modelling trachoma: a review focusing on the get 2020 goals€¦ · web viewin helping to reduce...
TRANSCRIPT
![Page 1: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/1.jpg)
Mathematical modelling of trachoma transmission, control and
elimination
Authors
Amy Pinsent*1, Isobel M. Blake||, Maria-Gloria Basáñez†, Manoj Gambhir*
Affiliations* Department of Epidemiology and Preventive Medicine, Monash University, Melbourne, Australia
|| MRC Centre for Outbreak Analysis and Modelling, Department of Infectious Disease Epidemiology, School of Public Health, Faculty of Medicine (St. Mary’s Campus), Imperial College London, London, United Kingdom
† London Centre for Neglected Tropical Disease Research, Department of Infectious Disease Epidemiology, School of Public Health, Faculty of Medicine (St. Mary’s Campus), Imperial College London, London, United Kingdom 1 Corresponding author: E-mail: [email protected]
Contents
1. Introduction
1.1 Clinical and epidemiological features
1.2 Trachoma control and elimination
1.3 The role of mathematical models
1.3.1 Deterministic and stochastic modelling
1.3.2 Modelling infection and modelling disease
1.4 Aims and objectives of this review
2. Methods
3. Results
3.1 Characteristics of identified studies
3.2 Deterministic and stochastic models
3.3 Data sets used and model fitting to data
1
![Page 2: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/2.jpg)
3.4 Transmission intensity and the basic reproduction number
3.4.1 Estimating the force of infection
3.4.2 Estimating the basic reproduction ratio
3.5 Acquired immunity: recovery rate and infectivity
3.6 Infection and active disease, modelling of disease sequelae
3.7 Simulation of interventions, forecasting infection and disease, and analysis of
their cost effectiveness
4. Discussion
4.1 Using modelling to determine the feasibility of the GET 2020 goals
4.2 Conclusions
Abstract
The World Health Organization has targeted the elimination of blinding trachoma by the
year 2020. To this end, the Global Elimination of blinding Trachoma (GET 2020) alliance
relies on a four-pronged approach, known as the SAFE strategy (S for trichiasis surgery; A for
antibiotic treatment; F for facial cleanliness, and E for environmental improvement). Well-
constructed and parameterised mathematical models provide useful tools that can be used
in policy making and forecasting in order to help control trachoma and understand the
feasibility of this large-scale elimination effort. As we approach this goal, the need to
understand the transmission dynamics of infection within areas of different endemicities, to
optimize available resources, and to identify which strategies are the most cost effective
becomes more pressing. In this study, we conducted a review of the modelling literature for
trachoma and identified 23 articles that included a mechanistic or statistical model of the
transmission, dynamics and/or control of (ocular) Chlamydia trachomatis. Insights into the
dynamics of trachoma transmission have been generated through both deterministic and
stochastic models. A large body of the modelling work conducted to date has shown that, to
varying degrees of effectiveness, antibiotic administration can reduce or interrupt trachoma
transmission. However, very little analysis has been conducted to consider the effect of non-
pharmaceutical interventions (and particularly the F and E components of the SAFE strategy)
2
![Page 3: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/3.jpg)
in helping to reduce transmission. Furthermore, very few of the models identified in the
literature review included a structure that permitted tracking of the prevalence of active
disease (in the absence of active infection) and the subsequent progression to disease
sequelae (the morbidity associated with trachoma and ultimately the target of GET 2020
goals). This represents a critical gap in the current trachoma modelling literature, which
makes it difficult to reliably link infection and disease. In addition it hinders the application
of modelling to assist the public health community in understanding whether trachoma
programmes are on track to reach the global elimination of blinding trachoma by 2020.
Another gap identified in this review was that of the 23 articles examined, only 1 considered
the cost effectiveness of the interventions implemented. We conclude that although good
progress has been made towards the development of modelling frameworks for trachoma
transmission, key components of disease sequelae representation and economic evaluation
of interventions are currently missing from the available literature. We recommend that
rapid advances in these areas should be urgently made to ensure that mathematical models
for trachoma transmission can robustly guide elimination efforts and quantify progress
towards GET 2020.
Keywords: Trachoma, Control, Elimination, GET2020, R0, Mathematical modelling
1. Introduction
1.1 Clinical and epidemiological features
Trachoma is one of 17 neglected tropical diseases (NTDs) prioritized by the World Health
Organization (WHO) for control and elimination through preventive chemotherapy or
intensified disease management strategies (WHO, 2015a). NTDs are mostly responsible for
chronic infections/conditions that can cause severe morbidity in affected individuals, leading
to long-term disability, but are deemed to be associated with relatively low mortality in
3
![Page 4: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/4.jpg)
comparison to acute, epidemic infectious diseases (WHO, 2015a; Hotez et al., 2014).
Transmission of NTDs is facilitated by living conditions that are associated with poverty, such
as poor housing and sanitation, and limited access to clean water and basic healthcare
(Hotez et al., 2009).
Trachoma is the leading global cause of infectious blindness, and is currently
estimated to affect 84 million people across 51 endemic countries (WHO, 2012). An
estimated 1.8 million people are visually impaired as a result of the disease, of which 0.5
million people are irreversibly blind (WHO, 2012; WHO, 2015b). Active inflammatory disease
—trachomatous follicular and trachomatous inflammatory (TF, TI, according to the WHO
simplified grading scheme (Taylor et al., 2014))—is caused by infection with the bacterium
Chlamydia trachomatis. Repeated infection with the bacteria leads to an
immunopathological response characterized by scarring of the inner part of the eyelid, and
an eventual curling-in of the eyelashes, which abrades the corneal surface. This can lead to
trachomatous trichiasis (TT), corneal opacity (CO) and blindness. Excess mortality is also
reported to be associated with blinding trachoma (Hotez et al., 2014). Estimates of the
Disability-Adjusted Life Years (DALYs) due to trachoma have been variable. The Global
Burden of Disease (GBD) 1990 study estimated the burden of trachoma (all ages) to be
144,000 (95% uncertainty interval (95% UI) 104,000–189,000) DALYs, whereas the GBD 2010
study reported a value of 334,000 (95% UI 243,000–438,000) (Murray et al., 2012). Other
authors have set this figure at least 1 million DALYs (Evans and Ranson, 1995) or as high as
3.6 million, with the highest proportion (72%) contributed by sub-Saharan Africa (Frick et al.,
2003). Among the major causes of blindness in 2010, trachoma represented 5.2% in sub-
4
![Page 5: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/5.jpg)
Saharan Africa (Naidoo et al., 2014). An accurate quantitative estimate of the burden of
trachoma remains, however, challenging due to several factors, including scarce data
availability—which limits the ability to estimate accurately the number of people infected—
and an unresolved issue as to whether trichiasis should be considered as a disabling disease
sequela (Burton and Mabey, 2009). The economic impact of trachoma in terms of lost
productivity is estimated to range between US $2.9 and 5.3 billion annually, rising to US $8
billion if trichiasis is included in the estimate (WHO, 2015b).
Countries reported to have the highest prevalence of infection are located in East
Africa and the Sahel belt; however, trachoma is also prevalent in Southeast Asia, the Middle
East, Indian sub-continent and Latin America (Burton and Mabey, 2009), although the
distribution and prevalence of infection are far more heterogeneous in these regions in
comparison to sub-Saharan Africa. While trachoma was previously prevalent in Europe and
North America only one hundred years ago, improvements in sanitary and living conditions
resulted in the gradual disappearance of infection (Burton and Mabey, 2009).
Infection with C. trachomatis is spread through two primary routes. The first is direct
personal contact which could be direct hand contact with an infected individual or through
contact with clothing which has contacted infectious discharge (Burton and Mabey, 2009).
The second route involves eye-seeking flies (e.g. Musca sorbens) which have contacted the
discharge from an infected persons eyes or nose (Emerson et al., 2004). For transmission of
infection to be sustained it must be consistently transmitted from person to person. The
severity of disease experienced by an infected individual varies with age and hence their
duration of exposure to infection (Bailey et al., 1999; Grassly et al., 2008). Infection with C.
5
![Page 6: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/6.jpg)
trachomatis becomes shorter in duration and reduced in frequency as individuals age;
therefore, the highest burden of C. trachomatis is observed in young children (Bailey et al.,
1999; Grassly et al., 2008). Repeated infection with age (continuous exposure) leads to
conjunctival scarring, ultimately leading to trachomatous trichiasis (TT), corneal opacity (CO)
and blindness as mentioned above (Burton and Mabey, 2009; West et al., 1991). Several
epidemiological surveys have suggested that severe sequelae in the form of TT and CO
disproportionally affect women in comparison to men, as a result of women having a higher
exposure to the reservoir source of infection, which is reported to be young children
(Courtright and West, 2004; West et al., 1991).
A number of risk factors for trachoma transmission have been identified, including
i) secretions from the eye which other individuals may come into contact with, and which
may also attract flies which help to facilitate transmission (Emerson et al., 2004; Ngondi et
al., 2008); ii) overcrowding within the household, which increases the frequency of contact
between individuals potentially leading to more frequent infection events (Abdou et al.,
2007; Ngondi et al., 2008); iii) limited supplies of clean water resulting in infrequent face
washing, general poor hygiene practice, and lack of easy access to latrines, which can lead
to a buildup of faecal matter in the environment which attracts eye-seeking flies (Emerson
et al., 2004). Transmission intensity of trachoma within a community is classified according
to the prevalence of active disease in 1 – 9 year olds. Communities are considered
hyperendemic if the prevalence of active disease in this age group is > 20%, mesoendemic if
prevalence is >10% but <20%, and hypoendemic if prevalence is <10% (Wright and Taylor,
2005).
6
![Page 7: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/7.jpg)
1.2 Trachoma control and elimination
The WHO has advocated for the elimination of blinding trachoma by 2020, and the Global
Alliance for the Elimination of Trachoma by 2020 (GET 2020) was established to develop
criteria to help to achieve this goal. These criteria are: 1) to reduce prevalence of TF and TI
(active disease) to <5% in 1–9 year olds across all endemic communities; 2) to reduce the
population prevalence of TT to <1 per 1,000 persons; 3) to enhance the use of the Facial
cleanliness (F) and Environmental improvement (E) components of the SAFE strategy
(Surgery for trichiasis, Antibiotics to combat the infection, Facial cleanliness, and
Environmental improvement) (West, 2003). The GET 2020 goals were developed to
eliminate blinding trachoma by 2020, not completely eliminate infection in the population.
Within the SAFE framework, as the public health burden of trachoma is reduced, infection
will be controlled, but not completely eliminated (hereafter, therefore, any reference to
elimination refers to the elimination of blinding trachoma, unless otherwise specified). The
WHO endorses the implementation of the full SAFE strategy in order to control and treat
trachoma. As mentioned above this comprises of four key components, more specifically
described as: i) surgery in order to correct trichiasis; ii) mass distribution of azithromycin as
the antibiotic of choice used to treat and clear active infection in the community (topical
tetracycline is used in very young children), iii) promotion of facial cleanliness in order to
reduce transmission via eye discharge, and iv) environmental modifications to improve living
conditions, ensuring that the environment no longer helps to facilitate the transmission of
infection (this can be in principle achieved through a number of avenues, such as facilitating
7
![Page 8: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/8.jpg)
the provision of clean water, increasing access to well-designed latrines, helping to reduce
the population density of flies, and reducing overcrowding) (West, 2003).
The London Declaration on Neglected Tropical Diseases (NTDs) (Uniting to Combat
NTDs, 2012) has led to a renewed commitment to control and/or eliminate these morbidity-
causing diseases from some of the poorest countries in the world. In addition, it is expected
that the recently established Neglected Tropical Diseases Modelling Consortium
(http://www.ntdmodelling.org/) will help to facilitate the exchange and discussion of ideas
across a range of NTDs and will foster and enhance collaboration between different
mathematical modelling groups. This will help to address many urgent policy issues
concerning the control and elimination of NTDs that can only be answered through the use
of quantitative tools (Basáñez and Anderson, 2015). Models are ideally suited to answering
a wide range of questions relating to the possible impact of various interventions within
populations affected by NTDs, e.g. which interventions to deploy, to whom, how often and
for how long should they be administered.
1.3 The role of mechanistic and statistical models
The epidemiology and dynamics of trachoma infection arise from a complex set of
contributory factors, including the natural history of infection and disease, which is
governed by an individual’s time-varying immunological response to and clearance of
infection. Transmission of trachoma can be altered by the behaviour and social contact
patterns of people within the community. At an even greater scale, the climate and local
ecology, may affect the transmissibility of the causative bacteria of trachoma,
8
![Page 9: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/9.jpg)
C. trachomatis. Integrating these effects across several hierarchical scales is a challenge for
any infectious disease.
However, through careful determination of key features of the epidemiology of
infection, models can be formulated as time-dependent mathematical expressions and
solved computationally. These include, but are not limited to, discrete event, agent-based,
or differential equation-based simulations. Equally, a quantitative understanding of data
generating processes can be provided through the development of statistical models.
Statistical models can help to understand the relationship between an outcome of interest
and external variables which may be important within the system, and this can be done
through techniques such as linear or logistic regression. Moreover, epidemiological data
rarely captures observations at every point in the transmission or infection cycle, thus
statistical models such as hidden Markov Models can be used to help provide insights into
the dynamic infection process where certain states of the infection process are not
observed, but the observed epidemiological outcome is dependent on these states.
Mathematical and statistical models, which seek to incorporate key features of the
population biology, demography and ecological covariates (often referred to as mechanistic
models) remain among the only quantitative methods capable of performing this kind of
integration.
1.3.1 Deterministic and stochastic modelling
Mechanistic models of disease transmission are commonly divided into either deterministic
or stochastic. Put simply, results from a deterministic model are fully determined by the
9
![Page 10: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/10.jpg)
parameter values chosen and the initial conditions and, therefore, model outputs are the
same every time the model is run (for a given parameter set). In contrast, stochastic models
contain inherent randomness (demographic and/or parametric) and the same parameter
sets (which will have distributions rather than only nominal values) and initial conditions
used will result in a range of output results. Deterministic models describe the average
behaviour of a system and, therefore, the average of many stochastic realisations tends to
approximate the deterministic solution.
The deterministic approach may miss some aspects of disease transmission,
especially in the context of small populations and low infection prevalence, where stochastic
fade-out or take-off may play an important role. However, the simplification gained through
the use of deterministic models allows simpler fitting of models to data to estimate
parameters of interest, a more transparent representation of complex natural histories of
infection, along with realistic population demography. Irrespective of the model structure
used, mathematical models are unlikely to be informative unless they are fitted to or
informed by high quality baseline and follow up surveillance data. The decision to develop a
deterministic or a stochastic model should depend upon the question under investigation.
For example, if data from a small community were analysed and the possibility of
elimination was being explored, a stochastic model would likely be considered more
suitable. However, if one were analysing impacts of different interventions at a population
level, a deterministic model may be more appropriate.
1.3.2 Modelling infection and modelling disease
10
![Page 11: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/11.jpg)
The vast majority of data collected from control programmes within endemic communities
have been obtained through clinical diagnosis of active disease or the disease sequelae
(Jimenez et al., 2015). More recently, testing for active infection has been performed in
clinical field studies using polymerase chain reaction (PCR) methods to identify active
infection from conjunctival swabs. However, laboratory tests can suffer from cross-
contamination across numerous stages of the DNA amplification process and the presence
of inhibitors of DNA amplification within samples can also cause amplification problems
(Solomon et al., 2004). Additionally, the collection of individual-level PCR data within a
community is prohibitively expensive, limiting its wide-spread use in surveillance. However,
studies that have been able to collect data on active infection and active disease prevalence
highlight the complex relationship between active infection and disease, whereby
individuals can have detectable active disease but do not have a PCR detectable infection
(Solomon et al., 2004). Hence, active disease can persist much longer than infection (Grassly
et al., 2008; Harding-Esch et al., 2009). This highlights the need for modelling studies to
explicitly account for the period of infection and active disease separately in order to insure
the prevalence of active disease is not underestimated.
The first two GET 2020 goals relate to the specific outcomes of active disease.
Therefore in order to understand if the GET 2020 goals are going to be achieved in different
localities, models of C. trachomatis transmission must also include progression towards
disease sequelae.
1.4 Aims and objectives of this review
11
![Page 12: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/12.jpg)
In this review we report previously published mechanistic and statistical mathematical
models that have been developed to provide insight into the transmission, dynamics and
control of trachoma. We compare the different mathematical model structures that have
been published, whether they are stochastic or deterministic and the type of data the
models were fitted to. We consider the availability of data on infection and disease
prevalence, and the estimation of epidemiological parameters such as the basic
reproduction number of the infection, R0. We evaluate how models have been used to
assess the impact of different interventions, and lastly how such models can be used as
disease forecasting tools to help to understand how achievable the GET 2020 goals are. We
then move to a discussion about which of these published studies contain results (or useful
modelling frameworks) pertaining to the GET 2020 goals. Lastly, we address what questions
require further elucidation before critical trachoma control-related questions can be
answered by mathematical models, and provide a perspective on future modelling
directions.
2. Methods
We performed a review of the literature to address and explore the mechanistic and
statistical models of trachoma transmission and control that have been developed to date.
Our search was performed through PubMed on the 24th June, 2015 with no restriction on
the year of publication or language applied. We employed a simple set of broad search
terms pairing the term “trachoma” with the following key words: [model OR modeling OR
modelling], a second search using the keyword [mathematical] was also performed.
12
![Page 13: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/13.jpg)
Studies were eligible for inclusion in the literature review if they developed a
mechanistic transmission model or a statistical model which quantified or modelled the
impact of an intervention, through regression analysis or by fitting different statistical
distributions to surveillance data. Statistical models also eligible for inclusion were those
estimating epidemiological parameters and inferring the true dynamics of infection though
Markov models. All study types were eligible if they met these inclusion criteria. Articles
were selected for inclusion by firstly reviewing the full title and abstracts of all studies that
were identified in the initial review searches. Studies that did not meet the inclusion criteria
after this step were excluded from the analysis. The full texts of potentially relevant articles
were then independently examined by AP, and data extracted were independently verified
by MG. MG was consulted by AP on queries relating to the inclusion or exclusion of studies
analysed in the review.
Once a set of eligible articles was established, data from each article were extracted.
The data extracted were as follows: i) the type and structure of the mechanistic or statistical
model applied and/or developed in the study; ii) why the study was performed, and the
data that were used to inform the modelling analysis presented; iii) whether the study
population was classified as a hyper-, meso- or hypoendemic and iv) key findings from the
study and how they related to the existing GET 2020 goals (Table 1). We then assessed if
epidemiological parameters had been reported, and the values estimated or used in the
model were noted. These were: the transmission rate or transmission coefficient (commonly
known as the beta parameter), the age-specific duration of infection (or its reciprocal, the
recovery rate), and the basic reproduction number R0. If R0 was not explicitly calculated, we
13
![Page 14: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/14.jpg)
used data on the transmission rate (parameter beta) and the recovery rate (denoted as
parameter gamma in this review) to estimate a crude R0 (where R0 is defined as the ratio
between these latter two parameters) (Ray et al., 2009) (Table 2).
Table 1 approximately here
3. Results
3.1 Characteristics of identified studies
Our initial two searches yielded 172 publications, with 18 publications being repeated across
the two searches. Of the 154 remaining publications, 24 articles were deemed appropriate
after reviewing the title and abstracts of all unique publications. The full text of these
articles was obtained and assessed. After this second screening stage, 23 articles were
deemed to meet the review’s inclusion criteria (Table 1). Twenty of these publications
presented mechanistic transmission models and 4 presented statistical models.
Of the 20 articles identified as describing transmission dynamics of trachoma, 9 used
a deterministic framework (including 4 catalytic models) and 9 used stochastic models, and
1 used both frameworks. One of the statistical models used linear and logistic regression, 2
used a latent/hidden Markov model and 1 fitted different statistical distributions to
surveillance data. We identified 11 studies that attempted to estimate the force of infection
(FOI) (the per susceptible incidence rate), the transmission rate parameter and/or the
recovery rate from infection. While only 5 calculated R0 explicitly, it was possible to calculate
14
![Page 15: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/15.jpg)
an approximate value of R0 from the data presented in 2 other studies (Table 2). Only 5
studies incorporated any form of acquired immunity as a result of successive infections. We
identified only 1 study that used serological data to make inferences on historical patterns
of transmission. All 23 studies had conclusions that could provide insight into the
achievability of the GET 2020 goals within the endemic locality analysed. However, we only
identified 1 study that fitted the model to epidemiological data collected before, during and
after an intervention had been implemented.
Table 2 approximately here
3.2 Deterministic and stochastic models
Two main research groups were identified who have developed mechanistic models of
trachoma, one group based at the University of California San Francisco’s (UCSF) Francis
Proctor Foundation and the second at Imperial College London (ICL). The mechanistic
models developed by the UCSF group have largely been stochastic, although their earlier
work was deterministic (Lee et al., 2005; Lietman et al., 1999). In these studies a single state
variable, P, representing the overall prevalence of infection in the community (or in a single
constant sized age group, commonly 0-5 year olds) is modelled (Lee et al., 2005; Ray et al.,
2007). The hazard of infection (also referred to as the force of infection (FOI), commonly
denoted by parameter lambda) for those who are uninfected is proportional to the current
prevalence of infection in the population, which is multiplied by a composite parameter
containing the contact rate and the probability of transmission upon contact with an
15
![Page 16: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/16.jpg)
infected person; this is often referred to as the transmission rate parameter beta mentioned
above. Many of the group’s studies set up a stochastic equation which permits determining
analytically the equilibrium distribution of villages with a given prevalence level, or
determining numerically the likely trajectory of the prevalence of infection for a given village
over time. These stochastic models capture and demonstrate the variability in prevalence
trajectories over time, given close baseline values and (antibiotic) treatment coverage levels
within small communities. In the context of the studies, designed by the UCSF group, these
small communities are often villages in Ethiopia. The prevalence levels of infection,
measured by the group’s field studies, can be represented by the stochastic model-
generated trajectories, with the mean prevalence calculated across several villages
appearing similar to a deterministic trajectory.
The mechanistic models developed at ICL have used primarily deterministic
frameworks (Gambhir et al., 2007; Gambhir et al., 2009; Gambhir et al., 2010), and have
focused on developing age-structured models which capture the age-specific profiles of
infection observed within communities of different endemicities. A backbone of these
models is a ladder of susceptible–infected–susceptible (SIS) infection (Figure 1a) to account
for key epidemiological features that change with age with each successive infection (Figure
1b). However, studies investigating the relative contribution of household and community
transmission lend themselves to be more naturally simulated stochastically (Figure 1c)
(Blake et al., 2009; Blake et al., 2010). We also identified a number of catalytic models
reported in the literature, the use of the catalytic model for epidemiological purposes was
initially presented by (Muench., 1934). These catalytic models explore the change in age-
16
![Page 17: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/17.jpg)
specific prevalence of infection, here the force of infection (FOI) can be estimated in a
population from age-specific disease prevalence data, or serological data. In these models
individuals experience an FOI as susceptible, they then progress into the diseased class
where they will stay until they recover or die (Figure 1 d).
3.3 Data sets used and model fitting to data
The availability and collection of routine high quality baseline (prior to intervention),
monitoring and evaluation (during the control programme), and (post-intervention)
surveillance infection data to inform mathematical modelling studies which aim to help and
advice control programmes has been extremely limited. Consequently, the data analysed by
modelling groups in the studies identified here have either been collected by the modelling
groups themselves (Lietman et al., 2011; Liu et al., 2013; Liu et al., 2014; Ray et al., 2007), or
through the careful development and management of collaborations with other research
groups who have conducted field studies (Blake et al., 2009; Gambhir et al., 2009; Gambhir
et al., 2010). Given the large range of settings around the world where trachoma is endemic
(with respect to climatic, ecological and social differences), it does not seem appropriate
that models parameterized primarily with data from Africa be used to make accurate and
informative projections about the transmission dynamics of infection occurring in other
endemic regions such as Southeast Asia or Latin America (where infection is also endemic)
(Figure 2). We identified 8 countries for which any modelling work had been conducted.
Datasets collected from Tanzania have been used in eleven studies analysed here. Datasets
from The Gambia have been used in six studies, from Ethiopia in 5 studies, Taiwan twice (no
17
![Page 18: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/18.jpg)
longer an endemic region), Egypt, Tunisia (no longer an endemic region), Malawi and
Australia and India just once each. However, the same dataset has repeatedly been used
across a number of studies. Data from Tanzania has come from 4 key regions: Morogoro,
Dodoma, Singida and Arusha. In The Gambia data has come from Jali, Berenday and Upper
Saloum. In Ethiopia, datasets have been analysed from the regions of Amhara and Gurage in
Ethiopia. Therefore, given the wide geographic distribution of trachoma (Figure 2) and the
considerable within-country heterogeneity in transmission http://www.trachomaatlas.org/), a
greater number of modelling studies should be looking to understand the dynamics of
infection at a more global scale, along with trying to gain insight into the spatial
heterogeneity in transmission that exists within the same country . However, current
progress remains limited given the scarce availability of data from regions outside of a few
key transmission areas, primarily in sub-Saharan Africa, where research (rather than
routine) studies have been conducted. Moreover, modelling is often the final activity
performed in such a partnership, and constitutes a secondary analysis of the data, following
a primary analysis that includes the evaluation of risk factors for disease, and a pre- and
post-intervention description of infection and disease prevalence. This has meant that
modelling and quantitative analysis have not been systematically used as an integrated part
of trachoma control programmes.
The datasets that have primarily been used for model calibration and model-based
analysis have come from long-term epidemiological studies in Upper Saloum district in The
Gambia (Burton et al., 2005), Rombo district in Tanzania (Solomon et al., 2003; Solomon et
al., 2004), Kongwa district in Tanzania (Burton et al., 2005), and Gurage zone of Ethiopia
18
![Page 19: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/19.jpg)
(Lietman et al., 2011; Lietman et al., 2015; Liu et al., 2013; Liu et al., 2014; Ray et al., 2007;
Ray et al., 2009). These data include longitudinal measurements of infection, clinical disease,
and associated risk factors. As most of these studies were begun at baselines where control
interventions had not yet been implemented, an ecological ‘steady-state’ (or endemic
equilibrium) constitutes an appropriate initial condition for modelling analyses.
Nevertheless, the vast majority of articles identified in this review estimate parameters
explicitly from a very small number of data sources. Studies for which infection, active
disease and disease sequelae status were all collected should be collated and shared in the
future in order to allow models to be simultaneously fitted to all of these (infection and
morbidity) data sources. In this study we only identified one article that explicitly attempted
to estimate the age-specific duration of active disease (Grassly et al., 2008) through a
hidden Markov model.
We identified one statistical study that used a substantial amount of programmatic
clinical disease data, from a wide time and geographical range collected by the International
Trachoma Initiative (ITI) (Jimenez et al., 2015). This large dataset included data on active
infection (through PCR data), prevalence data on active trachoma disease (TF or TF/TI) and
TT (Table 1). These data were used to investigate the conditions under which control
programmes have been successful. The importance of this study lies not only in its analysis
of a large dataset to draw its conclusions, but also in its presentation of these conclusions as
a ‘decision tree’, which is conceptually similar to the way in which the WHO presents its
recommendations for disease control policy.
19
![Page 20: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/20.jpg)
Mathematical models that are successfully fitted to epidemiological data (be it PCR
(infection) prevalence data or data on active disease prevalence), endeavour to estimate
just a few key epidemiological parameters. These are most commonly the transmission rate
(β), the basic reproduction number (R0), and the duration of an individual’s infectious period
(the reciprocal of the recovery rate, gamma). Whether the methodology applied to estimate
these parameters is approached through Maximum Likelihood Estimation (MLE) (Blake et
al., 2009; Gambhir et al., 2009) or a combination of MLE and Markov Chain Monte Carlo
(MCMC) methods (Lietman et al., 2011), the aim of these studies remains the same, which
is, to understand the intensity of transmission and the amount of effort required in order to
control infection.
3.4 Transmission intensity and the basic reproduction number
3.4.1 Estimating the force of infection
Intensity of transmission and hence the FOI experienced by a community can be estimated
through the careful collection of data on prevalence or through serological data in the form
of seroprevalence age profiles (Hens et al., 2012).
A study in the former category was reported by Sundaresan and Assad (1973) using
data from Taiwan collected in the 1960s. These authors applied deterministic FOI (catalytic)
modelling to provide insight into the changes in age-specific prevalence of trachoma over
time, and infer past patterns of transmission in the country. Using survey data from 1960–
1961—before the institution of a control, school-based programme with tetracycline
treatment (Assad et al., 1966)—trachoma prevalence increased with age in a fashion
compatible with the operation of a constant FOI (42/1,000 per year). However, using survey
20
![Page 21: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/21.jpg)
data from 1968–1969 revealed that the 0–9 year olds had experienced a substantial
decrease in the FOI (11/1,000 per year) (Sundaresan and Assaad, 1973).
The use of serological data to gain insight into the dynamics of transmission has
been a valuable tool across a number of infectious diseases (Badu et al., 2015; Hens et al.,
2010; Mladonicky et al., 2009; Wong et al., 2014; Williams and Keystone, 2012; Yildiz Zeyrek
et al., 2011). Analysis of serological data has also been particularly useful in post-elimination
or low transmission settings (Bousema et al., 2010; Corran et al., 2007; Oguttu et al., 2014).
However, our review only identified one article (a study conducted in Kongwa district,
Tanzania) that used a (reversible catalytic) model fitted to serological data to understand
how interventions may affect the serological status of a community (Figure 1d) (Martin et
al., 2015) (Table 1), and how these data can be used to shed light into the rate of acquisition
of infection and antibodies to infection. Reversible catalytic models (as illustrated by the
first two compartments of Figure 1 d) model seroprevalence in the population over time.
These models have two key parameters, the first is the seroconversion rate λ (Figure 1d),
the second is the seroreversion rate, denoted δ in Figure 1 d, which models the rate at
which immunity in the population is lost, hence the rate at which people serorevert. It is
therefore akin to the SIS model structure shown in Figure 1 a, however in this instance the
acquisition and loss of antibodies are modelled instead of the process of infection and
recovery. Martin et al. (Martin et al., 2015) were able to identify a change in the intensity of
transmission from serological data between 10 and 15 years prior, which was shown to
coincide with a mass antibiotic administration programme that had run between 2000 and
2002. While Martin et al. (Martin et al., 2015) were able to estimate a seroconversion rate,
21
![Page 22: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/22.jpg)
they were not able to estimate a seroreversion rate as they did not identify any individuals
in the dataset who seroreverted. As with other infectious diseases, a mounting evidence
base for serological testing, that is both sensitive and specific to trachoma infection, has the
potential to be used as a rich new data source for understanding the changing patterns of
transmission dynamics of trachoma (Martin et al., 2015).
3.4.2 Estimating the basic reproduction ratio
The most commonly-used parameter to summarize the transmission potential of an
infectious disease is the basic reproduction ratio or number, R0 (the average number of
secondary infectious cases that arise from an index case in a wholly susceptible population).
R0 was calculated (by the authors of the papers) in only 4 of the trachoma modelling studies
identified in this review, despite the fact that it could have been estimated in several of the
remaining models as the ratio of the transmission coefficient and the recovery rate from
infection (beta/gamma). This has been done by us and the results are presented in Table 2.
For the more complicated published models, involving age-structure and functional forms
for an individual’s acquired immunity and infectivity, the calculation of R0 is more involved
and can most easily be performed using the Next Generation Matrix method (Diekmann et
al., 2010). The acquisition and waning of immunity complicates the calculation of the
reproduction number, since R0’s definition refers to the beginning of an epidemic, i.e. in a
fully susceptible population (hence individuals have no prior immunity to infection). In these
circumstances, while the calculation process remains essentially the same, the parameter
22
![Page 23: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/23.jpg)
that is being calculated is correctly referred to as the effective reproduction number (Re); in
this review we identified only one study that estimated Re (Liu et al., 2014).
As evidenced by the total number of studies that explicitly calculated R0 in this
review, analysis and calculation of R0 in the trachoma modelling community has remained
very limited. However, recent work by Liu et al. (Liu et al., 2013) used a mechanistic model
to perform a statistical analysis of infection control and rebound data from several Ethiopian
villages to determine whether there was a change in the value of the reproduction number
following each round of mass antibiotic administration. Changes in the reproduction
number over time could potentially be caused by the decreasing efficacy of the antibiotic
drug over time, or the loss of short-term immunity in the population, as the stimulus of
exposure to C. trachomatis diminishes. Liu et al. (Liu et al., 2013) found no statistically
significant change in the reproduction number in the Ethiopian data set analysed and
therefore, these authors concluded that changes in drug efficacy or in immunity were not
operating at an appreciable level in the epidemiological settings evaluated in their study.
Comparing across the small number of studies that have attempted to calculate a
value of R0 or Re (or published values from which estimates could be obtained), we observed
limited heterogeneity in the range of estimates from communities classified as having the
same level of endemicity (Table 2). For example, Ray et al. (Ray et al., 2009) estimated R0 to
be 3.14 (95% CI, 2.51–3.77) for 16 hyperendemic communities in Ethiopia; in 2013 Liu et al.
(Liu et al., 2013) estimated an overall R0 of 1.39 (and yearly R0 values of 1.40, 1.38, 1.35)
across 32 communities of Tanzania; Ray et al. (Ray et al., 2007) provided values of beta
(0.047 per week) and gamma (0.017 per week) that yield a figure of 2.59 for the 16 villages
23
![Page 24: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/24.jpg)
mentioned above (which are located in the Gurage region of southern Ethiopia). We
estimated Re using the next generation matrix method (unpublished estimate) using a
previously published model and best fit parameters from another hyperendemic area in
Tanzania, and yielded a value of 3.2 (Gambhir et al., 2009), an estimate similar to that of Ray
et al. (Ray et al., 2009). Therefore, there is some consistency in the literature that values of
R0, even within high transmission settings are likely to be less than 5. These estimates have
been derived using different methodological approaches, suggesting that the estimate is not
specific to the methodology applied. However, it is worth noting that transmission of
trachoma in some regions is reported to be seasonal, and that this source of heterogeneity
has not been incorporated into any of the R0 estimates identified in this review. Therefore,
current estimates of R0 may be pertinent to the particular times the studies were conducted.
Very few studies have collected prevalence data of active trachoma and C. trachomatis
infection at multiple time points throughout consecutive years limiting the inference that
can be made on the role of seasonality on transmission.
24
![Page 25: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/25.jpg)
3.5 Acquired immunity: recovery rate and infectivity
25
![Page 26: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/26.jpg)
Immunity to trachoma with repeated infection is reported to decrease the duration of an
infected individual’s infectious period (Gambhir et al., 2009; Gambhir et al., 2010), and also
reduce the bacterial load harboured by an individual (West et al., 2005). However, in
addition to whether any acquired immunity affecting infectivity and/or recovery rates was
incorporated into the models reviewed here, we observed marked heterogeneity in the
rates of recovery from infection reported and used in the modelling studies analysed. The
models evaluated in this review that were published by the UCSF group did not, in general,
include the development of acquired immunity, with the exception of one study in which
the model assumed that older age groups developed partial immunity to (re)infection either
because of decreased susceptibility to infection, or because of clearing infection faster than
younger individuals (Lietman et al., 1999). In general, these models assume a single value
for the rate of recovery from infection, approximately equivalent to a mean duration of
infection of 6 months although there is some heterogeneity between their studies; in the
study by Lietman et al. (Lietman et al., 1999) duration of infection ranges from 9 to 17
weeks. This is a somewhat slower rate of recovery than that considered by the models of
the ICL group, which suggested that the mean duration of infection/infectivity after an
individual has experienced multiple prior infections was approximately 2.8 months—but
duration of infection for the first episode was 15 months—(Gambhir et al., 2009). Additional
work by the ICL group estimated the duration of infection to be 16–17 weeks (c. 4 months)
using frequent follow-up data from The Gambia (Grassly et al., 2008), which was
subsequently used in a number of other studies (Blake et al., 2009; Gambhir et al., 2010).
However, the assumed duration of infectivity can have substantial implications for the
26
![Page 27: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/27.jpg)
estimation of the feasibility of elimination of infection and the estimation of the
transmission rate β. For models that assume a longer duration of infection, the estimated
value of β is likely to be lower than for those assuming a shorter duration (although both
types of model would yield the same value of R0). From the perspective of trachoma control,
a lower value of β may suggest that the effort required to eliminate infection through
transmission reduction measures is less than that needed when a shorter infectious period
is assumed. However, the assumption of a longer infectious period would increase the
potential for infection to persist within a community, especially if movement of infected
individuals between villages is considered, as this could result in the re-introduction of
infection.
The models of the ICL group explicitly account for the development of immunity to
infection with an increasing number of infections experienced (Gambhir et al., 2009;
Gambhir et al., 2010). Acquired immunity to infection is modelled as a function of the
number of prior infections, i which is described by an exponentially saturating rise in the
recovery rate (or fall in the infectivity). These functions, which have three parameters, are
not the only choice of functional form accounting for acquired immunity; for example, a
logistic function or a Hill function which allow for saturating effects may be more
appropriate (Regoes et al., 2004). However, the available epidemiological data are at
present unlikely to permit distinguishing between these different functional forms. The
importance of this process and its implication for long-term transmission patterns remain
poorly understood.
27
![Page 28: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/28.jpg)
3.6 Infection and active disease, modelling of disease sequelae
Most infectious disease models are based on the idea that contact between
individuals leads to the transmission of a pathogen from one individual to another with a
given probability; they are, therefore, infection- and not disease-based. Surveillance for
trachoma control is based upon disease and therefore there is a need to link transmission
models to active trachoma and the later disease sequelae.
Due to the indirect relationship between active disease in trachoma and C.
trachomatis infection, PCR data (rather than data on active disease) have most often been
used to inform the vast majority of the models identified in this review. Trachoma models
have mostly been constructed as infection models, as demonstrated by the large number of
studies identified that have used the classic SIS model structure (Figure 1 a) (also commonly
used to model genital chlamydia infection). Notwithstanding the ubiquity of the SIS
backbone, we only identified a small number of modelling studies which explicitly included
age structure across the host population (Blake et al., 2010; Gambhir et al., 2009; Gambhir
et al., 2010; Grassly et al., 2008), and hence would have the ability to assess the impact of
infection (and disease) on an individual throughout life, this represents a critical gap in the
current literature. Whereas the infectious load and duration of active infection decreases
with age, repeated infection throughout an individual’s life results in the serious morbidities
associated with trachoma infection. This is particularly important as the first two GET 2020
goals seek to understand the impact of the infection dynamics on the long-term patterns of
active disease and the elimination of blinding trachoma.
28
![Page 29: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/29.jpg)
The successful economic evaluation of the impact of disease and disease-associated
morbidities through trachoma infection relies on accurately quantifying the burden of
disease (Lee et al., 2015). However, the development of mathematical models for trachoma
has remained primarily focused on the use of the SIS model structure, which does not lend
itself to the analysis and understanding of the long-term implications of numerous active
disease episodes with age. This represents a critical gap in the literature. However, we
identified 3 studies which used the ‘force of infection’ (FOI) modelling approach (Assaad and
Maxwell-Lyons, 1966; Parthasarathy, 1967; Sundaresan and Assaad, 1973) to describe age
profiles of total trachoma cases (simple catalytic model, with constant FOI) and of active
disease (two-stage catalytic model) in Taiwan before the institution of control measures in
the early 1960s. In these models individuals become infected and clinically diseased
according to an FOI (which was estimated to be 0.042 year–1); this translated into a
prevalence of trachoma that increased monotonically with age. By contrast, the prevalence
of active disease showed a humped pattern, reaching its maximum at 15–20 years of age
and decreasing afterwards; in this case the two-stage catalytic model estimated a rate of
acquisition and of loss of active disease of 0.050 year–1 and 0.072 year–1, respectively.
Through these models it was possible to gain insight into the dynamic nature of the burden
of active disease with age prior to the implementation of an intervention.
Models which are intended to be used in close conjunction with control programmes
should, therefore, include active disease and disease sequelae components if they are to be
truly useful as tools to support decision-making in the context of the GET 2020 goals. While
infection and active disease are the proximate causes of scarring and the resulting sequelae,
29
![Page 30: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/30.jpg)
the timescales associated with the development of scarring, trichiasis, corneal opacity and
blindness are different from one another, and should be accounted for when forecasting for
control programmes aiming to achieve the GET 2020 goals (Gambhir et al., 2009; Gambhir et
al., 2010; Gambhir et al., 2015; Shattock et al., 2015). To date, very few models have done
so. Gambhir et al. (Gambhir et al., 2010) illustrated a first attempt at adding active disease
to a simple age-structured infection-based model (Figure 3). The model used a series of
partial differential equations (PDEs) to explore the impact of treatment on active disease,
having previously fitted the model to disease sequelae data. The authors identified that in
their model the prevalence of active disease was always higher than the prevalence of
infection, and that following mass antibiotic treatment the decline in prevalence of active
disease lagged behind the immediate drop in infection prevalence (Gambhir et al., 2010).
The mechanistic model of Shattock et al. (Shattock et al., 2015) took this concept further by
developing a stochastic agent-based model to create a complex representation of the
natural history of infection (Figure 1e) and disease for every individual within a defined
community, and by then fitting this model to active disease data from multiple Australian
Aboriginal communities. Therefore, models which can be fitted to active disease data are
needed if we are to understand whether the GET 2020 goals are being achieved. Modelling
analysis of additional active disease datasets is a pressing need in order to disentangle the
complex relationship between the dynamics of infection across different transmission
settings and the prevalence of active disease/disease sequelae, and to help the public health
community to understand the feasibility of achieving the GET 2020 goals.
30
![Page 31: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/31.jpg)
Here we suggest 5 crucial pieces of data that could be collected and that would help
to ensure more accurate parameterization of subsequent models of trachoma infection and
disease: 1) collection of data that help to inform understanding of the age-specific duration
of active disease (TI/TF) following infection with C. trachomatis; 2) data on the duration of
overlap between active infection and active disease by age; 3) data on the age-specific
duration of infection alone; 4) data on how the bacterial load harboured with age varies
across different transmission settings; 5) data on how immunity to trachoma wanes in the
absence of frequent exposure.
3.7 Simulation of interventions, forecasting infection and disease, and
analysis of their cost effectiveness
Mechanistic modelling in the context of control interventions was explored in 13 of the 23
articles identified in this review, all of which examined the A (antibiotics) component of the
SAFE strategy. Lietman et al. (Lietman et al., 1999) showed that annual MDA was not
sufficient to eliminate infectious trachoma from hyperendemic communities and that the
frequency of treatment required depends on the initial doubling time of a re-introduced
epidemic; the latter does not decrease as prevalence of infection decreases during a
treatment programme. In order for the doubling time of a re-introduced epidemic to be
reduced, changes in the community must occur, such as the implementation of hygiene
measures (F and E components), that can help to reduce the overall level of transmission.
This study also highlighted the need for biannual mass antibiotic treatment in hyperendemic
communities. Gambhir et al. (Gambhir et al., 2010) demonstrated a similar finding. Ray et al.
31
![Page 32: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/32.jpg)
(Ray et al., 2007) investigated the impact of annual and biannual rounds of mass antibiotic
treatment and showed that 5 years of biannual treatment could be sufficient to eliminate
infection in 95% of the hyperendemic Ethiopian villages simulated in the study. In a follow
up paper, Ray et al. (Ray et al., 2009) used the same model—fitted to data from three
different transmission settings: Ethiopian (hyperendemic), Tanzanian (mesoendemic), and
Gambian (hypoendemic)—to demonstrate that an adjusted mass treatment strategy, in
which mass antibiotic administration was withheld from specific communities whose
infection prevalence levels fell below 5% (in children aged 1 – 5 years), had a comparable
impact to that of the WHO-endorsed strategy, and saved a considerable number of
antibiotic doses. However, the logistics and costs associated with testing infection at the
community level instead of the district level (outside of specific trial, research settings) can
be both expensive and challenging, making the practical implications of these findings
difficult to implement on a much wider, routine scale. To overcome these difficulties, Ray et
al. (2009) suggested that pooled PCR sampling for entire communities could be performed,
leading to substantial cost savings. However, a specific cost breakdown for the
implementation of such an approach would be useful for understanding whether it would
be feasible from a programmatic perspective and for modelling its cost effectiveness.
An important factor which can impact the effectiveness of antibiotic treatment is the
level of coverage that can be achieved within a community. If the level of coverage is poor
few people will get treated, therefore the impact on transmission going to be less than if the
level of coverage achieved was high. Within a trial setting, where much of the
epidemiological data about the success of MDA is collected the level of coverage achieved
32
![Page 33: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/33.jpg)
can be extremely high (>90%) (Lakew et al., 2009). However, outside of a trial setting the
level of coverage achieved is likely to be much lower, suggesting that the findings identified
in trials may not be fully reflective of the outcomes observed during routine rounds of MDA.
We did not identify any studies that assessed the sensitivity of their findings to the level of
MDA coverage achieved. This represents an important gap in the modelling literature that
needs to be explored more thoroughly in future studies if we are to make realistic forecasts
about the prospective impact of MDA.
One of the first questions public health workers ask disease modellers relates to the
forecasting of future disease levels. However, the majority of trachoma modelling work has
been conducted to improve the understanding of trachoma epidemiology, and not
specifically to forecast disease levels. Nevertheless, a well-parameterised mechanistic or
statistical model is often capable of being run forward in time, forecasting disease and
uncertainty bounds. Indeed, we identified 3 published studies that have been formulated in
order to make forecasts in the context of control interventions; these were the studies by
Gambhir et al. (Gambhir et al., 2015); Ray et al. (Ray et al., 2009) and Shattock et al.
(Shattock et al., 2015). Each of these 3 studies forecasts the prevalence of infection and/or
disease following various interventions (see section 3.7 Simulation of interventions and
analysis of their cost effectiveness). Shattock et al. (Shattock et al., 2015) fitted their
individual-based model to clinical disease prevalence surveillance data from 67 Australian
Aboriginal communities, and then investigated the possible future impact of different
treatment combinations on the long-term prevalence of trachoma infection. The authors
identified that under the current intervention strategies, control of infection within these
33
![Page 34: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/34.jpg)
communities would not be achieved by 2020. Ray et al. (Ray et al., 2009) used a stochastic
differential equation model to evaluate an antibiotic-sparing strategy that halted treatment
when community infection prevalence dropped below 5%, following the WHO
recommended 3 annual rounds of antibiotic administration. This was shown to be effective
in meso and hypoendemic communities (supported with evidence from trial data from
Tanzania (Solomon et al., 2008)). However, the feasibility of lowering the prevalence of
infection in hyperendemic communities was much lower. Gambhir et al. (Gambhir et al.,
2015) performed forecasts for a variety of treatment interventions within the SAFE strategy,
individually and in concert with one another, with the specific aim of determining whether
the GET 2020 goals could be met within the next 5, 10 or 20 years (as of 2015). These
studies highlight how modelling can be used to provide information and insightful
projections of public health importance in the context of the GET 2020 goals.
In a modified approach to the classic SIS models previously discussed, Blake et al.
(Blake et al., 2010) developed a model that incorporated both household and community
transmission which was used to explore cost-effective means of distributing antibiotics,
whereby treatment was tied to clinical disease diagnosis at the household level. The
modelling analysis by Blake et al. (Blake et al., 2010) is the only study that features a health
economics component, an area that is clearly in need of attention as rationales for budgets
to address blinding disease elimination are made in the future. Research groups working on
several of the other NTDs have integrated health economics evaluations of control
interventions, largely pharmaceutical, into mechanistic modelling frameworks resulting in
studies that can immediately feed into policy (e.g. onchocerciasis (Turner et al., 2014;
34
![Page 35: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/35.jpg)
Turner et al., 2015), lymphatic filariasis (Goldman et al., 2007), soil-transmitted
helminthiases (Turner et al., 2015), and schistosomiasis (Brooker et al., 2008; Guyatt et al.,
1994)). For a review outlining the use of mathematical modelling in economic evaluations of
interventions against the NTDs included in the London Declaration see Turner et al. (Turner
et al., 2014), and for a review of economic evaluation approaches and studies for the NTDs
see Lee et al. (Lee et al, 2015). Such approaches should also be taken forward in trachoma.
In principle, the results of the statistical modelling paper by Jimenez et al. (Jimenez
et al., 2015) could be used to understand the effectiveness of control interventions for a
variety of trachoma-endemic settings, whose previous intervention history varies from none
to the implementation of the full WHO recommended SAFE strategy over time. The
drawbacks of their statistical approach are that it only allows for a small number of location-
specific covariates to be defined for each community, and that while the results are often
intuitive (Table 1), the model is expressed as a set of correlations, without explicitly
accounting for the biological and epidemiological mechanisms underlying their community
classifications.
The majority of the modelling work identified in this review has sought to
understand and project the impact of antibiotic administration, whether in the modality of
mass treatment or through assessing the impact of targeted treatment strategies to only
those who present with active infection, or specific age groups (Blake et al., 2009; Gambhir
et al., 2010; Lakew et al., 2009; Lietman et al., 2011; Liu et al., 2014; Ray et al., 2007; Ray et
al., 2009).
35
![Page 36: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/36.jpg)
However, antibiotics represent only one arm of the SAFE strategy. As such, there is a
pressing need for research to determine how best to incorporate non-pharmaceutical
intervention (NPI) into mathematical models of trachoma. In this review we identified only
one study that explicitly assessed the impact of both MDA and NPIs within communities
(Shattock et al., 2015). Indeed, the third GET 2020 goal states that the F and E components
of the SAFE strategy should be enhanced. However, accurate quantification of the impact of
these interventions in helping to reduce trachoma transmission and incidence remains
limited and poorly understood, making it difficult to model accurately their potential impact.
Ejere et al. (Ejere et al., 2012) conducted a review of face washing promotion for the
prevention of active trachoma and concluded that although there is some evidence that
face washing combined with topical tetracycline can be effective in reducing severe
trachoma and increasing the prevalence of clean faces, face washing alone or in
combination with topical tetracycline did not significantly reduce active trachoma. Rabiu et
al. (Rabiu et al., 2012) performed a review of environmental interventions for preventing
active trachoma and concluded that insecticides against the mechanical fly carriers may be
effective in reducing trachoma but provision of latrines as a fly control measure did not
significantly reduced trachoma transmission. Generally there is a dearth of data to
determine the effectiveness of all aspects of environmental sanitation in the control of
trachoma.
Currently, the best available resource for understanding the possible impact of NPIs
on trachoma and incorporating these NPIs into trachoma models is the meta-analysis
presented by Stocks et al. (Stocks et al., 2014). This study collated epidemiological evidence
36
![Page 37: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/37.jpg)
for each of the components of water, sanitation, and hygiene (WASH) upon active trachoma
and infection with C. trachomatis. The Odds Ratios (OR) for each WASH component
provided by this article could be translated into modified parameter values to be
incorporated into appropriately defined mechanistic models; many interventions would
essentially alter the reproduction number by an amount directly calculable from the
relevant OR (generally by reducing the value of the reproduction number in a linear
manner).
4. Discussion
4.1 Using modelling to determine the feasibility of the GET 2020 goals
Studies focusing on the epidemiological modelling of trachoma have been steadily published
over the past 40 years. Collectively, they provide a considerable amount of accumulated
insight pertaining to the GET 2020 goals. The World Health Assembly passed resolution
51.11 in the year 1998 outlining the intention to end trachoma as a public health problem.
These goals included the global elimination of blinding trachoma by the year 2020
(GET 2020) (WHO, 2003). However, there has been limited involvement of modellers to help
to understand if and whether these goals are being/will be achieved within the proposed
timeframes.
The GET 2020 annual meeting is organized by the WHO and has taken place since
1997. Prior to the main meeting, a Trachoma Scientific Informal Workshop (TSIW) is held in
which the latest scientific results pertaining to GET 2020 are shared. The results shared by
this group comprise, for example, data from intervention trials of azithromycin antibiotic
distribution in a variety of trachoma-endemic settings (primarily in The Gambia, Tanzania,
37
![Page 38: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/38.jpg)
Ethopia, and more recently Niger (Cromwell., 2014) and Guinea Bissau (Last et al., 2014)),
and rates of success, reversion, and risk factors associated with surgery for trachomatous
trichiasis, among others (WHO, 2015). However, it is only recently that modelling is being
viewed as a tool to help inform policy and infection control strategies (see e.g. WHO 2015).
The vast majority of modelling studies identified in this review have not explicitly
referred to the GET 2020 goals. However, they all contain findings that pertain to the goals
(Table 1). Therefore, despite the high relevance of many of these articles to the questions
posed and goals proposed by GET 2020, modelling groups and their resulting studies have
largely failed to be integrated into the discussions leading to the development of control
programme policy. The body of model-based work that has been produced for trachoma is
not significantly different to that which has been published for the other NTDs. However,
the failure to integrate modelling into understanding the feasibility of achieving the
GET 2020 goals across a range of endemic settings may be due to the fact that trachoma has
previously been perceived as controllable and that the current intervention strategies are
working (House et al., 2007). As such, it may be felt that modelling studies tend only to
confirm intuition or are only taken into account as part of the decision-support apparatus of
public health workers when it is deemed that current interventions are not working.
As described, a wide range of models have been developed to explore a multitude of
hypotheses. However, it could be suggested that many studies seek not only to fit models to
data to support or refute a hypothesis, but also aim to develop and apply innovative
methodologies. This has potentially led to the publication of a wide range of studies that
from a programmatic perspective may all appear quite different, leading to confusion in the
38
![Page 39: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/39.jpg)
trachoma community. However, from the modelling perspective they are still seeking to
understand the same underlying processes. This in itself may be leading to a barrier
between modellers and those who work in the field.
4.2 Conclusions
Having assessed the modelling literature, we outline three key gaps that future modelling
work should address in order to help to understand if and where the GET 2020 goals can be
achieved. First, we would suggest that modellers work to develop mathematical models that
explicitly include an active disease component which accounts for the fact that even if an
individual does not have an active, PCR-detectable infection, they may still have active
disease. Mathematical models that are fitted to infection and active disease data and
validated with baseline and follow-up, longitudinal active disease data will be valuable for
forecasting purposes. Second, we would encourage that a better quantitative understanding
of the impact of the F and E interventions be developed; this will allow NPIs to be included
more accurately into mathematical models, facilitating the long-term impact of these
interventions to be explored more thoroughly. Finally, we would suggest that health
economic analysis be integrated into scenario analysis, as not only is the understanding of
the feasibility of the GET 2020 goals vital, but so is developing an understanding of the
financial implications of doing so.
Competing interests MG, MGB and IMB declare in accordance with the ICMJE conflict of
interest form that 7 of the articles identified and evaluated in this review articles have been
authored or co-authored by them. AP declares no competing interests.
39
![Page 40: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/40.jpg)
Authors’ contributions
AP and MG designed the study and prepared the first draft. IB and MGB critically review the
draft and contributed with intellectual input. All authors contributed to the writing of the
final version of the manuscript and approved the final, submitted version.
Acknowledgments
AP, MGB and MG gratefully acknowledge funding of the NTD Modelling Consortium by the
Bill and Melinda Gates Foundation in partnership with the Task Force for Global Health. The
views, opinions, assumptions or any other information set out in this article are solely those
of the authors. MG also acknowledges funding from the Australian NHMRC and Monash
University.
References
Abdou, A., Nassirou, B., Kadri, B., Moussa, F., Munoz, B.E., Opong, E., West, S.K., 2007. Prevalence and risk factors for trachoma and ocular Chlamydia trachomatis infection in Niger. Br. J. Ophthalmol. 91(1), 13–17.
Assaad, F.A. and Maxwell-Lyons, F., 1966. The use of catalytic models as tools for elucidating the clinical and epidemiological features of trachoma. Bull. World Health Organ. 34(3), 341–355.
Assaad, F.A., Sundaresan, T.K., Yang, C.Y., Yeh, L.J., 1971. Clinical evaluation of the Taiwan trachoma control programme. Bull. World Health Organ. 45(4), 491–509.
Badu, K., Gyan, B., Appawu, M., Mensah, D., Dodoo, D., Yan, G., Drakeley, C., Zhou, G., Owusu-Dabo, E., Koram, K.A., 2015. Serological evidence of vector and parasite exposure in Southern Ghana: the dynamics of malaria transmission intensity. Parasit Vectors 8, 251.
Bailey, R., Duong, T., Carpenter, R., Whittle, H., Mabey, D., 1999. The duration of human ocular Chlamydia trachomatis infection is age dependent. Epidemiol. Infect. 123(3), 479–486.
Basáñez, M.G., Anderson, R.M., 2015. Preface. Mathematical Models for Neglected Tropical Diseases: Essential Tools for Control and Elimination. Part A. Adv. Parasitol. 87, xiii–xviii.
Blake, I.M., Burton, M.J., Bailey, R.L., Solomon, A.W., West, S., Muñoz, B., Holland, M.J., Mabey, D.C., Gambhir, M., Basáñez, M.G., Grassly, N.C., 2009. Estimating
40
![Page 41: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/41.jpg)
household and community transmission of ocular Chlamydia trachomatis. PLoS Negl. Trop. Dis. 3(3), e401.
Blake, I.M., Burton, M.J., Solomon, A.W., West, S.K., Basáñez, M.G., Gambhir, M., Bailey, R.L., Mabey, D.C., Grassly, N.C., 2010. Targeting antibiotics to households for trachoma control. PLoS Negl. Trop. Dis. 4(11), e862.
Bousema, T., Youssef, R.M., Cook, J., Cox, J., Alegana, V.A., Amran, J., Noor, A.M., Snow, R.W., Drakeley, C., 2010. Serologic markers for detecting malaria in areas of low endemicity, Somalia, 2008. Emerg. Infect. Dis. 16(3), 392–399.
Brooker, S., Kabatereine, N.B., Fleming, F., Devlin, N., 2008. Cost and cost-effectiveness of nationwide school-based helminth control in Uganda: intra-country variation and effects of scaling-up. Health Policy Plan. 23(1), 24–35.
Burton, M.J., Holland, M.J., Makalo, P., Aryee, E.A., Alexander, N.D., Sillah, A., Faal, H., West, S.K., Foster, A., Johnson, G.J., et al 2005. Re-emergence of Chlamydia trachomatis infection after mass antibiotic treatment of a trachoma-endemic Gambian community: a longitudinal study. Lancet. 365(9467), 1321–1328.
Burton, M.J., and Mabey, D.C.W., 2009. The Global Burden of Trachoma: A Review. PLoS Negl Trop Dis. 3(10), e460.
Corran, P., Coleman, P., Riley, E., Drakeley, C., 2007. Serology: a robust indicator of malaria transmission intensity? Trends Parasitol. 23(12), 575–582.
Courtright, P. and West, S. K., 2004. Contribution of Sex-linked Biology and Gender Roles to Disparities with Trachoma. Emerg Inf Dis. 10(11), 2012–2016.
Cromwell, E.A., Amza, A., Kadri, B., Beidou, N., King, J.D., Sankara, D., Mosher, A.W., Hassan, S., Kane, S., Emerson, P.M., 2014. Trachoma prevalence in Niger: results of 31 disrtict surveys. Trans R Soc Trop Med Hyg. 108(1), 42-48.
Diekmann, O., Heesterbeek, J.A., Roberts, M.G., 2010. The construction of next-generation matrices for compartmental epidemic models. J R Soc Interface. 7(47), 873–885.
Emerson, P. M., Lindsay, S. W., Alexander, N., Bah, M., Dibba, S. M., Faal, H. B., Lowe, K. O., McAdam, K. P., Ratcliffe, A. A., Walraven, G. E., Bailey, R. L., 2004. Role of flies and provision of latrines in trachoma control: cluster-randomised controlled trial. Lancet. 363(9415), 1093–1098.
Ejere, H.O., Alhassan, M.B., Rabiu, M., 2012. Face washing promotion for preventing active trachoma. Cochrane Database Syst Rev 4: CD003659
Evans, T. G. and Ranson, M. K., 1995. The global burden of trachomatous visual impairment: II. Assessing burden. Int Ophthalmol. 19(5), 271–280.
Frick, K.D., Basilion, E.V., Hanson, C.L., Colchero, M.A., 2003. Estimating the burden and economic impact of trachomatous visual loss. Ophthalmic Epidemiol. 10(2), 121–132.
Gambhir, M., Basanez, M.G., Blake, I.M., Grassly, N.C., 2010. Modelling trachoma for control programmes. Adv Exp Med Biol. 673, 141–156.
Gambhir, M., Basanez, M.G., Burton, M.J., Solomon, A.W., Bailey, R.L., Holland, M.J., Blake, I.M., Donnelly, C.A., Jabr, I., Mabey, D.C., 2009. The development of an age-structured model for trachoma transmission dynamics, pathogenesis and control. PLoS Negl Trop Dis. 3(6), e462.
Gambhir, M., Basanez, M.G., Turner, F., Kumaresan, J., Grassly, N.C., 2007. Trachoma: transmission, infection, and control. Lancet Infect Dis. 7(6): 420–427.
Gambhir, M., Grassly, N.C., Burton, M.J., Solomon, A.W., Taylor, H.R., Blake, I.M., Basanez, M.G., 2015 in press. Estimating the Future Impact of a Multi-Pronged
41
![Page 42: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/42.jpg)
Intervention Strategy on Ocular Disease Sequelae Caused by Trachoma: A Modeling Study. Ophthalmic Epidemiol.
Goldman, A.S., Guisinger, V.H., Aikins, M., Amarillo, M.L.E., Belizario, V.Y., Garshong, B., Gyapong, J., Kabali, C., Kamal, H.A., Kanjilal, S., Kyelem, D., Lizardo, J., Malecela, M., Mubyazi, G., Nitièma, P. A., Ramzy, R. M. R., Streit, T. G., Wallace, A., Brady, A. M., Rheingans, R., Ottesen, E. A., Haddix, A. C., 2007. National Mass Drug Administration Costs for Lymphatic Filariasis Elimination. PLoS Negl Trop Dis. 1(1), e67.
Grassly, N.C., Ward, M.E., Ferris, S., Mabey, D.C., Bailey, R.L., 2008. The Natural History of Trachoma Infection and Disease in a Gambian Cohort with Frequent Follow-Up. PLoS Negl Trop Dis. 2(12), e341.
Guyatt, H., Evans, D., Lengeler, C., Tanner, M., 1994. Controlling schistosomiasis: the cost-effectiveness of alternative delivery strategies. Health Policy Plan. 9(4), 385–395.
Harding-Esch, E.M., Edwards, T., Sillah, A., Sarr, I., Roberts, C.H., Snell, P., Aryee, E., Molina, S., Holland, M.J., Mabey, D.C.W., Bailey, R.L., 2009. Active Trachoma and Ocular Chlamydia trachomatis Infection in Two Gambian Regions: On Course for Elimination by 2020? PLoS Negl Trop Dis. 3(12), e573.
Hens, N., Aerts, M., Faes, C., Shkedy, Z., Lejeune, O., Van Damme, P., Beutels, P., 2010. Seventy-five years of estimating the force of infection from current status data. Epidemiol. Infect. 138(6), 802–812.
Hens, N., Shkedy, Z., Aerts, M. Faes, C., Van Damme, P., Beutels, P., 2012. Modeling Infectious Disease Parameters Based on Serological and Social Contact Data. A Modern Statistical Perspective. Springer, London.
Hotez, P.J., Alvarado, M., Basáñez, M-G., Bolliger, I., Bourne, R., Boussinesq, M., Brooker, S.J., Brown, A.S., Buckle, G., Budke, C.M., Carabin, H., Coffeng, L.E.,Fèvre, E.M., Fürst, T., Halasa, Y.A., Jasrasaria, R., Johns, N.E., Keiser, J., King, C.H., Lozano, R., Murdoch, M.E., O'Hanlon, S., Pion, S.D., Pullan, R.L., Ramaiah, K.D., Roberts, T., Shepard, D.S., Smith, J.L., Stolk, W.A., Undurraga, E.A., Utzinger, J., Wang, M., Murray, C.J., Naghavi, M., 2014. The Global Burden of Disease Study 2010: Interpretation and Implications for the Neglected Tropical Diseases. PLoS Negl Trop Dis. 8(7), e2865.
Hotez, P.J., Feck, A., Savioli, L., Molyneux, D.H., 2009. Rescuing the bottom billion through control of neglected tropical diseases. Lancet. 373(9674), 1570–1575.
House, J., Gaynor, B., Taylor, H., Lietman, T.M., 2007. The real challenge: can we discover why trachoma is disappearing before it's gone? Int Ophthalmol Clin. 47(3), 63–76.
Jimenez, V., Gelderblom, H.C., Mann Flueckiger, R., Emerson, P.M., Haddad, D., 2015. Mass drug administration for trachoma: how long is not long enough? PLoS Negl Trop Dis. 9(3), e0003610.
Koukounari, A., Moustaki, I., Grassly, N.C., Blake, I.M., Basáñez, M.G., Gambhir, M., Mabey, D.C., Bailey, R.L., Burton, M.J., Solomon, A.W., Donnelly, C.A., 2013. Using a nonparametric multilevel latent Markov model to evaluate diagnostics for trachoma. Am J Epidemiol. 177(9), 913–922.
Lakew, T., House, J., Hong, K.C., Yi, E., Alemayehu, W., Melese, M., Zhou, Z., Ray, K., Chin, S., Romero, E., et al: 2009. Reduction and return of infectious trachoma in severely affected communities in Ethiopia. PLoS Negl Trop Dis. 3(2), e376.
42
![Page 43: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/43.jpg)
Last, A.R., Burr, S.E., Weiss, H.A., Harding-Esch, E.M., Cassama, E., Nabicassa, M., Mabey, D.C., Holland, M.J., Bailey, R.L., 2014. Risk factors for active trachoma and ocular Chlamydia trachomatis infection in treatment-naïve trachoma-hyperendemic communities of the Bijagós Archipelago, Guinea Bissau. PLoS Negl Trop Dis. 8(6),e2900.
Lee, B.Y., Bartsch, S.M., Gorham, K.M., 2015. Chapter Eight - Economic and Financial Evaluation of Neglected Tropical Diseases. Adv Parasitol. 87, 329–417.
Lee, D.C., Chidambaram, J.D., Porco, T.C., Lietman, T.M., 2005. Seasonal effects in the elimination of trachoma. Am J Trop Med Hyg. 72(4), 468-470.
Lietman, T.M., Porco, T.C., Dawson, C., Blower, S., 1999. Global elimination of trachoma: how frequently should we administer mass chemotherapy? Nat Med. 5(5), 572-576.
Lietman, T.M., Gebre, T., Abdou, A., Alemayehu, W., Emerson, P., Blumberg, S., Keenan, J.D., Porco, T.C., 2015. The distribution of the prevalence of ocular chlamydial infection in communities where trachoma is disappearing. Epidemics. 11, 85-91.
Lietman, T.M., Gebre, T., Ayele, B., Ray, K.J., Maher, M.C., See, C.W., Emerson, P.M., Porco, T.C., 2011. The epidemiological dynamics of infectious trachoma may facilitate elimination. Epidemics. 3(2), 119-124.
Liu, F.C., Porco, T.C., Ray, K.J., Bailey, R.L., Mkocha, H., Munoz, B., Quinn, T.C., Lietman, T.M., West, S.K., 2013. Assessment of transmission in trachoma programs over time suggests no short-term loss of immunity. PLoS Negl Trop Dis. 7(7), e2303.
Liu, F.C., Porco, T.C., Mkocha, H.A., Munoz, B., Ray, K.J., Bailey, R.L., Lietman, T.M., West, S.K., 2014. The efficacy of oral azithromycin in clearing ocular chlamydia: Mathematical modeling from a community-randomized trachoma trial. Epidemics. 6, 10-17.
London Declaration on Neglected Tropical Diseases, 2012. Uniting to combat neglected tropical diseases. Ending the neglect and reaching 2020 goals. Available: http://unitingtocombatntds.org/resource/london-declaration.
Martin, D.L., Bid, R., Sandi, F., Goodhew, E.B., Massae, P.A., Lasway, A., Philippin, H., Makupa, W., Molina, S., Holland, M.J., Mabey, D.C., Drakeley, C., Lammie, P.J., Solomon, A.W., 2015. Serology for trachoma surveillance after cessation of mass drug administration. PLoS Negl Trop Dis. 9(2), e0003555.
Mladonicky, J.M., King, J.D., Liang, J.L., Chambers, E., Pa'au, M., Schmaedick, M.A., Burkot, T.R., Bradley, M., Lammie, P.J., 2009. Assessing transmission of lymphatic filariasis using parasitologic, serologic, and entomologic tools after mass drug administration in American Samoa. Am J Trop Med Hyg. 80(5), 769-773.
Murray, C.J., Vos, T., Lozano, R., Naghavi, M., Flaxman, A.D., Michaud, C., Ezzati, M., Shibuya, K., Salomon, J.A., Abdalla, S., Aboyans, V., Abraham. J., Ackerman, I., Aggarwal, R., Ahn, S.Y., Ali, M.K., Alvarado, M., Anderson, H.R., Anderson, L.M., Andrews, K.G., Atkinson, C., Baddour, L.M., Bahalim, A.N., Barker-Collo, S., Barrero, L.H., Bartels, D.H., Basáñez, M.G., Baxter, A., Bell, M.L., Benjamin, E.J., Bennett, D., Bernabé, E., Bhalla, K., Bhandari, B., Bikbov, B., Bin Abdulhak, A., Birbeck, G., Black, J.A., Blencowe, H., Blore, J.D., Blyth, F., Bolliger, I., Bonaventure, A., Boufous, S., Bourne, R., Boussinesq, M., Braithwaite, T., Brayne, C., Bridgett, L., Brooker, S., Brooks, P., Brugha, T.S., Bryan-Hancock, C., Bucello, C., Buchbinder, R., Buckle, G., Budke, C.M., Burch, M., Burney, P., Burstein, R., Calabria, B., Campbell, B., Canter, C.E., Carabin, H., Carapetis, J., Carmona, L.,
43
![Page 44: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/44.jpg)
Cella, C., Charlson, F., Chen, H., Cheng, A.T., Chou, D., Chugh, S.S., Coffeng, L.E., Colan, S.D., Colquhoun, S., Colson, K.E., Condon, J., Connor, M.D., Cooper, L.T., Corriere, M., Cortinovis, M., de Vaccaro, K.C., Couser, W., Cowie, B.C., Criqui, M.H., Cross, M., Dabhadkar, K.C., Dahiya, M., Dahodwala, N., Damsere-Derry, J., Danaei, G., Davis, A., De Leo, D., Degenhardt, L., Dellavalle, R., Delossantos, A., Denenberg, J., Derrett, S., Des Jarlais, D.C., Dharmaratne, S.D., Dherani, M., Diaz-Torne, C., Dolk, H., Dorsey, E.R., Driscoll, T., Duber, H., Ebel, B., Edmond, K., Elbaz, A., Ali, S.E., Erskine, H., Erwin, P.J., Espindola, P., Ewoigbokhan, S.E., Farzadfar, F., Feigin, V., Felson, D.T., Ferrari, A., Ferri, C.P., Fèvre, E.M., Finucane, M.M., Flaxman, S., Flood, L., Foreman, K., Forouzanfar, M.H., Fowkes, F.G., Fransen, M., Freeman, M.K., Gabbe, B.J., Gabriel, S.E., Gakidou, E., Ganatra, H.A., Garcia, B., Gaspari, F., Gillum, R.F., Gmel, G., Gonzalez-Medina, D., Gosselin, R., Grainger, R., Grant, B., Groeger, J., Guillemin, F., Gunnell, D., Gupta, R., Haagsma, J., Hagan, H., Halasa, Y.A., Hall, W., Haring, D., Haro, J.M., Harrison, J.E., Havmoeller, R., Hay, R.J., Higashi, H., Hill, C., Hoen, B., Hoffman, H., Hotez, P.J., Hoy, D., Huang, J.J., Ibeanusi, S.E., Jacobsen, K.H., James, S.L., Jarvis, D., Jasrasaria, R., Jayaraman, S., Johns, N., Jonas, J.B., Karthikeyan, G., Kassebaum, N., Kawakami, N., Keren, A., Khoo, J.P., King, C.H., Knowlton, L.M., Kobusingye, O., Koranteng, A., Krishnamurthi, R., Laden, F., Lalloo, R., Laslett, L.L., Lathlean, T., Leasher, J.L., Lee, Y.Y., Leigh, J., Levinson, D., Lim, S.S., Limb, E., Lin, J.K., Lipnick, M., Lipshultz, S.E., Liu, W., Loane, M., Ohno, S.L., Lyons, R., Mabweijano, J., MacIntyre, M.F., Malekzadeh, R., Mallinger, L., Manivannan, S., Marcenes, W., March, L., Margolis, D.J., Marks, G.B., Marks, R., Matsumori, A., Matzopoulos, R., Mayosi, B.M., McAnulty, J.H., McDermott, M.M., McGill, N., McGrath, J., Medina-Mora, M.E., Meltzer, M., Mensah, G.A., Merriman, T.R., Meyer, A.C., Miglioli, V., Miller, M., Miller, T.R., Mitchell, P.B., Mock, C., Mocumbi, A.O., Moffitt, T.E., Mokdad, A.A., Monasta, L., Montico, M., Moradi-Lakeh, M., Moran, A., Morawska, L., Mori, R., Murdoch, M.E., Mwaniki, M.K., Naidoo, K., Nair, M.N., Naldi, L., Narayan, K.M., Nelson, P.K., Nelson, R.G., Nevitt, M.C., Newton, C.R., Nolte, S., Norman, P., Norman, R., O'Donnell, M., O'Hanlon, S., Olives, C., Omer, S.B., Ortblad, K., Osborne, R., Ozgediz, D., Page, A., Pahari, B., Pandian, J.D., Rivero, A.P., Patten, S.B., Pearce, N., Padilla, R.P., Perez-Ruiz, F., Perico, N., Pesudovs, K., Phillips, D., Phillips, M.R., Pierce, K., Pion, S., Polanczyk, G.V., Polinder, S., Pope CA 3rd., Popova, S., Porrini, E., Pourmalek, F., Prince, M., Pullan, R.L., Ramaiah, K.D., Ranganathan, D., Razavi, H., Regan, M., Rehm, J.T., Rein, D.B., Remuzzi, G., Richardson, K., Rivara, F.P., Roberts, T., Robinson, C., De Leòn, F.R., Ronfani, L., Room, R., Rosenfeld, L.C., Rushton, L., Sacco, R.L., Saha, S., Sampson, U., Sanchez-Riera, L., Sanman, E., Schwebel, D.C., Scott, J.G., Segui-Gomez, M., Shahraz, S., Shepard, D.S., Shin, H., Shivakoti, R., Singh, D., Singh, G.M., Singh, J.A., Singleton, J., Sleet, D.A., Sliwa, K., Smith, E., Smith, J.L., Stapelberg, N.J., Steer, A., Steiner, T., Stolk, W.A., Stovner, L.J., Sudfeld, C., Syed, S., Tamburlini, G., Tavakkoli, M., Taylor, H.R., Taylor, J.A., Taylor, W.J., Thomas, B., Thomson, W.M., Thurston, G.D., Tleyjeh, I.M., Tonelli, M., Towbin, J.A., Truelsen, T., Tsilimbaris, M.K., Ubeda, C., Undurraga, E.A., van der Werf, M.J., van Os. J., Vavilala, M.S., Venketasubramanian, N., Wang, M., Wang, W., Watt, K., Weatherall, D.J., Weinstock, M.A., Weintraub, R., Weisskopf, M.G., Weissman, M.M., White, R.A.,
44
![Page 45: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/45.jpg)
Whiteford, H., Wiebe, N., Wiersma, S.T, Wilkinson, J.D., Williams, H.C., Williams, S.R., Witt, E., Wolfe, F., Woolf, A.D., Wulf, S., Yeh, P.H., Zaidi, A.K., Zheng, Z.J., Zonies, D., Lopez, A.D., Al Mazroa, M.A., Memish, Z.A., 2012. Disability-adjusted life years (DALYs) for 291 diseases and injuries in 21 regions, 1990-2010: a systematic analysis for the Global Burden of Disease Study 2010. Lancet. 380(9859), 2197–2223.
Muench, H., 1934. Derivation of rates from summation data by the catalytic curve. JASA. 29(185), 25–38.
Naidoo, K., Gichuhi, S., Basáñez, M.G., Flaxman, S.R., Jonas, J.B., Keeffe, J., Leasher, J.L., Pesudovs, K., Price, H., Smith, J.L., Turner, H.C., White, R.A., Wong, T.Y., Resnikoff, S., Taylor, H.R., Bourne, R.R.; Vision Loss Expert Group of the Global Burden of Disease Study., 2014. Prevalence and causes of vision loss in sub-Saharan Africa: 1990-2010. Br. J. Ophthalmol. 98(5), 612–618.
Ngondi, J., Gebre, T., Shargie, E.B., Graves, P.M., Ejigsemahu, Y., Teferi, T., Genet, A., Mosher, A.W., Endeshaw, T., Zerihun, M., et al: 2008. Risk factors for active trachoma in children and trichiasis in adults: a household survey in Amhara Regional State, Ethiopia. Trans R Soc Trop Med Hyg. 102(5), 432–438.
Oguttu, D., Byamukama, E., Katholi, C.R., Habomugisha, P., Nahabwe, C., Ngabirano, M., Hassan, H.K., Lakwo, T., Katabarwa, M., Richards, F.O., Unnasch, T.R., 2014. Serosurveillance to monitor onchocerciasis elimination: the Ugandan experience. Am. J. Trop. Med. Hyg. 90(2), 339–345.
Parthasarathy NR., 1967. A simple catalytic model in trachoma epidemiology. J All India Ophthalmol Soc. 15(5):165–171.
Rabiu, M., Alhassan, M.B., Ejere, H.O., Evans, J.R., 2012. Environmental sanitary interventions for preventing active trachoma. Cochrane Database Syst Rev 2: CD004003
Rahman, S.A., West, S.K., Mkocha, H., Munoz, B., Porco, T.C., Keenan, J.D., Lietman, TM., 2015. The distribution of ocular Chlamydia prevalence across Tanzanian communities where trachoma is declining. PLoS Negl Trop Dis. 9(3), e0003682.
Ray, K.J., Lietman, T.M., Porco, T.C., Keenan, J.D., Bailey, R.L., Solomon, A.W., Burton, M.J., Harding-Esch, E., Holland, M.J., Mabey, D., 2009. When can antibiotic treatments for trachoma be discontinued? Graduating communities in three African countries. PLoS Negl Trop Dis. 3(6), e458.
Ray, K.J., Porco, T.C., Hong, K.C., Lee, D.C., Alemayehu, W., Melese, M., Lakew, T., Yi, E., House, J., Chidambaram, J.D., Whitcher, J.P., Gaynor, B.D., Lietman, T.M., 2007. A rationale for continuing mass antibiotic distributions for trachoma. BMC Infect Dis. 7:91.
Regoes, R.R., Wiuff, C., Zappala, R.M., Garner, K.N., Baquero, F., Levin, B.R., 2004. Pharmacodynamic functions: a multiparameter approach to the design of antibiotic treatment regimens. Antimicrob Agents Chemother. 48(10), 3670-3676.
Shattock, A.J., Gambhir, M., Taylor, H.R., Cowling, C.S., Kaldor, J.M., Wilson, D.P., 2015. Control of trachoma in Australia: a model based evaluation of current interventions. PLoS Negl Trop Dis. 9(4), e0003474.
Solomon, A.W., Harding-Esch, E., Alexander, N.D.E., Aguirre, A., Holland, M.J., Bailey, R.L., Foster, A., Mabey, D.C.W., Massae, P.A., Courtright, P., Shao, J.F., 2008. Two
45
![Page 46: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/46.jpg)
Doses of Azithromycin to Eliminate Trachoma in a Tanzanian Community. N Engl J Med. 358(17), 1870-1871.
Solomon, A.W., Holland, M.J., Alexander, N.D., Massae, P.A., Aguirre, A., Natividad-Sancho, A., Molina, S., Safari, S., Shao, J.F., Courtright, P., Peeling, R.W., West, S.K., Bailey, R.L., Foster, A., Mabey, D.C., 2004. Mass treatment with single-dose azithromycin for trachoma. N Engl J Med. 351(19), 1962-1971.
Solomon, A.W., Holland, M.J., Burton, M.J., West, S.K., Alexander, N.D., Aguirre, A., Massae, P.A., Mkocha, H., Munoz, B., Johnson, G.J., Peeling, R.W., Bailey, R.L., Foster, A., Mabey, D.C., 2003. Strategies for control of trachoma: observational study with quantitative PCR. Lancet. 362(9379), 198-204.
Solomon, A.W., Peeling, R.W., Foster, A., Mabey, D.C.W., 2004. Diagnosis and Assessment of Trachoma. Clin Microbiol Rev. 17(4), 982-1011.
Stocks, M.E., Ogden, S., Haddad, D., Addiss, D.G., McGuire, C., Freeman, M.C., 2014. Effect of water, sanitation, and hygiene on the prevention of trachoma: a systematic review and meta-analysis. PLoS Med. 11(2), e1001605.
Sundaresan, T. K., Assaad, F.A., 1973. The use of simple epidemiological models in the evaluation of disease control programmes: a case study of trachoma. Bull World Health Organ. 48(6), 709-714.
Taylor, H.R., Burton, M.J., Haddad, D., West, S., Wright, H., 2014. Trachoma. Lancet. 384(9960), 2142-2152.
Turner. H.C., Truscott, J.E., Hollingsworth, T.D., Bettis, A.A., Brooker, S.J., Anderson, R.M., 2015. Cost and cost-effectiveness of soil-transmitted helminth treatment programmes: systematic review and research needs. Parasit Vectors. 8: 355.
Turner, H.C., Walker, M., Attah, S.K., Opoku, N.O., Awadzi, K., Kuesel, A.C., Basanez, M.G., 2015. The potential impact of moxidectin on onchocerciasis elimination in Africa: an economic evaluation based on the Phase II clinical trial data. Parasit Vectors. 8: 167.
Turner, H.C., Walker, M., Churcher, T.S., Basanez, M.G., 2014. Modelling the impact of ivermectin on River Blindness and its burden of morbidity and mortality in African Savannah: EpiOncho projections. Parasit Vectors. 7: 241.
Turner, H.C., Walker, M., French, M.D., Blake, I.M., Churcher, T.S., Basanez, M.G., 2014. Neglected tools for neglected diseases: mathematical models in economic evaluations. Trends Parasitol. 30(12), 562-570.
West, S. K. 2003. Blinding trachoma: prevention with the safe strategy. Am J Trop Med Hyg 69(5 Suppl): 18-23.
West, S.K., Munoz, B., Mkocha, H., Holland, M.J., Aguirre, A., Solomon, A.W., Foster, A., Bailey, R.L., Mabey, D.C., 2005. Infection with Chlamydia trachomatis after mass treatment of a trachoma hyperendemic community in Tanzania: a longitudinal study. Lancet. 366(9493), 1296-1300.
West, S.K., Munoz, B., Turner, V.M., Mmbaga, B.B., Taylor, H.R., 1991. The epidemiology of trachoma in central Tanzania. Int J Epidemiol. 20(4), 1088-92.
World Health Organization (WHO). 2003. Report of the 2nd Global Scientific Meeting on Trachoma. Available: http://www.who.int/blindness/2nd%20GLOBAL%20SCIENTIFIC%20MEETING.pdf.
World Health Organization (WHO). 2012. Global WHO Alliance for the Elimination of Blinding Trachoma by 2020. Wkly Epidemiol Rec. 87(17), 161-8.
46
![Page 47: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/47.jpg)
World Health Organization (WHO). 2013. Trachoma. Status of endemicity for blinding trachoma, 2012. Available: http://apps.who.int/neglected_diseases/ntddata/trachoma/trachoma.html
World Health Organization (WHO). 2015. Global Elimination of Trachoma documents. Available: http://www.who.int/blindness/publications/get2020/en/.World Health Organization (WHO). 2015a. Neglected tropical diseases. Available:
http://www.who.int/neglected_diseases/diseases/en/World Health Organization (WHO). 2015b. Trachoma Fact sheet No 382. Available:
http://www.who.int/mediacentre/factsheets/fs382/en/Wilkins, P. P., Keystone, J. S., 2013. Schistosomiasis Serology Is Valuable and Reliable. Clin
Infect Dis. 56(2), 312.Wong, J., Hamel, M.J., Drakeley, C.J., Kariuki, S., Shi, Y.P., Lal, A.A., Nahlen, B.L.,
Bloland, P.B., Lindblade, K.A., Were, V., Otieno, K., Otieno, P., Odero, C., Slutsker, L., Vulule, J.M., Gimnig, J.E., 2014. Serological markers for monitoring historical changes in malaria transmission intensity in a highly endemic region of Western Kenya, 1994-2009. Malar J. 13: 451.
Wright, H.R., Taylor, H.R., 2005. Clinical examination and laboratory tests for estimation of trachoma prevalence in a remote setting: what are they really telling us? Lancet Infect. Dis. 5(5), 313–320.
Yildiz Zeyrek F, Palacpac N, Yuksel F, Yagi M, Honjo K, Fujita Y, Arisue N, Takeo S, Tanabe K, Horii T., Tsuboi, T., Ishii, K.J., Coban, C., 2011. Serologic Markers in Relation to Parasite Exposure History Help to Estimate Transmission Dynamics of Plasmodium vivax. PLoS ONE. 6(11), e28126.
47
![Page 48: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/48.jpg)
Figure legends
48
![Page 49: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/49.jpg)
Figure 1. Diagrammatic overview of the 4 key model structures and state variables that have
been used for modelling trachoma transmission. Here S represents an individual or
individuals who are susceptible to infection and I represents those who are infected with
trachoma. (a) Classic SIS model structure; individuals move into the infected compartment
at a rateλ (the force of infection), and recover at a rate γ (independent of the number of
previous infections), returning to the S compartment. (b) SIS ladder of infection model
(Gambhir et al., 2007); this framework tracks the number of infections (i) that individuals
experience throughout their lifetime as they age. They experience a force of infection
lambda λ λ and move into the I state; however, they recover at a rate γi dependent on the
number of previous infections they have had and move to state Si. Equally, the infectivity of
individuals decreases with each infection experienced. (c) Household model (Blake et al.,
2009; Blake et al., 2010); each oval represents either a household or a community;
transmission of infection is modelled within each household or community as in (a) with
local force of infection λL ; however, each community or household also provides and
experiences a (global) force of infection, λG from all other households or communities). (d)
Catalytic model (Assaad and Maxwell-Lyons, 1966; Parthasarathy, 1967; Sundareasn and
Assaad, 1973; Martin et al., 2015); individuals experience a force of infection (or
seroconversion rate), λ , and move into the infected (seropositive), or diseased state;
however, once they are classified as such they do not serorevert or recover. In the two-
stage catalytic model (Assaad and Maxwell-Lyons, 1968), individual’s progress from S state
49
![Page 50: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/50.jpg)
to present with active disease; individuals leave the active disease compartment (and
progress to other disease sequelae states). (E) An extension of the classic SIS model
structure, which explicitly models individuals in the disease state and after a certain period
of recovery allows them to be re-infected while they are still in the partially diseased state.
Figure 2. A map of the global distribution of trachoma endemic communities and those
currently under surveillance. Red indicates countries that are classified as endemic by the
WHO, and yellow indicates those currently under surveillance. Grey dots show a country
where a transmission model of trachoma has been fitted to data from that country,
although Taiwan and Tunisia are no longer classified as endemic regions. The size of the dots
is not proportional to the number of studies. However the total number of datasets
collected and used to inform modelling studies is not equal across all countries. Data
collected from Tanzania has been used in 11 modelling studies identified, data from The
Gambia has been used 6 times in the studies analysed here, data from Ethiopia has been
used 5 times, data from Taiwan has been used twice and data from Egypt, Tunisia, Malawi,
India and Australia have all been used once. Data on endemicity was inferred from
http://apps.who.int/neglected_diseases/ntddata/trachoma/trachoma.html .
Figure 3. Flow diagram of the modelling study that developed the ladder of infection model
(Figure 1b) in order to track and model progress to disease sequelae (Gambhir et al., 2009).
50
![Page 51: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/51.jpg)
Progression into TS (trachomatous scarring) and TT (trachomatous trichiasis) is determined
by a threshold number of repeated infections. Once individuals reach a threshold number of
infections they are considered to have TS, they recover from infection and then get re-
infected until they reach the threshold number of infections for TT (dashed arrows), from
here progression to CO (corneal opacity) occurs at a constant rate σ with no further
infections required due to the damaged already incurred by those with TT. Disease sequelae
classes are indicated with a dashed red box.
51
![Page 52: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/52.jpg)
Table 1. Studies identified through the literature review that have developed a mechanistic and statistical models of trachoma transmission and/or control
Reference Model type Study purpose Type of data used Levels of endemicity evaluated
Relevance for GET 2020 goals
Assaad and Maxwell-Lyons, 1966
Simple and two-stage catalytic model
Estimate FOI and rates of acquisition and loss of active disease before a control programme
Age prevalence of clinical trachoma cases and active disease in Taiwan (1960–1961)
Hyper Changes in transmission have arisen due to socio-economic change. These can be detected with a catalytic model, which showed a profound impact on transmission.
Parthasarathy, 1967
Catalytic model
Estimate the FOI in the population and elucidate the epidemiological pattern of trachoma in a high endemic area
Data collected from 15 Indian states between 1959 – 1963. Eye examination performed.
Not stated Illustrates catalytic models can be fitted to age-specific disease data. Estimates the FOI and identifies the maximum proportion of the population likely to have been infected, giving insight on the number of rounds of MDA that may be required.
Sundaresan and Assaad, 1973
Catalytic model
Determine how FOI changes after implementation of school-based treatment programme
Age prevalence of trachoma in Taiwan pre-control (1960–1961) and in 1968–1969
Hyper Estimating the FOI may be useful to measure changes in transmission as a result of control
Lietman et al., SIS Explore how frequently Endemic prevalence from Hypo, Meso, Achievable. Mass treat every
52
![Page 53: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/53.jpg)
1999 deterministic, partial immunity in older age groups
should mass antibiotic treatment be given for settings with different baseline endemicity levels
Egypt, Malawi, Tanzania, The Gambia and Tunisia
Hyper 12-24 months in hypo, every 6-12 mo. in hyper/meso; 100% coverage
Lee et al., 2005
SIS deterministic & stochastic, only state I (infecteds) is modelled
Investigate whether there are optimal times of the year for mass antibiotic distribution
Model prevalence over time. Rates of transmission, β , and recovery, γ calibrated with data for 24 Ethiopian villages
Hyper Easier to achieve elimination if treatment is delivered after high and before low season, when transmission rates are at their lowest
Gambhir et al., 2007
SIS deterministic, ladder of infection model
Develop a model to explain observed distribution of community infection. Explain heterogeneous response to treatment
Kongwa district, Tanzania Need to consider age infection profiles and different responders (heterogeneity in exposure, predisposition)
Ray et al., 2007
SIS stochastic Investigate how long should treatment be continued in high transmission settings to eliminate infection
16 Ethiopian villages. PCR data in children 1–5 years. Baseline and follow up data after 2, 6, 12, 18 and 24 months after mass drug administration
Hyper Biannual treatment implemented for 5 years will lead to elimination in 95% of villages. Need to consider distribution of infection that leads to hotspots and the role of re-introduction from untreated communities
Grassly et al., 2008
2 & 4 state Markov model. Looking at infected and active disease
Infer parameters on the natural history of infection, and how they relate to demographic and baseline immune
256 people from Gambian villages, Jali and Berending, signs of infection checked every two weeks over 6
Meso Average duration of infection in young children is long. This contributes to persistence of infection after treatment. Young children may need more
53
![Page 54: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/54.jpg)
measurements months. Presence of chlamydial lipopolysaccharide antigen tested by ELISA
follow-up.
Gambhir et al., 2009
SIS deterministic, ladder of infection model
Refine model to capture age-specific prevalence of infection and estimate the average number of infections for the development of disease sequelae
Data from The Gambia and Tanzania. Infection by qPCR. For hyperendemic setting age-profiles of infection, infection load (Tanzania) and rate of recovery (The Gambia)
Hypo, Meso, Hyper
Captures age-specific prevalence patterns. Indicates roughly the number of prior infections which are necessary for the development of TS and TT
Blake et al., 2009
SIS stochastic Examine the contribution of transmission between and within households
Cross-sectional data on prevalence of infection (by PCR) from Upper Saloum District and Jali village in The Gambia and Kahe Mpya and Maindi villages in Tanzania
Hypo, Meso, Hyper
Household transmission is an important contributor to incidence and repeat infections. Treatment only may not be sufficient. Not including household transmission can alter the expected result of mass treatment
Ray et al., 2009
SIS stochastic Investigate two different treatment strategies: 3 annual mass treatments in all communities (as per WHO) versus 3 annual mass treatments but
Data from 3 different regions. Collected at baseline, 3 and 6 months after treatment. Upper conjunctiva swab and DNA tested for with PCR.
Hypo, Meso, Hyper
Graduating communities from a programme when infection is reduced below 5% is a reasonable strategy and could reduce the amount of antibiotic distributed in some areas by
54
![Page 55: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/55.jpg)
stopping treatment in those communities whose prevalence falls <5%
In The Gambia 1–5 year olds tested; In Tanzania and Ethiopia data from 1–10 year olds are used
more than 2-fold
Blake et al., 2010
SIS stochastic.Differential transmission between children and adults
Investigate whether targeting antibiotics to households that have at least one member with active disease is effective in preventing infection
Calculate the cost effectiveness of targeted household treatment compared with mass antibiotic distribution
West and East Africa (Upper Saloum District and Jali village in The Gambia; Kahe Mpya sub-village and Maindi village in Tanzania). Presence of infection assessed with PCR. In Maindi village, quantitative PCR to indicate presence of infection. Clinical observations also used
Hypo, Meso, Hyper
Household-targeted treatment produces comparable results to mass treatment. Active disease not a 100% sensitive marker of infection. Probability of eliminating infection with 10 years of biannual treatment not very high
Goals not currently achievable
55
![Page 56: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/56.jpg)
Gambhir et al., 2010
SIS ladder of infection model, deterministic
Second model to look at a disease only state
Examine the effect of antibiotics on infection and disease, implications on programmes. Usefulness of modelling for control programme impact projections
Same data as in Blake et al., 2009
First model hyper
Second model mixture of endemicities
In low initial endemicity areas 1-2 annual mass treatments may be sufficient for elimination. In hyperendemic areas 3 annual treatments result in no lasting effect on infection or disease sequelae. Hard to achieve goals. Implement full SAFE to reduce transmission
Lietman et al., 2011
SIS stochastic Assess re-introduction of infection following treatment
Prevalence of ocular chlamydial infection by PCR in 24 Ethiopian communities after a single MDA round. Communities were re-treated, by design, after the 24-month survey
Hypo (Nepal), Meso (Tanzania), Hyper (Ethiopia)
Easier to achieve elimination of infection than previously thought when considering positive feedback (the hazard of a susceptible individual becoming infected per infectious case increases with prevalence). Re-introduction is slow enough to detect through surveillance
Koukounari et al., 2013
Latent Markov model
Infer the population prevalence of infection and active disease in addition to the sensitivity and specificity of 3 diagnostic tests
Same as Gambhir et al., 2009, but only use data for children <10 years of age.
Hyper The sensitivity and positive predictive values of clinical examination for infection were low in The Gambia but the sensitivity of TI and positive predictive value of the clinical exam for TF and TI were low in Tanzania. This could result in
56
![Page 57: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/57.jpg)
communities experiencing further rounds of treatment when it has been eliminated from communities .
Liu et al., 2013 SIS stochastic Estimate the change in R0 per year following treatment. Assess the evidence that treatment can cause loss of short-term immunity
32 communities cluster- randomized clinical trial in Tanzania. Children 0–5 years examined at baseline, and at 6–36 months after baseline. Swab for the presence of chlamydial DNA. Mass treatment at baseline, 12 and 24 months
Meso The lack of change in R0, small negative linear trend and no evidence of loss of immunity suggest repeated treatment can help to eliminate infection
Liu et al., 2014 SIS stochastic Assess the prevalence of infection within the community following treatment. Calculate the effective field efficacy of azithromycin in clearing ocular Chlamydia
Same data as in Liu et al., 2013
Meso Efficacy = 68% (95% CI: 57–75%); no decrease in efficacy during trial; 89% chance of elimination after 10 years of annual treatment at 95% coverage
57
![Page 58: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/58.jpg)
Martin et al., 2015
Catalytic model allowing for change in FOI
Use seroconversion rate to estimate changes in the force of infection after the implementation of mass antibiotic distribution
Tanzania community, Kahe Mpya sub-village, 575 people.Detection of conjunctival swab DNA and eye examination. Prevalence of TF, TI or both, and of TS/TT/CO.
Meso Absence of antibody responses in children born after implementation of mass treatment reflects lack of C. trachomatis transmission. FOI modelling suggests serology could play a role in post-treatment surveillance
Lietman et al., 2015
SIS stochastic Test the hypothesis that a geometric distribution describes the prevalence of infection in different communities where infection is disappearing
Gurage, Ethiopia, and the TANA study in Amhara Ethiopia; from 24 communities monitored for infection. TEF study, 1–5 year old children monitored biannually. TANA study, 0–9 year old children monitored, treated annually or biannually for 42 months. PCR data
Hyper Surveying infection in low transmission settings may be difficult. Community-level cross-sectional prevalence may be approximated by a geometric distribution. Its relatively heavy tail suggests that presence of an occasional high-prevalence community is to be expected, not necessarily reflecting transmission hotspots or programme failure
Jimenez et al., 2015
Linear and logistic regression statistical models
Determine how many annual mass treatments are needed to achieve elimination. Assess factors that affect the success of reaching
283 cross-sectional survey pairs with baseline and follow-up data, mass treatment conducted in 170 districts. Prevalence of
Hypo, Meso, Hyper
Annual treatment alone is insufficient. More information needed on the effects of baseline prevalence, therapeutic coverage and underlying environmental and
58
![Page 59: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/59.jpg)
elimination goals active trachoma (TF or TF/TI) and of trachomatous trichiasis (TT)
hygiene conditions. When trachoma prevalence >30%, 7 or more annual treatments may be required
Shattock et al., 2015
Multi-stage infection SIS, stochastic, individual- based model. Household and community transmission
Assess whether past trachoma intervention efforts have been effective. Evaluate what impact can be expected if current intervention strategies are maintained. Investigate whether shifts in strategy or increases in the intensity of its implementation may lead to improved results
67 Australian communities modelled using prevalence data. Data used for all 3 different transmission settings.
Hypo, Meso, Hyper
Current intervention strategy unlikely to achieve 2020 goals. The likelihood of achieving this goal can be significantly increased by large-scale antibiotic distribution programmes accompanied by screening, treatment, facial cleanliness and housing construction
Rahman et al., 2015
Statistical model
Test whether Chlamydia trachomatis prevalence across 75 Tanzanian communities where trachoma has been disappearing was exponentially distributed
75 communities in 8 districts in Tanzania. Survey infection swab and PCR. Pre-school children aged 5 years and under were surveyed
Prevalence in 1999: 17–79%; 2007/2008: 0–28%
Models correctly predict that infection prevalence across communities where trachoma is disappearing can be described by an exponential distribution
Gambhir et al., 2015
SIS deterministic, ladder of infection model
Explore the impact of each of the components of SAFE on disease sequelae. Investigate whether the prevalence
Maindi in Kongwa, Tanzania. Trachoma was assessed according to the WHO simplified grading scheme. Specimens were
Hyper Increased intensity of all interventions will reduce TT prevalence and CO incidence. Surgery against TT alone is a stopgap until transmission is
59
![Page 60: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/60.jpg)
of TT can be reduced to 1 in 1,000, and the incidence of CO to 1 in 10,000 per annum
collected for detection of ocular C. trachomatis by real-time PCR
reduced and mass antibiotic distribution is enhanced. Goals are achievable if all interventions are implemented simultaneously. High transmission settings may take 20 years to achieve GET 2020 goals
60
![Page 61: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/61.jpg)
Table 2. Values of basic (R0) or effective (Re) reproductive numbers reported in the articles identified in this literature review, or calculated (as beta/gamma) where estimates of the rate of transmission (parameter beta) and of the rate of recovery (parameter gamma) were provided.
61
Reference Beta (week–1 unless otherwise specified in original article)(95% CIs)
Gamma (week–1 unless otherwise specified)(95% CIs)
Parameters estimated in this study
Reproduction number (95% CIs)
Country and transmission setting
R0, Re, R*
1
Lee et al., 2005 0.20 month–1 0.10 month–1 Beta: No, fixedRecovery: No, fixed
2.00 Ethiopian (hyper) R02
Ray et al., 2007 0.047 0.017 Beta: Yes, MLERecovery: Yes, MLE
3.16 Ethiopian (hyper) R02
Blake et al., 20093
0.29 (0.16–0.51) 0.76 (0.39–1.40) 1.73 (1.18–2.37) 1.70 (1.15–2.46)
0.058 Beta: YesRecovery: No (1/17.2 weeks)
1.252.811.182.65
The Gambia (meso) The Gambia (hyper) Tanzania (meso) Tanzania (hyper)
R*1
Ray et al., 2009 0.0520.0330.039
0.052 (0.000–0.113) 0.037 (0.001–0.073)0.0123 (0.005–0.02)
Beta: Yes, MLERecovery: Yes, MLE
1.01 (0.50–1.27)0.89 (0.66–1.36)3.14 (2.51–3.77)
The Gambia (hypo) Tanzania (meso) Ethiopia (hyper)
R0
Lietman et al., 2011
0.014 (0.007–0.021)0.019 (0.016–0.052)0.014 (0.007–0.029)0.019 (0.015, 0.037)
0.017 (0.01–0.024)0.017 (0.013–0.041)0.014 (0.009–0.023)0.014 (0.011–0.024)
Beta: Yes, MLERecovery: Yes, MLE
1.151.161.111.84
Ethiopia (prevalence of ocular Chlamydia ranging from 5% to 75%)
R02
Liu et al., 2013 0.233 (0.210–0.258) month–1 0.167 month–1
(1/6 months)Beta: YesRecovery: No4
1.40 (1.26–1.55) R0 for first year, prior to interventions. Tanzania (hyper) at baseline
R0
Liu et al., 2014 0.229 (0.202–0.262) month–1 0.167 month–1
(1/6 months)Beta: YesRecovery: No
1.37 (1.21–1.57) Tanzania (hyper) at baseline of trial
Re
Gambhir et al., 20095
1.8 (1.6–2.1) year–1 2.4 (2.0–2.9) year–1
27.7 (21.8–35.1) or 186 year–1
0.066 (0.043–0.154) mo–1 (first infection, for R0) 0.357 (0.313–0.417) mo–1 (after many infections, Re)
Beta: YesRecovery: Yes, NGMM
1.101.403.206
The Gambia (hypo)Tanzania (meso) Tanzania (hyper)
Re2,5
Re2,5
Re2,5
![Page 62: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/62.jpg)
R0, Basic reproduction number; Re, Effective reproduction number; 1R*, Household basic reproduction number; 2Derived from parameters given in the article but not explicitly
calculated by the authors of the original study; 3values for beta areβG , Global (household/community) transmission coefficient; MLE, maximum likelihood estimation; 4Liu et al. (2013) estimate beta for each year of the study assuming a fixed rate of recovery ; 5values for Re calculated using the next generation matrix method (NGMM) (unpublished calculation) from the best fit parameters estimated in Gambhir et al., 2009.
62
![Page 63: Modelling trachoma: a review focusing on the GET 2020 goals€¦ · Web viewin helping to reduce transmission. Furthermore, very few of the models identified in the literature review](https://reader034.vdocument.in/reader034/viewer/2022050209/5f5c08ebcfcede00924c4f74/html5/thumbnails/63.jpg)
63