This Community is Archived.

This community is no longer active as of December 2018. Thanks to those who posted here and made this information available to others visiting the site.

CHWs in Malawi; Mobiles facilitate patient care

By Joshua Nesbit | 29 Jul, 2008

My name is Josh Nesbit, and I'm an undergraduate at Stanford. I'm currently at St. Gabriel's Hospital in Central Malawi, implementing a project aimed at expanding the role of CHWs in the catchment area ( The hospital has enrolled over 600 CHWs - all are volunteers. Many of the CHWs are recruited through the hospital's ART program.

I'm here with a laptop, 100 second-hand phones, and an open-source program, FrontlineSMS - which acts as a central SMS hub, coordinating the text-based communications network.

The pilot's been running for five weeks, and we're seeing functionality develop organically from the hospital's and CHWs' needs: the hospital now follows up on distant patients, ART and TB drug adherence is tracked, remote medical emergencies are reported and responded to by a mobile team, CHWs check medication dosages before administering drugs in the villages, communities’ medical questions are fielded, HIV/AIDS support groups effectively organize, and CHWs are alerted of mobile clinics’ and testing services’ schedules. Certain 'keyword' functions allow CHWs to get immediate responses re: drug dosages and proper administration; CHWs also receive additional communication credit through an automated system. Most of the communication, however, is free-form, linking the hospital to remote communities' needs - and connecting the hospital to their patients, via CHWs.

I am working to organize the healthcare data - actions taken by medical staff and CHWs as a result of communication - so that it will be useful to prospective NGOs, clinics, and implementers. Any ideas? I'm on the ground for another two weeks, before I return to Stanford.



Vinay Mohta Replied at 3:16 PM, 29 Jul 2008

I cross-posted a link to this question in the Technology community in case there are some folks there who may have ideas (

Vinay Mohta Replied at 3:25 PM, 29 Jul 2008

Josh, Here's some thoughts from a purely technology perspective:

how much information do you have in terms of the exchange - do you have data on who each text came from, which SMS is an answer to another SMS (some notion of a conversation), and is it correct to presume that the content of the SMS is completely unstructured text (e.g. no formatting of any sort)?

A couple of ideas - you could look at the timestamp of each SMS and group the ones close together into some notion of a "conversation" and characterize how often exchanges happen, do they mostly happen between the same people or do you have fairly broad usage amongst the entire population and so on. You could also do some basic text analysis of the content (e.g. most frequently used noun phrases, etc.) to determine (at a very simple yet useful) level what people are talking about.

What I cannot comment on is how useful NGO's, clinics, and implementers may find useful!

Michael Westerhaus Replied at 5:03 PM, 30 Jul 2008

Hi Josh - I'm a 3rd-year internal medicine resident in the Global Health track at Brigham and Women's Hospital. I spend most of my abroad time in northern Uganda and have also spent time at the PIH sites in Rwanda and Lesotho. My comments don't specifically pertain to organization of your data but are thoughts on the general structure of the program, which I think demands careful attention for the project to succeed. The utilization of community health worker's in this project is excellent as we're all learning more and more about the power of CHW's in delivering excellent care. The one red flag in the project that I think is worth carefully considering is whether the CHW's should be paid or volunteer. For far too long, it has been the status quo that CHW's volunteer, but shouldn't the work of CHW's be valued as a job? Viewed in this manner, utilizing CHW's provide both health and economic benefits to the community. This promotes long-lasting health interventions that don't fizzle when the initial enthusiasm and staff members are gone. Many will argue that this is not sustainable, but perhaps it is actually more sustainable. I realize that some of this may be out of your control, but certainly worth floating the idea as starting these sorts of conversations is what can lead to lasting change.

Best, Mike Westerhaus

Joshua Nesbit Replied at 1:47 PM, 1 Aug 2008

Thanks for the responses - it's unusual and very fortunate to have access to peripheral information while implementing a project.

Vinay - you're right, the SMS data is unstructured. There is a malleable 'form' function in FrontlineSMS that could wind up being very useful in the future, but none of the current data is formatted. I think you've presented some great ideas. There is plenty of data available on the exhanges - incoming/outging messages can read like conversations, if you view a contact's message history. I think that tracking these conversations - and placing them within the CHW hierarchy, to see who's communicating - could be very interesting. I'd already started organizing the SMS data based on the broad subject matter (e.g. request for care, patient update, CHW-to-CHW communication, etc.), but a more rigorous text analysis is a good suggestion. Also, thank you for the cross-post.

Michael - many thanks for the input. Actually, I've been struggling with this issue for quite some time. Since I began floating this idea around, about a year ago, I've come across many different reactions to the paid-worker/volunteer conundrum. I discussed with many advisors the possibility of forging that transition, as an integral part of my project. However, I am working with a very small budget, and this volunteer network has been in place for years. It seems that, as the CHWs take on more responsibility, it makes sense that they would be compensated. While some of the CHWs are linked to the hospital because they receive free ARVs here, no one with a functioning moral compass could justify their treatment as sufficient compensation. However, in terms of the sustainability of the current project, there are real incentives being provided. For example, CHWs are free to receive calls and texts from friends/family on their phone (communication's free, using Celtel's service, for those on the receiving end). Also, the CHWs will be given individualized solar panels, which will be used to charge their phones, other small appliances, and provide light - all, for free. I'm hoping that, as the program expands (and CHWs are performing rapid tests, for example), the CHWs will be solidified as paid healthcare workers. They certainly deserve the pay, given the distances they travel, hours they commit, and their all-important role in rural patient care. I haven't got the budget, and the hospital runs day-to-day operations on donations.

Vinay Mohta Replied at 2:41 PM, 1 Aug 2008

If you need any help or additional pointers on the text analysis
front, let me know. There are several great open-source natural-
language processing (NLP) packages available. I like Alias-i's
LingPipe package - not sure if their open-source / free license will
fit your needs (<>).
Another option that is easy to run "out-of-the-box" is GATE (<

Michael Westerhaus Replied at 12:56 PM, 2 Aug 2008

Hey Josh - Sounds great. It looks like you're putting plenty of thought into it. I realize that the budget restraints can be tough and that there is a great deal of oppositional in the global health "culture" to paying community health workers. Hopefully, if we all continue to push the idea, people will begin to more seriously consider the idea and eventually it will be the norm for a CHW's role to be seen as a paying job.

best of luck on your work,

Vinay Mohta Replied at 12:36 PM, 13 Aug 2008

Hi Josh,

Was wondering how you decided to proceed on organizing the SMS data so that it's useful for NGO's, clinics, implementers?


Joshua Nesbit Replied at 12:04 AM, 17 Aug 2008

Sorry for the delayed response - it's been a hectic week, traveling from Malawi to Stanford.

Mike - you might be interested in a series of video-interviews I'll be posting in the coming days. Before I left the hospital, I captured conversations with four English-speaking volunteers (CHWs) about their experience as health workers in Malawi - a lot of their comments focused on volunteerism. You'll be able to find the videos on the project's blog.

Vinay - many thanks for the helpful suggestions. In terms of making the information useful, I've chosen a more holistic approach (for the first round of analysis). As I stated in a previous post, I've categorized messages by their subject matter. Now, I'm looking at outcomes as a result of the communication (e.g. patient seen by medical staff, correct drug dose administered), and comparing them to what would have occurred absent the SMS system. I've also created a hierarchy of the CHWs and hospital staff, and I'm beginning to chart message frequency between individuals and groups, to see how who is using the network (and in what ways).

Thanks for the open source NLP packages - I may wait for more 'competent' friends/advisors to return to campus to move forward with that sort of analysis.

Sophie Beauvais Replied at 2:36 PM, 18 Aug 2008

Hi Josh,

I'm sure GHDonline members would learn a lot from your video-interviews too. Do you think you could add these as resources in the adherence community too either with the source file or with a link to your blog?

Also, maybe you would want to consider adding some of your working (de-identified) documents such as the hierarchy of CHWs and hospital staff or some of your notes on what happened with the SMS as opposed to without as resources in the "Adherence" community. I'm sure it would be invaluable information to members who would want to start SMS/CHWs projects too.

Many thanks, Sophie

Joshua Nesbit Replied at 1:31 AM, 22 Aug 2008

Hi Sophie and GHD,

Thanks for the ideas. I'll be sure to add the video-interviews, as I upload them.

I'm currently working on a guide, as recent discussions with advisors and partners have shifted to replication and expansion. Once that DIY guide and/or wiki has been created, I wiill be sure to share it with the GHDonline community.

As an aside, I am very open to collaboration. Feel free to post here or contact me ().

Thanks so much,


Joshua Nesbit Replied at 2:16 AM, 7 Oct 2008

Hi Everyone,

I've posted a guide, based on a series of FAQs, outlining recommendations for anyone interested in setting up an SMS network in order to facilitate patient care in a rural setting.

I hope it is a useful resource. If you have any questions whatsoever, feel free to email me.

All the best,


This Community is Archived.

This community is no longer active as of December 2018. Thanks to those who posted here and made this information available to others visiting the site.