Linking rattiness, geography and environmental degradation to spillover Leptospira infections in marginalised urban settings

Article

The manuscript will be made available when published.

Data

A codebook for all data can be found here.

1. Rat data

File: rat_data.csv

The cleaned dataset for the cross-sectional survey in Pau da Lima community, Salvador, Brazil. It contains information on the five rat abundance indices (traps, track plates, number of burrows, presence of faeces and presence of trails), and was collected between October and December 2014. It is georeferenced and includes all relevant covariate values used in the model described in the article. This dataset was collected by and belongs to the Instituto de Saúde Coletiva, Universidade Federal da Bahia (ISC-UFBA) and the Oswaldo Cruz Foundation. If you use this dataset please could you cite ISC-UFBA, Fiocruz and this published article (see license at bottom of page).

2. Human data

File: human_data_anon.csv

The cleaned dataset for the human cohort study in Pau da Lima community, Salvador, Brazil. Please note that household coordinates and valley ID have been removed to ensure participant anonymity. As a result some of the analyses in the R scripts in this repository will not be possible to conduct with this dataset. This dataset was collected by and belongs to the Instituto de Saúde Coletiva, Universidade Federal da Bahia (ISC-UFBA). If you use this dataset please could you cite ISC-UFBA and this published article (see license at bottom of page).

3. Prediction grid

File: prediction_grid.csv

A 2.5m by 2.5m prediction grid with covariate values within the Pau da Lima study area.

4. Model control parameters for rattiness model

This contains a set of parameters for controlling the rattiness model you wish to fit, each column is used as follows:

• rat: rattiness covariates you wish to include.
• rat_par0: starting values for rattiness parameters, order: regression coefficients, φ (scale of spatial correlation). Set to FALSE to fit model with starting values equal to zero.
• with_Ui: set to TRUE or FALSE to include/not include location-level iid random effects.
• psi0: starting value for ψ.

File: control_R-rat_explore_R_H_spat_withnugg.csv

5. Model control parameters for joint rattiness-infection model

This contains a set of parameters for controlling the joint rattiness-infection model you wish to fit, each column is used as follows:

• human: human covariates you wish to include.
• rat: rattiness covariates you wish to include.
• rat_par0: starting values for rattiness parameters, order: regression coefficients, φ (scale of spatial correlation). Set to FALSE to fit model with starting values equal to zero.
• par0.human: starting values for human regression coefficients. Set to TRUE to start with values estimated by glm() or FALSE to fit model with starting values equal to zero.
• multi.xi.on: set to TRUE to fit a model with multiple ξ parameters to test for interactions or FALSE to fit a model with a single ξ parameter.
• xi.var: the variable (factor) you wish to use to define different ξ parameter levels.
• xi0: starting values for ξ (include as many as xi.var levels).
• with_Ui: set to TRUE or FALSE to include/not include normally-distributed location-level iid random effects in rattiness.
• psi0: starting value for the scale of spatial correlation of rattiness, ψ.
• with_human_S: set to TRUE or FALSE to include/not include a spatial Gaussian process in the human data. (This was not included in the analysis in this paper because there was no residual spatial correlation in the human data)
• omega2.0: starting value for the variance of the spatial Gaussian process in the human data. (This was not included in the analysis in this paper)
• zeta0: starting value for the scale of spatial correlation of spatial Gaussian process in the human data (This was not included in the analysis in this paper)
• with_human_N: set to TRUE or FALSE to include/not include normally-distributed location-level iid random effects in the human data.
• omega2_nugg0: starting value for ω^2, the variance of the iid random effects in human data.

File: control_R1_3Xi.csv

Analysis scripts

R scripts to conduct the analysis described in this article are available here. Each script includes a brief description of its function.

1. Exploratory analysis of rat data

File: 1-rat-explore.R

2. Exploratory analysis of human data

File: 2-human-explore.R

3. Fit the joint rattiness-infection model and bootstrap for CIs

File: 3-fit-joint-model.R

4. Spatial prediction for rattiness and probability of infection

File: 4-spatial-prediction.R

Functions

A description of all of the functions used in this analysis are given here to enable other users to use the rattiness-infection framework for their own data. If you would like to adapt the

1. Non-spatial rattiness model

File: 1-rattiness_nonspatial_model_fns.R

2. Geostatistical rattiness model

File: 2-rattiness_spatial_model_fn.R

3.Spatial predictions for geostatistical rattiness model

File: 3-rattiness_spatial_model_predict_fn.R

4. Joint rattiness-infection model

File: 4-joint_model_fn.R

5. Extra functions used for fitting the joint rattiness-infection model and making predictions

File: 5-joint_model_extra_functions.R

6. Parametric bootstrapping for the joint rattiness-infection model

File: 6-joint_model_bootstrap_fn.R

7. Spatial prediction for joint rattiness-infection model (can predict for rattiness and probability of infection)

File: 7-joint_model_predict_fn.R

8. Scaling functions used for scaling of data

File: scaling_fns.R

Contact

If you have any questions about the project, data or software please contact max.eyre@lstmed.ac.uk.

Licenses

Data:

This work is licensed under a Creative Commons Attribution 4.0 International License.

Software:

License