Everything that I will talk about today has been used at the CDC field station in Kenya. We have a computer center and laboratory just outside of Kisumu, near Lake Victoria. Most of our field activities are conducted in the Asembo and Gem locations, approximately 50km to the west of Kisumu.

The countryside there is rural, with most people being either farmers or fishermen. Houses are commonly built in clusters for extended family use.

This is a very good place to study malaria, because there is intense, year-round transmission. Most malaria is p. falciparum. Infants bear the brunt of the morbidity/mortality.

It took a lot of people to collect all of this information. Over 600 people were involved in field data collection.

We’ve recently finished data collection and analysis of a large-scale bednet project that will provide much of the framework for today’s talk.

Study area: 450 sq km, 125,000 people.

This slide lists all of the components of the project.

As you can see, we collected a lot of data.

Multifaceted study: [walk thru headings NOT MOH]

It also took a lot of people to enter and edit the data. Over 50 people worked in the data management section.

Entering and cleaning the data was a very difficult and demanding job.

There were many problems.

Proper storage of forms was another tremendous burden. This is one of several areas used for storage of collected data.

We want to continue to collect a lot of data, but we want to avoid the problems that we have had in the past. So, we are currently exploring several approaches to do this. The GPS/mapping data collection, in place since the 1998 has been using virtually no paper forms at all. We collect longitude, latitude, altitude electronically from the GPS system, and enter ID numbers, home construction and Animal ownership information into a handheld PC (or PDA).

The earlier system used a proprietary DOS-based handheld computer. In early 2002, we began using Windows CE-based handheld computers/PDAs

Which we will focus on today.

As I mentioned earlier, we have completely switched field collection of GPS data to the latest GPS and handheld PC technology. We will now review the components of this system.

The GPS unit is in Maurice’s pocket. (point) But, this is what it looks like. It uses a small cell phone battery. The GPS circuit board is very small. The antenna plugs into a pocket in his cap. (point to both). A serial cable connects the GPS unit to the handheld pc (point).

The GPS unit gives us the location where the GPS unit is. That is not always the location that we really want.

Often it is inconvenient or not good GPS practice to stand next to the object being mapped.

So, we get the location information of where we are standing and use the LASER RANGE FINDER TO:

(1)Measure the distance to what we want to map

(2)Measure the compass direction of what we want to map

(3)Measure the change in altitude of what we want to map.

Once we have this,the GPS software automatically adjusts the GPS reading.

The Solar Battery recharger Keeps PDA batteries charged for all day use, Allowing us to keep the equipment in the field as long as we like. They are inexpensive and work with other devices.

Uses Windows-based Pocket PC 2002 Operating System.

Screen is highly visible in bright sun.

Most Windows-based PDA’s have this feature.

Screen backlighting is automatically adjusted for both indoor and outdoor use. Has both numeric and alphanumeric touch-screen data entry, the keyboard automatically adjusts to the type of data being entered.

On the screen, you can see the numeric keyboard.

Comparing The Old vs. the New

(1)The old GPS unit was the size of a brick & battery pack used required a heavy backpack with 2 very heavy old camcorder batteries that weighed a couple of pounds each.

The new unit fits in a pocket & uses a small cellphone battery. (point)

(2)The antenna was much larger than the new one.

(3)The old Data collection unit was much larger & DOS based. The new one is small, Win CE-based, and runs many other applications (Word, Excel, Outlook, etc).

(4)The new system costs 75% less.

So, you have seen the hardware, but how does the software work? These systems generally have two components: PC software to create applications, to store the master data set And (2) PDA software – to run applications, collect field data, transfer data back to the PC.

The PDA Data Entry Screens can have different kinds of edit checks.

These include Drop-down menus, radio buttons, and range checks.

Data Can be Sent to the PC, edited and returned to the PDA for further updates.

Example: Census files that are updated quarterly.

After the first round of data collection, most of the field work consists of updating records on people found in the first round.

For our GPS system, the software automatically creates a data dictionary. Creates a table for each feature, like a household. Then you decide what variables to collect, like ID numbers, and construction variables. The software allows you to create Menus, other data entry features Databases export directly to GIS with location data fields.

The key to making the system work is having an ID numbering system set up for each household.

The ID is recorded in the location database

Any other study that uses these household ID numbers can then be linked to map locations.

Of course, the software has components to facilitate GPS to PC data transfer, Calibration of GPS readings, and export of data and maps to the GIS software system.

You can also use this system for collecting survey data.

All that is needed is the PDA and the solar battery recharger.

No GPS or Laser unit.

Can use a variety of other database development tools for the Pocket PC

All produce Access-compatible databases.

Can also use Visual Basic extension for Windows CE. Can backup databases to compact flash RAM.

Only the CF card needs to be brought back from the field.

ID files on CF identifies the source of data to the Data Manager.

Software on PDA prevents incorrect CF installation.

ID written on CF identifies card to transport team.

CF card slides into a slot on the PDA.

Solar battery recharging allows unlimited field use. To date, the system has been used in Kenya, Haiti, American Samoa.

We have three GPS technicians.

One base station ($10K).

Three laser units ($3K each). A compass and a tape measure will work, also!

The GPS units are $500 each, the PDA’s are $500 each (with cables), software is $800, the solar battery recharger is $60.

GIS software is $1K to $3K (additional modules).

We can map/census a village in a day or two, depending upon how much information we collect for each household.

First, let’s review what we have mapped:

We’ve mapped approximately 15,000 compounds in Asembo and Gem covering over 450 sq km and 125,000 people.

One type of map takes household-specific information on malaria infections

And uses a smoothing procedure to identify areas of low, medium, or high

Occurrence of malaria infections for a particular time period.

Here is a similar map showing area of high/medium/low mosquito abundance.

It would be difficult to make these assessments using only household information (the dots).

A survey of vaccination coverage was conducted in a group of the villages.

We show the vaccination status for each household (blue dots=no) & use smoothing to show areas of LOW vaccination coverage where followup work needs to be focused. The same type of map could be Created for bednet coverage or for other factors important to a control program.

A cholera outbreak occurred in a group of communities.

Each dot shows where the ill people lived (point).

The color of each dot represents the

Week of onset for the disease in each house (dark=early, light = late).

The smoothing then shows us that the epidemic moved from east (darker) to west (point).

GPS data has been combined with Remote Sensing Data to study vector abundance as a function of distance to water/altitude measures (GPS) and vegetation/ground cover measures (remote sensing).

For childhood mortality, we again found that children in control villages that live within 300m of bednet villages got virtually the same benefit as their nearby neighbors with bednets. Both had roughly a 20% reduction.