Introduction of DHIS2
The initial phase of the DHIS2 started in August 2010 when the U.S. Centers for Disease Control and Prevention (CDC) contacted the University of Oslo in a bid to outsource the electronic HMIS (DHIS2) for Uganda. The University of Oslo then customized the DHIS2 for Uganda, and in January 2011, Uganda adopted the electronic HMIS. Six months later, a technical team comprising staff with background training in information technology, public health, statistics, and monitoring and evaluation attended the DHIS2 training at the East Africa Academy in Dar es Salaam, Tanzania. During the training, the team documented lessons from other countries that had already initiated implementation of DHIS2 to inform its implementation in Uganda. Some of the lessons included the need for: 1) setting up the server environment and customization of the system, 2) piloting the customized system in some districts and performing a national roll-out in a phased manner, 3) training of trainers (ToTs), 4) training of district personnel prior to roll-out, 5) importing data from formerly used systems, and 6) piloting the use of mobile phones in data collection using the DHIS2 . Full-scale implementation of the DHIS2 began in January 2012.
DHIS2 customization and setup
Prior to full-scale implementation of DHIS2, there was a need to customize it to suit the Ugandan environment. This was done by a joint technical team composed of representatives from MoH (some of whom had attended the East Africa Academy training) and the Health Information System Programme (HISP) of the University of Oslo. The customization process took a period of four months and involved: definition of data elements, data sets, dash boards, and designing of data entry forms. The team also generated validation rules to ensure accurate entry of records into the system. Examples of validation rules that were developed included: (i) a rule on the number of pregnant mothers attending antenatal care for the fourth visit which was coded as ‘less than or equal to the number of those who attended the first antenatal visit’ (ANC4 < =ANC1), (ii) a rule on the number of pregnant mothers receiving the second dose of Intermittent Preventive Therapy (IPT2) which was coded as ‘less than or equal to the number of those who received the first Intermittent Preventive Therapy (IPT1)’ (IPT2 < = IPT1); and (iii) a rule that the number of people receiving the third dosage of diphtheria, pertussis and tetanus which was coded ‘less than or equal to the number that received the first dosage’ (DPT3 < = DPT1). In 2011, geographical information system mapping was done for purposes of mapping health facilities and this was customized into the system over a period of four months. On the other hand, HMIS tools for reporting such as outpatient & inpatient reporting forms were developed and customized into the system over the same period.
DHIS2 was initially installed on a testing server for training purposes and later migrated to the production server hosted at MoH central server room. This was aimed at improving ownership and provision of centralized technical support at a single point of control. Districts can access DHIS2 via the internet in the server–client architecture. In anticipation of information and communications technology infrastructural and internet connectivity challenges, the system was set to work in a hybrid mode. With this mode, users can continue to access the system and input data even when there is loss of internet connectivity. Data entered into the system in a hybrid mode can then be uploaded into the main system once internet connectivity is reinstated.
Piloting the system
A few months after the technical team attended the DHIS2 East Africa Academy in Tanzania, MoH took the decision to roll-out DHIS2 to all districts. Training sessions were carried out in four pilot districts in Western Uganda (Kabarole, Kibaale, Kamwenge and Kyenjojo) in January 2012. The selection of the districts was informed by the presence of a project known as Saving Mothers Giving Life (SMGL). The SMGL project had already instituted a demand-driven mechanism for data collection at district level. The MoH then considered it prudent to pilot the system in the districts where the demand for data had been identified, as this would result in a rapid assessment of the performance of the system. The pilot phase was funded by the University of Oslo, but a few months later the MoH received additional funding from a PEPFAR/USAID project known as Monitoring Emergency Plan Progress (MEEPP) which supported the training of district-based staff in PEPFAR-funded districts.
National roll-out and training approach
In preparation for the national roll-out, the country was zoned into 12 sub-regions to ensure that all the health facilities identified were captured in at least one of the sub-regions, so that the sizes of the classes per regional training were manageable. Two people responsible for data management were drawn from each of the 13 regional referral hospitals, 39 general hospitals, and 164 Health Center IV facilities to attend the training. In addition, one person responsible for data management was selected for training from each of the 803 Health Center III facilities. Data managers at all the health facilities were trained using the revised paper-based HMIS, while District Health Officers, District Biostatisticians and HMIS focal persons at districts or health sub-districts were trained on the electronic HMIS (DHIS2), together with staff from district-based implementing partners, surveillance officers, and monitoring and evaluation specialists.
A total of 35 training workshops were organized across all the 112 districts, and each workshop lasted for 5 working days running from 8:00 am-5:00 PM. In total, 972 users were trained. Of these, 168 (17.3%) were Records Assistants, 112 (11.5%) were District Health Officers, 112 (11.5%) were HMIS-Focal Persons (HMIS-FPs), 107 (11%) were District Biostatisticians, while majority (473, 48.7%) included other categories of health workers. All trainings were facilitated by officials from MoH in collaboration with CDC-Uganda and the rapid roll-out was organized with concerted efforts involving different implementing partners. However, by March 2014, the system had been rolled-out up to district level, with the lower level health units continuing to report using the paper-based system.
During the process of rolling out the system to the districts, 53 districts received a donation of computers and their accessories from the Uganda Communications Commission (UCC) through its Rural Community Development Fund project to support effective utilization of the system. The donation included fully networked computers that were connected to the internet with a free annual subscription for each district. In addition to the support from UCC, MoH, with support from CDC, set aside a budget to facilitate monthly renewal of internet access and routine technical support supervision at the districts.
In order to import data that were available before the installation of DHIS2, we imported data for 2011/12 into the DHIS2 system using the DHIS2 import function. These data already existed in management software databases such as EpiInfo, web-enabled databases, and Microsoft Excel.
Indicator data extraction from DHIS2
In order to assess the progress made since the inception of DHIS2, we extracted outpatient (OPD), inpatient (IPD) and several health service coverage indicator data from DHIS2 for all the 112 districts during the period FY2012/13 and compared them with data imported into DHIS2 from databases that existed prior to the installation of DHIS2. Health service coverage indicators whose data were extracted included proportions of: pregnant women attending 4 antenatal care (ANC) sessions, pregnant women delivering in health facilities, one-year old children immunized against measles, pregnant women who completed intermittent preventive therapy (IPT2), human immunodeficiency virus (HIV)-exposed children accessing HIV testing services, and children under one year who were immunized with the third dose of pentavalent vaccine. We extracted data on two indicators of data quality, namely: completeness and timelines. We defined completeness as the proportion of health facility reports submitted to the district divided by the total number expected from the same district while timeliness in reporting was defined as the proportion of reports submitted by the deadline divided by actual reports received [9, 10]. Service usage reporting rates were obtained and compared against MoH’s 2010–2015 Health Sector Strategic Investment Plan (HSSIP) targets. We also reviewed workshop training reports to document challenges encountered in the DHIS2 implementation.