Imagine that a friend calls you on the phone. He just got his driver’s license today and tells you this:
“I am driving and noticed a light is on that looks like a gas pump. What does it mean?”
You might respond like this:
“It means that you are nearly out of gas.”
But if this is so simple, why is this conversation with your new-driver friend likely to include follow up questions-- the threshold for triggering the light, the miles per gallon, when the light was noticed, the nearest gas station and maybe a few others? The problem is that you have explained that the gas level is low, but knowing it is low actually doesn’t answer your friend’s original question: What does it mean?
Well, what does it mean?
Well….it depends. If your friend is calling you while parked at a gasoline station, it means very little. If your friend is far from a gasoline station, but driving a fuel-efficient car and the gas light just came one, it might be ok. If your friend is in the middle of the desert, driving a gas-guzzling ice cream truck and can’t remember when the light came on….
Low gas --> Low oxygen saturation
I thought of this example after a number of conversations with global health organizations interested in scaling oxygen saturation monitoring to identify neonatal hypoxia, avoid neonatal hyperoxia and diagnose pediatric pneumonia. Specifically, we were discussing clinical decision support and how our software, the NoviGuide, might help. I knew NoviGuide could tell a nurse that the 02Sat was low. Let’s start there:
Here is a series of screenshots from NoviGuide 2.0:
In this case NoviGuide has said, paraphrasing here, “You said the baby looks sick. Babies sometimes look sick when their oxygen levels are low. Even though you think this baby is breathing fine, you might be wrong and should check the oxygen” and when I put in a level it said “it is low.” It reminded me to look at the dashboard and then told me how to interpret the light. I believe this will help the nurse, but I still don’t think we know what the 02Sat means.
We need to introduce three wrinkles.
Wrinkle one: How far are you from oxygen? This one isn’t too bad. When NoviGuide is introduced, each health facility enters in its current equipment capabilities. Knowing available resources at each site has two benefits: NoviGuide can help healthcare workers use resources they have at hand and it can inversely recognize situations when a patient needs resources that are not available at the facility. When the latter condition occurs, we could alert the User and the referral center. Using oxygen as an example, something like:
Wrinkle two: What if the child has a low oxygen saturation level AND is already on oxygen? This one is trickier.
Let’s get into the weeds a bit—the oxygen delivery mechanism. Take two newborns—Baby A and Baby B. Both babies have low oxygen levels despite being on 2 liters of nasal cannula oxygen. Baby A is at a facility that only has a nasal cannula. Baby B is at a facility with both nasal cannula and CPAP. We can provide differentiated guidance. For Baby A, we might suggest considering a transfer if hypoxia persists. For Baby B, we might suggest switching from nasal cannula to CPAP. In either case, we would definitely recommend calling the physician and she might then consider additional diagnostics.
What if the baby is already on CPAP and has a low oxygen? Again, call the doctor, but how might we help that doctor (or that nurse if the doctor is far way)? We might provide guidance on troubleshooting the CPAP machine, resizing the mask, increasing the pressure or the oxygen concentration. In other words, oxygen guidance tailored to the oxygen delivery mechanism.
Wrinkle three: Let’s go back to our first case. A 3kg baby, term, sick-appearing, but with no respiratory distress who, when checked, had an 02Sat <88. Hmmm. Common things being common, the most likely explanation is that it is a problem with the lungs. But, if this was the pediatric board exam, this baby might have congenital heart disease. I put this example in intentionally; if the goal is to roll out oxygen saturation monitoring globally, then some of the cases with low oxygen will be cardiac. To make matters even more complicated, some of these cardiac cases will get worse with high oxygen levels. Yes, these are not as common as respiratory cases, but those groups doing global oxygen saturation rollouts need to at least think this through [primum non nocere]. There may even be opportunities to turn this challenge into an opportunity, facilitating the identification of congenital cardiac disease by creating algorithms for children who do not respond to oxygen or seem to get worse despite oxygen.
Ok, so where do we go from here? First, that it takes 1000 words to explain probably means it is a good target for clinical decision support—meaning if it is complicated to explain it will be complicated for a nurse, or health system, to execute. What makes a good target for clinical decision support software? I believe it is when the answer to the question “What is the treatment?” is “It depends.” Hypoxia fits that definition. The treatment might be transfer the baby, start oxygen, switch delivery mechanisms, etc. Second, it might be tempting to say, “Look, I just want to screen children and put them on oxygen-- basic cases.” I think we should resist that temptation: that strategy might deliver the most oxygen, but not prevent the most death. Lastly, oxygen saturation is a great data point. But it is much, much more powerful when combined with other data points. I wrote about some of those data points above—the nearest facility with oxygen, whether the patient is on oxygen already—but that is just scratching the surface. Whether the patient has a fever, is premature, lives in the Andes Mountains, is just visiting the Andes Mountains, has history of asthma . . . the list goes on.
It is nice to know if the oxygen light is low and it is certainly better than no warning light at all. But with clinical decision support software that can tell you what an oxygen saturation level truly means, we don’t need to stop there. Oxygen is an important piece of the puzzle in reducing child mortality; let’s use clinical decision support software to connect it to the other pieces.
Thanks for reading. Want to learn more? Contact us at: email@example.com