"The laboratory exercises have descriptions of the purpose of the exercise and instructions for using GIS to accomplish the purpose. These are applicable to both U.S. conditions and to developing country conditions. In several of the exercises, however, two different datasets are supplied and the user should choose whichever data set is closest to replicating the data in their own application setting. One data set is for a U.S. area and the other is for a developing country city."

In describing the synthetic data sets, the new "Data sets" button describes this data in this way:

"In learning situations it is often advantageous to use synthetic data. Synthetic data is data generated by simulation methods to conform to known characteristics. Because real (i.e. observed) data on infant births and deaths are expected to have inherent variability when measured for small geographic areas the synthetic data is generated to mimic this expected variability.

 

The reports on infant mortality in New Delhi and Ibadan document the difficulty of acquiring data from vital statistics records that is suitable for examining small-area variation in rates of infant mortality. In the case of New Delhi there are difficulties in assigning birth and death records from the official registers to digititized residential locations. Although this problem is potentially solvable by the development of suitable address-matching algorithms and datasets that record the locations of addresses, this problem has not yet been solved for New Delhi. In the case of Ibadan, Nigeria, research assistants with intimate knowledge of the address records of the city manually assigned vital statistics records to the 122 recognized geographic zones in the city. Records of births and deaths in Ibadan covered a smaller proportion of the total population, however, than in New Delhi. Hence population coverage was a greater problem in Ibadan than in New Delhi.

 

In generating this synthetic data we first obtained the birth and death rates for Delhi from the Annual Report of Health Department, Bureau of Health Intelligence, Health Department, Municipal Corporation of Delhi. We assumed that the infant death rates for slum areas are 30 percent higher than other areas. Using the crude birth rates and the populations of slum and non-slum areas, we projected the number of births in these areas. Using the simulated infant death rates we projected the number of deaths in these areas. The lab exercises (labs 3 and 4) use this data to compute maps of expected infant mortality rates in New Delhi.

 

The co-ordinates of the major slums were determined from the shape file of Delhi. The projected births and deaths in slums were randomly allocated to areas confined by slum boundaries. Other births and deaths similarly were allocated to areas outside the slum boundaries.

The specific parameters used to generate the synthetic data are described in the labs in which this data is used."

 

The instructional materials are on our updated web site at www.uiowa.edu/~gishlth  This site will continue to be revised as the recently received materials from New Delhi and Ibadan are incorporated in it.

Relevant Techniques, Training Modules and Data Sets

The training modules on our web site cover the following GIS-based techniques for computing, spatially analyzing and displaying health-related indicators at a small-area level.

1. Alternative approaches for geo-coding vital statistics data and methods for assessing the geometric accuracy and attribute completeness levels of the geocoding.

2. Mapping types appropriate to the above data characteristics. Measures of uncertainty associated with the levels and geographic entities mapped.

3. Methods for computing density estimates for vital statistics and reportable disease data, where available, (Bithell 1990). Important examples include small-area infant mortality rates, birth rates, fertility rates, age-adjusted death rates, (Rushton and Lolonis 1996; Rushton et al. 1996, Talbot et al. 1999). Methods for visualizing the above computations (ArcView scripts) and for displaying them on geospatial data fields to assist in communicating and interpreting them to the public as well as other concerned agencies.

4. Methods for computing statistical significance given the typically small number of observations in small areas reflecting the common rarity of events per unit of population in such areas, (Gatrell et al. 1996; Gelman and Price 1999; Lolonis and Rushton 1996).

5. Methods for computing geographic access to health services and visualizing the results (Armstrong et al. 1992; Densham 1994; Rushton, 1999).

Indian and Nigerian Participants

For India data sets for the training modules have been generated based on materials sent by staff at the National Institute of Urban Affairs, for selected small areas (poor communities) in New Delhi. The training modules integrated with the spatial databases of poor communities are targeted towards all those agencies and community groups concerned with the planning, delivery and management of basic services and urban poverty alleviation in the country.

In Nigeria, the city of Ibadan was our study area. More details of this site are contained in Dr. Ayeni's final project report which can be seen under the "Reports" button on our web site.

Discussion

It is clear from this collaborative project that the geographic elements of the materials needed to monitor the health of neighborhoods within developing country cities present considerable difficulties to anyone wishing to use GIS methods in health surveillance in these cities. The results of our collaboration show that the raw health records generally exist although the extent of coverage of the population was different between Nigeria and India.

The Indian materials appear to be more complete in their coverage but less easy to geocode to particular localities than the materials from Nigeria. In both cases, however, the difficulties were great though, in the long run, with the further development of basic geographic framework data (national spatial data infrastructure), improvements can be expected.