosquito is common in coastal and inland rural areas (24, 34).

There are four distinct stages of Aedes life cycle including egg, larvae, pupae, and adult. Figure 2.9 shows the life cycle of Aedes mosquito. Firstly, the eggs are laid on wet walls of containers with water. It can resist desiccation for several months. The next stage is hatching of larva. Larva will feed on microorganism to be able to grow until the stage of a fourth instar. Once the larva pass through the fourth instar, metamorphosis occurs into a pupa. Larvae and pupae prefer to live in clean water on the artificial containers such as water storage containers (tanks, jars, pots), rubber tires, or bottles (35). Unlike larvae, pupae do not feed. It just transforms until the flying mosquito is formed. The duration of the entire stages is 8-10 days under an optimal condition at a room temperature (36). However, the average lifespan for females mosquitoes is about 8 to 15 days with the flight range of 2.5 km per day in an open environment to less than 25 m in an urban environment (35). The dispersal is influenced by oviposition sites and blood meals, mostly limited to within 100 meters of the site of the emergence (3).

Source: http://www.cdc.gov/dengue/entomologyEcology/m_lifecycle.html.

Figure 2.9 Life cycle of A. aegypti

Some studies show a relation between meteorological factors such as temperature, humidity and rainfall with changes in the Aedes mosquito's life cycle. An increase in the temperature would lengthen the longevity of mosquito, shorten the extrinsic incubation period of DEN and also shorten the gonotrophic cycle of mosquito , resulting in more infected mosquito transmitting DEN (37). Longitudinal studies in southwestern Puerto Rico show a positive correlation between the rainfall and Aedes density (38) although other studies in Bangkok conclude that the abundance of adult Aedes was not related to the rainfall (37). These evidences show that an increase in Aedes density may be associated with factors other than rainfall, such as types of breeding sites and human behaviors. At last, Jansen CC and Beebe NW have concluded that meteorological factors alone would not change the geographical distribution of Aedes mosquito. The history of global dengue distribution was strongly related to urbanization, infrastructure development, socioeconomic situation and policy (39).

2.1.5.3 The host

Human beings and several species of lower primates are hosts for dengue infection. Indeed, human beings are the main urban reservoir of the viruses after being bitten by the infected female Aedes mosquito (3).

There are evidences that some factors contributed to an increased risk for severe dengue (DHF/DSS) in human including age, ethnicity and host genetic background, pre-existing immunity from a former dengue virus infection, sequence of infecting serotypes and time between infections (24). Mostly, epidemiology studies found that children under the age of 15 years old are at an increased risk for severe dengue (24, 33, 40). This may be related to increased capillary fragility and less ability of this age group to compensate for capillary leakage (3, 24). Other studies in Haiti and Cuba showed that Africans and their descents have genetic polymorphisms to protect from severe dengue (24, 41, 42).

There are three important stages in the transmission of dengue virus among individuals as shown in Figure 2.10.

Figure 2.10 Transmission of dengue virus

The female Aedes mosquito becomes infected when she bites an infected human during the viremic phase of illness. Thus, the virus infects the mosquito mid-gut and then the salivary glands. This extrinsic incubation takes about 8 to 10 days. Once the infectious mosquitoes bite a susceptible human, the transmission occurs. Virus incubation in human, called intrinsic incubation, takes about 3 to 14 days with an average of 4 to 7 days. Then the viremia will occur at the time or just before the onset of symptoms. The infectious period in human lasts about five days after the onset of illness. This period is the crucial period to maintaining the transmission cycle if an infected person is bitten by a female mosquito (3).

2.1.6 Clinical manifestation of dengue infection

Infection with any serotypes of dengue infection causes a ranging spectrum of illness: from asymptomatic or symptomatic with mild manifestation (DF) to severe manifestation (DHF/DSS). WHO (1997) has classified manifestation of dengue infection as shown in Figure 2.11.

Figure 2.11 Manifestation of Dengue Infection (3)

According to the WHO classification, symptomatic manifestation may present as mostly mild illness, such as a simple fever that indistinguishable from fever caused by other infections.

2.1.6.1 Dengue fever

Patients with dengue fever are generally presented with an acute biphasic fever, usually accompanied by headache, myalgias, arthralgias, rashes and leucopenia. Other symptoms such as severe muscle and joint pain (break-bone fever), particularly in adults, and occasionally unusual hemorrhagic may occur. In many endemic areas, clinical symptoms of DF are generally similar and must be differentiated from those of chikungunya fever, another arbovirus disease which also has similar epidemiology and an overlapped distribution in Asia and the Pacific, (13). A confirmed diagnosis is found by using virus isolation and/or serology.

2.1.6.2 Dengue Hemorrhagic Fever and Dengue Shock Syndrome

Referring to WHO's guidelines (13), this typical cases are characterized by four major clinical manifestations, including high fever (>39°C), hemorrhagic phenomena, hepatomegaly and often circulatory failure. Important laboratory findings are thrombocytopenia with haemoconcentration. The leakage of plasma is a major pathophysiological change in DHF that determines the severity of the disease in DHF and also differentiates it from DF. This pathophysiological change are manifested by elevated haematocrit (haemoconcentration), a serous effusion or hypoproteinaemia (3). Initial symptoms of DHF is a sudden high fever which is accompanied by facial flush and other non-specific viral infection symptoms such as headache, vomiting, and muscle or joint pains. Hemorrhagic phenomena are mostly shown as a positive tourniquet test; other spontaneous bleeding, such as petechiae, maculopapular, epistaxis, gum bleeding, and gastrointestinal haemorrhage may be present. Furthermore, the critical stage is usually started at the end of the acute phase which is around 2 to 7 days of fever when the patient can progress to shock and death if not treated properly. Fortunately, mild cases recover spontaneously or after receiving fluid and an electrolyte therapy. Additionally, dengue shock syndromes are shown as a rapid and weak pulse with narrowing of blood pressure shock, progress rapidly and then pass into profound shock. Within 12-24 hours, the patient either dies or recovers rapidly following a proper volume-replacement therapy.

Equally important for the diagnostic and management of the disease is the classification of DHF severity. According to the WHO (1997), the guidelines can be divided into four grades as follow:

Grade I : Fever accompanied by a non-specific constitutional symptom tourniquet test and/or easy bruising

Grade II : Spontaneous bleeding in addition to the manifestations of Grade I patients, usually in the forms of skin or other hemorrhages

Grade III : Circulatory failure manifested by a rapid, weak pulse and narrowing of blood pressure, with the presence of cold, clammy skin and restlessness

Grade IV : Profound shock with an undetectable blood pressure or pulse

The presence of thrombocytopenia with concurrent haemoconcentration differentiates grades I and II DHF from DF, whereas Grade III and IV are considered to be DSS. The course of dengue illness can be seen in Figure 2.12.

Figure 2.12 The course of dengue illness (1)

However, recent studies have reported difficulties in the use of this classification (43-46) and proposed for a new classification.

The clinical multicentre study across dengue-endemic region studied the evidence in order to develop the new dengue classification in the basis of severity. The study confirmed that differences among patients with severe dengue and non-severe dengue can be seen by using a set of clinical and/or laboratory parameters as seen in Figure 2.13.

Figure 2.13 Suggested dengue case classification and levels of severity (1)

Although this classification has been suggested by an expert group (Geneva, 2008) but it is still being tested in many countries by comparing to the existing WHO classification (1).

2.1.7 Case definition for DF and DHF/DSS

Considering a variability spectrum of the illness, The WHO (13) proposed case definition as shown below for DF and DHF.

Dengue Fever

Clinical description is an acute febrile illness of 2-7 days duration with two or more of the following manifestations: headache, retro-orbital pain, myalgia, arthralgia, rash, hemorrhagic manifestation, leucopenia

Laboratory criteria for diagnosis are one or more of the following:

Isolation of the dengue virus from serum, plasma, leucocytes, or autopsy samples

Demonstration of a fourfold or greater change in reciprocal IgG or IgM antibody titers to one or more dengue virus antigens in paired serum samples

Demonstration of dengue virus antigen in autopsy tissue, serum or cerebrospinal fluid samples by immunohistochemistry, immunofluorescence or ELISA

Detection of dengue virus genomic sequences in autopsy tissue serum or cerebrospinal fluid samples by polymerase chain reaction (PCR)

Case classification of Dengue Fever

Suspected: A case fulfilled with the clinical description

Probable: A case fulfilled with the clinical description with one or more of the following

Comparable Ig G ELISA titer or a positive Ig M antibody test on a late acute or convalescent-phase serum specimen)

Occurrence at the same location and time as other confirmed cases of dengue fever.

Confirmed: A case confirmed by laboratory criteria

Reportable: Any probable or confirmed case should be reported

Dengue Hemorrhagic Fever

A case is fever, or history of an acute fever, lasting 2-7 days, occasionally biphasic with the following must all be present:

Hemorrhagic tendencies by one or more of the following :

A positive tourniquet test

Petechiae, ecchymoses or purpura

Bleeding from the mucosa, gastrointestinal tract, injection sites or other locations

Haematemesis or melaena

Thrombocytopenia (100000 cells per mm3 or less)

Evidence of plasma leakage due to increased vascular permeability, manifested by at least one of the following:

A rise in the haematocrit ≥ than 20% above average for age, sex and population

A drop in the haematocrit ≥ 20 % of baseline following volume replacement treatment

Signs of plasma leakage such as pleural effusion, ascites and hypoproteinaemia

Dengue Shock Syndrome

A case is all of DHF criteria with the evidence of circulatory failure manifested by a rapid and weak pulse, and a narrow pulse pressure <20 mmHg or manifested by hypotension for age and cold, clammy skin and restlessness.

2.2 Diseases surveillance in Indonesia

Epidemiological surveillance plays a role in the development of effective and efficient control programs. It is defined as the ongoing systematic process of collection, recording analysis, interpretation, and dissemination of health data reflecting the current health status of a community.

Epidemiological surveillance of DHF in Indonesia is implemented mostly based on passive surveillance. According to the Epidemic Act (UU No 4/1984) and the Ministry of Health (Regulation No. 560/1989), all dengue infection cases should be reported to the local health authority.

The Health Ministry set a flow of reported DHF cases (47) as described in Figure 2.14. The hierarchy of routine as well as outbreak reports was submitted from the lowest level, i.e., sub districts, up to the highest level, i.e., the Ministry of Health. All suspected and DHF cases should be reported from all health care units to the District/Municipality Health Office within 24 hours with the copy to the local Primary Health Center (PHC). Suspected DHF reports are used for vigilance and follow-up actions to overcome the potential epidemic, while the case reports, in addition, are used as information to be passed in stages from the PHC to the Ministry of Health.

Epidemiology surveillance in kecamatan (sub district) is conducted by Puskesmas (Primary Health Center) activities, including collecting and recording data of suspected DHF, DF, and DHF/DSS cases; processing and providing a presentation of data for monitoring DHF outbreaks, epidemiological investigation for cases and routinely reporting cases to the District/Municipality Health Office. DF and DHF/DSS cases observed at the village level are weekly reported to the District/Municipality Health Office. In addition, the summary of confirmed morbidity and mortality --which is a part of controled activities of dengue infection at the sub-district level-- is reported monthly to the District/Municipality Health Office (Figure 2.14).

Subsequently, the District/Municipality/Regency Health Office conducts epidemiological surveillance based on passive surveillance from a health center and PHC as well. Here, the main surveillance activities are very similar to what PHC does but with a larger scope. The same stage for reporting cases tiered up until the highest level which is the Ministry of Health through the Directorate General of Diseases Control and Environmental Health.

"a" : suspected & cases reports from Health Center (24 hours) "d" : Weekly number of cases

"b" : individual data case reports (monthly) "e" : Outbreaks reports

"c" : cases + control program reports (monthly)

: reporting of information : feed back

Figure 2.14 DHF reporting flow chart

Case definition of dengue infection in Indonesia following WHO criteria for DF, DHF and DSS (13).

2.3 Time series

Time series refers to a set of observations taken sequentially at equal time intervals. As a matter of fact, many sets of data which are routinely collected in public health fields and medicine appear as time series, such as notifications of diseases, numbers of patients in hospitals, and blood pressure measurements for evaluating hypertension drugs. Uniquely, a dependency is more likely in time series. The present value may have correlation with the previous value. Therefore, time series can be used for understanding the structure of data, forecasting, monitoring and doing feedback control activity (48).

There are four components of time series:

Trend component: when observation data are taken over a long time period. It shows as a long-run increase or decrease over time.

Seasonal component: when the data are taken monthly or quarterly and observed within 1 year. The pattern is a regular wave in a short interval time.

Cyclical component: has a wave-like pattern in a long interval time.

Random component: is an unpredictable event due to random variations of nature or unusual events. This component is also called noise in the time series.

2.3.1 Stationarity

The stationary refers to unchanged probability structure of data series with time. Data should be stationary before conducting time series analysis. It has a constant mean, variance, and autocorrelation. In practice, the data mostly are not stationary. Therefore, transformation is obtained for such data series to yield stationary. The transformation such as logarithm, square roots or differencing may be appropriate to stabilize the mean and variance of the data.

2.3.2 ARIMA model

Autoregressive (AR) and Moving Average (MA) are the powerful approach for modeling time series. The AR approach assumes that the present value is linearly dependent on the previous value from the series and on the random shock. The value from this process is defined as a process of order p. Meanwhile, a moving average model implies the current value as a linear regression of the current and previous random shocks. This process produces a moving average of order q (49). Thus, the order of ARMA model denotes as ARMA (p,q) and suitable for a stationary series. If the data are not stationary, then differencing should be performed. This is called an integrated ARMA model of order p.d,q or ARIMA (p,d,q) model. The extension of this model is seasonal ARIMA which is suitable for a seasonal time series. The model is denoted by SARIMA (p,d,q)(P,D,Q) in which P.D,Q are seasonal orders.

Box and Jenkins developed a powerful method for generating ARIMA model that is also called the Box-Jenkins model. The model building consists of three main stages, which are identification, estimation and validation models (48). The important step when identifying a tentative model is to deal with stationary series and select the appropriate model by measuring the dependency structure of data. The first is an autocorrelation function (ACF), measuring the correlation between an observation in time "t" and an observation in a specific lag (lag k). The second is a partial autocorrelation function (PACF), measuring a correlation between an observation in time "t" and an observation in a specific lag (lag k) after removing the effect of time series in between "t" and " lag k". The AR model is appropriate if PACF shows statistical significance in lag k while ACF shows an exponential decay. Otherwise, MA model is appropriate if ACF shows a statistical significance in lag k while PACF is exponentially decayed. Then, the tentative models are estimated by using a maximum likelihood method or a conditional least squares method. The model is adequate if a residual prediction is stationary or white noise. In other words, ACF and PACF should be equal to zero. The best model can be chosen from the Akaike's Information Criterion (AIC) and the Schwartz's Bayesian Criterion (BIC). Models that give a minimum value for both criterions are to be chosen.

2.4 Spatial analysis in epidemiology

Elliot,P. and Wartenberg D. (2004) define spatial epidemiology as "the description and analysis of geographically indexed health data with respect to demographic, environmental, behavioral, socioeconomic, genetic, and infectious risk factors" (50). There are some purposes of spatial analysis in epidemiology such as a description of spatial patterns, a cluster detection and disease clustering, and an explanation or a prediction of disease risks (51). For this purpose, spatial analysis requires spatial data. Spatial data have two components: attribute data, which describe the variable, and geo-referenced feature data as points, line or area. The conceptual framework of spatial epidemiological data analysis is explained in Figure 2.15.

The important types of spatial data are individual data points and area data. The latter summarizes the group of individual data to a single value, and has a spatial location which can be defined as a spatial region using an administrative region at the proper level of resolution (51, 52). Spatial data can be obtained from many sources such as health data (vital statistics, notifiable diseases, disease registries, health surveys), census, environmental and natural resource data and remotely sensed data (53).

Figure 2.15 Conceptual framework of spatial epidemiological data analysis (51)

2.4.1 Geographic information systems (GIS)

The GIS are computer-based systems for integrating and analyzing spatial data. Waller and Gotway (2004) summarize some basic operations of GIS as follows (53):

Spatial Query is ability to query data elements with consideration to their locations and attribute values.

Layering is ability to link the values from several spatial data sets by location. This operation could provide a single map which consists of some information from all data set.

Buffering is ability to give the definition of the area within a specific distance of geo-referenced features.

GIS can be used in public health for various purposes, based on what people want to do with spatial data, such as: database management, visualization and mapping, and spatial analysis (54). For former purposes, database management systems (DBMS) are used to store, retrieve, and modify data into a database. However, the DBMS are still lack of functions of GIS for spatial data analysis involving statistical inference (53, 54).

2.4.2 Spatial visualization

There are some purposes of spatial visualization including not only identification of any patterns that might be present, but also the developing of hypotheses about factors that might influence to the pattern (51). Spatial visualization for public health purposes includes disease mapping. Points data can be visualized by plotting the locations using their Cartesian coordinates. This map shows the raw format of the data. The aggregated data can be visualized by choropleth maps. A disease map from the aggregated data may use either a relative risk, usually the standardized mortality or morbidity ratio (SMR), or statistical significance of the difference between disease risks in local areas and the overall risk average through the study areas (51).

2.4.2.1 Smoothing

A good map should be done without any noise and dominated freely by a particular variation to produce the stable estimates. Unfortunately, researcher found that using a raw rate or statistical significance can produce an extreme map due to an unstable rate (52, 55, 56). Furthermore, using SMR in small population or rare diseases and a significance level in large populations will produce the unstable estimates. Smoothing takes place to reduce the noise in map. Basically, information from neighboring area takes into account to produce a better rate estimate in a local area (53). Scientists proposed several methods for smoothing. Clayton and Kaldor (1987) proposed the Empirical Bayes technique: "the method involves estimations of the parameters of a random process that is assumed to generate the relative risk, and calculation of posterior expectations as the estimates of individual relative risks", for calculating of small area rates (56).

2.4.2.2 Modifiable Areal Unit Problem (MAUP)

The MAUP is defined as changing in a spatial pattern due to different units of a geographical area used in the analysis. This problem occurred when aggregated data at different scale areas are used for the analysis. Some analysts suggested using the smallest area units available in the data set for handling MAUP (51).

2.4.3 Spatial autocorrelation and clusters detection

Spatial data have a distinct nature called spatial dependence. Nearby locations are likely to have similar attributes as mentioned by Waldo Tobler," everything is related to everything else, but near things are more related than distant things" (57).Visualization of spatial data can show the presence of a spatial pattern which is classified as regular, random or clustered(51). Clustered in epidemiology can be defined as a spatial aggregation of health events. A disease cluster occurs if there is an excess number of disease events in a particular place and time after being adjusted for known risk factors such as demographic characteristics (51, 58). The correlation among the same types of measurement taken in different locations is called spatial autocorrelation (59). Spatial autocorrelation statistics are used to determine the magnitude of spatial similarity observed among neighboring values of attribute data through a study area in aggregated data (51). Moreover, spatial autocorrelation statistics are used to define either a global cluster or a local cluster as described below:

Global autocorrelation

Global indexes of spatial autocorrelation are used for the detection of global clustering. Using Moran's I coefficient, it summarizes the degree of spatial similarity that tends to occur over a nearby area (59). It can be defined as:

Where N = number of area unit

Wij = spatial weights between area i and j

Xi = attribute value of area i

Xj = attribute value of area j

= mean value in study region

I = Moran's I, ranges between +1 and -1

Interpretation if :

+1 : strong positive spatial autocorrelation (clustering of areas of a similar attribute value)

: no spatial autocorrelation or no clustering

-1 : strong negative spatial autocorrelation (clustering of areas of a dissimilar attribute value)

The spatial weights, defining the neighbor, can be done using either adjacency (contiguity) or distance. There are some methods based on contiguity including rook, queen, and higher-order (spatial lags) contiguity. Other methods are based on distance using a centroid location as an adjacent. Furthermore, the significance of Moran's I can be defined using Monte Carlo randomization (51).

Local Indicator of Spatial Association (LISA)

LISA detects the presence of local spatial autocorrelation in aggregated data by decomposing Moran's I statistics into each area of study. LISA statistics is used not only to detect the local clusters but also to test potential outliers in Global spatial patterns (60). Anselin (1995) defines LISA as (60):

Ii = Zi ∑Wij Zj

Where Zi and Zj : deviations from the mean for each area

Wij : spatial weights matrix in a row-standardized form

As a result, LISA statistics is present as four associations including: high-high, low-low, high-low, and low-high. Another output of LISA is a map of p-values in each area. However, Monte Carlo testing should be performed to assess distribution of LISA based on heterogenous non-Gausian random variables.

Relevant study

Numerous studies have been conducted using spatial and temporal analysis which provide more understanding about the distribution of dengue infection and provide the valuable evidence for public health activities (61-63).

Spatial autocorrelation analysis is widely used to identify a spatial pattern of dengue. Nakhapakorn and Jirakajohnkool (2006) investigated the relation between incidence and spatial patterns of dengue fever in northern Thailand. They concluded that spatial pattern of dengue have changed over time (64). Another study conducted by Hu., et al (2010) in Queensland, Australia. They identified the global and local cluster by Morans's I statistics and concluded that dengue cases have been expanded geographically over recent year in Queensland (65).

The relation of spatial distribution of dengue incidence and risk factors has been investigated using GIS and spatial analysis. Mondini and Neto (2008) assessed correlation of dengue incidence with potential risk factors such as socioeconomic, demographic and environmental factors in a Brazilian city. They found spatial dependence on dengue distribution and socioeconomy (66). A study in Taiwan by Wu., et al (2009) showed the association between risks of dengue infection and the level of urbanization and temperature (67). The spread of dengue infection as a contagious type was proven by Muttitanon., et al (2004) during DHF epidemics in Nakhon Pathom Province, Thailand. The result indicates that the epidemics occurred due to the spread of a new or rare dengue serotype (68).

In order to assess high-risk areas for the occurrence of dengue fever, studies conducted by Wu., et al (2006) in Taiwan , and Galli and Neto (2008) in Brazil applied temporal risk characteristics (temporal risk indices, frequency, and duration index) and LISA for spatial risk areas (69).

Other applications, modeling and forecasting of dengue occurrence, play as an important role for a dengue control program. Silawan., et al (2008) assessed the temporal dynamics and proposed a forecasting model of dengue fever in the northern province of Thailand (70). Study by Barbazan., et al (2002) determined spatial and temporal variations of DHF incidence in all provinces of Thailand to identify epidemics' months. This study can be used by a control program as a warning of epidemic outbreak (62).