The HMIS department in the EHIS directorate is mandated to manage the information systems used by the programs in Ministry of Public Health (MoPH). However, several donors have been supporting the programs to have vertical data management systems, and also population based surveys created a heterogenic environment of information systems. The scattered data management in the MoPH resulted in duplicated /triplicate data elements in the different data collection forms and reporting systems. Subsequently, this environment made the data access and analysis very difficult and cumbersome.
To overcome the heterogenic environment and facilitate D3 (Data Driven Decision making), MoPH come up with the idea of data warehousing. Ministry of Health has chosen DHIS2 as a data warehousing platform and HMIS department as leading of the process. The four (4) data sources selected for integration in the DHIS2 as a pilot included: HR, HMIS (EPI, TB, DDR, Blood-Bank), EMIS (Expenditure Management Information System) and Pharmaceutical.
These four major data sources we have selected includes different nature in terms of types of data which are:
- HMIS: Aggregate data
- HR: individual data
- Pharmaceutical: transactional data
- EMIS (Expenditure Management Information System): transactional data
We intentionally selected these different types of data sources to test the real scenario. So, we started from HMIS first and this was easy work because our HMIS was already in aggregate types of data and the system (database) was modeled in (EAV: Entity Attribute Value) which almost matches the model of DHIS2. So, data elements, data Sets and categories were already clear and defined within the HMIS database. We just developed a custom ETL (Extract Transform and Load) to extract data from our HMIS and load it into DHIS2 so Here with HMIS actually no/little transformation happened.
Having integrated our HMIS system, we went for EMIS. Our EMIS system is a transactional system which records expenses of MoPH and its partners based on unified Chart of Account. So, to facilitate the triangulation of data between data sources, data sharing and data use we identified a list of key indicators/data elements from EMIS and then we developed a procedure within the ETL to aggregate data from the EMIS.
After integrating these two systems, we continued to integrate the other two systems almost with the same methodology. While we went to test the Data warehouse (DHIS2) we faced the below challenges:
- Data elements from HMIS into DHIS2 had a little problem in mapping.
- After loading the data there were discrepancies of data between our HMIS and DHIS2.
- User requirement were not defined very well.
- Capacity of users.
To overcome the above challenges we took the below steps:
- We started to review the elements, give it contextual meaning and remap the elements.
- We established related committees from related departments to:
a. Look at their data in Data Warehouse (DHIS2) and cross match it with the current data source (system).
b. We revised our ETL based on findings.
- We started a series of meetings with focal points of different departments/programs to work on their list of indicators and map it in the system to meet user requirement
- We conducted two types of trainings:
a. Data Warehouse mentorship with all departments and programs
b. Data use workshops for all provinces and users.