The needs of those who directly interact with antibiotics are as important as those who influence policy-making. When consulting those with direct interaction, their input offers more practical and specific suggestions on the design of our device. (Please see indirect and direct stakehave direct influence concerning themholders:"the Talk and the Walk" under global policies for more information)

Dr. Sarah Thompson

Director of Infection Prevention & Control at Sheffield Children’s Hospital.

We interviewed Dr. Sarah Thompson to find out more about the reality of prescribing antibiotics.

The transcript of this interview can be found in the section "Prescribing culture and device accuracy."

Prescribing culture and device accuracy

Question 1. In the healthcare settings that you have worked in, have you or other healthcare staff been asked by patients for antibiotics?

“I’d go as far as saying that anyone who says the answer to that is no is probably not saying the truth… The general population comes into the GP practice and want to go away with something… the feel that they’ve got an infection that they need antibiotics. One of the Key problems is that if you’ve got a cold, how long does a cold normally last? 3,4,5,7 days? ...It’s gonna get better on its own. If you get your GP to give you antibiotics,...you start taking it on the second or third day… you’re gonna start feeling better anyway because you’re gonna be clearing your virus exactly the first time the first couple days you started taking your antibiotics… It reinforces (your thinking) that you need antibiotics…”

We were able to see from a stakeholder’s point of view that misuse of antibiotics exists in the UK. This confirms our human practices research in secondary sources.

Question 2. Given the medical importance of antibiotics, do you think prescribing antibiotics based on clinical symptoms alone is appropriate?

“Initially yes… for hospital patients, you are likely to be able to add more things to (the information for prescription decision), maybe blood test maybe X- Ray etc… But…you’re not gonna be able to confirm for a little while in most cases, so they start giving antibiotics. So I supposed from the remit of what you’re looking at, ultimately anything that can be quicker than culture for confirming and ruling out infections, it would be useful…”

We informed the team that as a general rule, if our device could give results faster than culture, it would be helpful to health care.

Question 3. Ideally, how accurate must the device be in order for it to be of use to inform your diagnosis/prescription?

"...there are patients that clearly need antibiotics and patients that don’t. So you care about the middle group, if we assume without the test, they are gonna get (antibiotics) anyway, if it is sensitive but not brilliantly specific, if you got a negative you can stop the antibiotics. ...if you got a (false) positive, they would get the antibiotics but they would have got it anyway. If you (also) got a false negative, then what can you trust? because you are having both false negatives and positives. Therefore, I think it will be most useful for being a rule-out test. And if that is what you are targeting, what you need is a very good sensitivity. You can have false positives but you have to minimize your false negatives.

"I’d say it’d have to be right majority of the time. You don't wanna be proven wrong and you don't wanna be not giving antibiotics to the people that have (bacterial) infections. As soon as that happens a few times, (our device) it would stop being used. So I would say that you would target sensitivity over specificity. And the false negative would be a problem."

The device should put emphasis on sensitivity rather specificity. Lipocalin-2 as a choice of biomarker is preferable because it is produced in response to a number bacterial infections. False negatives would potentially be more of a problem compared to false positives, as a false negative would allow a bacterial infection to remain untreated.

Device design, time and safety

Question 4. Do you think it would effectively reduce the number of antibiotic mis-prescriptions?

"If it works, yeah. There aren’t many point-of-care tests in microbiology...If there was something that exists that could reliably distinguish between any bacterial infection and say viral infection or no infection. "Then as a concept, it could be extremely useful."

This gives a clear validation to the concept of our device, this time from a stakeholder from “The Walk”. (For more, please see The Talk and The Walk under global policies.)

”...Getting the reliability would be difficult. .. blood culture, when people take blood, quite a few the positive cultures that we get are contaminants from the skin flora. So when they take the culture they are going through the skin, either because the skin wasn’t adequately cleaned or because the technique they used to obtain the sample or maybe they’ve touched something, introducing contamination. Actually, skin flora gets into the culture bottle, and indeed potentially any bottles that you are collecting for your test and if it is detecting any bacteria then it would be positive, not because they got an infection but because they've introduced bacteria into it."

With any test like this, the challenge would be the samples would have to be sterile. If you are detecting the presence of bacteria, there would have to be a sample (with data) that you would ordinarily get if there were no infection.”

Question 5. How much time ideally should the test take for it to be clinically practical?

“As quick as possible.”

Would two hours be ok?

“Probably. So you would be aiming at as short as possible. It kinds of depends on which environment you are targeting. If it is general practice, it has to be quick, because they only have about a 10 minute appointment. Yes it is better to know later that day, but it’s probably gonna be reasonably inconvenient for the GP. If you do a test and that takes two hours, particularly given how many people they see with infections again and again, you’re gonna have to re-contact them.”

“...You can imagine that you are gonna have to say to them that we will let you know if you need antibiotics. But if you assume a number of them will do, you would need to recontact them later that day, they need to come in and get a prescription and take that to the pharmacist, so that’s 3 visits in a day… So the added workload having to re-contact people.. depends on how many tests you do in a day, if it’s something that they would be using frequently then for a GP, waiting for the result means they would use less often.”

But actually in a hospital environment, like an emergency department, depends on the patients, if they come in and you see they are ok, they can go home and have antibiotics, then you’d want a quick test. But actually we keep a number of patients, especially children, in (the ED(emergency department), for a period of time for observation anyway. For example if it’s a urinary tract infection. They won’t let them go until they’ve given a sample and that could take awhile. And sometimes you will wait for the lab result to comeback to count red and white cells. So, there are definitely patients hanging around in the ED for their result. So clearly for a couple of hours would be fine for lots of patients. But people always want things faster. But being able to rule out a bacterial or a viral infection in 2 hours is a lot quicker than you can currently do it. The quicker it is, the more It would reduce a number of empirical prescription…”

We would like to make a device for point-of-care; for this, it needs to be able to give a result in under 10 minutes. Accordingly, we planned to use hemerythrin as a reporter system as using this protein may be considerably faster. If our device can produce a result in under 2 hours, it would be faster than current detection methods. We were made aware that point-of-care is not necessarily the only application for the device and patients would benefit from it even if it takes longer to get the result. If our design can produce results under 2 hours, it would still be quicker than current diagnostics.

Question 6. Our device would contain GM live bacteria. Would it be a problem to using the device in a clinical setting?

”Not inherently. If I was to be judging your project, my expectations from an infectious and control point of view would be that it needs to be easily cleanable from the outside of the device. If it’s gonna move from patient to patient, you need to make sure it doesn’t transmit bacteria. And with GM bacteria, as long as they don’t come out it doesn't matter.”

“You need to make it as easy as possible so people can use it and not make it hazardous, say, some bits pop out. It needs to be a robust and enclosed device where the bacteria cannot contaminate the device and the people using the device.”

“...make sure the bacteria wouldn’t contaminate the sample that you are testing or if the device is contaminated with those bacteria and they get into the sample then the sample you are testing will always be positive.Also, it would have to have a disposable element to it. Otherwise, whatever you are using to get the blood would then be contaminated and you would always get positive.”

The device needs to be easy to use and robust to prevent misuse of the device, which could pose a safety problem. The device should be composed of a cleanable material and any elements that come into contact with blood or bacteria should be readily disposable.

Blood for the device can only be drawn from a pin-prick if the device is to be used GP surgeries and clinicals. Larger volumes of blood can only be drawn in hospitals.