Mohammed Dalwai shares his idea for a Mobile Triage App at TEDGlobal 2014. Photo: Ryan Lash/TED

Every day, emergency room workers use triage to prioritize patient care — but exhausted personnel in under-resourced hospitals can easily make deadly errors in diagnostic tests and symptom scoring. South African emergency room doctor Mohammed Dalwai witnessed such avoidable tragedy firsthand while working with Médicins sans Frontières in Pakistan. He resolved never to let it happen again.

Dalwai urged MSF to apply a standard triaging system — the paper-based South African Triage Scale —  in his emergency room in Pakistan. This led to an 86% improvement in successful triaging, and to MSF adopting this standard in emergency rooms around the world. It also led to a big idea for Dalwai. Now, with The Open Medicine Project (TOMPSA), he and his team have made an app that is freely available. They are planning to roll it out across many regions.

Here, Dalwai tells the TED Blog about the app’s development, and its possible future uses — including the ability to track realtime data of disease outbreak.

How did you end up joining Médicins sans Frontières and creating the Mobile Triage App?

I actually always wanted to be a biomechanical engineer! But then I started studying medicine, and fell in love with it after the third year, when I began seeing patients. That was it for me. I finished med school at Stellenbosch University, and afterwards went into rural medicine. I went into the bush to work at Manguzi Hospital, on the border of Mozambique and South Africa.

There, I met an MSF doctor, who told me about the organization. The idea of going into low-resource settings and helping to make an impact in the system appealed to me, and I wanted to experience medicine outside of South Africa. So I went on multiple missions with MSF — to Pakistan, Afghanistan, Libya, Syria, Haiti and Sierra Leone.

It was in Pakistan in 2011, on my first assignment, that I saw patients dying due to incorrect triaging. One day, I lost a patient. A young woman, 22 years old, came in with abdominal pain. She was incorrectly triaged, and she waited for eight hours. She had something called an ectopic pregnancy — a pregnancy outside the uterus — and she was bleeding internally. When I found her, she was barely alive, and we tried everything to resuscitate her. But she died — and it really affected me. She was a woman, she was sidelined, she was put in a corner — no one cared, no one did the triage properly. If she’d been triaged correctly, we would have realized she was pregnant, and we would have prioritized her.

From that day on, I became determined to sort out the triage problem. I was part of a team that implemented the South African Triage Scale in my emergency room, and it was the first time it had been used in an MSF hospital. It was the first time the South African Triage Scale was ever implemented in Southeast Asia.

Villagers from Hhohho, Swaziland, wait outside to get their vitals taken before seeing a physician or dentist. Photo: Air Force Staff Sgt. Lesley Waters

What is the South African Triage Scale?

It’s a paper-based system based on a composite score — including complaints and vital signs — and one of the only triage scales made for the developing world to evaluate both adults and children. It was developed in a small but busy hospital in Cape Town in a low socio-economic area in response to massive patient loads, understaffing and high death rates. It was introduced in 2008, and shown to be effective when implemented.

MSF had never had a standard triage system in place before this. We lobbied hard for change and standardization. They let us try it, and we did a study that showed a successful implementation. It was at that point that MSF realized how valuable it was, and they started implementing it in every emergency center around the world.

But this is not necessarily a one-size-fits-all solution. The South African Triage Scale (SATS), being relatively new, has been tested extensively in South Africa, but not yet rigorously tested outside of the country. I’m now working on my PhD, documenting the SATS’s validity and reliability in other sectors and countries. For the last two years, I’ve been collecting data on the SATS and how it’s been implemented globally. We proved that it works in Pakistan, and we proved that it works in certain African countries. But Afghanistan and Haiti are different. What are those differences, and how can we adapt the system for local circumstances? In Sierra Leone, for instance, there was a massive malaria population, which has lower hemoglobin levels. Because of that, the triage scale wouldn’t pick up certain patients, so we would have to adjust one or two discriminators after research so that the triage scale is more sensitive for these people. Small things like that make a massive difference in patient care.

Why create a mobile app, when it sounds like the paper-based system works very well to correct the possibility of human error?

Even though the SATS works, it still needs to be implemented correctly across a variety of situations, so we need to standardize the format to further avoid human error. Health care workers are trained to various degrees across different countries. One of the easiest ways to standardize things is through technology. When I came home from Pakistan, I discussed my experiences with my friend Yaseen Khan. Together we decided we had to tackle health system problems using technology — and that’s how we formed The Open Medicine Project (TOMPSA).

When you look at the way the nurses or health care workers make mistakes, it’s usually one of two areas: it’s either they don’t understand the discriminator — so the first symptom that the patient comes in with. The paper-based version of the SATS offers no additional information, whereas a mobile app can. They also make mistakes in calculation. In the SATS, the vital signs are all linked to a composite score, and each one is different. So say, for example, you have a heart rate of 98 beats per minute, that’s zero point. If you have a heart rate of 101, that’s one point. It’s easy to make mistakes, and a massive number of errors are happening in that scoring system alone. So digitizing systems offers more information as prompts for medical care depending on the score. Nurses were forgetting to do pregnancy tests, for example.

The app is essentially a digital checklist. Checklists make massive differences in both the airline aviation industry as well as in medicine. You see the same thing with the WHO surgical checklist. It saves lives.

Can you, say, take someone’s blood pressure and have the reading go directly into the app for analysis?

We would love to do that, and that’s where we’re going with this in the future, as more devices become more integrated. At the moment, entering such information has to be manually based because we can’t afford the technology in our public hospitals. But as we move forward, or as we adapt it for the private sector or other markets, those are definitely technologies that can be added.

A nurse at Khayelitsha hospital, Cape Town, using the Mobile Triage App. Photo: Gregor Rohig

The app assumes that everybody’s holding a smartphone of some kind. Is this a barrier for hospitals in developing countries?

The Open Medicine Project has institutionalized it — so we actually put an iPad in hospitals, in the triage area, where all the patients stream past. All the nurses use that iPad. We do have an application that’s freely available, for use on your personal phone. But that’s more for training purposes, for people to get used to the application. The one in hospitals links to a printer. At the moment it prints out a little sticker so that it can integrate into a paper-based system. the next step is to send information to the hospital computer system.

The problem is that within the developing world, you can’t have a fully digital system. Everybody thinks you can, but to be honest with you, you can’t. Everything’s still paper-based, everything’s still analog. You have to accommodate for that in your solution. That’s where your big problems come in with these massive solutions that cost millions of dollars — they’re beautiful and they work really well, but they don’t work for us. It’s Africa. And for many other African countries, the key is: how do you combine the two? How do you mix technology with something that can easily integrate into a system that has already been there for years?

So really all you need for this is an iPad, the app, and a printer. You don’t even need the internet.

Yes. And we want to open source the code soon, so that we can have a massive collaboration of people in hospitals excited to build onto the app to meet their own needs — including finding ways to integrate the app with bigger hospital systems. Our goal is to scale this phenomenally across Africa.

What about MSF? Will they also use the mobile app to replace the paper SATS?

Yes, we hope they will. They can take advantage of the open source aspect as well. As long as developers have a base to work from, they will come up with some really cool ideas. One of the requests that came through was, “Can we have animations on the triage application to show the nurse when the patient is hemorrhaging?” That’s a brilliant idea — you can almost unify triage systems so that nurses from different regions learn a standard system.

Another thing that can be added is emergency guidelines from each country. The app can then become a knowledge base in any hospital that has an iPad. All these ideas are out there. It’s now, “OK, let’s put it out there, let’s get it up — then let’s go wild.”

A look at TOMPSA’s Mobile Triage App. Photo: Makkia da Costa

Is the application already done?

It’s already done, and freely available on the Google Play store and for iPad on the App Store. But that’s a version without the printing function. We’re working on a mobile printing version where it can be made open source, and then we’ll see where to go from there.

You mentioned in your talk at TEDGlobal 2014 the possibility of using the app for epidemiology. Can you explain more?

We set out to do a simple task, but then we started realizing the value that we can actually get from this — the diagnostic capability. You can start improving diagnosis by analyzing simple metrics like heart rate and blood pressure that can be used to calculate a shock index — an idea that a fellow emergency doctor gave us.

We also started looking at helping hospitals be more resource-efficient by analyzing data generated by their triage color system — sort of a surrogate marker for how sick a patient is. So if a patient is red, we know they’re going to use more resources. If they’re green, they’re not so sick, and will not use as many. Now, for the first time, because we have found an electronic format, we can actually analyze resource use. On a Friday night, the hospital has seen ten or eight red patients; on a Wednesday night, you only see one. But the staffing is exactly the same. Shouldn’t they change something to maximize effectiveness?

Taking it a step further, if you monitor certain alerts coming from hospitals in a region, simple algorithms can alert you to, say, a lot of children under five having diarrhea. Is it seasonal variation that’s normal, or is there something else going on? And then you can start looking at almost real-time epidemiology, which has never been done before.

Don’t we already monitor diseases in real time?

Yes, but it is based on mortality reports. If you monitor live trends, you can be proactive rather than reactive, and respond in time, before people die. Is there a water contamination issue, or a food poisoning outbreak? To me, that’s the key to public health, to medicine in the developing world. For too long, we’ve been only reactive. Only after hundreds of people die, we say, “Oh, I think there’s a problem here. Maybe we should go look at it.”

Wouldn’t this be also useful for the developed world?

It could. This is the nice thing — the technology can be translated for a lot of different purposes for different countries. Obviously the triage algorithms will be different in different countries, because there are four major systems in the world. But that can easily be sorted.

So if the triage system in Canada works fine, for example, we’d base the app on that. They have digitized systems already, but I don’t think they’ve ever understood the link to epidemiology. A lot of systems are also fragmented in the developed world. Electronic medical records don’t talk to each other. But if we link up anonymized, secure data to a central place to monitor the health of a nation or a population, epidemiologists could really start understanding and interpreting the data.

Does this mean that currently, epidemiologists only look at numbers of deaths rather than numbers of cases of symptoms coming through?

They do try to look at symptoms. For example, they look at how many patients contract TB per year, how many patients are HIV positive. But they look at diagnoses, not symptoms. That’s not bad, because diagnosis is what you need. But there’s no reliable way of actually recording that information.

For example, with TB notifications in South Africa, you have to fill in a form. You don’t know if that form ever gets to the post office, or to the places that they’re supposed to be, so numbers are underreported. And there’s no verification numbers that come in.

If you digitize the information, and make sure everything is recorded, you can see from which hospital — from which nurse — this information was sent. It’s an amazing verification mechanism. Improving the ability to report means you have better data. Better data means you can make better decisions. Better decisions mean better patient care, more lives saved. Boom.

Mohammed Dalwai provides a medical consultation to one of the 800–1000 migrants and refugees living amongst boats on an abandoned military base on the outskirts of Tripoli. Photo: MSF

Do you anticipate privacy issues?

I think that the biggest privacy issue is about patient data, and it’s an issue that’s always highly considered in our whole process. How do we ensure patient security of their data, and make sure that no one is compromised? In a way, the diagnosis is the sensitive issue. You don’t want to know if someone’s HIV positive, for example. But then, how do you secure the data?

You take certain steps to make sure that the data is secure, you take steps that comply with all the regulations, and then you try and do the best for the most people. Obviously there are going to be hackers that can hack into anything. But if you comply to all the regulations and you make sure that you have strict security, I think it’s important to be able to do that. We must also be careful to balance these issues with the good of what this can achieve.

How far are you prepared to take this vision? Is this something that you want to follow through alone, or would you like it to be fully open source?

It was never just me from the beginning. My co-founder and friend has walked this path with me, with many people joining us along the way. At the end of the day, I would like to see this as one of the triage systems or solutions that we can use across many developing countries, and even, if needed, in the developed world. The idea, for me, is I want to make sure my patient never has to die again. That I will never again experience what happened in Pakistan.

I want to see this kind of system help with the improvement of triage across the developing world. Even if others don’t use our solution, as long as there is a system that works, one that’s been validated, scientifically proven — let’s do this for our patients. Emergency medicine is growing as a field, and more and more people are accessing the health care system through the emergency room. That’s our gateway to hospital, and that’s why we need to make sure the gateway is effective. Triage is a vital component.