Electronic Health Records and the NYC Macroscope: Q&A with Carolyn Greene
New York City is currently developing a pilot public health program known as NYC Macroscope — the first domestic effort to aggregate electronic health record (EHR) data into a surveillance tool to inform public health decisions. The population health surveillance system will compile electronic health records from primary care practices to help city health officials monitor—and respond to—the real-time prevalence of conditions that impact public health. The project is the result of a partnership between the New York City Health Department and the CUNY School of Public Health, with support from the Robert Wood Johnson Foundation’s Pioneer Portfolio, and additional support from the de Beaumont Foundation, Robin Hood and the New York State Health Foundation.
NewPublicHealth spoke with Carolyn Greene, MD, Deputy Commissioner of the NYC Department of Health and Mental Hygiene’s Division of Epidemiology, about the plans and goals for the program.
NewPublicHealth: Tell us about NYC Macroscope and how it will work.
Carolyn Greene: NYC Macroscope is going to be New York City’s first electronic health record surveillance system. We have a program here at the health department called Primary Care Information Project (PCIP), and it’s one of the nation’s largest distributed electronic health record networks. PCIP began in 2005. It concentrated on primary care practices in high need areas where the Health Department really wanted to encourage providers to use electronic health records. The program has been extremely successful and they’ve completed many different activities to improve the quality of clinical care.
But in recent years, we’ve been asking the question: Can we use electronic health records for more than just patient care? Can we, in fact, use electronic health records to monitor the health of the population? Here at the Health Department, we have many different ways to conduct population health surveillance. We have surveys that we conduct by telephone, we have disease registries that we host, and we have our vital statistics registry on deaths and births. All these data are very important. But they are costly, resource-intensive and they often have a time lag from when the data are collected to when we can actually find the results from the data, so the advantages of an electronic health record surveillance system are many.
One advantage is that the architecture is already there. If you already have the electronic health records in place, you don’t have to find additional resources to collect the data because you’re already collecting the data through the EHR architecture. Other advantages are that potentially you can collect data in real time and potentially at low cost.
NPH: Do you see any potential disadvantages?
Greene: I think the first one is we always have to ask how representative the data will be in terms of representing the population as a whole. First of all, electronic health records only collect data on people who are in care and, because sicker patients go to the doctor more frequently, there’s a greater likelihood that we may be picking up more information on sicker patients. So we have questions about how representative are the data.
And then we also have questions about data quality. In our surveys we have well-trained data collectors who collect data in a standardized way. Well, in an electronic health record the data collectors are, in fact, providers. There are thousands of providers who may be entering data in different ways and sometimes may not be entering data at all. So this was our question—we want to build an electronic health record surveillance system for health, but how can we validate it? How can we look at those data that we draw from an electronic health record and see how valid they are when we compare them to a gold standard surveillance system? And that’s what we are going to do in this project.
We are developing this pilot electronic health record surveillance system to collect data on a set of health indicators— including blood pressure, cholesterol, blood sugar, diabetes, body mass index, depression, smoking status, flu vaccines—and then we’re going to compare what we get to gold-standard health data from a survey that we launched recently.
Through this voluntary survey, called the New York City Health and Nutrition Examination Survey or NYC HANES, we are systematically collecting information from a representative, random sample of about 2,000 adult New Yorkers. We’re not only asking them questions about their health, but we’re taking objective measurements including height, weight, blood pressure, pulse, as well as drawing blood, urine and saliva for laboratory testing. And what we will then do is compare the data we get from NYC HANES with data that we get from our pilot electronic health records surveillance system, which we are calling the NYC Macroscope.
NPH: Why is there such great potential in tapping into this real-time bank of comprehensive data for public health purposes?
Greene: What allows this project to be possible is the Primary Care Information Project technology called the Hub. The Hub is technology that allows us to query electronic health records at partner practices to ask a question such as, “In 2012, how many of your female patients age 30-40 had a body mass index of 30 or greater?,” which is the definition of obesity. And we can send out this query to all practices that are on the Hub and overnight we will receive aggregate data counts back to a secure centralized data site. So what this means is that we ask a very specific query of our electronic health records and the very next day, we will have counts back on that question.
NPH: How does that compare to the traditional time lapse between when you have a question about your population from a public health department and you would get the results back on how your community is doing?
Greene: I can answer with real evidence since we’ve been working toward our New York City Health and Nutrition Examination Survey. We began really planning for this survey about a year ago and it requires creating survey tools, creating protocols, recruiting field interviewers and field phlebotomists, coming up with standard operating procedures, and training all of the interviewers and phlebotomists.
Next, we’re going to have to recruit our study participants and even that process of sampling—of figuring out who we would sample to come up with a representative section—was extraordinarily complex. We have to recruit study participants, then, over a period of six months, we will engage in data collection. After that period, we will have a period of data cleaning and then we will begin analyzing the findings.
It’s a gold standard survey and very scientifically rigorous, but it is labor intensive and it can take up to a year and a half to two years to actually get the data. Now, that’s not always the case, there are other ways to collect data, of course, but in that example you can see just the resources and the time required.
What’s so great about electronic health records is that, if they can be validated, they not only allow us to look at indicators that we look at very well with some traditional methods of surveillance (such as body mass index and blood pressure), but they enhance our ability to track the prevalence, treatment and control of many chronic diseases. We’ll also be able to look at the use or the implementation of clinical preventive services.
But again, I should add that there are challenges here too. In a survey, we have a very clear way of how we’re asking a question, how we’re going to analyze the data. In electronic health records, for example, even to ask the question of how many patients in 2012 had high blood pressure, well, how are we going to define high blood pressure? Their blood pressure may be measured five times in a year. Do we look at the highest, the average, or the most recent blood pressure reading? There are all these questions that we have to ask. So, while the idea is very exciting, it really requires a lot of thought about how to do it in a way that it will make the data the most valid and most meaningful.
NPH: Do you think it’s important that the public health department is taking such a lead role in this?
Greene: Absolutely. One of the things that we’ve seen in the field in general is the importance of increasing connections between public health and clinical care, and I think this is one way for public health to interact with the clinical care system. And ultimately, in public health one of our jobs is to monitor the health of the population, and electronic health records offer a very useful tool to do this. In turn, it is our responsibility to feed data back to policymakers, to those who come up with clinical guidelines, to providers, to the public, and we hope that the data we gather and monitor using electronic health records will have an impact on clinical care. It’s a very nice feedback loop, where this is a tool that could potentially help us monitor the health of the population, but more than that we can use the data to then positively influence clinical care moving forward.
NPH: How do you see the role of different kinds of surveillance systems evolving, as EHR real-time surveillance becomes a reality?
Greene: As you can tell, I’m very excited by the NYC Macroscope and I’m excited by the potential of using electronic health records for population health surveillance. I think I should emphasize, though, that we don’t believe that electronic health records surveillance will ever replace the need to conduct some of our traditional surveillance. And I say this because I think we have the sense that electronic health record data will be particularly useful for some indicators and perhaps less useful for others.
Here are a couple of examples. We have a community health survey—an annual telephone survey—that we administer every year, and that survey is based on the behavioral risk factors surveillance system that is done nationally. We ask a lot of behavioral questions about nutritional intake, physical activity, sexual behavior, drug use, etc., and I think especially for those behavioral questions we’re not close to a point when electronic health records will have those data. So we’ll probably always want to be conducting these other forms of surveys in order to collect data on behaviors.
Another example, in our pilot we’re going to be looking at influenza vaccination, and again we suspect our data will not be very complete because now people can get their flu vaccine outside of their doctor’s office. We still will need to rely on other surveillance methods to figure out how we are doing in terms of coverage for flu vaccine. So those are some examples of how we’ll always want to supplement data from electronic health records with other surveillance tools, but I still think this offers new opportunities and will complement existing sources of data.
And I guess the one other thing I would add with the NYC Macroscope is how we hope to use it in the future. I think something that’s been very important to us in this project is that this is not a project about New York City alone. So throughout our planning phase, we have had a very diverse and talented Scientific Advisory Group working with us from other state and local health departments, from academia, from the federal government, CDC, HHS. Their advice has been very important because we’ve been very clear from the beginning that we want this project to inform other jurisdictions so that in the future other jurisdictions can think about how they can implement their own electronic health records surveillance systems.
NPH: Do you think this kind of project and overall the connection between the EHR data and population health surveillance can be an entree into more cohesive collaboration between the two sectors?
Greene: I think it does open up the possibility of increasing dialogue between the public health world and the clinical world. There’s some dialogue already, but having data that we are jointly invested in understanding, jointly invested in monitoring—and data that comes from the very electronic health records that clinicians are using—I think can only increase the conversation and the collaboration between these two fields.