Chotnsf.org



The Identification and Management of Information Problems by Patient-Care Teams in HospitalsPresenter: Alison Murphy, Penn State University Recorded on: April 22, 2015It's ready.>> Okay, great. Hi everyone. My name's Allison Murphy. I'm a PhD candidate from Penn State University. I work with Doctor Madhu Reddy, who's also part of the College of Information Sciences and Technology here at Penn State. This research study also include Doctor Jennifer Kraschnewski, who's from Penn State Hershey Medical Center.So I'm excited to talk to you today about our research study, which is currently ongoing. And it looks at the identification and management of patient information problems during morning rounds. So in order to introduce this topic and provide some context, I'd like to just talk about hospital work in general.So those of us that work in hospitals or do research in hospitals understand that hospital work is highly collaborative. So, individual outpatient care teams are very much dependent on each other to provide information and to make sure that their patient care tasks get done, because they're all dependent on one another.And this includes individuals like physicians, nurses, care coordinators, social workers, therapists, pharmacists, and many other clinical and non-clinical individuals. So, because patient care teams are so highly dependent on each other to provide accurate and updated information. So that they can make these informed decisions about patient care and to make sure that their patient care tasks can be completed effectively.However, there still are problems that occur with patient information, and these just exist in the hospitals. They can also have a negative effect on the patient care experience. However, as we know, they can also just be an inherent part of hospital work. And when I talk about patient information problems, or what I'll refer to as PIP's in this presentation.Is what we define as any issues related to clinical patient information that impact the patient cares team ability to perform work. So this may include a physician who goes into an electronic health record system in order to look at a patient's medication list, for their active medications. And it may be outdated or there may be wrong medications listed in there.So this would be an example of an information problem. Another information problem may be a nurse who's tasked with giving a patient a blood transfusion and perhaps the blood type is not listed. It's missing from the patient's record in the EHR system. And so therefore that would be missing patient information that can cause her additional work and impact her patient care task.So since patient care, I'm sorry, patient information problems are such an important part of hospital work, they just exist and they can have a drastic impact on the patient care teams themselves and the patient. We wanted to look in the background to see what;s currently being studied about patient information problems.And we specifically looked in the domains of medical informatics, which is more practitioner-based, looking at the hospital work itself. As well as Human-computer interaction, which looks at the technical side of designing systems that help users more effectively conduct their work. What we saw in the literature was that there's a lot of studies around the causes of PIP's, which is very important.Because if we can understand what's causing them, we can hopefully help prevent them, or at least mitigate some of the impacts that occur. And since we're looking in a more technical domain, there were really two main causes that were talked about by prior research. The first one was electronic health records system design.So this is simply just that way the systems are designed can sometimes lead to information problems. One example of this would be the fact that sometimes in drop-down values, one example that I saw in my own study, was when physicians are trying to select medication for a patient.And sometime medication names are very long or they include additional information such as the form, whether it's a tablet or a capsule or a liquid. And the string of information is so long that it sometimes gets truncated by the screen. And so the physician selecting that medication can't always see all of the information.So this would then provide for them an information problem and therefore they may actually select the wrong information because of this. The second cause that we saw researchers talk about was the EHR users themselves, which is simply the fact that sometimes people enter the wrong information. Either they enter incorrect information or sometimes they use a copy and paste function when they're trying to make their work more efficient.And they'll copy something from one patient's record Into another patient's record. And then try and change some of the things that are applicable for that record, but not everything is always changed. And then what this does is it creates a persistent problem that gets carried through other patient records.Another example given by the literature with EHR users is that at times just simply due to the sheer amount of work that needs to be done by clinicians, that they're very busy and they don't always have time to immediately update the formal record which in most cases would be the electronic health record.And so it leaves the system outdated for a period of time which impacts anybody else that's relying on that particular information to do their work. The second area that we saw a lot of researchers talk about were the impacts of patient information problems. And these, obviously, when they encounter a problem, it requires a lot of additional work for them to fix it, or at least to figure out if it is in fact wrong information or outdated information.Other times, it can actually lead to confusion about what treatments or procedures were done to patients. And throughout the process of additional work and confusion about treatments, this then leads to our third point, which is delayed patient care decisions. The researchers spoke about how having to work through all these patient information problems, delayed their ability to make decisions about treatment, about diagnosis, or even about discharge.And then finally, if these patient information problems are not even identified. If there's wrong or out-dated information and the person using that information does not realize it's wrong or out-dated. Decisions can be made that can then lead to medical errors that can harm patients. So based on understanding what current research is going on in the field of patient information problems, we have certain research motivations.Certain aspects that really motivate our own research in this project. The first is that we noticed current research focuses on the causes and impacts of PIP's. But there's really no general taxonomy that helps us classify the different types of PIP's. And what we mean by this is, a lot of times they would mention that there was wrong information or inaccurate or something wasn't the same.But we really don't know the entire scope of what are all the different types of information problems occurring and how do we kind of define and neutrally balance those different types. And what this can do is it can help better inform design of systems if we're aware of all of the different types.As well as just make the patient care teams more aware of the types of issues that can occur with their information. And so this leads to our first research objective, which is to create a PIP taxonomy. The second aspect that motivates our research is we saw there was a limited understanding of how PIP's are actually identified and managed.We talked a lot about how they were actually caused, which is important. But the fact is they still exist. There's still going to be a kind if wrong information or outdated or missing information. So we wanted to know how the patient care, who's so highly collaborative and very mobile and tasked with doing a lot of different things at once.When they identify that there actually is an information problem, how do they know that it's an information problem? And then what do they do about it afterwards? So how do they go about managing it. So our second research objective is to describe how PIP's are actually identified and managed so we can better understand that process.And then finally, as we know, due to High Tech Act, especially here in the United States, there's been a rapid implementation of electronic health record systems. So this transition from paper-based patient records to EHR electronic records. And this does, as studies show, change how patient care teams interview and share information.And so we also want to understand the role that technology plays in this. So, how does including a system change the way in which information problems are identified and managed? And can we possibly improve the design of these systems, which is our third objective, to provide improvements for the EHR design.To better help facilitate this collaborative patient care team's ability to identify and manage these information problems. So now that I've talked a little bit about the motivation of our study and our objective, I'll walk through our methodology, which is how we conducted our study. So, our setting for the study was I conducted a qualitative research in an inpatient area of a large teaching hospital here in Pennsylvania, and I shadowed five internal medicine teams during their morning rounds.So here's a list of the 116 individuals that I observed, and I primarily followed around these internal medical teams which had one attending physician and it had a number of medical residents and medical residents as a part of that team. And then throughout the process of morning rounds, their interaction with a lot of other members of the Patient care team including nurses can positions such as individual cardiology or neurology that has the consult on the patient as well as therapists, occupational therapists and physical therapists, care coordinators, social workers and we also have some pharmacists that rounded with our team.So the way in which I collected data as I mentioned it was a qualitative study we employed qualitative methods of shadowing and observation of 29 morning rounds. And it resulted in 155 hours of observations over the course of three months. And just to be specific about the activities that I observed, I went in very early with the residents to watch the hand-off discussions that occurred between the night flow and the day shift.And then, I walked with the residents and interns during their pre-rounds where they visit all of their patients that they're responsible for, to get preliminary information as those patients woke up and their vitals from the nurses. And then we met back In a resident workroom with the attending and then we went our a full round with the entire team, which typically included visiting anywhere between 12 to 16 patients throughout various units within the hospital.I then follow some back to the resident workroom and watch their EHR documentation process. And at times we also went to the emergency department when there were new inpatient admissions to watch that transfer process from the ED to the inpatient area. And then future data collections will also include some semi-structured interviews with participants so we can better understand the patient care team's perception of patient information problems as well.In terms of data analysis, the field study resulted in 280 pages of transcribed field notes and I analyzed this using the six phase thematic analysis by Braun & Clarke. If individuals aren't familiar with this, this is just a very rigorous process of becoming first familiar with your data by reading through it and transcribing it, and then I identified individual codes within our data.So here's some segments that relate to missing information. And then as I developed all these individual codes, I strip them into preliminary themes. So in this case, the preliminary themes ended up being the different communication modes that missing information ended up occurring in. And then as all of these preliminary themes start to evolve, we group them into final themes, which are defined and bounded.And one of our final themes, of course, was missing information which is a type of patient information problems. Now I'd like to talk a little bit about our results and then go into a discussion about how this can impact those in industry as well as research. So the first objective we had was to develop a PIP taxonomy.So we're currently still going through the analysis of the data but here's our taxonomy for now. It was 120 discrete PIPs were identified throughout all of the study data and they were classified into six different types. So here are the types. The first is wrong information, which is simply when information is not accurate and is identified as being not accurate.Second type is missing information. And this was information that the team clearly stated that they expected it to be there, but it was not there. The third type is incomplete where only part of the information is provided. And this frequently occurred when they had consulting physicians that came in to look at a patient for a particular reason.And at times their progress notes only included certain information, but the patient care team was missing other aspects of that information such as do recommended treatment, or diagnosis, or different tests that needed to be run and therefore, they had to follow up with the team. The fourth type is outdated information.And this was classified as information that frequently changes and is no longer valid and has not been updated. This also frequently occurred with nurses. So nurses are typically tasked with, throughout the day gathering vitals of the patient which includes input and output, the blood pressure, the glucose monitoring and any other vitals that need to be monitored for that patient.And since this is done so frequently they typically monitor this and they'll post it in the room using other forms of documentation. But, they don't always update it in the EHR system immediately. And so, sometimes, this is outdated information. But then the clinical team needs to follow up with the nurse on.The fifth type is unclear information, which is what we categorized as there being an uncertainty about the accuracy or meaning of information. So it wasn't that the information was immediately identified as being wrong or incomplete, it just wasn't quite clear what was being said. And then the final type is segregated information.This is where multiple pieces of related information are stored in different locations and are not easily viewed together, and this typically happened in the EHR system, which we also saw in prior literature where if individuals need it for instance they saw that the medication was ordered, but they didn't know who ordered it or why they ordered it.And so, they then had to go into the pharmacy system to look at who ordered it, when it was ordered, and sometimes even have to scan pages and pages of progress notes to understand why that medication was ordered. And just to give you a bit of the data from the field study, I'll give one example of wrong information directly from our field study.So here's a time when our team and our nurse exited a patient's room, and we had a pharmacist rounding with the team. And the pharmacist told the team that the patient was on the medication, Penicillamine, and she questioned whether or not this was correct or whether it was supposed to be Penicillin VK.So the team then gathered around the EHR and looked at the patient's medication list and patient history, and they realized that it was the wrong medication. The patient should've been on Penicillin VK. But this one medication had been carried over in the system from the patient's ED admission, and so they hadn't caught it.So, this was an example of when they identified wrong information. As part of the PIP taxonomy, we also looked at what mode of communication was the information problem identified during. So in this case you can see 69% of the information problems were identified when they were looking at the EHR system itself.Which makes sense since this particular hospital has had an information EHR system implemented for over ten years, and it's their primary source of patient information. However, it's also interesting to point out that a quarter, 25% of the PIPs were identified during verbal communication which was typically face-to-face with individuals or over the phone.And then 6% were also identified when they were looking at their hand-written notes that were on paper-based documentation. So what we can see from this the EHRsystem is where the majority of the PIP's are identified, which further motivates our study to take a look at the design of these EHR system to see if we can better design it to help with the identification and management of PIP.Okay, so now that we're familiar with what some of the PIPs are, I'd like to talk about how these particular patient care team members identified the patient information problems. So there were four ways in which they were identified. The first was when they were encountering conflicting information and this was simply when participants identified a discrepancy between two sources of information.The second was when they were making sense of information. And this is when they just questioned the accuracy of the patient's information based on either their medical training and expertise, or their experience with other patients, or even the specific knowledge of that patient's condition. So, when they were looking at things, something just didn't quite make sense and so they started to question it.The third area was interesting, we're generalizing unreliable information sources. And this is where participants actually associated certain hospital units with having a reputation for being unreliable sources of information. This was frequently spoken about with the emergency department, just due to how fast-paced the emergency department is and how quickly patients come in and out.Sometimes their documentation isn't always reliable, as well as the surgery ward. A lot of times surgeons' progress notes aren't always as detailed as a lot of the internal medicine teams need them to be, and therefore, they view it as somewhat unreliable sources of information. And the last one was simply when they were seeking specific information and they were not able to find the needed information.So once again, for the sake of time I'll just give one example from my field study. So with conflicting information, there was a time when an attending, during morning rounds, was asking a nurse why a patient was getting liquids, when the system clearly stated NPO except meds, which means they they were not supposed to be having liquids.So the nurse said that the resident had told her earlier that morning that the patient could drink liquids, but it was not updated in the system yet. So this was a case where the system was saying one piece of information and the nurse was saying something else. And so they were able to identify that there was an information problem in the system.We then tried to quantify these four different types of the ways in which they were identified, and look at them from the perspective of our taxonomy, our PIP types. And so I'd like to talk about just some of these categories here. The first is not finding needed information.You can see this was 42% of the time they identified a problem when they were seeking information and they couldn't find it, which makes sense. What we can see from this data is that this is only related to missing, incomplete or segregated data. So it does not have to do with wrong, outdated or unclear information.It was also, just a reminder, information that they actually expected to be there. It wasn't just information they knew may or may not be updated yet. They really did think that it was supposed to be there based on their conversation. And what I saw in my qualitative work, when I was observing them, is that this process of not being able to find information that they needed really did delay their work and a lot of the decisions that they had to make for that patient.So a lot of times they had to table their discussion during morning rounds and have to revisit it a couple hours later, when residents were able to follow up with the information that they needed. The second area I'd like to look at in this table is the encountering conflicting information.This was 32% of PIPs were identified this way. And again, based on my field study observations, this frequently occurred when there was communication with other members of the team. So the conflicting information was usually between the system and what somebody else was saying. And this really further exemplifies the importance of collaboration when comanaging patient data.And being able to communicate across, and having effective communication means across, the entire patient care team is so important to identify if there is conflicting information and to go about resolving that information. And once again, from the IT perspective, we want to see how systems can perhaps better facilitate this process.We also took a look and who were the roles that actually identified the PIP. So we'll see here that the residents and interns 41% of the time identified it, attendings 33% of the time. And this is perhaps the limitation of our study and that we're specifically following these clinical teams.So it's very expected that the residents, interns and attendings were identifying these because that was the direct team that I was following that time. But it's also important to note that the nurses did play a critical role. 14% of the time they did identify this. They would proactively come up to the team or call or page the team in order to question them about the accuracy of certain information.And especially when we look at wrong information, which tends to be one of the most critical categories in terms of impacts of wrong information on the patient or the workflow. The nurses actually identified the same amount of wrong information as residents and interns. So this again shows us the critical role that nurses play in identifying wrong information as well as the importance of all these teams really double-checking with each other and communicating any concerns about patient information.Okay, so now that we've talked a little bit about how they were identified, what did they actually do when they found out there was an information problem? And after doing cleaning on our data, we found five different ways in which they did this. They either fixed the problem, which is just when the participant had the knowledge as well as the system rights to go into the EHR and directly fix the PIP.Other times there were times where the participant could not fix the PIP themselves. This was typically due to lack of system rights or that they had to verify information perhaps with an attending or a clinician to make sure certain things were correct. So they then had to track down that person in order for them to fix the problem in the system.The third category was when they just needed to find more information about the PIP. So they had the rights to fix it in the system, but Or that they actually worked around the PIP, which was when they were trying to accomplish an immediate task and they did need that information, but they found a way to kind of just work around it to get their task accomplished anyway.And then ignoring the problem was when, again, they had an immediate task they were working on and it didn't necessarily deal with what they were doing. So they just continued on with their own work and did not address the PIP at that point. And again, this was while I was observing them, which was from the time of handoff from night shift up until their documentation in the afternoon.So I was not with them the entire day. They may have fixed these later on, but at that period of time when they identified it and during the time of morning rounds and documentation, they did not appear to attempt to fix any of these problems. So one of the issues with this, of course, is that it then persists in the system.So once again I'll give one example of this in terms of finding someone to fix the problem. This was a case during morning rounds that a nurse had to call and page our team in order to talk about a patient's Vancomycin order. So the nurse talked to the resident and said that he'd ordered the medication for tomorrow, which was a Sunday, but the patient was being discharged that day, on Saturday.And so the resident was confused and said that he thought he ordered it for today. And actually had a discussion with the nurse about how the system sometimes sets default values for medication administration, and that the default value was set for the next day and he forgot to change it.But he did say that the patient needed that medication administered before discharge, so he went into the system, since he had the rights to change that since he put the order in, and he changed it for the nurse. So again, this is an example of a nurse identifying that problem and then having to find the right resident to fix the problem.So once again, we quantified these according to the different ways in which they managed it and based on the PIP taxonomy of the six different types. So again, we'll take a look here at finding someone to fix the problem and finding more information. This actually accounted for 60% of how they had to manage the PIPs.And all of these times it required that they collaborate and communicate with others. So once again reinforcing the importance of making sure that there are good communication and collaboration tools available to these teams so that they can manage these problems more efficiently. Also, 10% of the time the PIP was not immediately fixed.Therefore, as I mentioned, it then persisted in the system, and this can potentially affect others who are relying on that information being updated in order to do their own path or to make decisions about their patient care path. Now I'd like to talk about what some of these results mean.The first thing we'll talk about is the PIP's impact on workflow. So as we saw in our background, a lot of medical informatics research talks about the impacts, and we wanted to see how our patient information problems that we saw had similar or different impacts on work flow and what we saw in terms of how they manage these PIPs was either that they managed it, meaning they fixed it, found someone to fix it, or found more information before it could be fixed.Or they didn't manage it at all, meaning that they worked around it or they ignored it. And we found that both of these results, these actions had an impact on their workflow. So when they were managing PIPs, it typically led to additional work and the specific additional work that we saw was it took a lot of time to track down other people.Because this was a very large hospital at times they had to physically go to other units or page individuals, and in this case their residents actually didn't have their own mobile phone. So they then needed to find phones that they would page themselves and a lot of times they were busy, so it was this back and forth of trying to track people down all day in order to get information.The second way in which we saw additional work with that it required them to search through a lot of work piece of information. And typically, this is due to a lack of integration of a lot of the information in hospitals. So either it was a different system they had to look in, like the case where if someone wanted to understand why a medication was ordered they had to go in the pharmacy system, or they then had to go to another area of the EHR.And look through pages of pretext that talked about the progress notes and decisions that were being made. So this took a lot of time for them to try in piece together a cohesive view of information that they were looking for. And both of the tracking and searching led to a lot of delaying their own patient care path, which also impacted the patients because sometimes decisions about treatments medication, tests that need to be run, diagnosis decisions, or even discharge were delayed for hours and sometimes a couple of days if they could not find the necessary information.So, the second way in which they manage PIPs was simply by not managing it. So they either found a way to work around it or ignored it. And this then led to persistent problems in what we're calling cascading effects on the entire team. And what we saw on our own research, which matches with a lot of what prior research talked about as well, is it lead to multiple people on the patient care team doing the same work because they didn't realize somebody else did it.Multiple teams ordering the same tests because it wasn't updated in the system. Confusion about the different treatment decisions whether it was medication or tests for procedures that needed to be done for that patient. And this also, from an organizational behavior standpoint, led to a lot of tension and frustration between the units enrolled, which also impacted their ability to communicate with one another.They were frustrated with each other, they tended not to reach out and ask questions, and therefore this kind of hindered the communication between the roles. Which as we saw from the findings, communication and collaboration was really a key part in identifying and managing piece information problems. So we wanted to take a look at how these impacts on workflow can potentially be helped through EHR design, since that's our area and focus in this project.So, we're still currently going through the data and analyzing it, but I like to share with you two of the design recommendations that we're currently discussing in our team. The first is the concept of decreasing data segregation. So we saw a lot of times that on trying to piece together information from a variety of different sources, and a majority of these times to dealt with medication or tests or procedures that were being ordered.And they simply just didn't know who ordered it, or why it was ordered. So we'd like to look at how we can design an EHR that can link standardized fields, such as what medication was ordered, with the free-text field that is associated with that. So, for instance, why was it ordered, and who ordered it, and when was it ordered?And we've seen in prior literature, especially in IT systems used in the medical field, that we can use hover boxes to temporary display information. And this also allows you to copy that information if they need it somewhere on that screen, or just simply for them to be able to view it and then they can close it, so that it's not crowding the screen in any way, it's just simply giving them additional information when they need it.So I'd like to just show you a brief example of what this might look like and again this is just a mock up, but here is an example of a fake medical record. For Buffalo Bill here, and he has amoxicillin prescribed to him. And if I'm a resident and I'd like to know who prescribed this and why they prescribed it, because it's not quite evident just from this part of the medical record, a way in which we can use hover boxes is simply by allowing them to click, perhaps, like click on it and allow them to view the order details.This would then allow a hover box to appear that pulls information and consolidates it from other areas of the system for integrated systems, such as the pharmacy system, to show who ordered it, even a unit that they belong to perhaps their pager numbers, they can be contacted if there's additional information that needs to be talked about.When they ordered it in a time, and even their reason from ordering which could typically comes from their progress note. This then allows them to quickly see it, so that they don't need to waste time going in to other systems and effecting their cohesive understanding of what's going on with that patient.And then, simply be able to make it disappear when their done knowing that information. So the second EHR design recommendation that we're looking at is flagging these patient information problems. So we talked about sometimes they just worked around it or they ignored it, or perhaps were just talking about it and there were sometimes hours before it was able to be fixed.So this could then lead to persistent problems in our system and it can affect other team members. So, simply the act of flagging this information would allow immediate recognition of the PIP. And it also provides a visible notification to other team members so that they know if they’re trying to make a decision on that information, that it’s in question.We’ve even seen that you can provide annotations, which offer explanations for why that information is being questioned as inaccurate. Once again, I'll show you an example of what this might look like on our fake record. So, if I'm a resident and I see that amoxicillin has been prescribed to Buffalo Bill and I don't think it's appropriate anymore, so it's questionable as to why this is still on here, but I'm not quite sure and I need to follow up with other people.Again, an individual might be able to right-click and flag the item. This would then visually display to anybody else, such as the nurse who's responsible for administering that medication, that this particular medication is in question right now. You could even perhaps, add a comment that automatically pulls the user's name and the time and date.And provide a comment as to why this is in question. So if they're not sure that this still needs to be prescribed to the patient, whether it's unnecessary, if their UTI had already cleared, and they no longer need amoxicillin on a daily basis. So if the nurse saw this, she could then follow up with whoever left the comment and say would you like me to still give it to them, or would you like me to hold off.So it's a simple way to help keep the entire team aware during the process of them trying to figure out how to manage that information problem. And it can also help prevent medical errors from occurring or unnecessary medication or procedures from happening. So as mentioned, we're still in the middle of this study.So our next steps for this is to continue with our semi-structured interviews with our participants. And also to better look through the data and come up with more EHR design recommendations like those that we talked about today, in order to provide recommendations to vendors in the design community for how we can better improve the identification and management of these PIPs, the information problems for the entire patient care team.So, thank you very much for listening to my talk, and I'm happy to take any questions or comments.>> Amy Julati, quick question, when you're looking at, can you hear me?>> Yes, I can hear you.>> Okay, when you're looking at kinda enhancements or whatever, do the EMRs kinda have data that times out, or can you set when something becomes stale.So for example a blood pressure, or those things that have to be updated routinely. Do they time-out and flag as invalid?>> From my experience in this particular hospital with this EMR, which is the experience that I have with EMR, it's always listed as active information but typically there is a timestamp.So they are aware with certain type of information, for instance, like blood pressure, input and output, they are aware of when the nurse last enter that information. So, it doesn't necessarily time out. It shows that information, but it does show when it was entered.>> Okay. It seems to me that you'd wanna be able to say, okay, what pressures are good for 12 hours.>> Exactly.>> And at 12 hours, this turns red or it blinks or something, so that it calls our attention to it being stale.>> Yeah.>> Oh>> Yeah, thank you. Any other questions?>> Well, thank you all and thank you very much for this great talk.I really enjoyed it especially cuz of qualitative research, but that's just myself.>> But thank you everyone for tuning in today's webinar. Please feel free to either email Allison or myself any questions you might have. And please make sure to join our next webinar, which will be next April 29th.Thank you all. Goodbye. ................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download