A new tool for an old fight: Using electronic surveillance and alerts to detect sepsis
Historically, sepsis has been one of the most important medical conditions in the United States, yet it is not properly appreciated by many in terms of its real significance. Unlike heart disease and cancer, where most people at least generally know about the tremendous number of deaths and the financial impacts they create, sepsis is somewhat of a mystery to the general public.
According to the National Institute of General Medical Studies, sepsis strikes more than a million Americans every year and between 28 and 50 percent of these people die – far more than the number of U.S. deaths from prostate cancer, breast cancer and AIDS combined. While a great deal of improvement has been made in treating high-profile conditions in the last 10-15 years, sepsis is lagging behind in where it could and should be in terms of care. It costs billions of dollars to the healthcare industry, but by-and-large, care for sepsis in the United States is sub-par.
To learn more about sepsis and its seemingly ambiguous presence within day-to-day healthcare environments, I spoke with Jim O’Brien, MD, MSc, Vice President of Quality and Patient Safety, Ohio Health Riverside Methodist Hospital, an 860-bed community teaching hospital in Columbus, Ohio.
O’Brien is a pulmonary critical care physician whose primary background is within the intensive care unit (ICU), which means that he’s treated a great many patients with sepsis. For the last three years, he has been serving as Vice President of Quality and Patient Safety. He has conducted a significant body of research, mostly around clinical effectiveness, epidemiology and retrospective cohort studies looking at risk factors and poor outcomes for sepsis. He currently serves as Chairman of the Board of Directors for Sepsis Alliance, a not-for-profit organization dedicated to saving lives by raising awareness of sepsis, both in the public and within healthcare, as a medical emergency.
(Editor’s note: To hear audio excerpts of this interview, click on the media player buttons that run throughout this article.)
Free: Please talk about how people view sepsis both inside the industry and outside the industry as well as the inexactness that’s around the diagnosis of sepsis?
O’Brien: First of all, there are challenges from within healthcare relative to sepsis. Most notably is the fact that we don’t have a diagnostic test that tells you whether or not someone has sepsis as opposed to other medical emergencies, like stroke or heart attack, which have pretty predictable characteristics and symptoms. We don’t have those indicators with sepsis. As a result, it’s really difficult for clinicians in determining whether or not sepsis is what’s truly causing problems for their patients.
A patient with sepsis might present with belly pain. Another one might present with shortness of breath. Another one might present with confusion. To try and lump those all into sepsis gets very challenging for clinicians as they’re working to see patients.
From a public perspective, we actually conducted a national awareness survey which we performed with Nielsen. We asked the American public how many have heard the word ‘sepsis’, and only 44 percent has ever heard the word. If a majority of Americans have never heard of this disease, they certainly don’t know to be concerned about it.
Free: What are some of the new tools emerging to help detect sepsis? Do you see help coming from vendors in the near future?
O’Brien: Hopefully, we’re going to get better diagnostic tests in the future that will be able to be incorporated into the medical decision making that clinicians have at the bedside. In the meantime, we are trying to leverage electronic data sources to identify septic patients earlier in the process. We are trying to combine some of those signs and symptoms to put a prompt in front of a clinician to ask, “Is this the diagnosis that you’ve excluded?”
One of the things that I think about are EKGs or electrocardiograms. We have actually gotten to the point where computers, without terribly complicated software, can read the EKG waves and actually suggest that the diagnosis is a heart attack. That diagnosis prompts the clinician then to consider that possible event. They may put it in the context of the clinical situation and say, “No, I don’t think that’s really going on,” but it at least puts it in front of them to decide whether this is the case or not.
By integrating data that is available to us, which includes things like signs, symptoms, vital signs, and lab work, and which is a lot of what platforms like Wolters Kluwer‘s allows us to do, we can have the same computerized assistance in place when dealing with sepsis. We may now integrate them all at once into a single alert or concern, and then also advance the data that goes into those algorithms to allow them to actually even be improved upon with future iterations.
Free: How did your work with Wolters Kluwer begin and then develop over time?
This is not a platform that we’ve tried in any of the hospitals in which I’ve got a day job. Most of that relates to the fact that we were in the process of installing a pretty expansive electronic medical record across 12 hospitals. So, all of our IT bandwidth was consumed in that. But, it’s something that I’ve had touch points with them along the way.
We also created a video a couple of years ago in trying to promote sepsis awareness, because one of the challenges for the healthcare industry is if they come up with a new tool, and I’ve seen this happen in a number of different areas, even if they come up with a new tool that seems to work, it’s very difficult for them to find the market when it comes to sepsis.
If I come up with a new therapy or diagnostic tool for heart attack, I know that I need to talk to the cardiologist who heads up the heart catheterization lab. They’re the people who are going to be able to implement this tool. Most hospitals don’t have a champion related to sepsis. Sepsis Alliance helps raise awareness within hospitals, as well as within the public, that creates a market then for new innovative solutions like POC Advisor.
To learn more about the work Dr. O’Brien mentioned Wolters Kluwer is conducting in the area of sepsis detection, I spoke with Stephen Claypool, MD, Medical Director, POC Advisor, Wolters Kluwer.
Claypool is a physician, board certified in internal medicine and medical informatics. He has been with Wolters Kluwer for about 15 years working in various roles. He and his current team work on various research projects, one of them being POC Advisor, the platform Dr. O’Brien mentioned during our conversation.
Free: I had the opportunity to speak with Dr. O’Brien at Ohio Health Riverside Methodist Hospital about his professional experiences with sepsis. He brought up the major issue of the lack of awareness that surrounds the disease. Why do you think sepsis has flown under the radar for so long?
Claypool: First of all, when some of the big measures were targeted by CMS in the 1990s and in the early 2000s, there was a debate on which conditions to tackle. Sepsis was considered, but it just missed the cut, so there wasn’t a lot of emphasis made. If there had been, I think more attention would have been brought to sepsis, 10-15 years ago. In addition, sepsis is a secondary complication of infection. The infection, the cause of sepsis, is really what’s actually identified as the problem, causing people to lose track of the disease.
For example, if you have pneumonia, and your case of pneumonia is really bad and you get sick and you die, you usually actually die from sepsis as a result. If you die from a bladder infection, or an infected wound, or even appendicitis, you usually are actually dying from sepsis. But that’s not what the family talks about. That’s not what the doctor talks about. They talk about, “This is just a really bad case of pneumonia. It’s gotten out of control,” and that’s what the public knows. They don’t know about sepsis. In fact, physicians don’t very commonly write in the chart ‘sepsis’, they just write the primary condition, so–
Free: It’s almost a battle of semantics.
Claypool: It is. It’s very much a battle of semantics. All the time, I’m having discussions with physicians about whether a patient has sepsis. And they’ll say, “No.” Physicians tend not to call something sepsis until a patient is getting really, really ill, and lands in the ICU, then they’ll finally use the word ‘sepsis.’
In reality, the academic definition of sepsis applies much, much sooner. In fact, most patients that are admitted to the hospital because of infection meet the criteria of sepsis, or they wouldn’t be admitted. That’s one of the reasons we admit people. You can treat somebody with pneumonia as an outpatient, and then there’s a point in time where a physician decides to admit him to the hospital, and it’s usually because they have sepsis, but they won’t call it that. Like I said, they won’t call it that until suddenly the patient is crashing and ends up in the ICU.
Free: Historically, what have been some of the problems with electronic surveillance alerts relative to detecting conditions like sepsis?
Claypool: To date, it has been difficult to have electronic surveillance for sepsis.
As we have mentioned, although it’s been defined, there’s still some grey area in terms of actually deciding which patients have sepsis or not. And there are a lot of other medical conditions that can cause the same type of abnormalities in lab tests and vital signs that sepsis causes. This means that it’s difficult to build an electronic system that’s really accurate as you can run into one of two problems. You can either have a problem where the electronic alerting system is very sensitive, which means it can catch every single case of sepsis, or most of them, but it’s very easy, if you ratchet up the sensitivity, to frequently have false positives. When it does fire an alert, that alert is frequently wrong, and that’s a problem because then staff get really annoyed by these false positive alerts and they don’t trust the symptom and they ignore it and they refuse to use it. Because of that, it’s hard to get systems to really move the needle. You can make the sensitivity less sensitive and try to make it more specific, so that you know when a rule fires, it’s accurate and people will trust it and they will, therefore, like it. But if you do that, the system gets less sensitive and you start missing cases. And if you start missing lots of cases, you’re not going to move the needle and save lives.
It’s hard to build a system that is really accurate. We knew that would be a challenge. We spent a lot of time trying to understand why this is hard, and the main reason that we realized you have inaccurate alerts is because all of these other medical conditions that happen to make patients look like they have sepsis.
Free: Let’s talk about how POC Advisor addresses sepsis detection. How does a hospital staff interact with the solution?
Claypool: Well, let me first say, in order to make a real difference with a condition like sepsis, you first of all need your teams to understand sepsis and to be able to recognize it. They need to understand the importance of treatment and what the treatment should be and they should have processes to act quickly. Then you also need the ability to identify it as early as possible, and quite a few different techniques in informatics can help with all of that process. So, education can help. Things like order sets and protocols can help. Processes to automatically contact certain people, like rapid response teams, can help. You can embed clinical decision support (CDS) in things like order sets and flow charts to give guidance on what should be done. All of that can help.
While there have been issues with other solutions, we believe one of the most impactful forms of CDS that can help is electronic surveillance and electronic alerting. Most of the definition of sepsis is composed of data parameters that can be assessed by a rules engine, and if a patient meets the criteria for sepsis, an engine can send an alert to the appropriate person and call attention to the fact that a patient may have sepsis. The care team evaluates that patient as soon as possible and initiates appropriate treatment. That CDS, then, can also guide them to the right treatment. Since the goal is to identify patients as early as possible and then start treatment as quickly as possible, CDS can really help with those components.
Free: So many organizations have different network infrastructures and policies from one another. How does the solution work within those various environments?
Claypool: With POC Advisor, we are using standard interfaces at hospitals, HL7 interfaces that send data back-and-forth between the EHR, lab and pharmacy plus the HL7 ADT interface for demographics. We pull that same data into our cloud system and we normalize it to our standard, running it in our own rules engine. The facility has to have those interfaces, but they tend to have the majority of them. If they don’t, they know how to build HL7 interfaces.
The second component is we feel really strongly about sending critical alerts to mobile devices at the point of care. Most facilities have their nursing staff carrying around some form of device. We built it so that it will work with pagers, regular phones, smartphones or tablets. As long as they’re carrying something, we can send it to their mobile devices. If they don’t have mobile devices, we can send alerts to a dashboard that is being monitored by a centralized staff member.
Free: Can you tell me a little bit more about Huntsville Hospital, the problems that they were having and how they were addressed?
Free: Since all of this activity is occurring in the cloud, is there not an opportunity for Wolters Kluwer in particular, to look at this gathered information and begin to create best practices for sepsis detection?
Claypool: Analytics is an important part of the opportunity. We have a lot of data. We have a lot of examples of what people are doing from their behavior. You can tie data, outcomes, interaction and intervention together and generate reports. We, in fact, do that already for sepsis. We send out reports showing, whether people are doing what they’re supposed to, and if they’re interpreting something properly, or whether there should be a change in process, but also a change in the system and the interaction.
So there is that big opportunity, and it extends not just to sepsis, but a lot of other conditions.
After speaking with Drs. O’Brien and Claypool, it is obvious that early detection and treatment of sepsis needs to become a higher priority for the U.S. healthcare system. It’s expensive. It’s killing patients.
Luckily, we are making progress. The last couple of years a lot more attention has been directed toward sepsis, but that’s just the starting point. Healthcare systems have to make it a priority. They need to spend energy on it. There has to be education and there have to be the proper processes and protocols put into place. Staffs must take the time to reach agreement on what those processes and protocols should be. Administrators must seek out strong collaborations with other organizations and vendors. In short, this is an industry issues and no single entity can tackle it successfully on its own.
To more closely examine these sorts of organizational considerations and how administrators choose vendor partners, I plan to interview Chuni (Ginny) Kwong, MD, Vice President, Chief Medical Information Officer, Halifax Health, to learn more about how her facility recently decided to utilize POC Advisor to tackle sepsis detection.
This interview will appear online in February.
Trackback from your site.