Estimating the basic reproduction number of dengue transmission during 2002-2007 outbreaks in Bandung, Indonesia Asep K. Supriatna# Department of Mathematics, Padjadjaran University, Jl. Raya Bandung-Sumedang Km 21, Sumedang 45363, Indonesia
Abstract In this paper a dengue transmission mathematical model has been developed. The model assumes that human population is divided into three sub-populations: susceptible, infective and recovered sub-population, while the mosquito population is divided into two sub-populations: susceptible and infective. It is also assumed that the survival rates for humans and mosquitoes decrease with age. Two basic reproduction numbers are derived, namely, the host-to-vector basic reproduction number and the host-to-host basic reproduction number. An illustration on how to estimate the basic reproduction numbers from dengue incidence data is presented using the dengue incidence data during the 2002–2007 dengue outbreaks in Bandung, Indonesia. Keywords: Dengue transmission; mathematical model; basic reproduction number; Bandung, Indonesia.
Introduction Indonesia is an archipelago comprising of more than 13 660 islands and stretching from 06° 08’ N to 11° 15’ S latitude and 94° 45’ to 141° 65’ E longitude. It has tropical climate in all its regions. Indonesia experiences hot, humid weather throughout the year and is, therefore, endemic for vector-borne diseases including dengue transmitted by Aedes aegypti and Aedes albopictus. Since its first appearance in Jakarta and Surabaya in 1968, dengue fever (DF) and its more severe form, #
dengue haemorrhagic fever (DHF), have been growing steadily, with low incidence rates initially,[1,2] to a significant high incidence rate. Other changing patterns of dengue epidemiology have now been recognized which pose challenges for the control of this infection and have made difficult the understanding and control of the disease, let alone eliminating it.[3] Although the disease is endemic throughout the country, most dengue infections have occurred in urban areas where approximately
E-mail: aksupriatna@bdg.centrin.net.id, ak_supriatna@unpad.ac.id 21
Dengue Bulletin – Volume 33, 2009
Estimating the basic reproduction number of dengue transmission in Bandung, Indonesia
one third of the 240 million people of Indonesia live. Almost 60% of them live on the island of Java. Bandung, as one of the biggest cities in Java has been experiencing severe dengue infection ever since it was reported for the first time in 1969.[4] A recent study in a cohort of adult population shows that all the four known serotypes are circulating in Bandung, with DENV-2 predominating.[5] For the prevention and control of dengue epidemic in Indonesia, in general, there are three critical strategies: (i) DF/DHF case and vector surveillance; (ii) disease management; and (iii) changing behaviours of communities and building partnerships in order to reduce the presence, or improve the management, of larval habitats.[6] Various applications of insecticides, including the application of ultralow volume (ULV), have been used in order to reduce the larval and adult populations of the vectors.[7,8] Nonetheless, the dengue incidence rates, number of reported cases and deaths, and affected areas under DHF during the period 1992–2004 were alarming. Simulation of the application of ULV shows that the abrupt change of vector mortality due to ULV will result in a delay of the endemic stage but will only slightly reduce the severity of the infection.[9] A similar result is also shown for a different scenario in which dengue re-infection is possible.[10] This indicates that different strategies for control are needed to improve the current situation for combating the disease. One possible control strategy is through vaccination.[11] Some scientists argue that in the preparation for vaccine trials it is important to understand the epidemiology of the disease in an adult population where most of the individuals have been exposed to one or more DENV infections.[5] Other theoretical studies have been made to anticipate the application of a tetravalent vaccine currently being
developed[12,13] in an attempt to eliminate the disease, such as the study on the effects of vaccination in single serotype and in multiple serotype dengue infections[14-16] and in an agestructured population.[17] One important epidemiological concept related to minimum vaccination coverage is the basic reproduction number.[18] The present paper discusses a method of how to compute this basic reproduction number, which, in dengue cases, is slightly difficult due to the presence of the indirect transmission of the disease and other complications such as different serotypes of the agents. We then apply the theory to estimate the value of the basic reproduction number during the period of dengue outbreaks (2002–2007) in Bandung. In the next section we develop a mathematical model for dengue transmission as a basis for the computation of the basic reproduction number used in the subsequent sections.
Mathematical model It is well known in literatures that in a directtransmission disease model the threshold number R0 can be estimated by the ratio of life expectancy and mean age at infection.[19,20] Examples are abundant, including estimation that considers a direct transmission in an age-structured model[21]. In this paper we will show that, if the survival functions of humans and mosquitoes are exponentially negative, then this estimation in the directtransmission model generalizes into the indirect-transmission model. The method used in this paper gives a practical way to obtain the estimate of the basic reproduction number. This finding is important since the minimal effort to eliminate the disease depends on the basic reproduction number through the relation pc > 1 − 1/ R0 .[18,19,22]
22
Dengue Bulletin – Volume 33, 2009
Estimating the basic reproduction number of dengue transmission in Bandung, Indonesia
The model discussed here is the generalization of the model in a study by Fred Brauer[21] to include recovery compartment and indirect transmission of the disease, such as in dengue disease transmission. To formulate the model, let us assume that the host population N H is divided into three compartments, namely, susceptible (S H ), infective ( I H ) and recovered assumed immune individuals ( RH ). We assume that vectors remain infective for life due to their short life period compared with the duration of the disease, and hence the vector population only comprises two compartments, namely, susceptible (SV ) and infective ( IV ). Recruitment rate for the host and the vector is BH and BV , with the death rate µ H and µV , respectively. The disease transmission probability from vector to host and from host to vector is given by b and b . Suppose H V there exist non-negative monotonically non-increasing functions of age, QH (a ) and QV (a ) , describing the fraction of host and vector population, respectively, who survive to the age of a or more, such that QH (0) = 1 and QV (0) = 1. Since the host life expectancy is finite, such that
with S H (t ) = S H (0)QH (t )e . And if we assume that the rate of recovery is γ , then the number of the infective host at time t is given by: t − b H IV ( s ) ds − ∫ γ ds 0 t −a I H (t ) = I H (t ) + ∫ BH QH (a ) 1 − e ∫t −a da e 0 , t t
0
−
∫0 b H IV ( s ) ds
t
(3)
. Furthermore, with I H (t ) = I H (0)QH (t )e since RH (t ) = N H (t ) − S H (t ) − I H (t ) , then we have: t − b H IV ( s ) ds − ∫ γ ds 0 t −a RH (t ) = RH (t ) + ∫ BH QH (a ) 1 − e ∫t −a 1 − e da 0 , t t
0
− γ ds
∫0
t
(4)
with RH (t ) = N H (t ) − S H (t ) − I H (t ). Equations (2) to (4) constitute transmission dynamics in the host population. Analogously, the transmission dynamics in the vector population is governed by the following equations: 0 NV (t ) = NV (t ) + ∫ BV QV (a )da , 0 0 with NV (t ) = NV (0)QV (t ) , ∞
0
0
0
0
(5)
aQH (a )da < ∞ , and assuming the initial conditions for the host population is given by N H (0), then we have: 0
∫
∞
∫
∞
0
QH (a )da = LH and
t − bV I H ( s ) ds 0 SV (t ) = SV (t ) + ∫ BV QV (a )e ∫t −a da , 0 − b 0 with SV (t ) = SV (0)QV (t )e ∫ 0 t
t
(6)
N H (t ) = N (t ) + ∫ BH QH (a )da < ∞ , 0 H 0
t
V IH
( s ) ds
, and
(1)
with N H (t ) = N H (0)QH (t ) . Furthermore, since the per capita rate of infection in the population at time t is b H IV (t ), then the number of those susceptible at time t is given by: t − b H IV ( s ) ds 0 S H (t ) = S H (t ) + ∫ BH QH (a )e ∫t −a da 0 , t
0
t − bV I H ( s ) ds 0 , (7) IV (t ) = IV (t ) + ∫ BV QV (a ) 1 − e ∫t −a da 0 0 with IV (t ) = IV (0)QV (t ) . t
(2)
0 (t ) It is clear that limt →∞ N H (t ) , limt →∞ S H 0 0 limt →∞ I H (t ) and limt →∞ RH (t ) are all equal to zero, so does the same thing for the vector. The full model for the host-vector SIR
0
Dengue Bulletin – Volume 33, 2009
23
Estimating the basic reproduction number of dengue transmission in Bandung, Indonesia
(susceptible-infective-recover) model is given by equations (2) to (4) and (6) to (7). If γ = 0, then the model reduces to a host-vector SI (susceptible-infective) model. It is shown in Appendix A that there is a non-trivial endemic * * equilibrium ( I H , IV ) satisfying: * IH = ∫ BH QH (a ) 1 − e − b H IV a e −γ a da , 0 *
infection is given by the inverse of the force of infection, hence a H = 1 1 aV = * * b H IV b V IH . and
Considering (8) and (9), consequently the threshold number in (10) can be written as: γ R0
∞
(8) (9)
=
(B b H
H
∫
∞ 0
γ aQH (a )da BV bV ∫ aQV (a )da 0 *
)(
∞
) *
∞ ∞ 1 − e − b H IV a 1 − e − bV I H a B Qγ (a ) da BV ∫ QV (a ) da * * H ∫0 H 0 I IV H
* IV = ∫ BV QV (a ) 1 − e − bV I H a da , 0 *
∞
=
(∫
(b I γ H
* H H
∞ 0
Q (a) 1 − e
(
∫
∞ 0
γ * (a )da bV IV aQH ∫ aQV (a)da 0 * − b H IV a
)( )da )(∫
∞
∞ 0
QV (a ) 1 − e
(
)
* − bV I H a
)da )
if R > 1 , in which: R = BH b H ∫ aQH (a )e γ 0 0
γ 0
(
∞
−γ a
da BV bV ∫ aQV (a )da 0
)(
∞
). (10)
which can be further simplified as: γ R0
=
endemic equilibrium point.[23] In the following section we show that the value of the basic reproduction number can be estimated as the ratio of life expectancy ( LH and LV ) and mean age at infection ( a H and aV ), which usually are available in the field or literature.
γ R0 also determins the stability of the resulting
It is also shown that the threshold number
1 1 1 1 1 1 1 1 2 2 2 2 aH Μ H aV µV aH Μ H aV µV = 1 1 − µV − Μ H µV + ΜH + 1 1 1 1 aV aH − − 1 1 ΜH Μ + µV µ + H V Μ H Μ H + 1 µV µV + 1 a a H V aH aV 1 1 1 1 1 1 µV + = 1 + 1 + ΜH + ΜH aV Μ H aH µV aV aH µV
=
Finally we have:
L L γ = 1 + H 1 + V R0 aH aV
,
(11)
The estimation of the basic reproduction number We assume that the survival rates for humans and mosquitoes decrease with age, satisfying QH (a ) = e − µ a and QV (a ) = e − µ a respectively. SI model is not relevant for dengue infection, but we do the calculation of SI model for certain reasons in the next section. To facilitate calculation for both SIR and SI models, we − ( µ +γ ) a γ −γ a = e −Μ a . d e f i n e QH (a) = QH (a)e = e Following [19] and [20], the mean age at H V H H
where LH = 1 and LV = 1 . ΜH µV γ Next, we define R0γ ≡ R0 . We note γ that R0γ and R0 have the same threshold γ number, i.e. R0 > 1 ⇔ R0γ > 1 . Observe also that R0γ conforms more to the definition of the basic reproduction number in [19, eq. 5.9 p. 75]. In this regard, the basic reproduction number is the multiplication factor in n generation measured on a “per generation” basis. Alternatively, R0γ can be interpreted as
24
Dengue Bulletin – Volume 33, 2009
Estimating the basic reproduction number of dengue transmission in Bandung, Indonesia
the geometric average of the expected number of new cases in human population caused by one infected mosquito during its entire period of infectiousness and the expected number of new cases in the mosquito population caused by one infected human during his/her entire period of infectiousness. However, some authors have pointed out that the basic reproduction number measured per generation basis is less useful when the control effort needed to eliminate a host-vector disease (such as dengue) is to be targeted at only the host population.[24] They argue that the expected number of secondarily infected hosts that results from a single infected host is the square of the host-vector basic reproduction number, since two generations are required to transmit an infection from host to host: the first generation due to infection from a host to the vector population and the second generation due to infection from infected vectors back γ to the host population. Hence, R0γ ≡ R0 is the host-to-vector basic reproduction number
However, if we derive the basic reproduction number using the next generation matrix in [19, eq. 5.9 p. 75; see also Appendix B], then we have:
1 B B 1 G R0 = b H H bV V . µ H ( µ H + γ ) µV µV
(13)
It can be verified that R0G is comparable to the known form of the basic reproduction number. [10] If we denote R00 as the basic reproduction number for the SI model, that γ is R0 with γ = 0 , then we have the following relation: γ G 0 , R0 < R0 < R0
(14)
with G R0 R0 µH + γ = 0 = > 1 , G γ µH R0 R0
(15)
and R0 is the host-to-host basic reproduction number. We note that using host-vector basic reproduction certainly underestimates the control effort needed to eliminate the disease, if the only target of control is host population, such as in host vaccination programme, since the resulting minimum vaccination coverage is given by: pc = 1 −
γ
which consequently is, G 0 γ R0 = R0 R0 .
(14)
1 1 < 1 − γ .[22] R0γ R0
The next section will illustrate the implementation of the theory developed in this section with the data during the 2002–2007 dengue outbreaks in Bandung.
Note that if we derive the basic reproduction number directly from (10) then we have: BH b H BV bV γ R0 = 2 2 ( µ H + γ ) µV .
The basic reproduction number in Bandung, Indonesia We collected data consisting of the number of suspected and/or confirmed dengue patients from three major hospitals in Bandung, Advent, Muhammadiyah and St. Yusuf
(12)
Dengue Bulletin – Volume 33, 2009
25
Estimating the basic reproduction number of dengue transmission in Bandung, Indonesia
Hospitals, during the dengue outbreaks from 2002 up to 2007. We also collected data from all existing puskesmas (community health centres). Both data sets consist of patients’ name and age at the time they registered at the hospital or puskesmas due to dengue illness symptoms. A total of 22 981 patients, comprising 12 030 patients from hospitals and 10 951 patients from puskesmas , were analysed.
The yearly patients’ distribution is shown in Figure 1. During the period 2002–2005, this yearly number of patients was approximately 0.15% of the total population of the city for the respective year. Figure 2 shows that during the period 2002–2007, the dengue incidence steadily increased for three different classes: babies/ toddlers, teenagers and adults. This figure reveals that marginal increase of the older
Figure 1: Total number of patients per year from hospitals (left) and from puskesmas (right)
Figure 2: Total number of patients per year for different classes: babies/toddlers, teenagers and adults
26
Dengue Bulletin – Volume 33, 2009
Estimating the basic reproduction number of dengue transmission in Bandung, Indonesia
Figure 3: The distribution of suspect and confirmed dengue patients
class is slightly greater that those for the younger classes. A similar trend was also observed in other regions of Indonesia [25] and was consistent with those reported on a nationwide scale that there is a shifting of age of infection to older classes.[3,26] This is true for other countries as well.[27,28] It remains unclear whether the data from hospitals and puskesmas overlap or are mutually exclusive. To handle this, we consider those data sets as samplings taken from two different places. We will compare the resulting basic reproduction numbers computed using those different data sets and our formula. To compute the basic reproduction number, our formula needs the value of the human’s average age at infection. This can be done directly from the data sets we collected from the hospitals and puskesmas as follows. We model the survival of the human after catching the disease by the Weibull distribution function given by f (t ) =
Figure 4: Mean age at infection decreases during the outbreak period 2002–2005 followed by increase in the periods afterwards
1 mean µ = aΓ + 1 . Statistical analysis b shows that our data fit to this distribution with the resulting parameters given in Table 2 Dengue Bulletin – Volume 33, 2009
bt a a
b −1
e
t − a
b
, with
along with the corresponding curves in Figures 3 and 4. The left part of Figure 3 shows the Weibull curves from hospital data sets 2002–2006. The curves from bottom to top, in increasing order, indicate the outbreaks in 2002, 2003, 2004, 2005 and 2006, consecutively. The figure reveals that there is a shifting of peaks from higher age at infection to lower age at infection and an increase of peaks during the periods 2002 to 2004, slight shifts to a higher age at infection in 2005 and then shifts again to a lower age at infection in 2006. The data from the puskesmas does not give a clear pattern.
27
Estimating the basic reproduction number of dengue transmission in Bandung, Indonesia
Table 1: The parameters of the Weibull human survival model during the outbreak period 2002–2007 Source of data set Year 2002 2003 Hospitals 2004 2005 2006 2004 Puskesmas 2005 2006 2007 Detail computation has been described by Rubiana.[29]
a 25.56 24.88 23.25 20.90 22.35 22.68 19.83 21.18 23.04
b 1.62 1.49 1.34 1.29 1.28 1.21 1.25 1.32 1.21
µ 22.89 22.48 21.35 19.33 20.70 21.29 18.46 19.50 21.63
Another human parameter we need to find the estimation of the basic reproduction number is the rate of recovery γ . The value of the average infectious period, that is, the inverse of the rate of recovery, is believed to be between three and eight.[30] Other scientists suggest that this average infectious period is six days,[31] and hence human recovery rate is 1/6 per unit time. We also need the Ae. aegypti life expectancy. There are numerous values of life expectancy used in literatures, ranging from three days – in the field – to more than 90 days – when reared in laboratory.[32] Macielde-Freita et al.[32] also pointed out that Ae. aegypti must survive for periods longer than the sum of the initial non-feeding period plus the virus’ extrinsic incubation period in order to be able to transmit the disease to another human. Nulliparous females usually do not blood-feed for ≥2 days and the extrinsic incubation period of dengue virus is at least
10 days.[33,34] Taking into account those facts, many authors argued that the life expectancy of the mosquitoes must be at least 12 days. We will assume that the value of this life expectancy in the field is approximately 14 days as many scientists believe.[35,36,37] We also assume that to enable transmission, the average at infection for the mosquitoes is as early as the second day of their adult life stage. The resulting basic reproduction numbers for various times of outbreak using different sampling data sets are figured out in Table 2.
Concluding remarks The estimation of dengue basic reproduction number in Bandung presented in this paper is considered to be among the first attempts to be made for such a study. The resulting basic reproduction numbers derived by the present γ G method, R0 and R0 in Table 2, are noticeably higher than the known estimated basic
28
Dengue Bulletin – Volume 33, 2009
Estimating the basic reproduction number of dengue transmission in Bandung, Indonesia
Table 2: The basic reproduction number during the outbreak period 2002–2007 Host-to-vector reproduction number Host-to-host reproduction number 0 R0
Source of data set
Year
aH 22.89 22.48 21.35 19.33 20.70 21.29 18.46 19.50 21.63
LH 72.50 72.52 72.54 72.56 72.58 72.54 72.56 72.58 73.39 0.1667 0.1667 0.1667 0.1667 0.1667 0.1667 0.1667 0.1667 0.1667
R00 5.7740 5.8145 5.9314 6.1669 6.0042 5.9378 6.2806 6.1463 5.9282
R0 2.8294 2.8295 2.8295 2.8296 2.8295 2.8295 2.8297 2.8296 2.8295
R0G 4.0419 4.0561 4.0967 4.1773 4.1218 4.0989 4.2157 4.1703 4.0956
R0 8.0057 8.0059 8.0062 8.0068 8.0064 8.0062 8.0071 8.0067 8.0061
G R0
2002 2003 Hospitals 2004 2005 2006 2004 Puskesmas 2005 2006 2007
33.3386 33.8078 35.1813 38.0300 36.0502 35.2579 39.4453 37.7764 35.1438
16.3371 16.4518 16.7829 17.4499 16.9891 16.8012 17.7719 17.3916 16.7739
aH = LH =
Human’s average age at infection are computed from the data sets Human’s life expectancy is provided by the Indonesian Bureau of Statistics
aV = 2 days LV = 14 days reproduction numbers for dengue infection from a neighbouring country, Singapore,[38] and from other countries.[31,39,40,41] In general, the estimates of dengue reproduction numbers vary considerably between studies while the reasons behind this variability are still not wellexplained.[42] However, we can argue that our results are certainly higher than those of others since we follow the argument by Roberts MG, et al.[24] that the host-vector basic reproduction number should be computed as a secondgeneration host-to-host basic reproduction number when it is only host population that is to be controlled. The result presented here is a crude approximation of the true value of the basic reproduction number since we ignored some details of the structure and transmission mechanism of the disease. Other factors that are worthy of being taken into account are the intrinsic and extrinsic incubation times. By taking these factors into consideration we predict that the resulting basic reproduction number will be discounted by the delay time of incubation periods (see [17] as an illustration). Other factors, such as the presence of multiple strains, may also increase the realism of the model, and hence, refine the estimate of the basic reproduction number.
Dengue Bulletin – Volume 33, 2009
29
Estimating the basic reproduction number of dengue transmission in Bandung, Indonesia
Acknowledgements This work was supported financially by the Indonesian Government (the Directorate of Higher Education) through Penelitian Hibah Kompetensi – Batch 1, 2008–2010. The author would like to thank an anonymous referee who read the earlier version of the paper and made constructive suggestions to improve the quality of the paper.
* * IH ) , (A.6) = ∫ BH QH (a ) 1 − e − b H IV a e −γ a da ≡ F1 ( IV 0 *
∞
* * IV ) . (A.7) = ∫ BV QV (a ) 1 − e − bV I H a da ≡ F2 ( I H 0 *
∞
Note that (A.6) can be written as a composition, ∞ 1− e− bV I * H s ds a − b H ∫ BV QV ( s ) ∞ −γ a * * IH ) = ∫ BH QH (a) 1 − e 0 = F1 F2 ( I H e da 0
.
(A.8)
Appendix A Since we are interested in the long-term behaviour of the model, let us consider the limiting condition of the system (1) to (7). Whenever t → ∞ , the system has the same dynamic behaviour as the following system: ∞ − b H IV ( s ) ds S H (t ) = ∫ BH QH (a )e ∫t −a da , 0 t
It is easy to see that (0, 0) is the trivial or disease-free equilibrium. To find the non-trivial endemic equilibrium we could observe that F1 F2 is bounded, satisfying d ( F1 F2 ) > 0 and dI H
d 2 ( F1 F2 ) < 0 , meaning that it is increasing 2 dI H
(A.1)
and concave down. Therefore, a unique * non-trivial value of I H exists if and only if d ( F1 F2 )(0) > 1 , which is equivalent to dI H H
∞ − b H IV ( s ) ds − γ a I H (t ) = ∫ BH QH (a ) 1 − e ∫t −a e da , (A.2) 0 t
∞ − b H IV ( s ) ds 1 − e −γ a da , RH (t ) = ∫ BH QH (a ) 1 − e ∫t −a 0 t
(B b H
∫
∞ 0
aQH (a )e −γ a da BV bV ∫ aQV (a )da > 1 0
)(
∞
)
(A.3)
or simply R > 1 , with R being the threshold number as in equation (10). γ 0 γ 0
∞ − bV I H ( s ) ds SV (t ) = ∫ BV QV (a )e ∫t −a da , 0
t
(A.4) (A.5)
Appendix B Let kij denote the expected number of new cases of type i , caused by one infected individual of type j , during the entire period of infectiousness. In this case, i, j ∈ { 1, 2}. Moreover, let type 1 be infection in host and type 2 be infection in vector, and by considering the exponential survival rates −µ a , then we have: QH (a ) = e − µ a and QV (a ) = e 1 1 * * k11 = 0 , k12 = b H S H , k21 = bV SV , (µH + γ H ) µV H V
∞ − bV I H ( s ) ds IV (t ) = ∫ BV QV (a ) 1 − e ∫t −a da . 0 t
Considering that we are only interested in the infective compartments without loss of generality, we can concentrate only on the sub-system (A.2) and (A.5), in which the equilibrium of the sub-system is given by * * * * and IV satisfying (IH , IV ) with I H
30
Dengue Bulletin – Volume 33, 2009
Estimating the basic reproduction number of dengue transmission in Bandung, Indonesia
k22 = 0 . In a completely virgin population, the B * steady states population are given by S H = H µH B and S = V . Theorem 5.3 of [19] implies that µV * V
matrix, with the dominant eigen value given by: B 1 BV 1 R0G = k12 k21 = b H H bV , µ H µV µV ( µ H + γ H )
(B.1)
0 K = (kij )= k12
k21 is the next generation 0
which is obviously the square root of the basic reproduction number in (13).
References [1] Hotta S, Miyasaki K, Takehara M, Matsumoto Y, Ishihama Y, Tokuchi M et al . Clinical and laboratory examinations on a case of “hemorrhagic fever” found in Surabaja, Indonesia, in 1968. Kobe J Med Sci. 1970;16(4):203-210. [2] Kho LK, Wulur H, Himawan T, Thaib S. Dengue haemorrhagic fever in Jakarta (follow up study). Paediatr Indones. 1972;12(1):1-14. [3] Setiati TE, Wagenaar JFP , de Kruif MD, Mairuhu ATA, van Gorp ECM, Soemantri A. Changing epidemiology of dengue haemorrhagic fever in Indonesia. Dengue Bulletin. 2006;30:1-14. [4] Rivai A, Hamzah S, Rahman O, Thaib S. Dengue and dengue haemorrhagic fever in Bandung. Paediatr Indones. 1972;12(1):40-48. [5] Porter KR, Beckett CG, Kosasih H, Tan RI, Alisjahbana B, Rudiman PI, Widjaja S, Listiyaningsih E, Ma’Roef CN, McArdle JL, Parwati I, Sudjana P, Jusuf H, Yuwono D, Wuryadi S. Epidemiology of dengue and dengue hemorrhagic fever in a cohort of adults living in Bandung, West Java, Indonesia. Am J Trop Med Hyg. 2005;72(1):60-66. [6] Kusriastuti R, Sutomo S. Evolution of dengue prevention and control programme in Indonesia. Dengue Bulletin. 2005;29:1-6. [7] Suroso T. Dengue haemorrhagic fever in Indonesia – epidemiological trends and development of control strategy. Dengue Bulletin. 1996;20:35-40. [8] World Health Organization, Regional Office for South-East Asia. Dengue/dengue haemorrhagic fever prevention and control programme in Indonesia: report of an external review, Jakarta, 5-19 June 2000 . Document SEA-Haem. Fever-73/SEA-VBC-79. New Delhi: WHO/ SEARO, 2001. [9] Esteva L, Vargas C. Analysis of a dengue disease transmission model. Math Biosci. 1998;150:131. [10] Soewono E, Supriatna AK. A Two-dimensional model for the transmission of dengue fever disease. Bull Malays Math Sc Soc. (Second Series) 2001;24:49-57. [11] C l a r k e T. B r e a k b o n e f e v e r. N a t u r e . 2002;416:672. [12] Kinney RM, Huang CY. Development of new vaccines against dengue fever and Jap ane se e nc e p h al itis. I nter virolog y. 2001;44:176-191. [13] Konishi E, Kosugi S, Imoto J. Dengue tetravalent DNA vaccine inducing neutralizing antibody and anamnestic responses to four serotypes in mice. Vaccine. 2006;24(12):2200-2207.
Dengue Bulletin – Volume 33, 2009
31
Estimating the basic reproduction number of dengue transmission in Bandung, Indonesia
[14] Soewono E, Supriatna AK. Paradox of vaccination predicted by a simple dengue disease model . In: M.C. Joshi et al. Eds. Industrial Mathematics. New Delhi: Narosa, 2006. p. 459. [15] Nuraini N. Mathematical models for external and internal dengue disease transmission. Ph.D. Dissertation, ITB Bandung – Indonesia (unpublished, in Indonesian), 2008. [16] Supriatna AK, Nuraini N, Soewono E. Mathematical models of dengue transmission and control: a survey. In: Basak Ganim and Adam Reis (eds.), Dengue Virus: Detection, Diagnosis and Control . New York: Nova Science Publishers. (in press). [17] Supriatna AK, Soewono E, van Gils SA. A twoage-classes dengue transmission model. Math Biosci. 2008;216:114-121. [18] Anderson RM, May RM. Infectious diseases of humans. Oxford: Oxford University Press 1991. [19] Diekmann O, Heesterbeek JAP , Mathematical epidemiology of infectious diseases. New York: John Wiley and Sons, 2000. [20] Nishiura H. Mathematical and statistical analyses of the spread of dengue. Dengue Bulletin. 2006;30:51-67. [21] Brauer F. A model for an SI disease in an agestructured population. Disc Cont Dyn Sys-Ser B. 2002;2:257-264. [22] Scherer A, McLean A. Mathematical model of vaccination. British Medical Bulletin. 2002;62:187-199. [23] Supriatna AK, Soewono E. Analysis of an SIR dengue transmission model with agedependent survival rates (in prep.) [24] Roberts MG, Heesterbeek JAP , A new method for estimating the effort required to control an infectious disease. Proc Roy Soc Lond. 2003;B270:1359.
[25] Corwin AL, Larasati RP , Bangs MJ, Wuryadi S, Arjoso S, Sukri N, Listyaningsih E, Hartati S, Namursa R, Anwar Z, Chandra S, Loho B, Ahmad H, Campbell JR, Porter KR, Epidemic dengue transmission in southern Sumatra, Indonesia. Trans R Soc Trop Med Hyg. 2001;95:257. [26] Suwandono A, Kosasih H, Nurhayati, Kusriastuti R, Harun S, Ma’roef C, Wuryadi S, Herianto B, Yuwono D, Porter KR, Beckett CG, Blair PJ. Four dengue virus serotypes found circulating during an outbreak of dengue fever and dengue haemorrhagic fever in Jakarta, Indonesia, during 2004. Trans R Soc Trop Med Hyg. 2006;100(9):855-62. [27] Ooi EE, Goh KT, Gubler DJ, Dengue prevention and 35 years of vector control in Singapore. Emerg Infect Dis. 2006;12:887. [28] Patumamond J, Tawichasri C, Nopparat S, Dengue hemorrhagic fever in Uttaradict, Thailand. Emerg Infect Dis. 2003;9:1348. [29] Rubiana I. Age-structured model of dengue patients in Bandung during 2002-2007 dengue outbreaks. S1 Thesis, Dept. Mathematics, University Padjadjaran 2008 (unpublished, in Indonesian). [30] Halstead SB. Dengue . In: Tropical and geographical medicine. Warren KS, Mahmoud AAF. Eds. New York: McGraw-Hill, 1990. p. 675-684. [31] Marques CA, Forattini 0P , Massad E. The basic reproduction number for dengue fever in Sao Paulo state, Brazil: 1990- 1991 epidemic. Trans R Soc Trop Med Hyg. 1994;88:58-59. [32] Maciel-de-Freita R, Codeco CT, Lourenco-deOliveira R. Body size-associated survival and dispersal rates of Aedes aegypti in Rio de Janeiro. Med Vet Entomol. 2007;21:284-292. [33] Kuno G. Review of the factors modulating dengue transmission. Epidemiological Review. 1995;17:321-335 .
32
Dengue Bulletin – Volume 33, 2009