L'analyse de la couverture géographique vous permet de prendre en compte la disponibilité des services (c'est-à-dire la capacité des structures de santé à accueillir des patients) et d'en tenir compte en plus des contraintes d'accessibilité physique, afin de définir l'aire de captage associée à chaque structure de santé.

En d'autres termes, l'aire de captage va:

• s'étendre jusqu'à ce que la durée maximale du voyage (par exemple 120 minutes) soit atteinte si la population de ce bassin est inférieure ou égale à la capacité estimée du centre de santé.
• ne pas atteindre le temps de trajet maximal (par exemple, 120 minutes) si la population de ce bassin dépasse la capacité estimée de la structure de santé. Dans un tel cas, l'aire de captage sera plus petite et couvrira la taille de la population correspondant à la capacité estimée de la formation sanitaire.

La capacité de couverture maximale de chaque structure de santé doit être incluse dans un champ (colonne) séparé de la table attributaire du shapefile des structures de santé (voir section 3.3.1.6). Une fois que c'est le cas, ces informations peuvent être directement utilisées par AccessMod lors de l'analyse.

Les deux captures d'écran suivantes expliquent comment entrer des données et définir des paramètres pour l'analyse:

Données en entrée:

Sous:

(1) “Select population layer (raster)”, sélectionnez la couche raster contenant la distribution spatiale de la population cible que vous souhaitez utiliser dans la liste déroulante (celle nommée "population" dans le présent exercice).

(2) “Select merged land cover layer (raster)”, sélectionnez la couche raster contenant l'occupation du sol fusionnée résultant de l'utilisation du premier outil (voir la section 5.5.2) - nommée "land cover merged".

(3) “Select scenario table (table)", sélectionnez la table de scénario de voyage que vous souhaitez utiliser (pour l'exercice, nous vous recommandons de sélectionner la table de scénario créée lors de l'analyse précédente (voir la section 5.5.3).

(4) "Select existing health facilities layer (vector)”, sélectionnez la couche vectorielle de la couche des structures de santé existantes nommée "facility" dans le jeu de données d'exemple

(5) “Select facility ID field (unique)”, sélectionnez le champ de la table attributaire de la couche des structures de santé qui contient l'identificateur unique de chaque structure de santé.

(6) “Select facility name field (text)”, sélectionnez le champ de la table attributaire de la couche des structures de santé qui contient le nom de chaque structure de santé.

(7) “Select facilities capacity field (numeric)”, sélectionnez le champ de la table attributaire de la couche des structures de santé qui contient la couverture, la capacité maximale de chaque structure de santé.

(8) "Select zones layer", si vous avez coché l'option "Generate zonal statistics (sélectionnez la couche des zones dans le panneau de données en entrée)" dans la section "paramètres d'analyse" (voir le point 7 dana la capture d'écran suivante), sélectionnez ici la couche contenant les limites des zones à utiliser pour extraire les statistiques pertinentes. Aux fins de l'exercice, la couche s'appelle "zones".

(9)  "Select zone unique ID (integer)", sélectionnez le champ dans la table attributaire de la couche de limites de zone qui contient l'identificateur unique de chaque zone.

(10) "Select zone name (text), sélectionnez le champ dans la table attributaire de la couche de limites de zone qui contient le nom de chaque zone

Scénario de voyage:

 (11) En ce qui concerne l’analyse de l’accessibilité (voir la section 5.5.3), cette section permet d’importer le contenu d’une table de scénario externe et / ou de modifier manuellement les informations reportées dans les colonnes pour l’étiquette, la vitesse et / ou le mode.

Ensuite, les données peuvent être entrées pour la deuxième partie du panneau d'analyse, comme suit:

Paramètres d'analyse:

Sous:

(1) Type of analysis: Comme pour l'analyse précédente, indiquez si vous souhaitez adopter une approche «anisotrope» ou «isotrope» pour l'analyse. Nous avons choisi «anisotrope» dans le présent exercice afin de prendre en compte l’impact de la pente sur la vitesse pour les modes de transport WALKING et BICYCLING.

(2)  Direction of travel: Sélectionnez le sens de déplacement des patients. Veuillez noter que ce champ n'apparaît que si l'option "anisotrope" a été sélectionnée ci-dessus. Aux fins de l'exercice, choisissez "Vers les structures".

(3) Facilities processing order: L'analyse de la couverture géographique est effectuée dans un ordre de traitement séquentiel: une structure de santé est prise en compte et traitée, puis la structure suivante est prise en compte, et ainsi de suite, jusqu'à ce que toutes les structures de santé aient été traitées. A chaque itération, la population identifiée comme appartenant au bassin calculé d'une structure de santé spécifique est soustraite de la couche de population globale. Un tel ordre de traitement permet à AccessMod de simuler à la fois la saturation des services dans les zones peuplées et / ou les préférences des patients potentiels lorsque les patients sont à la portée de nombreuses structures de santé différentes. Pour cette raison, l'ordre dans lequel chaque structure de santé est prise en compte dans l'analyse peut influencer les résultats (bien qu'il soit très difficile de prédire comment). Veuillez vous reporter à l’Annexe 4 pour des détails sur la manière dont cette analyse est effectuée.

Plusieurs choix sont disponibles pour déterminer l'ordre dans lequel les structures de santé sont prises en compte. Pour les besoins de l'exercice, vous pouvez sélectionner la première option: ""A field in the health facility layer". Le but ici est de commencer par le centre qui a la plus grande capacité, puis de continuer dans un ordre décroissant.

Deux autres choix possibles pour l'ordre de traitement sont actuellement disponibles:

- "The population living within a given travel time from the facilities": en donnant un temps de trajet maximal (en minutes), AccessMod calcule d’abord l'aire de captage pour un temps donné autour de chaque structure de santé, puis détermine la population vivant dans chacune de ces aires de captage et utilise ces estimations pour déterminer l’ordre de traitement des structures de santé. Cette approche donnerait la priorité aux structures de santé situées dans des zones densément peuplées et faciles d'accès.

- "The population living within a circular buffer around the facilities": en donnant un rayon circulaire de zone tampon (en mètres), AccessMod calculera d’abord une zone tampon circulaire autour de chaque structure de santé, déterminera la population vivant dans chacune de ces zones tampons et utilisera ces chiffres pour déterminer l’ordre de traitement des structures de santé. De nouveau, cette approche donnerait la priorité aux structures de santé situées dans des zones densément peuplées et faciles d'accès.

(4)  The field appearing in this step will change depending on the processing order option selected under step 3, namely:

• "Select field from the facility layer": this field appears if you have selected "A field in the health facility layer" option under step (3). In this case, select the field from the health facility layer that contains the integer values you want to use for defining the processing order - this option is being used for the exercise followed here and the field containing the maximum coverage capacity has been chosen in this case.
• "Given travel time [minutes]": this field appears if you have selected "The population living within a given travel time from the facilities" option under step (3). In this case, enter the travel time in minutes to be used by AccessMod to compute the extent of the catchment area for each health facility.
• "Buffer radius [meters]": this field appears if you have selected "The population living within a circular buffer around the facilities" option under step (3). In this case, enter the radius in meters to be used by AccessMod to compute the extent of the buffer for each health facility.

 (5) "Processing order": Select here the processing order based on the option that has been selected under point 3. For the purpose of the exercise, select a descending order, in order for AccessMod to first process facilities with the highest maximum coverage capacity.

(6) "Maximum travel time [minutes]": specify the maximum travel time (in minutes) that should be used by AccessMod to define the maximum reach of the catchment area attached to each health facility. For the exercise, specify 120 minutes.

(7)  "Options": Several additional options are available:

- "Compute catchment area layer": Select this option if you wish to obtain a vector layer containing the individual catchment areas (polygons) attached to each facility during the analysis. This option is checked by default.

- "Remove the covered population at each iteration": Select this option to remove the population attached to the health facility from the population distribution grid at each iteration. This option is checked by default and should remain as such to avoid that the same population is attached to more than one facility. Un-selecting this option is nevertheless useful if you want to estimate the population located within a given travel time of a set of facilities without account for the overlap between catchment areas.

- "Compute map of population cells on barriers": Select this option to create an output raster file containing the cells in which a population resides but where the cells fall on a barrier. This population will not be taken into account in the analysis, and it is therefore often necessary to modify the input population distribution layer in raster format prior to the analysis to avoid this issue (see Appendix 1). This option is selected by default and you can keep it that way for the present exercise

- "Generate zonal statistics (select zones layer in data input panel"): Select this option to automatically obtain the percentage of population being covered by sub national level zones through the analysis. This option is unselected by default. Once selected, a new field labeled "Select zones layer (vector)" appears in the "Data input" section (see point 8 attached to the previous screenshot) for you to select the layer in question. For the purpose of the exercise, check the box, (select the "zone for stat" layer, "cat" as the field containing the unique ID and "admin_name" as the field containing the zone names).

(8) Add short tags:  Indicate short tags to be attached to the different outputs of the analysis. We will use "geographic analysis 120m" for the present analysis. Avoid very long name as it has been found to sometime prevent Excel to open the output generated xls files.

Sélection de structure de santé

 (9)  As in the previous analysis, you can select the set of facilities for which the analysis will be performed. Keep all facilities selected for the exercise.

Validation:

 (10) The validation module should indicate that all fields have been correctly filled in (with a green “OK”). If this is the case, you can hit the "Compute" button to launch the analysis. If this is not the case, the "Compute" button will still be in red and you will have to go through the warning and error message to find out what needs to be adjusted.

A transparent window with some text and a progress bar will appear in front of the panel while the analysis is being conducted. Please wait until this window disappears to continue using AccessMod. You will notice that the analysis is slower than the previous one, because of the iterative way of processing each health facility.

Once this is done, go back to the Data module to check the eight output datasets that have been generated:

The geographic coverage analysis generates the following datasets:

1. scenario processed class: Table containing the travel scenario that has been processed.
2. result geographic coverage analysis class: Table containing the results of the geographic coverage analysis.
3. zonal coverage class: Table containing the results of the zonal statistics analysis in case this option has been checked.
4. population on barriers class: Raster format layer containing the spatial distribution of the target population on barriers.
5. population residual class: Raster format layer containing the spatial distribution of the residual population.
6. catchment class: Vector format layer containing the extent of the catchment area for each facility.

Next, just as you did above for the first part of the exercise, you now need to archive the results, export and unzip them in order to open and visualize the results. All of this can be done in the "Data" module (see Section 5.4).

We will describe here only the new types of output generated by AccessMod compared to the previous analytical step. Let us start with the geospatial data that have been generated.

The first type of data is the vector format layer containing the extent of the catchment areas (polygon) attached to each health facility. This layer, found in the "shape_catchment_geographic_analysis_120m" folder, should be opened in a GIS software and would appear as below:

Each catchment area can be linked to the corresponding health facility through the unique identifier indicated in the attribute table of both layers. Please just note that the header of the column containing the unique identifier in the attribute table of the catchment area layer contains "_join" at the end of it ("cat_join" in the case of the present exercise as per the screenshot below) compare to the header of the column containing that same identifier in the attribute table of the health facility layer ("cat" in the case of the present exercise).

The raster format residual population layer stored in the "raster_population_residual_geographic_analysis_120m" folder contains the spatial distribution of the target population that remains uncovered after performing of the analysis. This layer can, for example, be used as an input for the scaling up analysis (see Section 5.5.7), and it looks like this once open in a GIS software:

The geographic analysis also generates two additional Excel files compared to the accessibility analysis, namely:

 Health facility specific statistics

 The health facility statistics file is named: "table_result_geographic_coverage_analysis_geographic_analysis_120m.xlsx". This file contains the columns shown below (note that the table is split in two in the below screenshot, and that we have sorted the data by decreasing value in the column "amRankComputed" - 5th column from the left in the upper part of the screnshot):

• cat: unique identifier of the health facility as per to the field selected from the attribute table of the health facility layer.
• name: name of the health facility as per the field selected from the attribute table of the health facility layer
• capacity: maximum coverage capacity of the health facility expressed as the number of people that the facility can serve (as per the field selected from the attribute table of the health facility layer).
• amRankValues_capacity: values for the parameter used to define the processing order. In the present exercise, this column contains the maximum coverage capacity of each facility as this is the option that was selected above. Please see the boxed note below for more information regarding the label and content of this column when using the other options.
• amRankComputed: processing order applied during the analysis based on the content of the "amRankValues_capacity" field and the direction of processing (ascending or descending), as selected by the user.
• amTravelTimeMax: maximum travel time in minutes, as set by the user for the analysis.
• amPopTravelTimeMax: total population located within the maximum travel time set for the analysis (amTravelTimeMax).
• amCorrPopTime: Pearson correlation coefficient between the set of travelling times (t, from 0 to the maximum travel time) and the corresponding covered population within this time step (i.e. the sum of population in all cells that are located between t and t+1 of travelling time). This correlation measure gives a rough estimate of how the population is distributed through space as we move outward from the health facility. As an example, a large positive value (e.g. 0.707 in the third line of the above table) means that the population is relatively uniformly distributed as you expand outwards from the facility. A strong negative correlation would mean that there is more population close to the health facility than there is far away from it. A correlation close to zero means that there is no specific tendency in how the population is spread within the catchment. This correlation should really just be used as a relative indicator, as we do not provide the statistical significance of the correlation (you should use proper GIS analysis outside AccessMod if you need statistical insights about how the population is spatially located within the catchments).
• amTravelTimeCatchment: travel time to reach the maximum extent of the catchment area attached to the health facility. This value is either:
  • Equal to the maximum travel time set for the analysis (amTravelTimeMax) when the maximum coverage capacity of the facility has not been reached within the set time (120 minutes in this example). In the current exercise, this is the case for the Queen Elizabeth Hospital.
  • Smaller than the maximum travel time set for the analysis (amTravelTimeMax) when the maximum coverage capacity of the facility is reached before reaching amTravelTimeMax. This is the case of The Medka Health Centre in the exercise.
• amPopCatchmentTotal: population located within the catchment area for the travel time reported in the amTravelTimeCatchment field
• amCapacityRealised: part of the maximum coverage capacity of the health facility that is being used (realized) based on the total population located in the catchment area for the given travel time. This value is:
  • Equal to the maximum coverage capacity of the health facility (capacity) when this value is reached before the maximum travel time (amTravelTimeMax). An example is the Medka Health Centre in the present exercise.
  • Smaller than the maximum coverage capacity of the health facility (capacity) when such capacity is not reached within the maximum travel time (amTravelTimeMax). An example is the Chiradzulu District Hospital is in this the present exercise.
• amCapacityResidual: part of the maximum coverage capacity of the health facility that is not being used. This value is calculated by making the difference between the maximum coverage capacity (capacity) and the realized capacity (amCapacityRealised). This value is therefore equal to 0 when the maximum coverage capacity is being reached within the given travel time.
• amPopCatchmentDiff: Part of the total population located in the final catchment area (amPopCatchmentTotal) which does not experience facility coverage by lack of sufficient coverage capacity in the health facility. This value is calculated by making the difference between the total population in the catchment area (amPopCatchmentTotal) and the realized capacity (amCapacityRealised). This value is therefore:
  • Equal to 0 when the maximum coverage capacity is not reached within the given travel time
  • Greater than 0 when the maximum coverage capacity is reached within the given travel time
• amPopCoveredPercent: cumulative geographic coverage expressed as a percentage. This value corresponds to the percentage of the initial population distribution grid (i.e. the layer selected in the "Select population layer (raster)" field) that has been covered after having processes the health facility in question.
  
  
  

The header of the columns containing the unique identifier, the name, and the maximum coverage capacity of each health facility in the resulting table will be the same as the header of the field you have selected from the attribute table of the health facility layer.

 The spelling of the header and content of the column containing the values used for defining the processing order will be as follows:

• amRankValues_X with "X" being the label of the field selected when using the defining the processing order according to the "A field in the health facility layer" option
• amRankValues_popTravelTimeXmin with "X" being the travel time expressed in minutes when using the defining the processing order according to the "The population living within a given travel time from the facilities" option
• amRankValues_popDistanceXm with "X" being the distance expressed in meters when using the defining the processing order according to the "The population living within a circular buffer around the facilities" option

This table provides a set of important information that can be used to analyze and improve the performance of the health service delivery network being considered. More specifically, this table identifies the health facilities that:

1. Could cover a larger population within a set travel time (e.g, 2 hours) if their coverage capacity was extended. In our example, this applies to Namadzi Health Centre and Mdeka Health Centre.
2. See their coverage capacity being underutilized because of limited population residing within e.g., 2 hours of travel time. This is the case for all the facilities for which the maximum coverage capacity is not being reached within the maximum considered travel time.
3. See their neighboring population being already covered by other health facilities based on the selected processing order. This is the case of the Nkula Clinic in the present exercise for which none of the coverage capacity has been utilized. This could present evidence in favor of reallocating resources between facilities (on the assumption that patient preferences are indifferent).

 In addition to that, the cumulative geographic coverage reported in the "amPopCoveredPercent" columns provides the percentage of the population covered for the given travel time when taking the coverage capacity of the health facilities into account (in the example above, this percentage amounts to 49.21%), and by subtraction the percentage of the population not covered (50.79%). The cumulative geographic coverage also allows you to evaluate how coverage expands after each iteration in the analysis.

 Zonal Statistics Results

 If you checked the "Generate zonal statistics (select zones layer in data input section)" in the "Analysis settings", then the analysis will generate one last table containing the distribution of geographic coverage obtained at the zone level after conducting the analysis.

 Once opened in Excel, this file

(named “table_zonal_coverage_geographic_analysis_120m.xlsx”) in the present exercise, contains the following columns:  

• cat: unique identifier of the zone, as per the field selected from the attribute table of the zone layer.
• name: name of the zone, as per the field selected from the attribute table of the zone layer
• amPopSum: Total population located in the zone
• amPopCovered: Population being covered through the geographic coverage analysis, therefore attached to the existing health facilities
• amPopCoveredPercent: Percentage of the total population being covered through the geographic coverage analysis

 This table is particularly useful to identify potential inequities in geographic coverage at the sub-national level. In the current exercise, for example, we can observe important disparities in coverage between zones. One of them even has 0% of geographic coverage (North West).