Going Wireless From stand-alone to web-based Software

Going Wireless

From stand-alone to web-based Software increasingly available as online services rather than installed applications on computers – Gmail, yahoo, googledocs, dropbox, facebook, DHIS 2 Access to data from any online computer This has implications for HIS/IHIA 3

Stand-alone deployment – Hard to manage across many users – How to maintain the data definitions, share data, get access to data etc? – Reinstall deleted software, upgrades, bug-fixes, etc. 4

Challenges of stand-alone implementation § Maintaining servers on LANs distributed around the country is challenging and costly § Power supply interruptions § Workstation problems can be dealt with by local IT companies, but DHIS on the server requires more specialized competence § Even with hardware working 100%, keeping the entire HMIS metadata in synch between so many systems over time is an uphill battle => comparability loss § Software: virus and mal-ware infections, bad security practices (USB-sticks) Each of these factors point to the non-sustainability of a distributed technical architecture and the resulting pressure to centralize 5

Online Deployment o Web browser only requirement o Computer can be reset to fix problems o No data lost in case of disk crash

Great promises of ”cloud computing” Only one installation of the software and database + backups – All changes instantly apply to all users – No need to travel around the country to update and synchronize software and database – Users can get access to peer data for comparison – Capacity to maintain the server is centralized – External experts can be given access to help solve issues But where will the data reside? WARNING: Technical expertise to maintain the server is crucial and needed from day 1 – all users depend on the server running 7

With Patient Data the situation is much more critical! 8

Modes of online deployment Internationally In country MOH hosted Govt. hosted (not MOH) Privately hosted Direct ownership. Data security? National soverginity Within country. Better capacity than Mo. H? Cheaper. National soverginity More robust. 24/7 support. In country. Cheaper Elasticity Need Capacity to keep it robust and secure Bureaucracy between departments, planning cycles Running much the same infrastructure as Mo. H. Might outsource More robust. 24/7 support. Cheaper. Elasticity. Minimal investment up front Other laws apply 9

Three examples Kenya Rwanda Ghana hosting privately abroad hosting within Mo. H hosting at a national private ISP 10

Kenya “Due to poor Internet connectivity and inadequate capacity of the servers at the Ministry of Health headquarters, a reliable central server using cloud computing was set up” “Cloud computing" in this context meant a third part commercial Linux hosting company with its primary site in London, UK. Since Sep 2011 used in all districts (~250) Online using mobile Internet (USB modems) 11

Rwanda Smaller than Kenya. 11 million living within a land area of 25, 000 square kilometers. Approximately 550 registered health facilities spread across 30 districts Original plan in mid-2011 was to follow the Kenya example. MOH e-Health Coordinator intervened - data had to be stored within the country! DHIS 2 was set up within the MOH 12

13

Ghana Scale similar to Kenya. . . 170 districts and 4000 facilities New server specifically for running DHIS Like Kenya, there were perceived difficulties in locating server within MOH Decision was made to physically host the server with a local Accra Internet Service Provider 14

Learnings from the deployment cases Despite challenges, the online deployments are viewed as successful by national stakeholders Reporting rates are high, users are active, data is visible in ways it wasn't before Handover of full control of the servers to the country teams remains an outstanding concern in all cases (Rwanda is furthest along this path) Kenya recently moved their DHIS 2 hosting to Safaricom in Nairobi 15

Managing risks § Data is held by government in trust on behalf of citizens. § Ghana and Kenya both prefers internal hosting. Their choice to outsource hosting is pragmatic. § Centralized data storage has increased dependencies - mobile operators, ISPs, hosting providers, technical support (HISP) § Ghana and Rwanda risk hardware failure § Kenya risk in terms of governance and sovereignty § Outsourcing hardware to the “cloud" can obviate the need for internal technical competence and infrastructure, but generates requirements for new IS management capacity § The storage of patient data raises security challenges 16

Some Concerns Are tradeoffs and total-cost-of-ownership issues understood? Regulatory and policy environment regarding governance of health data is often far behind the technological possibilities. Does it matter if HMIS data is hosted in London or Chicago? Important to have a viable exit strategy with vendors – generally means maintain control of the data (e. g. avoid premium charges or subscriptions required to access own data!) 17

Individual based data – increasing importance Insurance schemes Pregnant mother and child tracking Various m. Health initiatives (programme tracking (TB/HIV) Implications – Civil Registration & Vital Statistics (CRVS) becomes increasingly important – Need for CRVS to speak with HIS/IHIA 18

CRVS: some key functions § Create birth, death, cause of death, live birth statistics § Check routine data for errors and correctness § Integrate CR with other data sources (surveys, sample registration, etc. ) § Unique identifiers § Patient access to their own health data 19