Why care about digital information systems? Millions receive good care without them

A community health worker interviewing a client about her family's health.  Most of the notes she jots down never
make it into the program's information system, but she'll remember them.  Gazipur, Bangladesh.

Sure, there's a lot to be gained from collecting information digitally. But it's often expensive, time-consuming, and even risky to implement one.  So it's worth asking: why is it really necessary?

BRAC has been delivering health services for decades in rural Bangladesh.  Typically a community health worker (shashtya shebika) looks after about 200 households surrounding her home.  She visits each one monthly, offering basic health information, products and services.  Once or twice a month she's visited by her supervisor, a more specialized community health worker (shasthya kormi).  During these visits, the pair will follow up on any issues that were identified in the recent household visits and provide check-ups to pregnant mothers and infants.  The supervising community health worker supervises around 10 local community health workers, and rotates between them.

Occasionally government community health workers come to the village, offering immunizations, family planning services, or other resources.  The local community health worker can connect them with the right households, based on her knowledge of their demographics and health needs.

At the end of every month, the specialized community health workers sit down with other local BRAC staff to compile reports for the month.  All of them bring their registers and diaries where they've recorded information throughout the month, for themselves and those that they supervise. Together, they produce a consolidated report for the entire office--the sum of the activities and outcomes for all community health workers and clients served by the local team.  All of this is done on paper. The branch reports are sent as a hard copy to a district office, where the local manager aggregates the reports from all local branch offices into one "district" report, which he enters into excel and emails to the head office.  These reports are used for a variety of purposes, including management, monitoring, planning, and reporting to donors and the government.  I don't want to get into it here, but it's a common question, so I'll just mention it: there are several mechanisms in place to cross-check the data and monitor its quality.

At the central level, BRAC can only see aggregate numbers -- how many pregnant mothers were identified this month?  How many cases of tuberculosis were identified?  It can drill down to the district level (total population 2.5 million), but without a lot of extra compilation (i.e. going back to the branch-level reports on paper), it's tough to get a sense of what an individual community health worker is doing from the system.

Yet if you call the branch manager and ask him about a given community health worker in his area, or a specific client, chances are that he'll know a great deal of information off the top of his head.  In fact, he's probably met the client at least once, so knows her personally. This is what has made BRAC, and other large service organizations working at the grassroots level, able to work effectively without the digital information systems that have now come into fashion -- there is an extensive local information system in the heads of frontline staff, never captured formally, that enables them to do their work effectively.

This is how the system works. We've got a network of over 100,000 community health workers across Bangladesh interacting with millions of households each month.  It's indisputably operating at scale.  And there are many other organizations also reaching thousands, if not millions, that have similar set-ups and also get along just fine.

The immediate reaction of many "experts" when they hear about large-scale, paper-based information system is, "wow, there's no way the program can be good if that's what the formal information system looks like."  But I've seen many of these experts come and eat their words.  The truth is, there are very effective ways to get by without loading all the data into computers, assuming our key requirements, including an extensive grassroots presence and an informal information network, are in place.  Look at the dabba wallahs in Mumbai, where 5,000 workers deliver 200,000 lunches daily across the city. With 100 years of experience, they don't need to "upgrade" to Fedex's information system; they've got their own worked out.

So if effective, large-scale service delivery is possible using a paper-based information system with only aggregate numbers available centrally, where are the potential benefits of a digital information system?  There are four that I see as really exciting.

A community health worker conducts a health forum with expecting mothers in Hazaribagh, Dhaka.  Even though most grow up outside of Dhaka, several mention that they attended BRAC schools and have known BRAC most of their lives.

1) Improving productivity of frontline staff and managers. 

Well-designed information systems make data entry more efficient, therefore reducing the amount of time that staff have to spend consolidating their reports at the end of the month.  This means that they have more time for direct care, training, or other activities.  The system should also give them the ability to better understand their clients, better plan their schedules, and to identify those households that need additional support.

2) Make monitoring much more specific and effective.  

Good monitoring is crucial to keeping service quality high.  Without a centralized system, it's difficult to spot potential troublemakers (branches or people), so you have to resort to less-effective and less-efficient tactics.  With all the data, the monitoring team can start to look for outliers, suspicious patterns, and other trends that they want to investigate.  And predictive analytics takes it up another notch.  One of my favorite examples of data put to work to improve the "hit rate" is from New York City, where a bunch of data geeks helped the city's inspectors go from 13% of follow-up visits resulting in a vacancy warrant, to 70%.  That's a huge improvement in time allocation and results.  The BRAC Water and Sanitation Program has developed a very interesting "participatory monitoring system" that's entirely digital.

3) Systemically reward people for jobs well done, and learn from them.  

In the private sector, many sales teams rely heavily on performance-based rewards.  These are similarly important in many social sector programs with extension workers. With a local information system, this is still possible at the individual employee level, but makes it much harder to implement organization-wide standards or recognition systems.  And it also makes it hard for the organization to make sense of the big picture -- how much money are my best performers making?  Are all services provided in all communities?  Where is activity highest, and where may we need to boost our coverage?  I'm also a big proponent of looking at the positive deviants to capture what they do and incorporate it into training and communication materials.  Having systems that let you look at the work of thousands of community health workers in one place, makes that easier.

4) Do research on the fly.  

When making decisions on policy, priorities, or resource allocations, information truly is power. There's nothing better than being able to ask questions like, "But are adolescent girls buying sanitary napkins?" or "Is the average age of first pregnancy changing?" and having a big arsenal of data to look to for guidance.  All leaders think of these questions, and often investigate them, but the relative cost to get an answer is much lower when you have a good digital system (otherwise, someone is wading through stacks and stacks of paper, probably stored at branches across the country).  So you can get more answers more quickly, while expending fewer resources.  And it's also easier to ask even more questions, generating even more insights, quickly, thereby influencing decision making. Another local organization called Shiree has created a public database on all the households in extreme poverty that they work with, and it's an amazing resource -- not just for them, but for people in other organizations (like me) who are trying to make decisions and can benefit from information.

So there's no reason to freak out about paper-based information systems -- there are a lot of things that they do well, and I'd argue MORE evidence in their favor than the newer digital systems that seem susceptible to "pilot-itis".  If developed and implemented appropriately, going digital offers a lot of scope for improvements in the overall management system.  But chances are, one of the major contributors to successful programs is the deep knowledge that frontline workers have of their communities, and this doesn't depend on the information system.  This will remain a critical success factor even with a new information system.  The real question is: how can some of that local intelligence get into the system to inform decision-making at other levels?