Spatial Analysis of Jembrana Disease in South Kalimantan Province

SUJONI

GRADUATE SCHOOL

BOGOR AGRICULTURAL UNIVERSITY

BOGOR

I, Sujoni, hereby stated that this thesis entitled:

SPATIAL ANALYSIS OF JEMBRANA DISEASE IN SOUTH KALIMANTAN PROVINVE

is a result of my own work under the supervision of advisory board during the period of December 2010 until July 2011 and that it has not been published ever. The content of this thesis has been examined by the advisory board and external examiner.

Bogor, August 2011

Sujoni

SUJONI. Spatial Analysis of Jembrana Disease in South Kalimantan Province. Under supervision of SURIA DARMA TARIGAN and BAMBANG

PURWANTARA.

Jembrana Disease (JD) is an acute infectious disease in Bali cattle that caused by Jembrana Disease Virus. It is causes high economical losses and endemic in several provinces in Indonesia. Studies on the epidemiology of JD rarely consider the spatial dimension of disease prevalence. Geographic Information System (GIS) has been increasingly used in spatial epidemiology to analyze the disease pattern based on the geographical data. This study presents the spatial analysis of JD to provide information about the distribution of JD in South Kalimantan province. Serological data were obtained based on the surveillance throughout the province and screened using PCR diagnostic technique during 2008 to 2010 to determine JD seropositive. JD was found mostly in the northern, southern, and western parts of the province. The seroprevalence of JD was higher in district of Banjarbaru, Banjar, and Tanah Laut. Using spatial scan statistic, the distribution of JD was spatial clustered in specific area. This elevated risk within the cluster was significant (p<0.001). JD seropositive positively associated with cattle density and distance to the main rivers and negatively associated with cattle density and elevation. It indicates that JD seropositive was higher in lowland and the area with higher cattle density.

SUJONI.Analisa Spasial Penyakit Jembrana di Provinsi Kalimantan Selatan.

Dibawah bimbingan oleh SURIA DARMA TARIGAN dan BAMBANG PURWANTARA.

Penyakit jembrana (JD) adalah penyakit infeksius pada sapi Bali yang disebabkan oleh Virus Penyakit Jembrana. JD menimbulkan kerugian ekonomi yang tinggi dan endemis di pelbagai daerah di Indonesia. Studi terhadap epidemiologi JD jarang mempertimbangkan dimensi spasial terhadap prevalensi JD. Sistem Informasi Geografis (SIG) telah banyak digunakan dalam bidang epidemiologi spatial untuk menganalisa pola penyakit berdasarkan data geografi. Penelitian ini menjelaskan analisa spasial JD untuk memberikan informasi tentang distribusi JD di provinsi Kalimantan Selatan. Data serologis JD diperoleh berdasarkan kegiatan surveillance di seluruh Kalimantan Selatan dan diuji dengan PCR untuk menentukan seropositive JD dalam periode 2008 sampai 2010. Seropositive JD sebagian besar ditemukan didaerah utara, selatan, dan barat dari provinsi Kalimantan Selatan. Seroprevalence JD yang tinggi terdapat di kabupaten Banjarbaru, Banjar, dan Tanah Laut. Dengan menggunakan spatial scan statistic, distribusi JD terkelompok pada daerah tertentu. Peningkatan resiko penyakit didalam cluster terjadi secara significant (p<0.001). Terdapat hubungan yang positif antara seropositive JD dengan elevasi dan kepdatan ternak. Hal ini menunjukkan bahwa seopositive JD banyak ditemukan di dataran rendah dan daerah yang mempunyai kepadatan sapi yang tinggi.

Kata kunci: analisis spasial, penyakit jembrana, SIG, provinsi Kalimantan Selatan

Province. Under supervision of SURIA DARMA TARIGAN and BAMBANG

PURWANTARA.

Jembrana Disease (JD) is an acute and severe disease of Bali cattle (Bos javanicus) that is caused by Jembrana Disease Virus (JDV). It is endemic in some

provinces of Indonesia such as Bali, South Sumatera, Lampung, Bengkulu, West Sumatera, and South Kalimantan. JD causes high economical losses due to high morbidities and mortalities. JD is one of strategic animal disease due to specific on Bali cattle and is found only in Inodonesia. The epidemiological factors contributing to occurrence of JD are unknown. There are no effective drugs to JD treatment. JD has become main concern of animal disease eradication program of Animal Husbandry Office of South Kalimantan Province. The first case in South Kalimantan was found in 1993 and endemic areas are Kotabaru, Tanah Laut, Tanah Bumbu, and Barito Kuala. It is causes high losses due to mortality and morbidity rates are very high reaching. There are still some outbreaks in South Kalimantan every year and cause losses suffered by farmers.

GIS have become an important tool in modern animal disease control. GIS technology is used for spatial distribution and analysis for the several diseases eradication program. In the area of disease surveillance GIS can be used to produce maps of disease occurrence. GIS combined with methods of spatial analysis provide powerful new tools for understanding the epidemiology of diseases and for improving disease prevention and control. The main objective of this study is to conduct a spatial analysis of JD in South Kalimantan Province. This study presents a spatial distribution of JD, identify the JD cluster, generate JD mapping, and identifying factors associated with the spatial distribution of JD in South Kalimantan Province.

The study area is South Kalimantan province which is located in southern of Borneo Island at 114° 19" – 116° 33" E and 1° 21" – 1° 10" S. South Kalimantan Province consists of 151 sub districts in 11 districts and two municipalities. The JD serological data used in this study were obtained from Animal Husbandry Office of South Kalimantan Province, based on the surveillance activities during 2008 to 2010. All samples were screened with Polymerase Chain Reaction (PCR) or Enzyme-Linked Immunosorbent Assay (ELISA) followed by PCR to determine the JD seropositive. ArcGIS 9.2 software with extension spatial analyst and geostatistical analyst was used to analyze and visualize the JD distribution, while SaTScan 9.0 used to identify the JD cluster.

The ordinary kriging is used to generate spatial continuous map of JD in South Kalimantan province based on the prevalence rate. The kriging method was ordinary, semivariogram model was spherical, and search radius type was variable. Cluster analysis was set with the maximum cluster size of 50% of the total population at risk and the type of analysis was purely spatial. The probability model was Discrete Poisson Model and scan for areas with high rates with maximum number of replication was 999. A regression analysis was used to identify factors that influence the risk of JD being present or absent at specific location using binary data i.e. positive (disease present) and negative (disease

covered 31 sub districts and 10 districts in South Kalimantan province. There are 571 samples (33.31%) were screened by ELISA, 122 samples (7.12%) were screened by PCR, and 1021 samples (59.57%) were screened by ELISA followed by PCR. Among the total of 1143 samples which were screened by PCR or ELISA and followed by PCR, 57 samples (4.99%) were positive to JD based on the PCR assay. There are eight districts had been reported at least one positive serological result to JD. The overall prevalence of JD in South Kalimantan was 4.99%. Sample testing positive to JD mostly located in the southern, western, and northern parts of the province. Among the total 13 districts in South Kalimantan Province, 8 districts had at least one positive serological result to JD. There are 5 districts that were not classified as endemic area, 5 districts were low endemic area, and 3 districts were medium endemic area. Based on the risk area, there are 5 districts were not classified, 6 districts were low risk area, 1 district was medium risk area, and 1 district were high risk area.

Semivariogram parameter was used to generate JD mapping based on the Kriging analysis. The fitted semivariogram had a major range of 1.86 km, a nugget of 0.488, and a sill of 1.266. It was described that between two points observed was spatial autocorrelation with the maximal distance 1.86 km, while the sample locations that separated by distance more than 1.86 km was not spatially correlated. Based on the ordinary kriging analysis, most of high risk areas were located in the north, south, and west of the province (Tabalong, Balangan, Barito Kuala, Banjarbaru, and Tanah Laut), while the lower risk areas were located in central part of the province.

Using spatial scan statistic with the maximum spatial cluster size of <50% of the total population, two clusters identified in 2008, one cluster identified in 2009, and two clusters identified in 2010. In 2008, a primary cluster was defined 10.95 km around the Martapura sub district, while secondary cluster located at Kuranji sub district. The relative risk (RR) within primary cluster was 79.90, with an observed number of cases of 7 compared with 1 expected case. The relative risk (RR) within secondary cluster was 10.77, with an observed number of cases of 2 compared with 1 expected case. In 2009, one cluster was defined 30.45 km around Bumi Makmur sub district. The RR of this cluster was 95.14 with an observed of cases of 14 compared with 2 expected cases. In 2010, a primary cluster located at Tanta sub district and a secondary cluster located at Barambai sub district with RR equal to 257.39 and 52.87 respectively. This elevated risk within the cluster was significant (p<0.001). JD cluster were identified each year are located in different area. It shows that the occurrence of JD has spread to areas where previously free from JD. However, JD cluster generally located at southern, western, and northern of the province. Overall in the period of 2008 to 2010, the primary cluster was identified at Tanta sub district and secondary cluster was identified at 48.20 km around Tamban sub district. The RR of primary cluster was 108.68 with an observed number of cases of 12 compared with 1 case, and the RR of secondary cluster was 10.94 with an observed number of cases of 34 compared with 7 cases.

showed that JD seropositive positively associated with cattle density and distance to river, and negatively associated with elevation. The results indicate that JD seropositive was higher in the area with higher cattle density than area with lower cattle density. The results also indicate that JD seropositive was higher in low land area and decreased in high land. The areas near the main rivers indicate the lower JD seropositive. The analysis also generates the following statistically significant association with JD seropositive: cattle density (O.R. 1.0140, 95% C.I. 1.0001, 1.0281), elevation (O.R. 09965, 95% C.I. 0.9921, 1.0010), and distance to the river (O.R. 1.0002, 95% C.C. 1.0001, 1.0003). It indicates that cattle density and distance to the main rivers were significantly influence to the distribution of JD, while elevation was not significantly affected to the occurrence of JD seropositive.

It can be concluded that JD seropositive mostly located at the north, south, and west parts of the province. The spatial distribution of JD in South Kalimantan province was spatially clustered in specific area. JD mapping using Kriging method shows the distribution of JD was higher in the district of Tabalong, Balangan, Barito Kuala, Banjarbaru, and Tanah Laut. To better further analysis, it should be to collect sample proportional with the population and quite distribute in all areas. It is also important to further investigate the association between other factors with the increasing of JD incidence to better understanding of JD patters, such as environmental, socio-economic, cultural, or other factors.

mentioning the sources;

a. Citation only permitted for the sake of education, research, scientific writing, report writing, critical writing or reviewing scientific problem. b. Citation does not inflict the name and honor of Bogor Agricultural

University

2. It is prohibited to republish and reproduced all part of this thesis without written permission from Bogor Agricultural University.

SUJONI

A thesis submitted for the Degree of Master of Science in Information Technology for Natural Resources Management Study Program

GRADUATE SCHOOL

BOGOR AGRICULTURAL UNIVERSITY

BOGOR

Student Name : Sujoni Student ID : G051090081

Study Program Master of Science in Information Technology for Natural Resource Management

Approved by, Advisory Board

Endorsed by,

Date of Examination: Date of Graduation: August 1st, 2011

Dr. Ir. Suria Darma Tarigan, M.Sc Supervisor

Program Coordinator

Dr. Ir. Hartrisari Hardjomidjojo, DEA

Dr. drh. Bambang Purwantara, M.Sc Co-Supervisor

Dean of the Graduate School

studying at MIT and finished the thesis. This thesis was written as a requirement to complete the master program at Master of Science in Information Technology for Natural Resources Management (MIT), Bogor Agricultural University. I would like to express my gratitude to:

1. Ministry of Agriculture of the Republic Indonesia, especially to the Human Resources Department for granted my master at MIT IPB.

2. Governor of South Kalimantan, the Head of Animal Husbandry Office of South Kalimantan Province, and the Principal of School of Agricultural Development of Pelaihari for the permission and opportunity to study at IPB.

3. Dr. Ir. Suria Tarigan, M.Sc. and Dr. Drh. Bambang Purwantara, M.Sc. as my supervisor and my co-supervisor for all inputs, corrections, and their guidance during the writing of this thesis research. Thank you also to Drh. Surachmi Setiyaningsih, Ph.D. as external examiner for her suggestions to improve this thesis.

4. My deep appreciation would like to be expressed to all my family, especially my beloved father and my beloved mother for all their support and patient. Special thanks also addressed to Eka Risma for her understanding and patience.

5. The Head of Department of Animal Health and Veterinary Public Health, Animal Husbandry Office of South Kalimantan province, especially to Drh Ni Wayan Sri Armiati who helped in JD data collection.

6. The MIT Coordinator, lecturers, all MIT staff, and employers for their helping during studying at MIT.

7. My 2009 classmates and all MIT students, especially to Lasti Pitriani and Serge Claudio Rafanoharana for their kind cooperation.

8. All teachers, staff, employees, alumni, and students of SPP Pelaihari for any extraordinary encouragement. Special thanks to Drh. Lidya Subiyakto who has helped all the administration.

9. All parties who are not mentioned one by one. Thank you for all helping.

I realize that this paper is far from perfect; hence we need all input and suggestion for improvement this thesis. Hopefully this thesis can be useful especially in the use of GIS applications in veterinary medicine field in Indonesia. Amen.

Bogor, August 2011

Sujoni was born in Jombang, East Java, Indonesia on January 27th, 1979, child of couple Matsa’i and Sutah. He was finished his Undergraduate Degree in Veterinary Medicine in Airlangga University in Surabaya in 2002 and Veterinarian Education in 2003. Now he is a teacher at Agricultural Vocational School in Pelaihari, South Kalimantan province. He received scholarship from Agricultural Ministry of Republic of Indonesia in Master of Science in Information Technology for Natural Resources Management, Bogor Agricultural University in 2009, and completed his master study in 2011. His final project for thesis is Spatial Analysis of Jembrana Disease in South Kalimantan Province.

Table of Contents ………... xiii

List of Tables……….. xiv

List of Figures ………... xv

List of Appendices ………. xvi

I. INTRODUCTION ……… 1

1.1 Background ………. 1

1.2 Objective ……….. 3

II. LITERATURE REVIEW ……… 4

2.1 Jembrana Disease ……… 4

2.2 Spatial Analysis Using Geographic Information System ……… 6

2.3 Spatial Analysis in Veterinary Epidemiology ………. 7

2.4 Methods for Spatial Analysis of Animal Disease ……… 9

2.4.1 Data Visualization ……… 9

2.4.2 Exploratory Analysis ………... 11

2.4.3 Spatial Modeling ………. 13

III. METHODOLOGY ……….. 14

3.1 Time and Location ……….. 14

3.2 Data and Tools ………. 15

3.3 Research Framework ………... 16

3.4 Spatial Distribution of Jembrana Disease ……… 17

3.5 Jembrana Disease Mapping ………. 17

3.6 Spatial Cluster Analysis ……….. 20

3.7 Factors Associated with the Spatial Distribution of Jembrana Disease ………. 21 IV. RESULT AND DISCUSSION ……… 23

4.1 Study of Cattle Population ……….. 23

4.2 Spatial Distribution of Jembrana Disease ……… 24

4.3 Jembrana Disease Mapping ………. 28

4.4 Spatial Cluster Analysis ……….. 31

4.5 Factors Associated with the Spatial Distribution of Jembrana Disease ………. 36

V. CONCLUSION AND RECOMMENDATION ………. 38

5.1 Conclusion ………... 38

5.2 Recommendation ………. 38

REFERENCES ………. 39

Page

1 Data Requirement ……… 15

2 The Number of JD Seropositive and Seroprevalence based on the PCR in South Kalimantan Province during 2008 – 2010 ……… 25

3 The Number of JD Seropositive and Seroprevalence at District Level in South Kalimantan Province during 2008 – 2010 .…………. 25

4 Cluster of JD in South Kalimantan Province with Maximum Spatial Size Cluster of 50% of the Total Population Period of 2008

-2010……….. 32

5 The Regression Coefficients, Standard Error, and Odds Ratio of Variables Influencing the JD Seropositive ……….. 37

1 The Clinical Sign of JD in Bali Cattle ………. 5

2 Conceptual Framework of Spatial Epidemiological Data Analysis … 9 3 Study Area ………... 14

4 The Study Framework ………. 16

5 The Graph of Semivariogram ………. 19

6 Cattle Population in South Kalimantan Province ……… 23

7 The Distribution of JD Sample Location during 2008 – 2010………. 24

8 The Jembrana Disease Seropositive in South Kalimantan Province ... 26

9 The Jembrana Disease Seroprevalence in South Kalimantan Province ………... 27

10 The Semivariogram Graph of JD Seropositive ………... 29

11 The Jembrana Disease Mapping in South Kalimantan Province …… 29

12 The Standard Error of JD Mapping using Kriging Method ………… 30

13 Jembrana Disease Cluster with the Maximum Spatial Cluster Size of 50% of the Total Population Period of 2008………... 33

14 Jembrana Disease Cluster with the Maximum Spatial Cluster Size of 50% of the Total Population Period of 2009……… 33

15 Jembrana Disease Cluster with the Maximum Spatial Cluster Size of 50% of the Total Population Period of 2010……… 34

16 Jembrana Disease Cluster with the Maximum Spatial Cluster Size of 50% of the Total Population Period of 2008 – 2010 ………... 34

1 The Association of JD Seropositive and Elevation ………. 44 2 The Association of JD Seropositive and Cattle Density ………. 44 3 Location of JD Seropositive within 500 meters and 1000 meters

from Main Rivers ……… 45

4 Location of JD Seropositive within 5000 meters and 10000 meters from Main Rivers ……… 46

5 The Cluster Analysis Processing using SaTScan with the Maximum Spatial Size Cluster of 50% of the Total Population Period of 2008... 47

6 The Cluster Analysis Processing using SaTScan with the Maximum Spatial Size Cluster of 50% of the Total Population Period of 2009... 50

7 The Cluster Analysis Processing using SaTScan with the Maximum Spatial Size Cluster of 50% of the Total Population Period of 2010... 53

8 The Cluster Analysis Processing using SaTScan with the Maximum Spatial Size Cluster of 50% of the Total Population Period of 2008 to 2010 ... 56

I. INTRODUCTION

1.1 Background

South Kalimantan province is one of area in Indonesia where the population of Bali cattle is high supporting the vision of the Animal Husbandry Office as producer of beef cattle in Kalimantan Island. Several challenges have to be faced to increasing the beef cattle population, including the animal disease outbreak that affected the beef productivity directly and indirectly. One of animal disease that usually attacks Bali cattle is jembrana disease (JD). JD is a major threat to the success of the various Bali cattle distribution programs and consequently to the attempts to alleviate poverty and increase food production in Indonesia.

Jembrana Disease (JD) is an acute and severe disease of Bali cattle (Bos javanicus) that is caused by Jembrana Disease Virus (JDV) which is a member of

the lentivirus genus of the family Retroviridae. It is endemic in parts of Indonesia (Setiyaningsih 2006) and resulting in heavy economic losses because of the high mortalities (Kusumawati et al. 2010). JD causes high economical losses reaching

US $3 million (Suwiti 2010). The major clinical signs of JD are depression, anorexia, fever, enlargement of the superficial lymph node, and blood sweating (Subronto 2003). Infection of Bali cattle with JDV results in a case fatality rate of approximately 20%, and the remainders survive with no recurrence of disease (Soesanto et al. 1990).

JD is one of strategic diseases due to specific on the Bali cattle and is found only in Indonesia. There are no effective drugs for JDV treatment. Disease prevention is mainly by vaccination, although not provide maximum protection against cattle (Suwiti 2010). The epidemiological factors contributing to occurrence of JD are unknown. Movement of livestock, especially Bali cattle, probably becomes one factor that affecting the JD outbreak. The JD incidence occurs all year round with sporadic outbreaks and usually occurs at the end of dry season or at the beginning of the wet season.

JD has become main concern of animal disease eradication program of Animal Husbandry Office of South Kalimantan Province. The first case in South

Kalimantan was found in 1993 in Tanah Laut and is now endemic in the areas of Kotabaru, Tanah Laut, Tanah Bumbu, and Barito Kuala. There are still some outbreaks in South Kalimantan every year and cause losses suffered by farmers (DISNAK KALSEL 2009). The losses of this disease include the death of cattle and additional costs for treatment. Local government also spends much money and time to prevent and eradicate this disease through vaccination and conduct public awareness for farmers.

Geographic Information System (GIS) are now used in many different areas including town planning, ecology, and utility management, reflecting the importance of the spatial dimension to most processes occurring in the world (Pfeiffer et al. 2008). GIS has been widely used in several major field or areas.

GIS is used for surveying and mapping, forest resource inventory, harvest planning, wildlife management and conservation, mining and mineral exploration. In public health, GIS is used for pattern and spread disease, distribution and delivery of health services (Lo and Yeung 2007).

GIS have become an important tool in modern animal disease control. The potential applications for GIS in animal disease control range from use in epidemiological field studies and simulation to use in animal disease surveillance. The main two areas of use in epidemiological field studies include the visual display of geographical patterns and spatial analysis. In the area of disease surveillance GIS can be used to produce maps of disease occurrence and it can be part of a sophisticated animal disease information system. In the field of veterinary epidemiology, GIS has been used widely in field research for visual appraisal and to provide data for advanced spatial analyses (Pfeiffer et al. 1994).

GIS combined with methods of spatial analysis provide powerful new tools for understanding the epidemiology of diseases and for improving disease prevention and control (Chaput et al. 2002). GIS technology is used for spatial

distribution and analysis for the several diseases eradication program. For example, Allepuz (2008) explained the spatial analysis of Aujeszky’s disease eradication in Catalonia, Spain. Haghdoost et al. (2007) used GIS to explain the

spatial distribution of brucellosis in endemic district in Iran. Chen et al. (2007)

environmental in China. Chhetri et al. (2010) identified the factors that associated

with spatial clustering of foot-and-mouth disease in Nepal.

Disease mapping and spatial analysis to explore the nature of such spatial distributions has been widely used in risk modeling to identify high-risk areas or geographically – related risk areas. An objective of such efforts is to identify high risk areas that could subsequently be targeted for control, eradication, and prevention action (Chhetri et al. 2010). Spatial analysis deals with the exploration,

description and analysis of data taking into account their geographical distribution (Saez and Saurina 2007 in Allepuz 2008).

Spatial analysis involves the analysis of data representing geographical features which have a location attribute such as absolute location (coordinates) or relative positioning (distance). Disease occurrence produces a spatial point pattern where each point pattern could be the result of infectiousness or environmental factors. The objectives of spatial analysis are to identify areas of locally increased risk and of factors resulting in spatial interaction which cause, for example increased transmission probabilities (Pfeiffer and Morris 1994).

In Indonesia, studies on the epidemiology of animal disease rarely consider the spatial dimension of disease prevalence. This study presents a study combining surveillance, laboratory diagnostic method, and GIS for spatial analysis of JD pattern to better understand its epidemiology. This study would help in developing detection, surveillance, and control strategies of JD in South Kalimantan province, and supporting the local government to conduct the JD eradication program.

1.2 Objective

The main objective of this study is to conduct a spatial analysis of JD in South Kalimantan Province. This study presents a spatial distribution of JD, identify the JD cluster, generate JD mapping, and identifying factors associated with the spatial distribution of JD in South Kalimantan Province.

II. LITERATURE REVIEW

2.1 Jembrana Disease

Jembrana Disease (JD) also known as Ramadewa disease is an acute and severe disease of Bali cattle and endemic in parts of Indonesia (Setiyaningsih 2006). The first epidemic was occurred in 1964 at Jembrana District, Bali province. It was estimated to have killed at least 30,000 head of cattle from the total approximately 300,000 head of Bali cattle. A few years later, it is regarded as endemic throughout the island of Bali, South Sumatera, Lampung, Bengkulu, West Sumatera, and South Kalimantan (Subronto 2003). JD was reported to affect Bali cattle (Bos javanicussyn Bos sondaicus) and buffaloes (Bubalis bubalis), but

only Bali cattle were reported in the later outbreaks in Bali and elsewhere in Indonesia. The disease was diagnosed firstly as a rinderpest–like disease and subsequently name jembrana disease (Soeharsono and Temadja 1996).

JD is caused by Jembrana Disease Virus (JDV). JDV is a lentivirus belonging to the Retroviridae family of viruses (Burkala et al. 1996). JDV is

arranged by several major proteins with approximately molecule weight 45kD, 42kD, 33kD, 26kD, and 16kD consistently. Minor protein sometime found with molecule weight 100kD and 15kD (Wilcox et al. 1993). JDV is filterable, the

estimated particle size being 100 – 200 nm. It is inactivated after exposure to 55°C for 15 minutes and to extremes of pH (3.0 and 11.2). JDV is resistant to the action of sodium deoxycholate (1:1000), diethyl ether, and range of antibiotics. It

LV UHDGLO\ LQDFWLYDWHG E\ IRUPDOGHK\GH DQG ȕ-propiolactone. Infectivity in meat

persisted up to 36 hours at 22 - 25°C and for 72 hours at approximate 4°C. JDV stored well for several months at –7°C (Ramachandran 1996).

The acute disease associated with JDV infection has a short incubation period of 5 to 12 days and duration of about 7 days, during which affected animals show signs of fever, lymphadenopathy and lymphopenia (Soesanto et al. 1990).

The disease is atypical of many lentivirus infections. It produces an acute clinical disease persisting for up to 12 days, after a short incubation period of less than 12 days (Soeharsono et al. 1995a). During the acute disease the titer of infectious

virus in blood is high and can be detected in saliva and milk (Soeharsono et al.

1995b).

Animal that infected with JD suffering fever, approximately 40 - 42°C, followed by fatigue, anorexia, lacrymation, hyper salivation, and secretion of mucous. Infected animal also show the enlargement of the superficial lymph node especially prescapularis lymph node, prefemoralis lymph node, and parotidea lymph node. About 23% of infected animals indicate hemorrhaging and erosion of mucosa at nasal, tongue, and mouth. The clinical patognomonis of JD is blood sweating which appear during fever and occur for 2-3 days (Subronto 2003). Another sign is enlargement of spleen. Abortion was recorded in 49% of pregnant cows affected with JD. Abortion occurred at all stages of pregnancy (Putra et al.

1983). Figure 1 shows the clinical signs of JD in Bali cattle.

(a) (b)

(c) (d)

Figure 1. The Clinical Sign of JD in Bali cattle: (a) enlargement of the superficial

lymph node, (b) blood sweating, (c) enlargement of spleen, and (d) erosion at tongue (personal documentation).

Consistent clinical signs include fever, lethargy and lymphodenopathy. Haematological changes during the acute phase include elevated blood urea concentrations, decreased plasma protein, leucopenia mainly due to a lymphopenia, eosinopenia, and thrombocytopenia (Soesanto et al. 1990). The

mortality rate in the experimentally infected cattle was 17% (Soeharsono et al.

1996a).

The mechanism of transmission of JD is poorly understood. There is evidence of direct transmission from acutely affected animals in close contact with susceptible cattle, possibly by virus in these secretions infecting cattle by conjunctival, intranasal or oral routes. During the acute disease the titer of infectious virus in blood is high and can be detected in saliva and milk. It is probable that the virus is also transmitted mechanically by Haematophagous

arthropods (Soeharsono et al. 1995b). The blood sucking insect could transmit

JDV from infected animal during fever to susceptible animal (Putra and Sulistyana 1995). Recovered cattle are also a potential source of infection. Recovered cattle are persistently viraemic but the titre of infectious virus in blood by 60 days after recovery from the acute disease is only about 101 ID/ml (Soeharsono et al. 1996b).

Commonly, JD can be diagnosed by enzyme-linked immunosorbent assay (ELISA) and Western Blotting (WB) to detect the JD antibody. Antibody was not detected by ELISA in majority of infected cattle until 11 weeks after infection and a maximum antibody response was detected 23 to 33 weeks after infection. Antibody was still detectable 59 weeks after infection (Hartaningsih et al. 1994).

Another technique is polymerase chain reaction (PCR) that can be used to confirm both field and laboratory diagnosis of JD in Bali cattle (Tenaya and Hartaningsih 2005).

2.2 Spatial Analysis using Geographic Information System

GIS stands for geographic information system is a computerized system that helps in maintaining and displaying data about geographic space. GIS provides four capabilities in data capture and preparation, data management (storage and maintenance), data manipulation and analysis, and data presentation

(de By 2004). GIS is computer-based system specially designed and implemented for two subtle but interrelate purposes: managing geospatial data and using these data to solve spatial problems (Lo and Yeung 2007). GIS is capable of assembling, storing, manipulating, and displaying map and database information. GIS allows us to view, understand, question, interpret, and visualize data in a very informative way. Data management is more effective in order to support any management in any sector by decision and policy makers (Wicaksono et al. 2010).

The construction and functions of GIS were explained by dividing them into subsystems of input, processing, analysis, and output. GIS is made up of four components, namely, data, technology, application, and people. Geospatial data record the locations and characteristics of natural features or human activities that occur on earth’s surface. GIS technology can be explained in terms of hardware and software. The application component of GIS can be explained from three perspectives: areas of application, nature of application, and approaches of implementation. The people component is defined as GIS users (Lo and Yeung 2007).

Spatial analysis using GIS includes a wide range of operations. Typically they relate to analyses within or between layers of geographical data provided by the GIS. In spatial analysis three different types of spatial data can be analyzed: point data, geostatistical data, and lattice data (Pfeiffer et al. 1994). Point patterns

are the data set may consist of locations only, or it may be a marked point process, with data values associated with each location. Geostatistical data represents continuous variation of a feature attribute such as rainfall or temperature (Allepuz 2008). Lattice data represents discrete variation in space based on regular or irregular units. These units can be for example farm or administrative boundaries (Pfeiffer et al. 1994).

2.3 Spatial Analysis in Veterinary Epidemiology

Epidemiology is the study of disease in populations and of factors that determine its occurrence. Veterinary epidemiology involves observing animal populations and making inferences from the observations (Thrusfield 1986). Classic epidemiological analysis focused mainly on the animal dimension,

whereas time and space were usually explored using fairly basic methods. Transmission of an infectious agent requires direct or indirect contact between the source of infection and the susceptible animal, which means that spatial proximity has to be considered as a key factor when determining the risk of infection for individual animals or herds (Durr and Gatrell 2004).

The term of spatial epidemiology is defined as sub discipline of epidemiology whose primary purpose is to describe and explain the spatial pattern of disease (Durr and Gatrell 2004). Spatial epidemiology is the description and analysis of the geographic, or spatial, variations in disease with respect to demographic, environmental, behavioral, socioeconomic, genetic, and infectious risk factors (Elliot and Wartenberg 2005). Some of the analytical tools used in spatial epidemiology include disease mapping, geographic correlation studies to determine if spatial patterns are associated with particular risk factors, and disease cluster detection.

The use of the tools of GIS, spatial statistics and remote sensing is generally necessary for spatial epidemiology (Durr and Gatrell 2004). Herbreteau (2006) explained that the purposes of spatial analysis in disease epidemiology are; the epidemiological surveillance, with disease mapping of reported incidences, and further active surveillance, involving collection of animal health and animal population information; the explanatory understanding of animal population and disease dynamics, by identifying patterns in the spatial-temporal distribution of diseases and identifying risk factors or causes of the diseases; and for the diseases prevention, by predicting outbreaks and assisting in decision making.

The objectives of spatial epidemiological analysis are the description of spatial patterns, identification of disease clusters, and explanation or prediction of disease risk. The specific analytical objectives then lead to three groups of analytical methods: visualization, exploration, and modeling. Visualization is probably the most commonly used spatial analysis method, resulting in maps that describe spatial patterns. Exploration of spatial data involves the use of statistical methods to determine whether observed patterns are random in space. Modeling introduces the concept of cause effect relationships using both spatial and

non-spatial data sources to explain or predict non-spatial patterns (Pfeiffer et al. 2008).

Figure 2 show the conceptual framework of spatial epidemiological data analysis.

Figure 2. Conceptual Framework of Spatial Epidemiological Data Analysis (Pfeiffer et al. 2008)

2.4 Methods for Spatial Analysis of Animal Disease

The objectives of spatial data analysis in animal disease are the description of spatial patterns, the identification of disease clusters and the explanation or prediction of disease risk. Most currently available statistical methods will represent polygon data using the centroid point location together with any associated attributes. A framework for the spatial analysis of epidemiological data includes of the following groups of analytical methods: data visualization, exploratory analysis, and spatial modeling (Durr and Gatrell 2004).

2.4.1 Data Visualization

The most commonly applied spatial analysis technique in research and surveillance of animal diseases is data visualization. This involves generating maps to present the spatial and temporal patterns of disease occurrence, which are then used to develop hypotheses about possible cause – effect relationships (Durr and Gatrell 2004). There are three kind of spatial data type, i.e. point data, aggregated (lattice) data, and continuous (geostatistical) data.

The location of the disease occurrence is generally visualized using a point on the map. The visual analysis of point data includes the simple map display of the point locations and the use of smoothing methods to generate surface representations of point density. Spatial smoothing can be achieved through estimation of localized averages by using a spatial filter or by applying a mathematical function such as kernel smoothing (Durr and Gatrell 2004).

Point maps are the simplest way to visualize disease event information when the locations of events are known. The oldest and most frequently used method to visualize point data is to plot the locations of the study subjects using their Cartesian coordinates. Although point maps are the simplest way to visualize disease event information when the locations of events are known, they present problems where there are large numbers of events or multiple events at the same location. The use of different symbols to represent attribute values is one solution (Stevenson 2009).

The process of aggregation involves summarizing a group of individual data points into single value to produce, for example, a total, mean, median, or standard deviation. This summary statistic may then be assigned a spatial location, often a discrete area such as state, county, or some other administration region. Disease counts can be expressed as a function of the population size to provide estimates of prevalence, incidence risk, or incidence rate per unit area. Choropleth maps are the most commonly used means for visualizing data in this format. A choropleth map shows information by colouring each component area with colour, providing an indication of the magnitude of the variable of interest (Pfeiffer et al.

2008).

Spatially continuous data such as rainfall, humidity, air pollution, or soil mineral concentrations may be estimated at all possible locations within a region of interest. In epidemiology continuous variables of the type cited above may be used as covariates for predicting disease risk. In the simplest situation, continuous data may be summarized by area unit and plotted as a choropleth map (Pfeiffer et al. 2008). Continuous data also can be visualized using interpolation that predicts

are inverse distance weighting (IDW), kernel smoothing, and kriging (Allepuz 2008).

2.4.2 Exploratory Analysis

Exploratory analysis has the specific objective of using a statistical hypothesis – testing framework for the identification of spatial clusters of disease. The term clusters refers to locations at which disease occurrence is higher or lower than would have been expected if disease were randomly distributed in space. The statistical methods can be grouped into global and local statistics depending on whether they generate a single statistic for the whole area or statistics for individual locations within that area (Durr and Gatrell 2004).

Clustering of a disease can occur for a variety of reasons including the infectious spread of disease, the occurrence of disease vectors in specific locations, the clustering of a risk factor or combination of risk factors, or the existence of potential health hazards such as localized pollution sources scattered throughout a region, each creating an increased risk of disease in its immediate vicinity. The investigation of possible disease clustering is fundamental to epidemiology, with one of the aims being to determine whether the clustering is statistically significant and worthy or further investigation, or whether it is likely to be a change occurrence, or is simply a reflection of the distribution of the population at risk (Pfeiffer et al. 2008).

Statistical tests applied to the detection of spatial clusters can be either global cluster detection tests, were a summary statistic identifies whether or not clustering is present in a region under investigation, or local cluster detection tests which seek to define the spatial location of clusters within a given region. The cluster detection tests that are available differ in that some use complete population counts to characterize the population at risk whereas other use a sample of controls (Stevenson 2009).

Global cluster detection tests can be done by using Moran’s I. Moran’s I statistic gives a formal indication of the degree of linear association between a vector of observed values and a weighted average of its neighbouring values (Stevenson 2009). Moran’s I is approximately normally distributed and has a

expected value of – 1/(N – 1) (where N equals the number of area units within a study region), when no correlation exists between neighbouring values. The expected value of the coefficient therefore approaches zero as N increases. A Moran’s I of zero indicates the null hypothesis of no clustering, a positive Moran’s I indicates positive spatial autocorrelation (i.e. clustering of areas of similar attribute values), while negative coefficient indicates negative spatial autocorrelation (i.e. that neighbouring areas tend to have dissimilar attribute values) (Pfeiffer et al. 2008).

The Moran’s I statistic is calculated as follow:

Moran’s I =

Where:

n: the number of polygons in the study area wij: the values of the spatial proximity matrix

yi: the attribute under investigation

ǔWKHPHDQRIWKHDWWULEXWHXQGHULQYHVWLJDWLRQ

Local cluster detection test can be conducted by Kulldorff’s spatial scan statistic. Spatial scan statistic uses a likelihood ratio test for the number of cases found in the study region population (the null hypothesis) to a model that has different disease risk depending on being inside or outside a circular zone (Stevenson 2009). The test can be used for spatially aggregated data as well as when the exact geographic coordinates are known for each individual. Therefore it can be used for lattice or point spatial data (Allepuz 2008). When data is aggregated into census districts the measure will be concentrated at the central coordinates of those districts. The scan statistic is commonly used to test if a one dimensional point process is purely random, or if any clusters can be detected by using a variable circular window (Kulldorf 1997).

The number of cases is compared to the background population data and the expected number of cases in each unit is proportional to the size of the

population at risk. Circle centers are defined either by the case and control/population data or by specifying an array of grid coordinates. Secondary clusters are computed, based on the degree of overlap allowed in the cluster circles, and includes the options no geographical overlap, and no cluster centers in other cluster (Pfeiffer et al. 2008).

2.4.3 Spatial Modeling

Spatial statistical modeling can be aimed at investigating possible causal effects which are considered to be associated with the disease occurrence (Allepuz 2008). The aims of epidemiological modeling of spatial data are to explain or predict the occurrence of disease. Modeling of spatial data can fulfill the production of risk maps or expert systems rules that can directly guide the decision-making process (Durr and Gatrell 2004). Various static or dynamic relationship defined by the underlying models are used to derive new output maps from a set of input maps. It is also important to investigate the potential effects of error propagation. Statistical methods such as regression analysis, weight of evidence or neural network techniques are then used to provide the weightings, combining the inputs to generate output maps (Bonham-Carter 1994 in Durr and Gatrell 2004). Modeling in epidemiological describe the effect of a set of explanatory variables on the spatial distribution of a particular outcome.

The dependent variable in epidemiological regression models typically consist of either points location of cases and non-cases or aggregated area data representing the number of cases in an area, given a certain size population at risk. Statistical regression methods suitable for incorporating spatial dependence with these types of dependent variables include generalized linear mixed models (GLMM) and Bayesian estimation methods. An alternative approach to the regression approaches presented is kriging, which is based on the mathematical modeling of the local spatial dependence using information obtained from a variogram (Durr and Gatrell 2004).

III.

METHODOLOGY

3.1 Time and Location

This study was conducted from December 2010 to May 2011. Data processing and analysis was accomplished at Remote Sensing and GIS Laboratory of Master of Science in Information Technology (MIT) Bogor Agricultural University, while data collection and field survey were conducted in South Kalimantan province.

Geographically, South Kalimantan province located in southern of Borneo Island at 114° 19" – 116° 33" E and 1° 21" – 1° 10" S. It is bounded by East Kalimantan in Northern, Central Kalimantan in Western, Makassar Strait in eastern and Java Sea in southern. The total area is 37.377,53 km2 and mostly covered by forest and peat land. South Kalimantan Province consists of 151 sub districts in 11 districts and two municipalities (PEMPROV KALSEL 2007). The annual mean minimum and maximum temperature is 21.5° C and 34.7° C with average 27.0° C, while the annual mean humidity is 82.2% (BPS 2010).

3.2 Data and Tools

The cattle population and JD serological data used in this study were obtained from Animal Husbandry Office of South Kalimantan Province. Cattle population was based on the 2009 annual report, and JD serological data was based on the surveillance activities from suspected animal during 2008 to 2010 in endemic area and in area with higher of cattle population. All samples were transported to the laboratory and screened with Enzyme-Linked Immunosorbent Assay (ELISA) or Polymerase Chain Reaction (PCR). To conduct the spatial analysis of JD, only samples that have been examined by PCR or ELISA followed by PCR diagnostic assays were involved in this study to determine the JD seropositive. Hence, JD case was defined as the samples that positive to PCR assay.

The sub district level polygon base map at 1:250.000 scale was obtained. The geographical coordinates of the central points of each village and sub district was created. JD seropositive was assumed to be located at the centre of each village. JD seropositive and cattle population were imported to the base map and converted into shape file and visualize using ArcGIS 9.2 (ESRI Inc.). All vector data were geo-referenced to the Universal Transverse Mercator (UTM) coordinate system zone 50S with World Geodetic System (WGS) 1984 Datum. Digital Elevation Model (DEM) 90 meter also downloaded from Consultative Group on International Agricultural Research (CGIAR).

Table 1. Data Requirement

No Data Source Year

1 Base Map (RBI) Bapeda Tanah Laut 2007

2 Elevation SRTM-DEM 90m 2009

3 Jembrana Disease Serological Data Disnak Kalsel* 2008 – 2010 4 Cattle Population Disnak Kalsel* 2009

* Animal Husbandry Office of South Kalimantan Province

In order to carry out the data processing and analysis, software used in this study are Microsoft Office 2007 to make final report, ArcGIS version 9.2 with extension Spatial Analyst and Geostatistical analyst (ESRI Inc.) to analyze data

and visualize the distribution of JD, and SaTScan version 9.0. (available at http://www.satscan.org) to identify JD cluster.

3.3 Research Framework

The main objective of this study is conducted a spatial analysis of JD in South Kalimantan province. This study consists of four main activities i.e. exploratory analysis to describe the spatial distribution of JD, identification of JD cluster, produce the spatial JD map based on the ordinary kriging analysis, and identifying factors associated with the spatial distribution of JD. Figure 4 depicts the general methodology of this study.

3.4 Spatial Distribution of Jembrana Disease

To conduct the GIS analysis on the spatial distribution of JD, an endemic area is defined as district that having at least one seropositive of JD based on the PCR assay during 2008 to 2010. The aggregate data on district level were used to mapping the JD seropositive and seroprevalence in South Kalimantan province.

Based on the JD seropositive data, all districts were grouped into four classes: non endemic area, low endemic area with JD seropositive between 1 – 10 cases, medium endemic area with the JD seropositive between 11 – 30 cases, and high endemic area with JD seropositive more than 30 cases.

A hazard map of JD was produced to assess the risk of JD in South Kalimantan Province. The hazard map represents the JD seroprevalence at each district. The crude prevalence of JD was defined as the number of positive samples (JD seropositive) over the total number of samples and was calculated by district. Districts grouped into four categories: free area, low risk area with the seroprevalence less than 10%, medium risk area with seroprevalence between 10 to 20%, and high risk area with the seroprevalence more than 20%.

3.5 Jembrana Disease Mapping

Disease mapping is considered as exploratory analysis used to get an impression of the geographical or spatial distribution of disease or the corresponding risk (Berke 2004). Disease mapping usually chooses certain spatial interpolation method, and then creates a continuous surface of disease distribution according to geographically distributed sampling data of disease (Zhong et al.

2005).

The ordinary kriging is used to generate spatial continuous map of JD in South Kalimantan province. Kriging is an interpolation technique in which the surrounding measured values are weighted to derive a predicted value for an unmeasured location (Wittich 2007). It is determines the spatial distribution at unsampled points and provides optimal unbiased estimates with known estimation variances based on a model of the spatial variation (Fischer et al. 1996).

The basic model for kriging is based on the following equation:

Ž(s0) =

where Z(si) is the measured value at the _{i WK SRLQW , « 1 ȜL DQ}

unknown weight for the measured value at the ith point, s0the prediction location,

and N the number of observed points (Pfeiffer et al. 2008).

Kriging is determined by the semivariogram, the distance to the prediction location and the spatial relationship between measurements around the prediction location. The empirical semivariogram shows the spatial dependence in the variable of interest as a scatterplot. Distance (spatial lag) is presented on the x –

axis and semivariance on the y-axis. The semivariance is calculated as follows

(Pfeiffer et al. 2008):

ȖK 2

where Z(Si) is the measured value at the ith location, N(h) is the set of

distinct pairs of values separated by distance i – j = h, |N(h)| is the number of

distinct pairs in N(h), zi and zjare data values at locations i and j,respectively.

The semivariogram, a graph of semivariance plotted against separation distance h, conveys information about the continuity and spatial variability of the process. If observations close together are more alike than those farther apart, the semivariance increases as the separation distance increases, reflecting the decline of spatial autocorrelation with the distance (Chen et al. 2007).

The semivariogram is characterized by particular parameter i.e. sill ( ), range ( ), and nugget ( ). The sill is the semivariogram value where the model first flattens or upper bound of semivariogram. The range is the distance at which the semivariogram reaches the sill value (SAS Institute 2011). Sample locations that separated by distance closer than the range are spatially correlated, while the distance of sample location farther than the range are not. In theory, the semivariogram value at the origin (0 lag) should be zero. If it is significantly different from zero for lags very close to zero, then this semivariogram value is

referred to as the nugget. The nugget represents variability at distances smaller than the typical sample spacing measurement error (Bohling 2005).

Figure 5. The Graph of Semivariogram (SAS Institute 2011)

The semivariogram is central to geostatistic and essential for most geostatistical analysis. It compares the similarity pairs of points a given distance and direction apart (the lag) (Fischer et al. 1996). The range identified the

maximum distance at which spatial autocorrelation was detected among the sampled counties. The nugget quantified the minimum variability at a lag distance of zero, whereas the sill quantified maximum variability among spatially independent samples (Yabsley et al. 2005). The nugget describes the spatially

uncorrelated variation in the data. The larger of nugget value show the less spatial dependence (Pfeiffer et al. 2008).

The module of ArcGIS 9.2 is used to carry out the ordinary kriging analysis and developed the prediction map based on the number of JD seropositive. The kriging method was ordinary, semivariogram model was spherical, search radius type was variable, and the number of points was 12. The analysis extent was same as South Kalimantan province, and for advanced parameters was setting based on the semivariogram result.

3.6 Spatial Cluster Analysis

The aim of spatial cluster is to detect JD cluster in South Kalimantan province. It is to perform the geographical surveillance of disease, to test whether the spatial distribution of JD was homogeneously distributed over the region or clustered in space, and to evaluate any identified spatial disease cluster for statistical significance.

The SaTScan software version 9.0 was used to identify and locate significant spatial clusters of JD in South Kalimantan province. SaTScan is a free software that analyzes spatial, temporal, and space-time data using the spatial, temporal, or space-time scan statistics. It is designed for any of the following interrelated purposes: 1) perform geographical surveillance of disease, to detect spatial or space-time disease clusters, and to see if they are statistically significant; 2) test whether a disease is randomly distributed over space, over time or over space and time; 3) evaluate the statistical significance of disease clusters alarms; and 4) perform prospective real-time or time-periodic disease surveillance for the early detection of disease outbreaks. The spatial scan statistic imposes a circular window on the map. The circle is centered on each of the points (Kulldorff 2010).

SaTScan uses a likelihood test for the number of cases found in the study region population (the null hypothesis) to a model that has different disease risk depending on being inside or outside a circular zone. The test can be applied to both point and area data (Stevenson 2009). The program tests the hypothesis that animals within a particular window have the same risk of being seropositive as animals outside the window. Primary cluster is cluster with the largest likelihood ratio, while secondary cluster is clusters with smaller likelihood ratio (Jaime et al.

2005).

To conduct JD cluster analysis, JD cases are defined as the number of JD seropositive in the sub district level and the population at risk is the total cattle number per sub district. Data were assumed to be located at the center of each sub district. Data were created and imported to the SaTScan. Data consist of case files (sub district, number of cases), population files (sub district, year, number of cattle population), and coordinate files (sub district, x-coordinate, y-coordinate).

In this study, the type of analysis was purely spatial; probability model was Discrete Poisson Model, and scan for areas with high rates. With the discrete Poisson model, the number of cases in each location is Poisson-distributed. Under the null hypothesis, and when there are no covariates, the expected number of cases in each area is proportional to its population size (Kulldorff 2010). Cluster analysis was set with the maximum cluster size of <50% of the population at risk with a circular window space. Significant clusters were identified using Monte Carlo simulation with maximum number of replication was 999. Primary and secondary cluster were searched and described by center and radius of the cluster. No overlapping of the circles was allowed. Clusters were imported and mapped using ArcGIS software (ESRI Inc).

3.7 Factors Associated with the Spatial Distribution of Jembrana Disease

JD usually occurs at the end of dry season or at the beginning of the wet season. However, it is important to identify the relationship between JD seropositive and its related factors. A multiple regression analysis was used to identify factors that influence the risk of JD being present or absent at specific location using binary data i.e. positive (disease present) and negative (disease absent).

The analysis used regression by Pezzulo (2011) version 05.07.20, free statistic software for research development and education. The program generates the coefficients of a prediction formula (standard error of estimate and significance levels) and odd ratios with 95% confidence levels. The odd of an event is defined as the possibility of the outcome event occurring divided by the probability of the event not occurring. The odd ratio for a predictor tells the relative amount by which the odd of the outcome increase (OR greater than 1.0) or decrease (OR less than 1.0) when the value of the predictor value is increase by 1.0 unit (Pezzulo 2011).

There are one non-geographical variable (cattle density) and two geographical variables (elevation and distance to the river) were included in multiple regression. The map of elevation and cattle density were overlay with the location of JD seropositive to investigate the association between JD seropositive

and its related factors. A buffer of area was also created around main rivers in South Kalimantan to determine distance sample location to river.

Cattle density is defined as the number of cattle in kilometer square. It was grouped into five classes: less than 5 heads / km2, 5 – 10 heads / km2, 10 – 20 heads / km2 , 20 – 30 heads / km2, and more than 30 heads / km2. Elevation data were obtained as digital elevation model and grouped into five classes: less than 25 meters, 25 – 50 meters, 50 – 100 meters, 100 – 300 meters, more than 300 meters above sea level. To determine the distance to the river, a buffering was created with 500 meters, 1000 meters, 5000 meters, and 10,000 meters around the main rivers.

IV. RESULTS AND DISCUSSION

4.1 Study of Cattle Population

South Kalimantan province is one of the regions with the number of beef cattle is quite high in the Kalimantan Island. The population of cattle in this area was 218,065 animals in 2009 with the density of cattle were 5.83 animals per kilometer square. The highest cattle population is Tanah Laut that is equal to 80,533 animals or approximately 36.93% of total population, followed by Tanah Bumbu district with a population of 30,556 (14.01%) and Banjar district with a population of 18,413 (8.44%). There are 13 sub districts with a cattle density more than 30 heads per kilometer square, 9 sub districts with a cattle density between 20 – 30 heads/km2, 17 sub districts with a cattle density between 10 – 20 heads/km2, 9 sub districts with a cattle density between 5 – 10 heads/km2, and 93 sub districts with cattle density less than 5 heads/km2(Figure 6).

4.2 Spatial Distribution of Jembrana Disease

During 2008 to 2010, 1714 samples were collected from 46 villages, covered 31 sub districts and 10 districts in South Kalimantan province. JD sample were not collected in 120 sub districts. Sample collection focused on the endemic area and in the area with higher of Bali cattle population. The distribution of JD sample location during this period is shown in Figure 7. The sample collection mainly conducted at district of Barito Kuala, Tanah Bumbu, Tabalong, Kotabaru, and Tanah Laut. There were no any sample collection in district of Hulu Sungai Tengah, Hulu Sungai Utara, and Banjarmasin.

Figure 7. The Distribution of JD Sample Location during 2008 – 2010

There are 571 samples (33.31%) were screened by ELISA, 122 samples (7.12%) were screened by PCR, and 1021 samples (59.57%) were screened by ELISA followed by PCR. Among the total of 1143 samples which were screened by PCR or ELISA and followed by PCR, 57 samples (4.99%) were positive to JD

based on the PCR assay. The results of the sample testing are shown in Table 2. The JD prevalence in South Kalimantan province tends to decrease in this period.

Table 2. The Number of JD Seropositive and Seroprevalence based on the PCR in South Kalimantan Province during 2008 – 2010

Year Sample Positive Negative Prevalence (%)

2008 54 11 43 20.37

2009 293 15 278 5.12

2010 796 31 765 3.89

There are eight districts had been reported at least one positive serological result to JD. The highest JD seropositive was in Barito Kuala, followed by Tanah Laut and Tabalong, while the lowest JD seropositive was in Tapin and Kotabaru. There are five districts that are no samples tested by PCR or ELISA followed by PCR, i.e. Hulu Sungai Selatan, Hulu Sungai Tengah, Hulu Sungai Utara, Balangan, and Banjarmasin.

Table 3. The Number of JD Seropositive and Seroprevalence at District Level in South Kalimantan Province during 2008 – 2010.

District Sample Positive Negative Prevalence (%)

Tanah Laut 149 14 135 9.40

Barito Kuala 492 17 475 3.46

Tapin 4 2 2 50.0

Tanah Bumbu 263 3 260 1.14

Kotabaru 6 1 5 16.67

Banjar 22 3 19 13.64

Tabalong 193 12 181 6.22

Banjarbaru 14 4 10 28.57

Kalimantan Selatan 1143 57 1086 4.99

The seroprevalence of JD at district level ranged from 1.14% to 50.00% (Table 3). However, due to the limitation of sample, the JD seroprevalence in Tapin and Kotabaru was excluded in this analysis. The number of samples in this area was not representative. The highest JD seroprevalence was located at Banjarbaru, followed by Banjar and Tanah Laut, while the lowest JD

seroprevalence was Tanah Bumbu. The overall prevalence of JD in South Kalimantan was 4.99%.

Figure 8. The Jembrana Disease Seropositive in South Kalimantan Province

Among the total 13 districts in South Kalimantan Province, 8 districts (61.54%) had at least one positive serological result to JD (Figure 8). There are 5 districts that were not classified as endemic area (occupied by 16.12% of the total cattle population), 5 districts were low endemic area (occupied by 36.72% of the total cattle population), and 3 districts were medium endemic area (occupied by 47.15% of the total cattle population). Sample testing positive to JD mostly located in the southern, western, and northern parts of the province (Figure 8).

Based on the risk area, there are 5 districts were not classified (occupied by 16.12% of the total cattle population), 6 districts were low risk area (occupied by 73.87% of the total cattle population), 1 district was medium risk area (occupied by 8.45% of the total cattle population), and 1 district were high risk area (occupied by 1.57% of the total cattle population) (Figure 9). District of

Tapin and Kotabaru were assumed to be classified as low risk area due to limited sample.

Figure 9. The Jembrana Disease Seroprevalence in South Kalimantan Province

JD is endemic in several districts in South Kalimantan province since the first outbreak. Based on the clinical, pathologic, and virologic, JD was found in district of Tanah Laut and Tanah Bumbu. The results of this study showed that JD serologically found throughout the district which have taken samples both blood and tissue sample. It is allow that JD seropositive was also found in all districts in South Kalimantan province. During this period, JD has spread to other districts especially Barito Kuala, Banjarbaru, Banjar, Tapin, and Tabalong (Figure 9).

Uncontrolled movement of cattle was probably a factor in the spread of JD. Ditjennak (2005) explained that regional autonomy and cattle trading played a role in the spread of JD in South Kalimantan province. There are several livestock check-points on the border that was not functioning properly. The suspected cattle allow entering new area that free from JD due to the infected cattle was not

usually show the clinical signs of JD. The JD virus was transmitted by suspected or carrier cattle to healthy cattle in non endemic area.

ELISA is a common diagnostic method for JD serological test but can lead to false positive results of vaccination. Vaccination activities that have not been implemented uniformly in all areas make it difficult to distinguish JD cases. Confirmation by PCR test must be implemented to test the causative agent of JD with better accuracy. PCR can be used to confirm both field and laboratory diagnosis of JD in Bali cattle with good accuracy (Tenaya and Hartaningsih 2005). Therefore, the analysis of JD spatial distribution in this study used PCR technique to determine JD seropositive.

This spatial analysis was limited to the described endemic area based on the active and passive surveillance. From a total of 1714 samples, 64 samples (3.73%) come from active surveillance. However, the understanding of the spatial distribution of JD would be enhanced if the non endemic area were also included in the active and passive surveillance. It is also important to consider the sample size is proportional to its cattle population in the next investigation.

4.3 Jembrana Disease Mapping

The spatial correlation of JD is shown by semivariogram based on the JD seropositive (Figure 10). Each dot in the semivariogram represents a pair of location. The x-axis refers to the distance between two observed points (h) and y -D[LVUHIHUVWRWKHVHPLYDULDQFHȖ7KHILWWHGVHPLYDUiogram had a major range of

1.86 km, a nugget of 0.488, and a sill of 1.266. The range shows the maximum distance at which spatial autocorrelation was detected among the sampled sub districts. The increasing of semivariance value shows the decreasing of the spatial autocorrelation. With the sill value was 1.266, it was describe that between two points observed was spatial autocorrelation with the maximal distance 1.86 km, while the sample locations that separated by distance more than 1.86 km was not spatially correlated.

The JD mapping was created using kriging method to describe the JD distribution at unsampled points based on the semivariogram. The observed point data was the JD seropositive at sub district level that represents at least one JD

seropositive. Figure 11 shows the JD mapping in South Kalimantan province based on the ordinary kriging analysis. Most of high risk areas were located in the north, and south of the province, while the lower risk areas were located in central part of the province. Higher JD seropositive spread in Tabalong, Balangan, Hulu Sungai Utara, Hulu Sungai Tengah, Kotabaru, Barito Kuala, Banjarbaru, and Tanah Laut.

Figure 10. The Semivariogram Graph of JD Seropositive

Kriging method allowed better visualization and understanding of JD distribution. Kriging allows graphical investigation of spatial autocorrelation by using statistical models and creating isopleths maps of predicted values and the error of prediction (Wittich 2007). By setting semivariogram function, kriging can be used to comprehensively consider structural changes and randomness of the variables (Chen et al. 2007).

The result of JD mapping using kriging method showed that kriging interpolation can be used to predict or estimate the JD seropositive on un-sampled points. This is important in the epidemiology activities to present JD seropositive in all areas. Applying the kriging method allowed better visualization and understanding of JD distribution. The results of JD mapping can be used as a reference by local government to design a new program in JD control and eradication more effectively and efficiently.

Figure 12. The Standard Error of JD Mapping using Kriging Method

The kriging method allows validating the predicted values by using cross validation which expressed by standard error. The error map is useful to investigate the predictive performance of the kriging step but less useful for

analytical inferences (Berke 2004). Figure 12 shows the standard error of JD mapping based on the kriging method. The darker colour indicates the higher prediction error, while lighter colour indicates the lower prediction error. The larger error value indicating the error prediction due to the lower number of observed points in this area. High standard error found in the areas that are not taken JD samples. It shows that the sampling during surveillance is not distributed in all areas. Surveillance activities in South Kalimantan are still limited to endemic areas and areas that are potentially infected with JD. Among the total 151 sub districts in South Kalimantan province, JD samplings were only conducted in 31 sub districts of both active and passive surveillance.

The determination of the location and sample size are very important in JD mapping. JD mapping using kriging method will give better results if sampling size is proportional to the cattle population in all areas, because un-proportional sample allows the misinterpretation of JD distribution. It is highly recommended JD surveillance performed on all areas that have a high population of Bali cattle to obtain a more representative JD mapping.

4.4 Spatial Cluster Analysis

SaTScan results for the purely spatial scanning for clusters with high rates using the Poisson model gave coordinates location, radius, observed cases, expected cases, relative risk, likelihood ratio, and p – values for primary and secondary clusters. Using spatial scan statistic with the maximum spatial cluster size of <50% of the total population, two clusters identified in 2008, one cluster identified in 2009, and two clusters identified in 2010.

In 2008, a primary cluster was defined 10.95 km around the Martapura sub district, while secondary cluster located at Kuranji sub district (Figure 13). The relative risk (RR) within primary cluster was 79.90, with an observed number of cases of 7 compared with 1 expected case. The relative risk (RR) within secondary cluster was 10.77, with an observed number of cases of 2 compared with 1 expected case (Table 4). In 2009, one cluster was defined 30.45 km around Bumi Makmur sub district (Figure 14). The RR of this cluster was 95.14 with an observed of cases of 14 compared with 2 expected cases. In 2010, a primary

cluster located at Tanta sub district and a secondary cluster located at Barambai sub district with RR equal to 257.39 and 52.87 respectively (Figure 15). This elevated risk within the cluster was significant (p<0.001).

Table 4. Cluster of JD in South Kalimantan Province with Maximum Spatial Size Cluster of 50% of the Total Population Period of 2008 -2010 Year Location Radius

(km) Observed cases Expected cases RR* Likelihood Ratio P-value 2008 114°52'E

3º23'S 10.95 7 1 79.90 19.78 <0.001

115°40'E

3º28'S 0 2 1 10.77 2.76 <0.001

2009 114°36'E _3º31'S 30.45 14 2 95.14 25.21 <0.001

2010

115º22'E

2º13'S 0 12 1 257.39 51.51 <0.001

114º38'E

2º59'S 0 10 1 52.87 27.88 <0.001

*Relative Ratio (RR): ratio of the estimated risk within the cluster and the estimated risk outside the cluster.

JD cluster were identified each year are located in different area. It shows that the occurrence of JD has spread to areas where previously free from JD. However, JD cluster generally located at southern, western, and northern of the province. Overall in the period of 2008 to 2010, the most likely cluster (primary cluster) was identified at Tanta sub district and secondary cluster was identified at 48.20 km around Tamban sub district. The RR of primary cluster was 108.68 with an observed number of cases of 12 compared with 1 case, and the RR of secondary cluster was 10.94 with an observed number of cases of 34 compared with 7 cases (Figure 16).

Spatial cluster analysis identified a statistically significant cluster in area with high JD seropositive. JD cluster can be detected in the northern area of the province i.e. in Tanta sub district. JD cluster were also found in the western and southern area which includes district of Barito Kuala, Banjar, Banjarbaru, Banjarmasin, and Tanah Laut. The identified cluster contained 12.15% of the total population. JD cluster based on the surveillance during 2008 to 2010 showed that JD had spread to Tabalong, Banjarbaru, and Banjar. There are elevated risks of JD in this area.

Figure 13. Jembrana Disease Cluster with the Maximum Spatial Cluster Size of 50% of the Total Population Period of 2008

Figure 14. Jembrana Disease Cluster with the Maximum Spatial Cluster Size of 50% of the Total Population Period of 2009

Figure 15. Jembrana Disease Cluster with the Maximum Spatial Cluster Size of 50% of the Total Population Period of 2010

Figure 16. Jembrana Disease Cluster with the Maximum Spatial Cluster Size of 50% of the Total Population Period of 2008 - 2010

The identification of JD cluster is an important for JD surveillance. Cluster detection can help identifying environmental factors associated with disease and thus guide investigation of the etiology of diseases (Aamodt 2006). The identification of JD cluster will explain for the JD disease occurrence over the province due to the geographic factors or other factors. In an epidemiological, it is important that cluster detection tests adjust for the structure and spatial distribution of the underlying population at risk (Stevenson 2009).

In this study, the maximum spatial cluster size was set <50% of the total population to identifying the JD spatial cluster. The SaTScan recommended the upper limit as a percent of the population at risk. The spatial scan statistic allows the maximum spatial cluster size of <50% of the total population, because exploring for high risk of clustering in areas that include >50% of the population at risk is equivalent to explore for areas at low risk outside the cluster (Chhetri et al. 2010). A cluster of larger size would indicate areas of exceptionally low rates

outside the circle rather than an area of exceptionally high rate within the circle (Kulldorff 2010).

The cluster analysis evidenced that the geographical distribution of JD in South Kalimantan province was clustered in the specific area. There are similarities of JD clusters location in South Kalimantan province. Geographically, identified clusters located in the north and west of the province and in the low land area. The location of clusters detected by SaTScan generally matched those areas with higher seroprevalence. JD cluster found in the endemic area where the seroprevalence was higher in this area. Area within the clusters was endemic area and the number of cattle also higher compared with other area outside the cluster.

Overall, the spatial scan statistic is a useful tool to identify cluster in eradication program. Cluster with significantly high incidence of JD identified will be helpful of investigating the underlying causes of increased risk in the identified areas. The identification of spatial cluster may also to implement of new program in JD eradication more effectively and efficiently. JD preventive and eradication program should be a priority in the area within the clusters.

4.5 Factors Associated with the Spatial Distribution of Jembrana Disease

Most of JD seropositive distributed at the area where located in the low land. Among the total of 57 JD seropositive, 21 JD seropositive (36.84%) located at the area with elevation less than 25 meters above sea level, 22 JD seropositive (38.60%) located at the area with elevation between 25 – 50 meters above sea level, 14 JD seropositive (24.56%) located at the area with elevation between 50 – 100 meters above sea level. This is understandable due to more than half the area of South Kalimantan province is in low land with the elevation less than 100 meters above sea level.

Generally, the incidence of infectious disease is usually related to the cattle population. The higher cattle population allows high risk of infectious disease from suspected animal to healthy animal. A total of 18 JD seropositive (31.58%) were reported at the sub district with cattle density less than 5 heads/km2, one JD seropositive (1.75%) were reported at the sub district with cattle density between 5 – 10 heads/km2, 10 JD seropositive (17.54%) were reported at the sub district with cattle density between 10 – 20 heads/km2, one JD seropositive (1.75%) was reported at the sub district with cattle density between 20 – 30 heads/km2, and 27 (47.37%) were reported at the sub district with cattle density higher than 30 heads/km2.

In this study, the JD location also was analyzed with the distance to the main rivers. There were 13 JD seropositive (37.81%) located at area from 500 meters from rivers, 9 JD seropositive (15.79%) located at area from 1000 meters from rivers, 23 JD seropositive (40.35%) located at area from 5000 meters from rivers, and 12 JD seropositive (21.05%) located at area from 10,000 meters from rivers.

The relationship between JD seropositive and its related variables have been identified using regression analysis. Table 5 shows the results of regression analysis of JD seropositive. JD seropositive was positively associated with cattle density and distance to the main rivers and negatively associated with elevation. The results indicate that JD seropositive was higher in the area with higher cattle density than area with lower cattle density. The results also indicate that JD

Appendix 7. The Cluster Analysis Processing using SaTScan with the Maximum

Spatial Size Cluster of 50% of the Total Population Period of 2010

_____________________________ SaTScan v9.1.1

_____________________________

Program run on: Thu Jun 09 20:56:53 2011 Purely Spatial analysis

scanning for clusters with high rates using the Discrete Poisson model.

________________________________________________________________ SUMMARY OF DATA

Study period...: 2010/1/1 to 2010/12/31 Number of locations...: 151

Total population...: 221018 Total number of cases...: 31 Annual cases / 100000...: 14.0

________________________________________________________________ MOST LIKELY CLUSTER

1.Location IDs included.: Tanta

Coordinates / radius..: (9.7537e+006,321033) / 0 Population...: 541

Number of cases...: 12 Expected cases...: 0.076 Annual cases / 100000.: 2219.6 Observed / expected...: 158.14 Relative risk...: 257.39 Log likelihood ratio..: 51.507143

P-value...: < 0.000000000000000010 SECONDARY CLUSTERS

2.Location IDs included.: Barambai

Coordinates / radius..: (9.66822e+006,238457) / 0 Population...: 1973

Number of cases...: 10 Expected cases...: 0.28 Annual cases / 100000.: 507.2 Observed / expected...: 36.14 Relative risk...: 52.87 Log likelihood ratio..: 27.882414 P-value...: 0.0000000013

________________________________________________________________ PARAMETER SETTINGS

Input

---Case File : C:\Documents and

Population File : C:\Documents and

Settings\USER_GIS10\import55711318286157937.pop Coordinates File : C:\Documents and

Settings\USER_GIS10\import2190798369923768154.geo Time Precision : Year

Start Time : 2009/1/1 End Time : 2009/12/31 Coordinates : Cartesian Analysis

---Type of Analysis : Purely Spatial Probability Model : Discrete Poisson Scan for Areas with : High Rates

Output

---Results File : F:\cluster2010

Cluster File : F:\cluster2010.col.txt Cluster File : F:\cluster2010.col.dbf Stratified Cluster File : F:\cluster2010.sci.txt Stratified Cluster File : F:\cluster2010.cci.dbf Location File : F:\cluster2010.gis.txt Location File : F:\cluster2010.gis.dbf Relative Risks File : F:\cluster2010.rr.txt Relative Risks File : F:\cluster2010.rr.dbf Simulated LLRs File : F:\cluster2010.llr.txt Simulated LLRs File : F:\cluster2010.llr.dbf Data Checking

---Temporal Data Check : Check to ensure that all cases and controls are within the specified temporal study period.

Geographical Data Check : Check to ensure that all observations (cases, controls and populations) are within the specified geographical area.

Spatial Neighbors

---Use Non-Euclidian Neighbors file : No Use Meta Locations File : No

Multiple Coordinates Type : Allow only set of coordinates per location ID.

Spatial Window

---Maximum Spatial Cluster Size : 50 percent of population at risk Window Shape : Circular

Isotonic Scan : No Space And Time Adjustments

---Adjust for known relative risks : No Inference

---P-Value Reporting : Default Combination Adjusting for More Likely Clusters : No

Number of Replications : 999 Clusters Reported

---Criteria for Reporting Secondary Clusters : No Geographical Overlap

Additional Output

---Report Critical Values : No Report Monte Carlo Rank : No Print ASCII Column Headers : No Run Options

---Processer Usage : All Available Proccessors Logging Analysis : Yes

Suppress Warnings : No

________________________________________________________________ Program completed : Thu Jun 09 20:57:06 2011

Total Running Time : 13 seconds Processor Usage : 2 processors

Appendix 8. The Cluster Analysis Processing using SaTScan with the Maximum

Spatial Size Cluster of 50% of the Total Population Period of 2008

to 2010

_____________________________ SaTScan v9.1.1

_____________________________

Program run on: Thu Jun 09 20:31:45 2011 Purely Spatial analysis

scanning for clusters with high rates using the Discrete Poisson model.

________________________________________________________________ SUMMARY OF DATA

Study period...: 2009/1/1 to 2009/12/31 Number of locations...: 151

Total population...: 221018 Total number of cases...: 57 Annual cases / 100000...: 25.8

________________________________________________________________ MOST LIKELY CLUSTER

1.Location IDs included.: Tanta

Coordinates / radius..: (9.7537e+006,321033) / 0 Population...: 541

Number of cases...: 12 Expected cases...: 0.14 Annual cases / 100000.: 2219.6 Observed / expected...: 86.01 Relative risk...: 108.68 Log likelihood ratio..: 42.926016

P-value...: 0.0000000000000013 SECONDARY CLUSTERS

2.Location IDs included.: Tamban, Mekarsari, BJMBarat, Tabunganen, BJMTimur, BJMSelatan, BJMUtara, AnjirMuara, AluhAluh, Alalak, TatahMakmur, BJMTengah,

AnjirPasar, KertakHanyar, Mandastana, Gambut, BeruntungBaru, Belawang,

SungaiTabuk, BumiMakmur, LiangAnggang, Wanaraya, RantauBadauh, Jejangkit, Kurau, MartapuraBarat, LandasanUlin, Barambai, BjbUtara, Cerbon,

MartapuraTimur, BatiBati, Martapura, BjbSelatan, TambangUlang

Coordinates / radius..: (9.63203e+006,218863) / 48201.61 Population...: 26311

Number of cases...: 34 Expected cases...: 6.79 Annual cases / 100000.: 129.3 Observed / expected...: 5.01

Relative risk...: 10.94 Log likelihood ratio..: 36.834652

P-value...: 0.00000000000028

________________________________________________________________ PARAMETER SETTINGS

Input

---Case File : C:\Documents and

Settings\USER_GIS10\import7461930406283018454.cas Population File : C:\Documents and

Settings\USER_GIS10\import55711318286157937.pop Coordinates File : C:\Documents and

Settings\USER_GIS10\import2190798369923768154.geo Time Precision : Year

Start Time : 2009/1/1 End Time : 2009/12/31 Coordinates : Cartesian Analysis

---Type of Analysis : Purely Spatial Probability Model : Discrete Poisson Scan for Areas with : High Rates

Output

---Results File : F:\final50

Cluster File : F:\final50„ol.txt Cluster File : F:\final50„ol.dbf Stratified Cluster File : F:\final50ci.txt Stratified Cluster File : F:\final50„ci.dbf Location File : F:\final503e-313is.txt Location File : F:\final503e-313is.dbf Relative Risks File : F:\final50rr.txt

Relative Risks File : F:\final50rr.dbf Simulated LLRs File : F:\final50r.txt Simulated LLRs File : F:\final50r.dbf Data Checking

---Temporal Data Check : Check to ensure that all cases and controls are within the specified temporal study period.

Geographical Data Check : Check to ensure that all observations (cases, controls and populations) are within the specified geographical area.

Spatial Neighbors

---Use Non-Euclidian Neighbors file : No Use Meta Locations File : No

Multiple Coordinates Type : Allow only set of coordinates per location ID.

Spatial Window

---Maximum Spatial Cluster Size : 50 percent of population at risk Window Shape : Circular

Isotonic Scan : No Space And Time Adjustments

---Adjust for known relative risks : No Inference

---P-Value Reporting : Default Combination Adjusting for More Likely Clusters : No

Number of Replications : 999 Clusters Reported

---Criteria for Reporting Secondary Clusters : No Geographical Overlap

Additional Output

---Report Critical Values : No Report Monte Carlo Rank : No Print ASCII Column Headers : No Run Options

---Processer Usage : All Available Proccessors Logging Analysis : Yes

Suppress Warnings : No

________________________________________________________________ Program completed : Thu Jun 09 20:32:01 2011

Total Running Time : 16 seconds Processor Usage : 2 processors