One of the stated reasons for implementing mHealth projects is that they are more cost effective than the status quo of using paper or nothing at all to support health programming in developing countries.
Yet there is one glaring question that comes forth if we position our work as a cost-effective intervention:
How much do mHealth projects cost?
Thanks to the Malawi Ministry of Health and Population, we now have one answer. They commissioned mHealth 360 Analysis, a mobile health landscape analysis and technical feasibility study based on the 31 mHealth projects officially registered with the Ministry.
First, the 12 registered projects that submitted costing data had an average $158,931 annual cost and ran for 5 years, meaning the average mhealth project had a total cost of $794,657. Then the mobile health projects reached an average of 459 users per year.
Finally, this results in in an average cost per active user per year of $346.26.
What is in $346.26 annual mHealth cost?
While this is an interesting number, what does it really mean? We have very little context to understand how this cost number was calculated, beyond a generic average.
mHealth Cost Drivers
The study claims that software development costs accounts for the largest portion of budgets, an average of $115,076 over 5 years, or 15% of the $794,657 average total mHealth project budget. Motioning and evaluation costs are second at $78,597 or about 10% of the total costs.
Interestingly, the average cost per device per user is $96 with an average of 6% of hardware being replaced each year due to hardware being lost, stolen, or damaged. That 6% replacement rate is lower than the 10% I would’ve expected and contradicts a common perception that mobile phones will be lost or damaged by mHealth project constituents.
mHealth User Types
The study found that 23 out of 31 mHealth projects focused on Health Surveillance Assistants and Community Health Workers, with supervisors, district staff, and project-specific staff in the next largest cohort of users. However, there is no correlation mentioned between the 23 projects focusing on HSAs and CHWs, and the 12 projects with costing data.
The majority of mobile health projects focused on Reproductive Maternal Neonatal and Child Health (RMNCH), and Integrated Management of Childhood Illness (IMCI), and supplying surveillance, promotion, prevention, and curative services.
Is $346.26 Cost Effective mHealth Spend?
In addition to a vague calculation, we are also challenged to understand the context we might use this number to evaluate the cost-effectiveness of a mobile health activity.
Status Quo Cost
Of course the first test would be to understand the cost of not doing an mHealth intervention. One could argue that the status quo costs $0, so a $345.26 mHealth intervention is very expensive compared with an average GNI of $320 per person.
However, what cost would you assign the death or debilitation of a Malawian? If a mHealth intervention improved the lives of 360 Malawians enough to keep them in the workforce, then one could argue that an mHealth intervention is at least break-even.
Alternatives Cost
A better question might be to ask how else the projects could achieve similar outcomes. One might be able to argue that a radio program replace SMS texts to phones with greater cost-effectiveness.
However, surveillance activities might be more difficult without mobile phones, and analog CHW decision support tools and paper-based reporting are much less effective than digital solutions.
Direct Comparison Costs
I can easily argue that this study lumped together several different types of interventions, which confuses potential comparisons.
For example, 6 costed projects used SMS text messages to communicate with constituents, while 8 projects were mobile application based interventions. Just at this level of differentiation, we see a large disparity in costs per users:
- SMS projects averaged $507,247 in total costs and reached 7,772 users for a $65 average cost per user.
- Mobile app projects averaged $1,159,959 in total costs and reached 2,244 users for a $517 average cost per user.
Next, there is a huge goal and impact difference between a Reproductive Maternal Neonatal and Child Health project that focuses on mother education though sending SMS text messages while she is pregnant, and a decision support and activity reporting mobile application for community health workers to monitor a mother and child’s health from conception through the first year of a child’s life.
I would argue that the disparity is so great that one should only make direct comparisons between projects with the same stakeholders and goals, using similar modalities and methodologies.
We Still Do Not Know True mHealth Costs
While I commend the Malawi Ministry of Health and Population for this study, and it builds on related work to find out What Does It Really Cost to Run an ICT for Education Program? I find that both studies raise more questions than they answer.
In fact, there are so many unanswered questions, that I cannot agree with the study’s cost conclusions.
I see the $346.26 average cost per user per year as a vague data point that lacks the needed context to be useful, and I worry that it could be used incorrectly as a benchmark for what a mHealth project should cost.
What do you think? I’d love your thoughts in the comments.
+1 on looking at the cost of status quo!
As the team that worked on the Malawi mHealth Landscape Analysis, we want to respond to the recent ICTWorks post about the cost of mHealth.
The purpose of our mHealth 360 analysis is to develop an inventory of mobile health technology systems currently implemented in Malawi, and provide concrete recommendations for Malawi’s Ministry of Health to assess and evaluate the evidence when formulating policies, standards, and strategies in mhealth.
There is indeed insufficient literature guiding the costing of mHealth applications. Our study produced an estimated cost of mHealth interventions by generating a financial indicator based on a ballpark estimate of resources used for mHealth in Malawi. We understand the limitations of this approach and intend to refine it moving forward. We hope these numbers can move us closer to understanding how much mhealth costs, demonstrate how we can conduct better costing studies, and uncover methods for reducing inefficiencies and demonstrating the return on investment of mhealth.
While this data is insufficiently precise to allow for future budgeting, it offers a snapshot of the situation in the field. As a first step, there is value in identifying how partners are spending in various categories in order to understand the order of magnitude of funding necessary for mHealth as we plan and strategize in the future.
At the recent World Health Assembly, the Malawi ministerial statement stated that “31 different mobile applications under implementation to support health service delivery in Malawi with some of the geographical locations having as many as half of these applications being implemented in the same district reflects a fundamental challenge in the implementation of mHealth in developing countries. It is a clear demonstration of lack of harmonization among implementing partners resulting in duplicate efforts and inefficiencies in use of investments.”
The analysis reveals a project’s average lifespan to be 5 years, coincidentally, most donor funding is on a 5-year timeline. Just as projects are starting to ramp up, the funding dies down. For governments to absorb projects, realistic planning and budgeting for scale must begin from the onset of the project with a clear return on investment demonstrated. We need consistent, timely financial data to make that happen.
We believe there needs to be a strong methodology upfront to understand how plans and expenditures are used and on what inputs they are based. In the current status quo, we don’t validate the total cost of ownership and the budgets, we just know that the money is spent. Currently, the global mHealth community also does not compare actual resource use with actual scope and scale achieved by each project over time, which precludes us from having quality empirical data for making future funding decisions. We need to change that.
While we agree that the value of mHealth interventions are not only monetary, it is challenging to get partners to share their expenditure data. Only 12 of the 31 projects who responded to the national registration were willing to provide cost data and much of what was provided was incomplete. Moving toward a global standard around how best to document true cost for mHealth, then getting everyone to share this information so it can be used to improve how we’re actually paying for services in mHealth — that is the end state that we hope to achieve.
In-depth after-the-fact costing studies are not the way to track costing for mHealth. We need something that is more routine and lightweight. It requires relentless coordination, lots of elbow grease, and consistent openness with data. Using the project data that already exists, expenditure analysis can lead us to the answers of how much mHealth cost.