Data Governance Talking Points: Simple Lessons from the Trenches
Dale Sanders: Thanks to everybody for joining today. Much appreciated, and hopefully will be a great use of your time. We’ll do our best.
Before I share my screen, I’m going to ask Sarah to pop up a couple of polls for some early feedback from the attendees today about their perception of data governance in the healthcare industry. So Sarah, if you want to pop up our data collection tool, that would be great.
Sarah Stokes: Yep, for sure. This is the first poll question that Dale would like to hear from you guys on today. It’s, “Who do you think should chair the data governance committee?” Your options are the CIO, chief medical officer, the COO, the CFO, or other. We’ll give you just a moment to respond.
Great, the votes are really pouring in.
Dale Sanders: Very good.
Sarah Stokes: And they’re still coming.
Okay. It looks like they’ve slowed down, so we’re going to go ahead and close that poll and share the results. It looks like 43% said the CIO, so that’s where we’re getting the majority. 13% said the chief medical officer. 16% said the COO. Only 1% said the CFO, and 28% said other. What do you think about that, Dale?
Dale Sanders: I think that’s interesting, friend. I would agree. My recommendation has always been that the CIO is the natural leader for that. Some CIOs truthfully operate at more the infrastructure level, not at the data level of organizations, and I’ve seen situations in which the CIO leading the function of data governance is not always effective. So it’s really important if the CIO is going to lead that, that they need to be sort of a data enabled and an application enabled personality.
It was interesting to note that the second most popular response was ‘other’ at 28%. So what I would ask of the audience is in the questions panel go ahead and offer your thoughts, and let’s talk about those towards the end of the day about who are other candidates that would make good leaders for sharing data governance in a healthcare setting. I think that would be very interesting. And we’re already starting to see those pour in. Okay. Thanks everybody for doing that. We’ll take a look at those in greater detail towards the end of the webinar.
Okay. We’ve got some interesting data there. We’re going to pop up one more poll to collect some feedback from the audience. Sarah, if you would bring that up, that would be great.
Sarah Stokes: Yep.
All right. Here’s our second poll question for everyone. Why does data governance struggle in healthcare? Your options are still growing as a data governance industry, data is not perceived as a strategic asset by C levels, too many other priorities, wrong data governance leadership and strategy, or other. Again, we’ll give you a few moments to respond.
Dale Sanders: And of course, the scientific pollsters will criticize me for leading the audience with that question, as if there is a struggle. But I think most of us would agree that healthcare struggles with data governance. But again, they’ll be [inaudible 00:06:21] to talk about that with the audience. We’ll leave time for some of that dialogue towards the end of the webinar.
Sarah Stokes: All right. And people are still voting, so I’m just leaving it for one more second. Okay. It looks like it’s tapering off here, so we’re going to close the poll and share the results.
We had a pretty even spread here.
Dale Sanders: Wow.
Sarah Stokes: 23% reported still growing as a data and digital industry. 28% reported data is not received as a strategic asset by C levels, so that had the slight majority. 24% said too many other priorities. 23% said wrong data governance leadership and strategy. And 1% only this time said other.
Dale Sanders: Wow.
I’m not sure what we do with that data. That would suggest that we have a fair number of issues to address in order to reduce the struggle.
Sarah Stokes: Most definitely.
Dale Sanders: Let me absorb this just a little bit. Still growing, that makes a lot of sense. Not perceived as valuable, yeah. All right. Let’s come back to these later. That’s very interesting. I think I would have expected more skewed results to either too many other priorities, which is kind of an indication that it’s not a high level priority for C levels yet. Yeah, anyway. All right. Thanks everyone, very informative.
Okay. Let’s go into this. This is a little unusual. I’m going to follow the talking paper that I’ve used to facilitate these discussions directly. What I’m trying to do is mimic exactly the conversations that I’ve had with clients of Health Catalyst as well as a couple of organizations that are not official clients at Health Catalyst. Our feedback about this talking paper in this style was positive, so I’m just trying to recreate that today. So, let’s kick that off.
I started off by mentioning that … Thank you. Chris is reminding me to share my screen. There we go.
I started off by mentioning that technology of data is important, but data governance success is often more about the soft, human side of data, and the leadership around data. A lot of times I think we try to apply technology as a tool for data governance. In fact, there are some very expensive tools on the market that portray themselves as enablers of data governance. But the reality is, I think success or failure in data governance is more about leadership and soft skills. You’ll see me and hear me talking about that as a common theme in today’s discussion.
All right. One of the first things that I think we have to realize in healthcare as we enter into the notion of data governance is that we’re missing data in healthcare, and the data we have is poor. I think we all generally agree that 80% of healthcare outcomes are attributable to events outside the four walls of healthcare. It’s what I call the dark zone of data, the dark matter of data sometimes. In the universe, we know that dark matter has an influence on galaxies and other astronomical events and entities, but we can’t quite put our finger on what that dark matter is doing, and it’s the same sort of thing that I see in data in healthcare.
I think what that suggests is that when we engage with clinicians and we pretend as if we have all the data that we need to understand healthcare, we wrong about that. And I’ll talk a little bit more about that later. The other is from a data governance perspective, we need to acquire the data to more clearly understand what we’re doing. Data acquisition is a critically important part of every data governance function. When I go around the country, the blessing of my current role is I get to see a lot of different clients as well as non-clients in the industry, in this space. One of the chronically missing parts of the data governance function is a strategic data acquisition roadmap.
Sarah Stokes: Hey, Dale. Sorry to interrupt. Could you just zoom in just a titch on that paper. We’re having a few requests for that.
Dale Sanders: Yeah. You bet, friend. You bet. Thank you.
Sarah Stokes: Thank you.
Dale Sanders: Is that getting better?
Sarah Stokes: Yes, much better. Thank you. That’s perfect.
Dale Sanders: I do want to talk about that with the audience as we progress through the discussion today. One of the most significant missing pieces of the data governance function in healthcare is a data acquisition roadmap to round out our understanding of patients as well as the healthcare delivery process.
Point three here, executive sponsorship. Of course, it’s a worn out cliché, but it will stay a cliché until this changes, and we have to have executive sponsorship and appreciation that data is our new asset of the future that will outlive all of us. I joked with full truth that the oldest living thing on most campuses now is actually data. We now have data that goes back 30, 40, 50 years. A lot of facilities have turned over in that timeframe. A lot of people have turned over in that timeframe, but the data that we have about healthcare now is becoming the longest living asset that we have, and so we need to apply traditional asset management to the concept of data and maximize its value to the organization.
I advocate that the triple aim of data governance is data quality, data literacy, and data utilization. I define data quality as completeness times validity. That gets back to the completeness factor. It gets back to the discussion earlier that most of what effects healthcare outcomes falls outside the traditional four walls of healthcare delivery. If we’re going to take ownership in a population health perspective, we have to increase the completeness of our data and move out into the community with social determinants of health as well as bio-integrated sensors, so that we’re sampling the data about our patients and their lives on a more frequent bases, and therefore increasing the completeness of the data that we have about them. And of course, the validity of data is also very important in healthcare and validating the data. The equation has the encompass both variables, completeness and validity.
Validity is a much harder thing to validate. Sorry for the recursive pun. But that’s just the truth of it. Validity is a much harder thing to validating healthcare. Completeness can be measured computably, and that gets back to the construction of a data acquisition roadmap as a part of the data governance function. And of course, putting in place the tools, and the processes, and the data stewards to ensure that the validity of the data that we are collecting, and analyzing, and utilizing for decisions is maximized, and it’s a combination of computable events as well as subjective assessments of validity. That’s probably worth a whole separate topic to talk about what data quality is, and how you measure completeness, and how you measure validity within the context of a data governance strategy.
Data literacy, of course, is the utilization of data. And do we have within the workforce the skills, and ability, and the willingness to use data to improve our decisions. And do we understand the pros and cons of using data in different contexts? And the appropriate use of data as well as the misappropriate use of data is also important.
The data governance function has to progress the notion of data literacy in the organization through hiring practices, through changes in job descriptions, through education. Take the same approach that we would to normal definitions of literacy, reading literacy, and apply that to data. You need a strategy as a data governance organization to address that, and how you’re going to achieve what levels of literacy, and what the expectations are of the workforce at every level of the workforce when it comes to the utilization of data.
And then, of course, assuming that you have data in place that’s high quality, reasonable quality, understandable quality. You have a literate workforce when it comes to data, then it’s important that data be utilized on an ongoing basis in the proper way. I’ll talk about proper and improper uses of data later on, but it’s again back to the human side of data. I think for the most part we are misappropriating the use of data in healthcare to the detriment of certainly our clinicians who are burned out from being over-measured by metrics that have little or no clinical validity in particular.
It’s kind of interesting. Our drive to become more data driven in healthcare is a good thing, but the way that we’re going about it, I believe, is significantly off-base, and it’s having a very negative impact on the culture, and especially the moral and the welfare of clinicians, physicians in particular.
All right. Now we’ve kind of talked about the triple aim of data governance. I want to talk about some of the vital signs that I see when I encounter what I believe is a healthy data governance function.
It’s really important to define the values and the principles of your data governance first. That is maximize the use of value … or the value in the use of data to provide personalized care at the lowest cost possible. Then focus on your org chart, your leadership, and processes. So, define the values and the principles of data governance. What are you trying to achieve from a values and principles level, and then start building out the org chart, the leadership, and the data governance processes. But if you don’t have these governing values and principles to begin with, it’s I think a fool’s errand to try to organize committees, org charts, and processes without the guidance of those values and principles.
Here’s a good example of something that organizations don’t handle well most of the time when it comes to the definition of principles. And that is the tension between information security and information utilization. You need to understand and you need to govern that tension. I would suggest that most organizations in healthcare are inappropriately limiting the use of data out of misplaced fears of HIPAA violations. I believe that these misplaced fears of HIPAA violations are slowing down data utilization, slowing down the benefit of data to patient care, and also standing in the way of innovation.
One of the ways that I address this is to help organizations decide where on the continuum of data security and data utilization do you want to reside. It boils down to an understanding in the culture of leadership that’s risk tolerant around the use of data. More conservative organizations will put a lot of, I would say, misplaced barriers around the use of data, around the exposure of data.
More risk tolerant organizations, and I would put places like Intermountain, and Geisinger, Allina, Mission Healthcare, I would put them in category of being more risk tolerant when it comes to the utilization of data because they realize that underutilized data is denying its value to the mission of healthcare, and so they tend to have a more risk tolerant, or I would say, a more risk informed approach to healthcare data governance and utilization as a result.
I encourage people to be more risk tolerant in this regard. Because the interesting thing is, most of our HIPAA violations of significance in healthcare don’t come from inappropriate data governance processes. They come from inappropriate technical safeguards around our data. So, close those barn doors first before you close the barn doors of utilization because it’s those open doors of utilization that lead to better care and lower costs for patients, and less frustration for researchers, less frustration for clinicians.
One of the things I joke about is that sometimes I wish I could take data governance teams on a mountain climbing expedition and help them become more comfortable with calculated risk. You certainly can do a lot of things with proper training and equipment to mitigate risk, but still realize the value and the fun of mountain climbing. So, encouraging risk tolerance is an important part of what I try to advocate.
Another thing that I advocate is picking four to five demonstration projects at the early phases of data governance to exercise your relationships, your processes, and the mission of data governance.
By the way, I want to give credit to my friends and colleagues in the Alberta health environment. I’ve worked with them for a number of years on a very large provincial project about creating a provincial wide resource for everything from personalized care analytics to population health analytics. The project is still going on. It’s an incredibly complicated project to coordinate all that at the provincial level, but it’s not technically complicated. It’s organizationally complicated because of the number of parties that were involved in the development of that provincial resource. And to that team’s credit, we developed four demonstration projects associated with that, that proved to be invaluable to the development of the data governance processes and muscle power associated with that very complicated project.
Some of those … And by the way, the projects range from releasing data to the public that was de-identified, that you could argue the public in the province should have always had access to, and just being more comfortable being transparent with existing de-identified data. To more specific projects such as the code development across many different organizations and the development of patient registries that included clinical information as well social determinants of health, so that a diabetic patient from a clinical perspective was in the same cohort of patients from a social determinants of health perspective. So, patient registry is not defined by just the typical criteria of clinical definitions, but also including dimensions of social determinants of health in that as well.
Anyway, I’m a strong advocate of those four to five demonstration projects as a way to build muscle memory in the process. And also, the important thing is building working relationships to all the people involved. That’s critically important.
Another item here is the soft side of being data driven. Those of you that have heard me speak and opine about this topic, this won’t be the first time you’ve heard me mention it. Daniel Pink wrote a very good book a few years ago about what motivates human beings. It’s kind of a modern version Maslow’s Hierarchy.
It boils down to every human’s desire to pursue mastery, autonomy, and purpose. That is, are you mastering a skill, and you feel good about that skill? Are you in a situation that allows you to build the mastery of a skill that you can be proud of? Do you operate within an environment that’s essentially autonomous? We all want to be involved in supportive and nurturing environments, but for the most part we want to be left alone. We don’t want to be micromanaged. We don’t want to depend on other people for our success and for the mastery of the skill that we’re pursuing. And then finally, do you feel as if you’re attached to a sense of purpose that’s greater than yourself? Are you a member of a team or a mission that has a purpose that’s greater than yourself? Can you feel good about what you’re doing in the larger context of humanity and the planet? I would argue that, in the case of physicians especially, our industry is applying data in a way that diminishes their sense of each.
If you look at what we’re doing to physicians with our pursuit of being a data driven industry, I think we’re taking away from their mastery, autonomy, and purpose in every dimension. So in the data governance process, I think you have to develop a strategy for engaging the organization with data that answers the question, “What can we do to increase your sense of mastery, autonomy, and purpose in the job that you hold with the data that we can provide?” And if we can’t provide the data that you need to have a greater sense of those three, what’s the data that you need? And how can we provide that?
If I were still a practicing CIO/CAO, I would have a conversation with my clinicians along the lines of something like this. I would say, “Look. I know that we have hundreds if not thousands of external performance measures that you’re bombarded with every day. It’s the nature of the industry that we’re in right now. We have to fulfill those reporting requirements for the sake of financial success and financial viability. We get that. We’re going to do what we can to protect you from that deluge of measures that’s sucking the life out of your mastery, autonomy, and purpose. We’re going to do our best to protect you from that. In the meantime, we want to provide you with whatever data in whatever form you need that will help you feel better about your job.”
And I think that’s an important part of data governance and a data governance strategy is to know how to engage, not just with clinicians, but especially with physicians, but also administrators, general day to do managers in a way that they feel better about what they’re doing in these three dimensions of human motivation.
Finally in this area, the softer side of things, is endorsement from the CEO that data as a critical asset is really important. They need to send and reinforce the message in their behavior. I’ve actually written emails and developed what amounts to marketing campaigns for CEOs to send out and share. Messaging along the lines of what I have here. And that is, from the CEO, “We are a data driven culture. Digital health is coming. We trail almost all industries in our use of data to improve our mission. Our data governance committee is responsible for helping us ensure that we collect, utilize, and acquire data as a strategic asset to help improve the health of our patients, lower our costs, and improve margins.” Simple statements like that, and then the behavior and the prioritization that backs up that statement is a critical part of engaging from the very beginning.
So, you certainly don’t want your CEO to be involved in the nitty-gritty details of data governance. In fact, that’s one of the reasons that data governance fails is you engage executives in a level of detail that they don’t have time to manage, and then they check out of the process altogether. But what you do want is that CEO to play a part in defining the values and the principles, and reinforce the importance of the data governance committee to everyone in the organization that this is a new era in healthcare.
I believe that the person leading the data governance effort has to have a solid vision and expertise for leveraging data. That’s one reason that I hedged. Sometimes CIOs are great at that. Sometimes CIOs are not great at that. And being a former CIO, I have seen both, and I’ve watched my career progress, somewhat by accident, but also by design. I was essentially an infrastructure-based CIO for a long time in my position in the Air Force as an officer.
But I soon recognized that it wasn’t the infrastructure that really mattered to senior decision makers and to the mission that I wanted to be associated with. It was important, but it was the data that was important to senior decision makers. It was the data that was important to the mission of the military that I wanted to be associated. And so I did a very deliberate pivot in my career to start operating more at the data and the application levels of the technology stack, and less so at the infrastructure level.
So for those CIOs in the audience who are kind of wondering about their role in the process, it’s entirely within your purview to make that pivot in your career, if you haven’t done so already. The interesting thing is the public cloud is making the infrastructure more and more a commodity anyway. In the old days of being a CIO, it was quite complicated to keep that infrastructure running, and the skillset was very important, but that skillset is more and more a commodity. So I highly encourage CIOs to step into this role because it can be very satisfying, but also because you’re the natural leaders for this. You’re the horizontal, information aware, and data aware skillset that can tie all of this together, so I highly encourage CIOs to step into that role.
What’s interesting is I see a lot of organizations starting to appoint chief data officers and chief analytics officers, and I think that’s a symptom of a problem. I think the problem is the organization’s CIO hasn’t filled that need, and so the natural tendency is then to create a new role that operates at the data level and maybe the analytics level.
These folks who lead the data governance function have to have the skills that motivate people around the vision of data. They have to be respected in the organization for their skills, and they have to be a motivational leader. We’ll talk a little bit more about this, but there’s nothing more political than data in any organization. There’s turf wars around data. There’s paranoia about sharing data. There’s paranoia about misuse of data. Data is the most complicated part of the IT stack, I believe. And so the person that leads that, the people that lead the data vision in the organization need to be respected. They need to have diplomatic skills, and they need to have motivational skills as well. So, very important. It’s a leadership issue.
They realize that also data governance is just a means to an end. It’s not the ends in itself. A lot of times what I see is consultants will come in who kind of live and breathe the data governance message, but they turn data governance into the end state.
Dale Sanders: Hornets message, but they turn data governance into the end state, but it’s really just a means to an end. It’s not an in state, and so in the extremes that I see data governance operating, either too little or too much. Most of the time there’s been someone, who is over applying data governance processes to the point of detriment to the organization, and just remember is kind of back to governance in general, right?
The founding fathers of the United States had a philosophy, that government to the least extent necessary for the greatest common good. Now that’s not always an easy thing to define the greatest common good, but we certainly have the ability. I think to recognize when we’ve gone too far. So, that’s … and I’ll talk a little bit more about this player. The extremes that I see the organization, is operating in is either too little data governance or too much, and it’s not the end goal. Data governance is not the end goal. What we’re shooting for is the maximization of a data as a tool to improve health care, and the lives of patients and clinicians.
Okay, effective data governance is like cybersecurity, when it’s working, it’s ambient. You actually don’t notice it, when it’s not working, you need a process for reacting and resolving. So, rather than making data governance an enormously important, kind of front facing issue in organizations… What I generally try to encourage is, let’s send out a message from the CEO of the data governance is really important. We’re going to become more data centric. We’re going to put in place today to governance committee, but then just let it become more of an ambient process. Letting people know that you’ve got to data governance function in place.
These are the values and the principles around which it operates, and if there’s a need to resolve a problem around data quality, data utilization, data literacy, you now have a governance body in place that you can appeal to for help. So, it’s a delicate balance, just like cybersecurity. You basically want it quiet and ambient in the organization, and then a process. A leadership structure in place so that when it’s not working, people know where to go to resolve the issue.
One of the important things, back to the earlier statement, data governance cultures that are successful don’t look for something to govern. They wait for something to govern to come to them. A lot of times consultants will come into organizations, and immediately start looking around for issues to govern, and what that creates is governance for the sake of governance.
The things that need to be governed, from a data governance perspective are usually quite obvious. You don’t have to look very hard for them, they’ll usually. If you put in place a data governance committee, and you charter and advertise that body, the data governance needs that the organization will come up to you. You won’t have to look for them. More and more, though I see organizations looking for something to govern, to justify the existence of the data governance committee, and the data governance function. That’s a no win situation. If you start governing for the sake of governing, you’ll burn people out, you’ll disillusion people, and the data governance function will fall by the wayside.
In these successful organizations. Data stores play a very, very important role there, facilitators not watchdogs. I’ll use the term watchdogs on purpose, because a lot of times what I see are the establishment of data stewards. But those data stewards, immediately put up a chain link fence with barbed wire around their data, rather than facilitating and helping people with access to the data. They wear their new title as a data steward, as a new expression of their power, and they become watchdogs not facilitators.
The best data stewards are risk tolerant, they’re helpful, they feel good about utilizing their data, not paranoid about it. They know their data content, and they help others to turn it into enterprise value. One of the things that I did as a practicing CIO and CA is, I would give bonuses in some form or another financial reward to data stewards, to recognize they’re very important role in the organization. So, these bonuses and these rewards weren’t much, but I would take it out of my budget. I would reward them on a regular basis to recognize the importance of their role. I think that’s an important part. Because a lot of times your best data stewards, are people that have been working in the trenches of data collection for 20 or 30 years. They understand their data content inside and out, they understand the limitations of the data content they work with.
They’ve generally been kind of underappreciated. But in a new digital world, the role of those people as data stewards, becomes more and more critical. So, I think it’s important to help them feel better about their job, right back to mastery, autonomy and purpose. It allows them to extend their sense of mastery in a new dimension, and I’ve had the great benefit of watching two people literally blossom in front of me, who have been toiling away, maintaining a cancer registry for example. When they become appointed as a data steward, and they’re appreciated now for what they’ve been doing for so many years. In the broader context of the value that data, you can see their job satisfaction, and their sense of self-esteem grow significantly, and it’s a very rewarding thing to watch that.
The best organizations have a strategic data acquisition roadmap. So, I go back to this all the time. If you round out the data that we need, in order to provide a combination of personalized care as well as population health. It ranges from better clinical data using bio integrated sensors, and things like that rather than the three times per year, that we currently collect data on patients, right? Average of three clinical encounters per person per year in the US, that sort of digital sampling rate right now, we have to sample or data more often. Therefore, you need some kind of bio integrated sensor strategy in the organization, and allow patients to subscribe, participate in that data sharing environment.
If we truly want to provide population health and precision health, you need genomics data, you need epigenetic data, you need microbiome data, you need social determinants of health data. All of that data, needs to be updated on a frequency that’s appropriate for his volatility. So, it’s not a one and done thing. In every data has its refresh date on it best used by date, and you need to understand that and put that in your data acquisition roadmap. Partner with the vendors, and the technology that can provide that full understanding of a patient in their complete ecosystem of data.
These successful organizations, have a formal Data Literacy Training Program. Things as advances Inter-mountain’s advanced training program, for example, that Brent James Lead for many, many years. Some form of a formal Data Literacy Training Program is critically important. The nice thing is, you can borrow a lot of content online now, to provide that Data Literacy Training. You don’t always have to build your own, like Brent did at Inter-mountain. But it’s really important, that you have various levels of Data Literacy Training available to your employees, that you hire according to the folks who have those skills in existence as well.
The other thing that’s interesting about successful companies in the most advanced in data governance, is they’ve moved and they’ve evolved from just data governance to algorithm governance, which is a natural state of evolution. In fact, I’ve been wondering if we should change the name from data governance to something else, but I think data governance is so embedded in it may be too hard to change that.
It’s important that you include the notion of algorithm governance, in the evolution of what you’re doing. Everything from relatively mundane, and boring definitions like length of stay, or the local definition of what constitutes a diabetic patient. Including dimensions of their social determinants of Elsa that more precisely understand the total context of a diabetic patient. All the way up to deep learning algorithms, predictive models, and that sort of thing that’s quickly evolving. So, design ahead your data governance function, realizing that for a while if you’re struggling, just be content with the traditional definitions of data governance, and get that lined up well. But design something in design membership and skills, and the recognition that algorithms, are going to become more and more important.
If nothing else, it’s important to keep up with the algorithms associated with compulsory measures. That in itself is a challenge in most organizations, so compulsory measure management, compulsory measures being that combination of the things that we have to abide by for reimbursement. CMS compliance, Joint Commission Compliance. To things that we should comply with such as: Professional Association, Society of Thoracic Surgeons, National Registry and for Myocardial Infarction, that kind of thing.
There are algorithms and data governance, associated with those that need to be governed as well. Especially all the value sets in the measure sets, and things coming out of CMS and Joint Commission, and the private payer versions of those. That in itself is a commerce and complicated process today, and it amounts to not data governance, but algorithm governance. Those have to be kept updated, you have to understand their definitions within the context of your organization, and that’s an important part of the data governments function is to include that.
Okay, governance structure. I would advocate and I think this is kind of, a standard definition of what amounts to a democratic governance mindset. That is, you centralize the principles and the values, and then you decentralize execution and adherence, with emphasis on the cultural requirement. Required for decentralized adherence, because if you don’t want to have a culture, that aligns around the notion, of centralized principles but decentralized execution adherence. If you don’t have that common sense of community, to follow data governance principles out in the field, then the data function will decay.
That applies, I think to any kind of decentralized governance function actually. It’s the community’s willingness to participate in those principles, and values that makes it successful or not. Which is why, data governance is largely a leadership issue, not a technology issue.
One of the things I strongly advocate is to make the data governance committee a sub-committee of something in existence, and so quite often you’ll have an executive committee in healthcare. That executive committee in healthcare, normally has a couple of sub-committees already in place. There’s usually a HR sub-committee, there’s usually a finance committee, there might be a branding and public image stuff committee, there’s an audit and privacy sub-committee. The same sort of structure, can be used to extend governance out to data governance. So, I highly recommend folding data governance into the existing structure of corporate governance in general. As CIO, I think I mentioned this earlier, I have consistently been the chair of this sub-committee. It was a natural fit, but you have to pick the executive that’s right for your culture.
I’ve talked a little bit about, what’s right for the culture, sometimes CEOs fit, sometimes they don’t. The important thing is to find a leader that’s data driven. It’s data savvy, respected in the organization, and can drive and lead change, and understands the human soft side of data governance. I strongly advocate not overloading the sub-committee with C levels. A couple is fine, three is starting to get too much.
The reality is they just don’t have time, that should be a working group. So, the CIO can be a proxy for the rest of the C level organization quite often, and if you’re a good CIO or trusted CIO. The other C levels in the executive committee, will trust you with that duty. So, there might be another say a CMO, CQO on the committee, but at the working level.
What I’ve often seen is they will, over time as trust builds those other C levels will delegate more and more, of their proxy of their vote to the CIO. So, that they don’t have to be involved day to day in, and by doing that, you increase the agility of the data governance function. Because as we all know, scheduling these meetings with C levels is quite often the thing that slows down the most progress. So, trust is really important, don’t overload the C levels, and if you do overload the C levels, work towards the notion that over time a single C level, like the CIO will own most of the responsibility for making decisions, and progressing the mission. Give that CIO and give that person that trust to resolve tactical issues, kind of district court versus Supreme Court Judge Roll.
Realizing that the executive committee, and the Data Governance Committee in general is the Supreme Court from which more complicated issues, higher risk decisions can be escalated. But while you’re trying to do is push decisions down to the lowest level possible, and make this leadership of the data governance function as adaptable and is agile, and lean as you possibly can.
Talked about this earlier, the data stewards are facilitators, not barriers to data access literacy and utilization. This is interesting. I left the reference in here, I didn’t mean to. This is actually referenced this manager’s manager and Atlas. It’s a reference to a couple of tools, that we provide our clients to facilitate this. I’ll reference the generic functions not advocating or selling these products. I’m Sorry about that. I’m back to the issue, of how many measures we have to manage from a compulsory standpoint today. We have a product called Measures Manager, that supports what amounts to a centralized content management system. Glossary of compulsory measures, that our team keeps updated according to what’s happening in the industry. Takes that burden off of local organizations.
Quite often I think those, have you been in the trenches, the definitions of these compulsory measures are spread all over the organization, that leads to inconsistent analytic results in consistent reports. A very labor intensive process to keep that updated. So, we’ve just centralize that in a content management system. I have highly advocated all of you folks doing the same thing. Centralize the measurement of that turning into a Wikipedia style application, so that everyone can keep it updated, and there’s a central repository, to keep a handle on all these measures.
Atlas, is another. It’s a metadata tool we designed specifically for healthcare. I would just caution everyone that, there are data governance tools that are cross industry, that overemphasized computable metadata, rather than subjective metadata. Again, this tool is essentially a metadata repository, that helps you document all of the data, that you have in an organization.
Our product is sort of tightly bound to our underlying data operating system, but I would just caution everyone. However, you address the metadata management problem in your organization, as a function of data governance. That you be careful to understand the distinction between computable metadata, that has its limits and value, and subjective metadata. That can help guide people, and guide data analysts to the best use and the maximum value of data.
The traditional tools available for metadata management in healthcare, or in any industry are quite expensive. I would argue that … In fact, I’ve reach to the point that, when I was still in the trenches of this as a CIO. I stopped buying those tools are too expensive, and the value that computable metadata to my mission was not high enough to justify the expense.
Okay, another attribute of success, and healthy data governance functions as data analysts. Who are keen to ensure the accuracy reports, and avoid analytic comedy rooms and synonyms. So, they understand that producing the same report with a slightly different name creates confusion. Likewise, a different report with the same name, creates confusion in the organization. So, data analysts if they’re not centralized, and most organizations are not centralized.
I advocate the data governance function, bring those data analysts together on a regular basis to collaborate, and work across organizational boundaries to eliminate, but I call analytic confusion in the organization. Establish standards for the way data analysts, will present their reports, so that you create what amounts to UL seal of approval before report. So, that the consumer knows that the report has been well vetted, and has been validated by the community of data analysts in the organization, and that’s the information can be trusted.
There are lessons to be learned here, from my days in the National Security Agency, and the Military. Where senior decision makers, wanted to know a lot about the background of the intelligence reports. How the data was collected, who was involved in vetting the data, what was the reliability of the source. All of the context around the decision, that might be made based upon the intelligence report.
Those same kinds of concepts apply to healthcare data analysis. If you have an interest in expanding your knowledge of how the CIA, and the NSA and other intelligence agencies handle that. There’s quite a bit of online material, that how they classify the credibility of information, that’s presented to senior decision makers. I think there’s a lot of lessons to be learned, that can be applied to big expansion of skills base for healthcare data analysts.
Now bouncing out, getting beyond the attributes of successful data governance. I’m going to talk about a few other things. The tension that exists between information security, and data governance is healthy. But as I mentioned earlier, I believe the most successful data driven organizations in healthcare, have a tolerance for data utilization. That feels risky, I think to many organizations in the industry. But in reality it’s calculated risk, and they understand these organizations understand that, data that’s unused, perfectly locked up, and perfectly secure is the least valuable data to an organization.
Someone on the data governance function, needs to constantly pull against the tension of information security. Which would for the most part, has put a lot of restrictions on the use of data in healthcare again. I think HIPAA, is chronically overflight in healthcare to the detriment of patient care, and physician satisfaction. So, having the courage to take calculated risks around data utilization, is a critically important attribute to the successful organizations that I see in the industry.
Problems with data governance in general, it swings between too much and too little. I rarely see organizations that hit equilibrium and stay there. I think the important thing is, to minimize the oscillation in that swing, but quite often what I see is a dramatic swing and the oscillation. So, an organizational established the data governance function, they’ll overkill it. People will check out of it a year later, there’s really almost no existence of data governance functioning in the organization. That’s a consequence of applying too much too soon.
The other extreme is there’s no data governance function, there’s no vision around it. There’s nobody to which you can appeal data quality problems, and it’s essentially an architect. So, while I advocate governing to the least extent necessary, it’s important to try to find balance. Quite often what I see, is data governance being driven from a very techie perspective. What’s the standard format for gender? I can’t tell you how many times I’ve been in senior data governance meetings, where this is an hour long discussion, and I don’t deny its importance. I’m just not so sure that, you need to engage three C levels in that debate.
I think you can solve that at a lower level in the organization, and this kind of specific things, not just this one in particular, you can solve this sort of techie perspectives for data types and naming conventions. That are lower level, and then push those out for formal approval if you want to. But you don’t need to engage C levels in these kinds of discussions, because they’ll burn out, and they’ll check out a data governance function.
The other thing that I see is in the pursuit of becoming data governance, a data driven. We continue to put too many clicks on the backs of physicians and nurses, to collect more and more data without a sense for the return on investment, associated with that data. Every click cost something, every click … Every data entry, keyboard entry costs something, and it costs something to collect it from the labor perspective.
It costs something to collect it from a technical perspective, and it costs something to curate it and analyze it. What I see quite often is recreational data collection, as opposed to a return on investment mindset towards data collection. I personally would say, if I were in an organization today leading analytics and data. I’d say we’re not adding any more clicks to the backs of physicians and nurses unless we can prove, the value of that and the cost benefit trade off. No more recreational data collection.
I want to advocate for a concept that I call it digital mission. That is this role that sits between the clinician and patients. By the way, this is something that I were, if I were engaged in the operational details of healthcare and data governance, I’d be advocating this position. I would probably set aside people to fulfill this role on my staff. It’s sort of extension of informatics, but it’s also I would think kind of modern version of the traditional informatics definition. It’s a role that would sit between the clinician and the patients, and responsible for managing the data profile, data quality, and the refresh rate of data for patients. It could be an extension of the care manager’s job, it’s not too far from what they’re doing already.
I believe that very strongly. We need to start developing data profiles for patients. Based upon the unique needs, and types of patients that we’re encountering. So, I’m not talking about the traditional quality measures associated with a diabetic patient, for example, I think we’re almost over killing that profiling. I’m talking about a broader view of the data that we need about a patient to ensure, the maximum outcome for that patient. The longest length of life, and the highest quality of life. So, I would argue that there are different data profiles, and I’ll give a simple example, right?
The data profile required. The telemetry required about a patient associated with a total knee replacement, and their rehab is dramatically different, than the data profile required to manage congestive heart failure. I really believe strongly, that we have to start creating these data profiles, and then its wild bio integrated. Other sensors are being developed to help us collect this data, when patients aren’t within the four walls of our healthcare delivery organization. We need to collect this data in a way that we can, buy paper if necessary, but we can’t sit back passively. While this data sits out there in the dark matter of space, and we need it in order to improve the precision, and the personalized care that we’re providing.
So, I’m going to do a little more thinking, and sketch out this notion of a digitization. Maybe we’ll do another Webinar on that later, but I’d appreciate everyone else contributing to that as well. I’ll close the part of the discussion and then hopefully there’ll be lots of questions. I think we scheduled this for 90 minutes. I’ll close by referencing a few other online resources for additional study. I wrote a paper a few years ago called Demystifying healthcare data governance, that still gets a lot of traffic. Still gets a lot of compliments, very grateful for those compliments. So, something about that paper still resonates. It’s a relatively simple read. I would encourage you, if you’ve had a further interest in details to go out and take a look at.
Mackenzie has a great framework for what they call digital assessments and assessing what is your digital quotient. I think it can be a very valuable tool for informing a data governance function as well as strategy. There’s plenty of information online and at the Mackenzie website, but you can also engage Mackenzie for consulting agreement and engagement, and have them come out and do these assessments. But there’s almost enough on their website to conduct the assessment yourself. It’s a very pragmatic way to go about the assessment of your digital capabilities, and your DQ is they call it.
Then of course from that score and that overall assessment of your DQ, you should be able to plot a digital and data governance strategy going forward. Then another reference that I advocate, is the Hymns Adoption Model for Analytics Maturity. It was based on a paper that a few of us wrote at health catalyst, and number of years ago called the Analytics Adoption Model. That’s if you want someone, an external entity to come in, and do an overall sort of data governance, and analytics maturity assessment.
I highly recommend the services of the hymns folks who have taken the paper that I wrote and made it a scalable, applicable tool, across the entire healthcare industry. I strongly advocate its use, similar to what we achieved, and we applied in the industry around EHR adoption. Okay everyone, that’s it for the talking paper. Thanks for hanging in there for an hour and now we’re going to look at questions.
Sarah Stokes: Do you mind if I jump in with our closing poll question really quick before you do that?
Dale Sanders: Sure. You bet, friend go ahead.
Sarah Stokes: Okay, everyone. Thank you again for joining us today. Right before we jump into the Q&A here, we just have our closing poll question. A well, today’s topic was an educational Webinar focused on assessing, and enhancing your data governance strategy. Some attendees would like to know more about health catalyst products and services. If you would like to learn more, please answer this poll question. We can just leave this open for a moment while you just get going on those questions.
Dale, if you want to jump on in.
Dale Sanders: Sure. So, lots of input initially about alternatives to the CIO, and things in it. I think we covered some of it. So, lots of suggestions that a Chief Data Officer could fulfill the chair role, which I’m seeing a lot of that Chief Quality Officer, Chief Analytics Officer, CMIO. I do see CMIOs fulfilling that role quite effectively, actually the Chief Medical Informatics Officers. So yeah, I think pretty standard, alternatives there.
Let’s go through some of these other questions and comments. Let’s see here. What happened to timeliness is an element of data quality. Dean Olsen asked that … Yeah, I think that’s a valid question, Dean. I tend to lump, data timeliness is an element of validity. So, you could call it out as a separate variable in that equation, but if it’s untimely data, I don’t consider it valid data. If it doesn’t fit the timeliness of the decision making, if it’s stale data, for example, outdated, that kind of thing. Jason Shum mentions that we have data from 1905. That’s a really good point, Jason. Lots of data is getting older and older, as am I.
Question here from Bill Russel. “What outside data set should Health System acquire in order to deliver on the promise of population health?”
Well, Bill, I have a cartoon that I’ve carried around for years that sketches out the healthcare data ecosystem with a patient in the center. That would certainly inform what I would suggest is an acquisition plan. The obvious choice is there. You’ve got to get claims data on the patients that you’re responsible for, and I’m surprised at how hard it is to get claims data from payers. They still make it very hard to get that claims data.
That’s an interesting example, but there’s census data, there are state level, all-payer claims databases, there are data sets from CMS. All of those can inform your local healthcare delivery system in general, but they can’t inform your system specifically.
So it depends on how you’re going to find outside data, I suppose. If it means the data that you need that you don’t have now, it’s important to the mission of delivering care to your patients, I would call it local outside data as opposed to, I think our traditional definition of outside data, which means anything that is collected externally from the organization that we might be able to use.
More and more I’m trying to dissuade people from using the notion of outside data and just start thinking about the ecosystem of data that you need that you don’t have right now, and I regret now not having that cartoon as an example today in the presentation.
I actually could share it on my screen. I have it in another presentation here. Let me do that really quick. I’m going to pop out. I’m working on another lecture here I think that I’ll be giving it at a boot camp. I’m seeing my old boot camp in a couple of weeks.
I’m going to pop out and show this cartoon for a second. Sarah, can you see my screen there?
Sarah Stokes: Yep. I sure can.
Dale Sanders: Okay. So in this part of the lecture, I’m talking about the importance that, AI needs breadth and depth of data, featured domain engineering, feature engineering rather, in the domain of healthcare requires rows and columns, records and facts.
So it’s not enough to have, for example, CMS will share data on millions and millions of patients, but the reality is the number of facts that CMS has on those patients is pretty limited. So you can use a narrower data set to some degree with AI, but what you really need is just more facts about patients.
So this is the cartoon that I’ve been carrying around since somewhere around 2010. Well, actually, no. I was at Northwestern when I drew this. So earlier than that, 2008 I guess.
This is the data that you need to start collecting as a healthcare delivery organization. If you really want to understand the patient in the middle. We’re stuck right now in the lower left quadrant of this diagram from around claims and healthcare, traditional healthcare and counter data. We have a hard time getting our hands on claims data.
The only biometric data that we’re collecting beyond the four walls of the ICU, for the most part, is Fitbit data, which has almost no value. We rarely ubiquitously collect patient collected outcomes data. The only organization that I know that routinely collects robust patient reported outcomes data is Partner’s Healthcare. It would probably make sense to collect consumer data, although now that Facebook and others have tainted the waters of data utilization and data exploitation, that one’s going to set us back. I don’t think we’re going to be able to use consumer purchasing data to inform our healthcare analytics.
We can bring in and we can collect socioeconomic data. We should be doing that as a part of the patient intake process and then routinely updating that. Genomics data should be a common part of what we collect. If the genetics is next, then microbiomes data is next, there are probably things that I’m not even thinking about here.
What in the world is going on. So, back to Bill’s question. That’s one form of strategic data acquisition road map that I think is important for the data governance function to pursue. Interested in your thoughts about what I’m missing there.
Let me make one more point here is that we don’t even collect this data on the patients who are seeking treatment much less on patients that are relatively healthy that never come in for an encounter. We have to create some kind of incentive for patients that are healthy to contribute their data to the healthcare ecosystem so that we can better understand what’s leading them to be healthy so that we can apply those lessons to the patients that aren’t healthy.
So, our training set for healthy patients is essentially nonexistent in healthcare if you’re using an AI term. Our training set for healthy patients is essentially nonexistent, so we have to create financial incentives in our data government strategy or other incentives, maybe it’s social connectivity incentives, for healthy patients to contribute their data to the ecosystem for the benefit of others.
I’ve said in the past that contributing your data to healthcare will become as important as donating your blood. I’d better get off that soapbox.
Romeo Berry asks, “can a salutation on the 80% missed data be shared later with the presentation?”
There’s varied sources for that. 80% of healthcare outcomes are attributable to factors outside of healthcare delivery. The one that I referenced was from the University of Wisconsin’s School of Population Health. I think if you google that, I think you’ll find it fairly easily.
Dan River asks, “I’ve always professed a need for reporting governance in addition to data governance. Do you see a need to have a separate reporting governance program?”
That’s a great point, friend, and I would agree that some kind of governance needs to lay over the top of the data to ensure the UL seal of approval … now I’m dating myself, the Good Housekeeping seal of approval.
I have traditionally encompassed reporting governance within the data governance function as a form of utilization, but I think your point is accurate and that is, it’s not enough to govern the data, you have to govern the consistent use of the data in a way that I referenced.
CIA and NSA intelligence communities are actually very good at this kind of thing. They have really formal vocabularies and very formal processes for describing the validity of a record. I highly recommend learning from them.
Let’s see. What else here. Bill asks, “When you talk about corporate governance, that is usually done at the board level. Is that where you think this fits?”
No, not really, Bill. I’m not talking about that level of governance. When I’m talking about corporate governance, I’m talking about a level below the board. So what amounts to the executive team in the organization, the sea level executive team in the organization, is the right place for the data governance subcommittee to report. Probably not at the board level, although there might be some forward thinking boards that would want something like that if they believe that the digitization in the healthcare delivery system that they are governing is important. I can see that becoming a subcommittee.
I can’t think of a case in which I’ve seen that, but it’s an interesting thought.
Ed Gray asks, “Do you have any additional advice to help data governance discussions remain focused and not constantly digress into endless HIPAA and cyber security concerns?”
Well, Ed, it gets down to leadership again, and if the leadership of the information security and the data governance functions don’t understand this tension and the importance that for every data that you don’t use is a dollar in value that you probably denied from patient care and admission in the organization.
There has to be an appreciation for that, but again, one of the benefits of my background having grown up in the military and NSA is, I grew up in an environment that was incredibly paranoid about security of information, but we also appreciated the importance of information sharing in a very tense, high-risk environment.
So what I don’t think a lot of the HIPAA and security paranoia folks appreciate is that if you don’t use data, you’re cutting into the mission of healthcare delivery in a very significant way. You’re denying value to patients, you’re denying financial value of the organization. Someone in the leadership team needs to stretch the importance of that.
My teams, I’m blessed by this complement, but my teams have complimented me in the past that I provide them air cover as a CIO and a CAO in the utilization of data. I’m willing to stand up and defend the utilization of data for the betterment of the mission and the betterment of patient care, realizing that I also have a responsibility as a CIO to the security of data.
So I don’t know any other advice to give you other than that, but it just has to be an important leadership and cultural trait.
Let me see here. Let me scan down. There’s a question from Jennifer Burn. Jennifer, I hope you’re still on, friend. “What are the ramifications of the integration of clinical research as a care option with respect to data governance?”
That’s a topic we’ve been discussing lately, isn’t it Jennifer. I don’t know that I have a good answer to that. What are the ramifications of integrating clinical research as a care option with respect to data governance? You know what? Jennifer, friend, I am going to have to think about that. Thanks for stumping me. I don’t have a great answer for that. I’m going to have to think about that a little more.
I think it’s a really important point. I think that it’s such a new concept. I’ll lean back on my Northwestern experience for just a minute, and to some degree my LDS experience. We certainly included the IRB membership. So the head of the IRB at Northwestern, for example, sat on our data governance function.
So we certainly appreciated the importance of the traditional IRB process in the data governance role at the traditional academic medical center, but you’re suggesting taking it a step further, and that is integrating clinical research as a care option, proactively offering participation in clinical research as a deliberate option for care and improvement.
I can’t answer that. I want to think about that some more. So thanks for giving me to churn on at 3:00 a.m. in the morning.
Okay. Let’s go on to another one here. Tammy Gray makes a really good comment here. I think it was in reference to the overabundance of clicks on clinicians right now in collecting data. She calls it CYA data collection, and I would agree. I see a lot of that. But again, CYA originates in risk aversion. If you’re tolerant with just being truthful and being who you are, you’ll worry less about CYA data collection and who you are as an organization.
So being able to stand up to CYA data collection and challenge the common sense of that, you know I advocate to everyone all the time, don’t surrender to violations of common sense. Do not stop until you overcome the violations of common sense. We surrender all the time to violations of common sense in healthcare. Don’t do that, and we’re doing that when it comes to data collection.
Okay, let me scroll through a couple of others here. Amy Cahn says, “Data governance seems like the secret sauce in moving skeptical clinicians to fundamentally adoption of data as a critical tool to serve patients. How can I help facilitate this goal in making this patient centered data home dashboard the new stethoscope for practitioners?”
Well, that’s an awesome metaphor friend Amy. I love that. Patient centered data home dashboard, the new stethoscope for practitioners. I do see evidence of physicians using dashboards of that type more and more in clinical delivery of care. It’s typically in organizations that are less worried about fee-for-service volumes and more worried about risk mitigation, both at the patient level as well as the population level.
I love this metaphor of the patient centered data home is the new stethoscope, at both a population and a patient level. I’m seeing more and more of that, so I feel hopeful about that. I think again, it goes back to whether it’s a high-volume fee-for-service environment or if there’s a significant risk at stake in a population health and an ACO type setting.
Financial economics drives a lot of the behavior I think, and also, just giving physicians a pause and a break during the day, stop measuring them by RVU’s and start measuring them by outcomes, which that ought to be a part of your data governance strategy as well, how do we treat physicians with the respect they’re due around outcomes as opposed to RVU’s?
If we keep measuring Physicians around RVU’s, we’re going to continue to drive volume over quality. I think that starts to address, how do you move skeptical clinicians?
Well, frankly, I would be skeptical, too, if I were a clinician in today’s world, because every time someone approaches clinicians in most conversations today about data, it’s about taking away their sense of autonomy. It’s about measuring them. It’s about watching everything they do instead of watching what they achieve.
So I’d be skeptical too, and frankly, it’s just acknowledging that to clinicians and changing the dialogue is the first step. The second step is in practicing the behavior. Recognize that we’re in this weird world right now with actually a little bit of glimmer of common sense emerging from Sema Verma and CMS to try and eliminate this over measurement of meaningless clinical process measure as opposed to focusing more on outcomes, but change the nature of the dialogue you have with clinicians, acknowledge their point of view, have some empathy for the situation they exist within as it relates to data, and then actually do something and give them data that they want to see that makes them feel better about the mission that their pursuing.
Okay. Thanks for all these great questions. I really appreciate them. I want to go back one other earlier comment. Allison Mayo makes an interesting point here. A doctor of behavioral health would be a great leader for data governance. That’s an interesting thought. Interesting thought.
Tiffany Lemon asks, “What is the role of data governance for oversight in the design of building a data capture within the electronic health record?”
Well, it’s a fundamentally important part of the data governance committee. The data governance committee, if we were really thinking ahead, we would have created a data governance committee before we started rolling out EHR’s and anticipated the need to influence and strategically design the deployment of DOCHR’s to achieve the analytics and the clinical outcome improvements, the financial outcomes, required by the organization.
I can’t think of any HR that was ever deployed that way. I’ve never heard of one, I’ve never been associated with one. So absolutely, positively, the data governance function has to encompass not just the back end of data, but also the front end of data collection. That’s the role of data stewards to help with that.
Okay. Oh, Liz Glass asks a really good question. “How do you think a patient satisfaction survey fits in your drawing?”
I lapped patient satisfaction data in the healthcare encountered data plan, Liz. I consider that a byproduct of a healthcare encounter. So there’s a lot of granular data that fits into that. That’s just my definition of it.
Patient Sat is just a dimension of that cloud. Let’s see. Peter Attia makes an interesting point here. “Regarding data from healthy patients, an indirect way to get this might be through incentives offered by insurance companies for patients to contribute their data.”
I agree with that. If there were sufficient financial incentives to me, and also a guarantee that that data was going to be used to improve my health in some fashion, I would certainly be willing to contribute my data to a trusted healthcare delivery organization or a trusted third party that would help with that.
I see insurance companies now offering devices in discounts, rather, for devices you place on cars for teenage drivers that monitor the driving behaviors of, not just teenagers but I think it’s primarily teenage drivers, and if you contribute the data and if you show good driving behaviors, then you’ll get a reduction in your premium.
I think something like that has to be in the future of healthcare delivery as well. I’ll share another thing too. One of the things I witnessed with my wife during her first pregnancy with our little girl, this was now four and a half years ago, my wife had kind of two forms of digital interaction during her pregnancy.
She interacted with the healthcare delivery system and the midwives and the OB that was associated with her pregnancy, The traditional healthcare delivery team.
There was a portal for her to interact through the healthcare delivery system and potentially, those members of the care team.
As it turns out, there was really zero interaction with the care team through that patient portal. The best that she could get out of that was lab results.
The other part of her digital interaction about her pregnancy, her health, was with the social connections she had with other mothers in her age group at a similar gestation rate, similar gestation period.
It was a 95% to 5% division of her time. 95% of her time, and I would say what affected her overall emotional, physical, and behavioral health during her pregnancy came through the social interaction of patients like her, same age, similar geography, same gestation period. Almost zero interaction with the healthcare delivery system and their traditional delivery team.
The way to engage patients, really, I think is not necessarily around lab results, though that’s important, but what I think most of us want is to have an interaction with social patients that are like us. Combine that with traditional healthcare delivery interactions as well and bring those two worlds together.
I think there could be a way to incentivize patients to participate, and you might not have to do anything more than provide them a social network of patients like them in order to get them to contribute their data. That might be more effective than money. Who knows.
Okay. Let’s see. Ed asks a question. Let’s see. We’ve got four minutes here. “Our organization already has a generic data governance subcommittee that is not focused in healthcare data. Would you recommend adding a subcommittee to it or assemble a separate data governance committee focused specifically on healthcare?”
Yeah. I would suggest the latter, Ed. Yeah, I would suggest you take advantage of those existing bodies if you can, governance bodies.
Chris Minick asks, “Can you say more about that idea of the soft side of metadata captured? Maybe give a concrete example of soft side vs. computable.”
Yeah, sure, because great question. So computable meta data is number of rows, cardinality in a column, number of no-values. Those are computable. When was the last updated? Who accessed it? Those are all kind of computable.
The soft side, the non-computable data, is the narrative that goes into that. So a data steward that can describe maybe changes in system configuration, maybe changes in policy around the collection of data as a historical artifact. That’ captured in the narrative of the data content that you can’t compute.
So, it’s the narrative storytelling of the pros and cons, the data history, what an analyst needs to be aware of that affected the data. All these things that can’t be computed around the history of the data.
You can almost think of it as a data health record, a DHR, where you have computable vital sign information, vital sign data collection, things that you can compute from devices and diagnostics of valid data, but there’s always a clinician’s note associated with the subjective assessment of a patient.
That’s the same kind of thing that a data steward can offer in the subjected description of data that they’re responsible for. If the history of the data is the non-tangibles and non-computable but yet very important attributes of that data.
Okay. We’re running out of time. We’ve got two minutes. “Does Intermountain Geisinger data governance coined as knowledge from your insight governance as a continuum of evolvement of governance?”
I don’t know for sure, Ananthan. I can’t remember what they’re calling it, but it’s not unusual to see some of these data governance functions evolve towards knowledge governance. I personally am not a big fan of that. I think it gets too esoteric too quickly, and I would rather take this thing that is already struggling, data governance, not complicated, not overplay it, take care of that first.
I think if you lay the foundation of good data governance, I think over time it naturally extends itself into these more ambitious governance roles I think that your describing which you could call knowledge governance or inside governance, but initially, I keeping it simple, and recognizable, and less academic is more important. That would be my style.
Hi, Chris. I see you’re still here friend. That’s interesting, Chris Harper. “You briefly mentioned Fitbit not being valued. Can you talk more about that in terms of current sensory capabilities of wearables and future sensory?”
Well, thanks friend. I will do that, actually. You’re prompting me, I’ll end with a slide. Let me point you back to the deck. I had the good fortune od being acquainted with John Rogers at Northwestern University who I think might be the world’s leader in the development of bio integrated sensors.
If you watch what his team is doing, it’s phenomenal, and I think it’s an indication of the future of data collections for patients in the biometrics space. So his team is constructing these microns thin, 1″ square, skin pliable sensors that contain a Bluetooth antennae, a CPU, the physiologic monitoring package specific to that wafer, and wireless powered systems, wirelessly charged dual power systems.
So that’s taking the notion of Fitbit and expanding it to orders at magnitude more valuable and also more complex of course, but the progression and what’s going on, there’s actually the UCSD folks are doing something similar. They now have these three dimensional wafers that can jam even more physiologic monitors onto these, and they’re passive.
You stick them on your skin, you stick those on a child’s skin in the NICU and suddenly, you’re not tethered to data, you don’t have to enter data anymore, and these things are already in clinical trials, sports teams are already wearing them.
So this is coming and it’s coming fast, and I think that’s the answer of the future, it’s current and future. It’s the replacement of the whole concept of Fitbits.
All right. We have to end now. I appreciate this very much everyone. Thanks for sticking around for 90 minutes. Best of luck to everyone and do good out there and help us with healthcare. Have a great day. Bye, bye.