Upgrading MoH DHIS2 Version

Hi, Most countries where I work use DHIS2 for their HMIS, and have used DHIS2 for several years. Their current version is several version behind the most recent and therefore they can not take advantage of the benefits in the more recent versions. There may be support among funders/ppartners to assist MOH with upgrading, and I am wondering if there is experience in this group that would help us understand potential barriers (e.g… a reluctance to upgrade) and important steps (e.g. server upgrade, training, etc.) .

Please share any experiences with upgrading DHIS2 for an HMIS. Would be very interested to hear any specific details about: What hurdles need to be addressed? What resources may be needed? What training is required at both MoH-implementation level and end-user level.



Hello, It is true that we cannot instantly upgrade to latest version due to lack of technical expertise, none of the IT have knowledge of server administration and other DHIS2 technical. Moreover, funding constraint to attend training and workshop. Mostly relying on consultant for upgrade. Require server administration, implementation and technical training, mostly importantly funding/training support.

Thank you

1 Like

Dear @John_Painter

Thanks for sharing this important topic, and indeed many countries are using versions behind.

To my findings the barriers to upgrade are not major even when countries have experts and resources to retrain its a challenges.

One major challenge has been having a test/training server where all the testing and training can happen before upgrade. They need to test extensively before they can can upgrade production to eliminate fears. Training is not a major barrier, because its easy for the users carry on with the new versions.

Also if there are custom Apps running on the old versions, its been difficult to get the developers of those Apps to upgrade them, because of the API changes after 2.28.


1 Like

Hi John,
The inability to upgrade to new versions has many reasons and is also a symptom of other issues in a country’s HMIS. Generally someone/funder arrives and ‘sells’ DHIS2 with all the benefits that can be had. BUT it is seldom spelt out, or reinforced that this is only the first step and that ongoing support is required. And as Sangay has pointed out, there is no local expertise to help with upgrading or testing (from Prosper), so the default remains. The other issue is that there is frequently no proper plan for hosting DHIS2, and all it involves. The servers purchased become too small too quickly due frequently to the (gender and age) dis-aggregation demanded, as well as the maximum data sets countries set out to collect. Another issue is that tools like WHO DQ app, Indicator Dictionary and HMIS Search are not installed.
The solution - the understanding that a DHIS2 HMIS is not a one off and needs to be nursed and cared for, Data Integrity Checks done and general database maintenance become part of the PLAN for DHIS2 support. A country also needs to PLAN for the required upgrades at least yearly. But you will seldom see this in any HMIS policy document. There is also some training that may go along with these updates, and how that is done also needs to be considered. Would make for an interesting panel discussion at the next Oslo Academy…
Thank you for starting this discussion


Great and very important question, John. Norah I think is spot on. Any information system needs to have routine support funded and built in to work plans. DHIS2 is no exception. This is, unfortunately, the ‘boring’ stuff that doesn’t excite many people because it doesn’t bring in shiny new things that can be talked about or for which you can get funding. The exception, of course, is upgrading DHIS2 itself because that does now bring in many new features.

At the very least there needs to be an appreciation of the infrastructure layers upon which information products are built. Without this, the whole edifice could come tumbling down. Without sound and continuous investment in the infrastructure, the fun and interesting stuff will always be vulnerable.

I quite like the AeHIN GAPS framework (Governance, Architecture, People, Standards) that Uggowitzer, et. al refer to in their recent paper (Marrying engineering with health policy to bring digital health to scale).

I think GAPS applies to each level of the infrastructure. Infrastructure includes the ICT hardware, the application (software) and the metadata. (btw, the state of metadata in some of the DHIS2 instances I’ve seen recently seem to me to be approaching a data version of the big ball of mud we used to talk about years ago when discussing software).

The WHO guideline on digital interventions for health system strengthening published last week also discusses the essential “foundational layer” around ICT and an enabling environment, similar to GAP – see figure 2 on page xiii of the document.

1 Like

Dear John

In Bangladesh we are running very large DHIS2 instances as National HMIS system and community HMIS system and combined size is more than 330 GB. We also upgrading version with some gap as that’s require filed staff orientation/training and checking all data entry options. Upgrading several version is very challenging.and requires patience and expertise. In HISP Bangladesh, we have expertise and successfully do conversion these databases. We might be able to help you depending on from which version you are converting. You server and database environment is also very important. Please let us know if we could do any help.


Hannan Khan
HISP Bangladesh

1 Like

Dear Hazim

Yes AeHIN GAPS framework is good, but implementation and coordination is a challenge. Most countries we are working, find that there is a significant resource shortage to implement basic systems. The shiny project are very good to pilot with lots of resources, but long term sustainability is where you should look at first. Every country I work with find similar issues. In a low resource setting environment, you have to show success and possibility of success only then you can have government investment. So this is similar question like ‘chicken and egg’.

About the metadata, if you want to start with perfect metadata it will never happen, its a myth. Start what you have in hand and gradually convince the associated people to correct it and correct / cut the shit. Other wise you will end-up with ‘Pilotities’.

Hope you understand reality.



1 Like

Dear Hannan,

Yes, I agree there is a real “chicken and egg” problem here, and of course long-term sustainability is what is needed. Thinking needs to change from implementing “projects” (time-limited) to providing “services” (continuous, not time-bound).

Of course, services always evolve, as do the metadata!


1 Like