Hi everyone! I’m exploring existing data collection platforms that are free, open source, and widely used by many organizations/health ministries for routine disease surveillance (not ad-hoc surveillance for use during outbreaks or emergency contexts only). What characteristics should these platforms have? So far, I can think of the following:
Free and open source
Has technical support and is being regularly updated
Can handle large datasets like COVID-19
Easily adaptable to add/remove diseases and syndromes for surveillance
Has the ability to update entries (not single/initial entry only) such as the ability to search for cases and update their laboratory result and case classification
Multiple levels of user access (hospitals/clinics can only see their data, provincial surveillance units can see data for all hospitals under their jurisdiction, national health ministry can see all the data)
So far, I have only seen DHIS2 that fits these criteria. I’ve used ODK-based tools like Kobo and they seem to only collect data one-time with users unable to search and update data unless they go through the dashboard which is a bit cumbersome. I also tried Go.Data and it seems more suited for outbreak investigation/response rather than routine surveillance.
Looking forward to insights from other members here. Cheers!
Regarding DHIS2: it was initially developed for surveillance via aggregate data transfer, and the tracker module (which does take individual patient-level data) was added later. Because of this, there still seem to be some issues to iron out - I haven’t tried this myself but other users tell me that it is difficult to create new forms for example. There are also still some challenges for collecting GIS data using the DHIS2 tracker. In the implementations that I have worked with, the ability to link lab data was not available (but that may be due to the local setup).
Regarding Go.Data: although I agree that it is not suitable for routine surveillance (as that is not what it was designed for) there are some features in the design that I like and think would be useful in a routine disease surveillance system.
I have also looked at LimeSurvey (originally used in Germany but they have switched to something else now) - it is free and can be installed on a local server (you don’t have to use the cloud version) but I found that (at least without some further development) there are some limitations for certain question types.
I would add the following criteria:
Some universal / core variables that would apply across all diseases
Ability to easily create and update or edit data collection forms
Ability to link lab with clinical and epi data in different ways (via bulk upload/variable mapping from a LIMS, or single entry of lab data to an existing case by the lab, or where unique common identifiers are not available, the option to use different fuzzy matching algorithms to bulk upload with a quarantine for non-matches that could be checked later).
Ability to include different algorithms for event detection and event-based surveillance modules
Ability to track actions done after detection of an event / possible outbreak.
Ability to create and modify key performance indicators, easy development of summary dashboards
Database should be easy to query with e.g. SQL
Data export format should be as tidy data - one row = one observation
Data collection possible offline on smart phones or computers with synching when internet available afterwards
Possible to enter GIS coordinates on the spot as well as pre-mapped coordinates for geographic regions
For the most part I think it is worth working with existing systems (such as DHIS2) to extend their capabilities and cover the current gaps. Interoperability with other systems is I think important too. At present with DHIS2, what I’ve noticed is that when specific solutions are applied in one country, there isn’t always good dissemination that that has become available for other countries to apply as well - so something about better communication of what the community has developed already could help.
In addition to what colleagues proposed, kindly look at Research Electronic Data Capture (REDCap). I think the tool attributes or criteria have been answered already.
Hi @patrickbkeating1 ! Thank you for sharing your paper. It was very comprehensive and I discovered many tools which I was not even aware of. The associated app was very helpful especially with the filter functions.
If I may make a suggestion, could a filter option for longitudinal data collection be included perhaps under the special features tab? That could help users narrow down their options if they need a tool for just a one-off data collection like a survey, or something where they need to update certain variables like laboratory results or patient outcomes for a surveillance system.
Thank you for the detailed response, @amy.mikhail !
I have just started the introductory course at the DHIS2 academy to learn more about this platform.
Yes, Go.Data has very interesting features that I hope can be extended for routine surveillance such as the ability to quickly add new diseases/syndromes under surveillance (currently have to add it as an “outbreak”), the laboratory module, as well as the close contact module where contacts can be converted to cases if they meet the case definition. If the system can be expanded for more “routine” purposes, I believe it could be a really viable option.
I checked out LimeSurvey and it seems it is operating on a subscription basis now? I couldn’t find the option for running it on a local or self-hosted server.
I just found out that there an online user community for DHIS2. Is that used for dissemination of specific or new solutions or is each deployment operated in silo?
Hi @ngomanqobile ! Thank you for sharing about REDCap. I have not personally tried it, but I will definitely look into it. How does it compare to other tools like ODK or Kobo Toolbox? Does it have functionality for longitudinal data or repeated data collection, like updating laboratory results or close contacts?
For Go.Data, what I have been doing is storing the settings for diseases that are not actively being surveilled yet as templates - then you can just use the template when an active outbreak / surveillance needs occur. I think the line between large outbreaks and surveillance tends to get kind of blurry anyway - for the setup I was working on for cholera, I was using clusters, events and geography to define outbreaks that needed to be investigated on a more individual level, so effectively it turned into a cholera surveillance system where outbreaks can be defined within it. Clusters are not currently used much by the Go.Data community, so I have shared my use case with the Go.Data team and suggested how they might be more functional. Currently clusters mean a group of people that have contact with each other, so a cluster can only exist if it has an index case and at present it is not possible to create one on the fly in a phone app. You can create events on the fly, however.
For LimeSurvey - the subscription prices are for using their cloud service, and they have another company that provides technical support. However if you don’t want to use their cloud, you can install LimeSurvey on your own computer or on your own server. I don’t think it is as straight-forward to do the server configuration of LimeSurvey as Go.Data, but to be fair I didn’t try with LimeSurvey. I was looking into using it for GI bacteria, and for a trial / pilot we payed for technical support to create the configuration that we put on our own server. The link to download is here: Downloads - LimeSurvey | Open Source Survey Tool
Re the DHIS2 online community, it is the same idea as Go.Data - a place for people to share their solutions that they are applying to their own setup / installation. I haven’t looked at it much, but I think it is largely user driven (as these setups are supposed to be - since the idea of open source is that anyone can develop a solution to something and then make it shareable with others). You might find for example that someone else has created a configuration very similar to what you need, and if they have made the details of how they did it / their code available, then you can just tweak / adapt it to your own context.
Hi @amy.mikhail ! Thank you for the detailed response. Could you elaborate on how you use the clusters and events features of Go.Data? I have not really tried it much. My experience in Go.Data was during the early part of COVID-19 where we used it for ensuring good contact follow-up and viewing transmission chains.
Hi @iancgmd sorry for the long delay in reply! However the new version of Go.Data (just out) allows you more flexibility on creating events, so worth checking out.
My use case was for cholera in Haiti - ‘events’ were (possibly contaminated) water points and clusters were cases identified within the Case Area Targeted Intervention (CATI) approach, usually linked to a single alert. I think these features can be really useful in any kind of geographically wide-spread outbreak that has ‘sub-outbreaks’ within it that need different management or interventions due to how they are being spread.