Electronic Health Records: PHR Opportunities for Public Health – Part 2
In this podcast, Dr. Ken Mandl discusses electronic health records and personally-controlled health records. Dr. Mandl leads the IndivoHealth personally-controlled health record project, the original reference model for the Microsoft, Google, and Dossia personal health records (PHRs or PCHRs). He has successfully used PHRs for immunization and influenza, leads efforts in real-time surveillance systems, and is currently adapting personal health records for longitudinal and genomic research. The lecture was given at CDC on June 19, 2009. Created: 9/10/2009 by Coordinating Center for Health Information Service (CCHIS), Healthy Healthcare Settings Goal Team, Office of Strategy and Innovation.
Date Released: 6/3/2010. Series Name: Public Health Lecture Series.
Electronic Health Records: PHR Opportunities for Public Health – Part 2
[Announcer] This podcast is presented by the Centers for Disease Control and Prevention. CDC - safer, healthier people.
[Dr. Mandl] Behavior change. This is the big one. This is the tough one. Will personally controlled health records work here? I don't know. There are some models of -- not the portable electronic health records, but of more traditional personal health records. For example, Aetna has a claims-based record that is linked to -- that the patient has access to, and they have a company called ActiveHealth which actually uses it to actively manage their complicated cases. Okay? And they'll use the data that are available in claims and the data that aren't in claims -- which is a lot of data -- they'll get from the patients or actually even ask the patients to go to their doctor and get the number that they need from their echocardiogram. We need this number. Put that in. I want to understand your cardiac function, and then I'm gonna run this decision-support algorithm, because I've got complete data to run it.
So, there are ways to think about using this very actively. We built a bunch of tools for a CDC-funded project so that we could start to link patients into a model of behavior change, and we did it around something we thought was simple -- immunization for influenza, and knowledge, attitudes, and beliefs around influenza immunization, which, during the five years that we -- since we started that project, you know, the public's attitudes towards how important influenza is have gone up and down. Right now, people think it's actually quite important.
So, the -- we built things so a survey tool, so you can actually reach out through this channel of communication and get to your patients, get structured data back from them, store that data in the record, alongside data that's coming in through these subscriptions. We can go through -- there's a decision engine, so we can actually, based on the data that's in your record, we can run rules, decision-support rules locally, and see, you know, are you up-to-date, are you in a high-risk group, should you be immunized, you know, did you have it last year -- we can look across the data from the health system and the data from the individual and run those rules, and then tailored and targeted messaging where we can actually message back according to those rules. When those rules fire, and we can fire them in terms of some way to get back to a patient.
Now, it turns out that communicating back to patients is not a simple thing. Turns out people really don't like things like alerts very much, and it turns out, also, that as enormously important as the CDC initiatives are, there will be competition for bandwidth along these channels, so when patients do have, you know, an iPhone app that's helping them with their health, all of a sudden, there's going to be a lot of them out there, and there's really going to be some competition, I think, for that. And if you think about it on an iPhone model -- and by the way, this stuff might really run on an iPhone. That's separate. But you might be getting into people's pockets and being able to engage them in interesting ways, and there are some prototype applications out there that do this kind of thing. You're gonna be competing, okay, against very moneyed interests in this area, and it's going to be very important to think about how to do that.
So, we have a couple papers that describe the experience in this. One is where we went into Hewlett-Packard and actually tried to influence immunizations in randomized controlled trials. And the other is we went into MIT and we learned a lot of very interesting lessons and got, you know, amazing feedback and actually produced some statistically significant results, too, even though the numbers in the studies were not extremely large. I think it's a very promising technology, because it's linked into the health system in an interesting way. If you link it into the health system -- in the public health system in interesting ways, I think people really do pay attention to what's going on.
I'll spend the last few minutes talking about cohort research in a trial model. This is a little more complicated, and this is some of the work that we're gonna be launching at Children's this summer. So, one of the papers that I sent Dr. Popovich was "Tectonic Shifts in the Health Information Economy." So, this is a paper published last year. And the tectonic shift is that the data are flowing -- slowly right now, but I think they will -- out of traditional electronic health records systems, out of claims databases, out of institutionally controlled records systems -- into patient-controlled health records.
So, what happens is, let's say at New York Presbyterian, where Microsoft HealthVault is set up, or Children's Hospital, where Indivo is set up, or Walmart, where Indivo Dossia is set up, or Cleveland Clinic, where Google Health is set up, those data that were just at Cleveland Clinic and all handled through IRB and data-use agreements and institutional control and turf and control and silos are now in the hands of populations of patients.
So, that kind of raises a couple of interesting opportunities. One is that, well, I can now, if I -- depending on the policies of my personal health-record platform, I could actually just go dip in and look at trends across those data. Or, if I use a consented model, I could potentially engage patients. There are ways to now reach out to communities of patients that go across institutional walls.
And I can tell you that the Google and Microsoft plans clearly include having these kind of data available on a very large scale. There's significant monetary value in doing that if you do it the right way. And what we need to do also as a public health agency, I think, is to be sure that the outcomes are in some way linked to health improvement as we go through, and think about the policies from that perspective.
So, what can you do here? You can be just a guy out there in a population and end up with a personally controlled health record, so you got it populated by New York Presbyterian, or by Beth Israel Deaconess, or by Children's, and you can contribute those data to a population database if you link -- one of your applications is an application that does that. Okay? So this is a way to actually link in people who are just out there in the real world. You can approach them because you're from that institution and you want everyone in that institution, or you can just go out into the world and do it, and I think this is what we're gonna see next. So, Google Health will do it, HealthVault will do it, and we're doing this at Children's. Let me tell you a little bit about what we're doing at Children's.
We're trying to understand how to follow our patients longitudinally. So, in healthcare, what do we do? We study very small numbers of patients in things called clinical trials. Or we study -- or we follow, meticulously follow some cohorts. We conclude, we make inferences, we run statistics, and then we apply those results to everyone in the population. And the fit is not always perfect. Okay? The FDA has learned this, and with some hard lessons. Okay? The small trials for efficacy do not always address the issues of safety when this thing is out there and when the drugs are out there and the post-marketing phase. This technology actually gives you the opportunity to track this in the post-marketing phase as well, right?
So, let's think about real discovery. You know, big public health here. Genes, environment -- we'll include microbiome in the environment -- phenotype, and healthcare, as things we need to know about each person in order to understand the complex interplay that produces health that's visible at the individual or at the population level. To do this, you need very large Ns, large numbers of people in your studies. You need data capture beyond the health system, because the health system only has some of the information that you need. Very little information about the home environment will be found with even through meticulous record review of health system data.
So, from a genetic perspective, this is really, what do we do with all the Human Genome Project results, okay? Right now they're kind of -- you know, there are many discoveries being made, but is there something we can do beyond this? Can we really start to leverage this and make the proper linkages?
So let's just take a quick look at three models for doing longitudinal cohort work. One is a health-system only model. One is patients or consumers only. And one is sort of a health system plus patients. I'll tell you briefly about another project called I2B2, which is a very successful project that you probably should hear about, too, at some point, for recruiting patients for clinical trials. And particularly for recruiting samples for cohort study. And what it does is it goes into the electronic medical records. It started at Partners, and now it's being deployed at 18 medical centers nationally, academic medical centers, and beyond. It looks across the electronic records and identifies, for example, bipolar disorder. This is a 20 million dollar NIH grant. It identifies bipolar disorder, okay, and using natural language processing of the notes, it makes a table of all the people who have bipolar disorder, and if one of those people shows up in the laboratory, and there's discarded blood left over when the sample has been processed for clinical use, that sample can become part of a cohort study. This has reduced the cost of sample acquisition from hundreds of dollars per sample to under 10 dollars per sample, and is a very interesting and disruptive technology.
But -- but, so, I could just stop there and deploy that, but at Children's, because we always like to do things that are much harder and more complicated and take much longer and fail much more often in the early stages, we're taking a more complicated approach. And it's in order to get some, I think, some big wins at the end. One is that in the I2B2 system, and also, by the way, in cross-sectional survey study that the CDC is so good at, you get that phenotype at one point in time. Okay?
So whatever that phenotype was when you did the survey or you did the reading of the record, that was it. Also, it turns out that these data are getting more complic -- these research results and the ethics around research results, and if the researcher knows something that pertains to the patient, does the researcher have to tell -- I'll tell you, those lines are shifting, and they're shifting fast.
So, all of a sudden, if I learn that my patient's got something from the MRI that I did as a study or from the DNA that I analyzed in my study, I might actually need to tell the patient about it, or the consumer, or the study participant. And these mechanisms don't give you a way to do that. And there's no patient engagement. I am not using this to engage my patients in any way. So, I -- in the remaining minute, I will just tell you about two other models they're for. One is what I call sort of recreational genomics. There are companies that will sell you back your genome and some interpretation of it. This completely disintermediates the healthcare system right now. There's no MDs involved. They were almost shut down in the state of California, but now they're back and doing, you know, some very interesting things.
There are some issues, like the incidentalome, as my colleague Zak Kohane called it, and the incidentalome refers to the fact that when you study, when you measure 500,000 tests, you get a lot of false positives. And you get – and when there's 500,000, even if you have a test that's 99 percent sensitive and 99 percent specific, you're gonna have hundreds and hundreds of false positives per chip. So, one issue.
It's a lot noisier than, for example, ICD codes. Is the PCHR the answer to these problems? Well, from my perspective, we're coming to the hospital. We have an open PCHR system, we're approaching research unlike the big platform vendors, Google and Microsoft, through IRB-approved protocols. We're ensuring full clinical-grade protection, because we're used to acting as a HIPAA-covered entity. And in this study, which is available through Science, as well, in this model, we actually are giving -- we're actually using personal health records to engage every patient who comes through the door to link samples to phenotype, and not just one phenotype, but a phenotype over time. Okay? And we are returning early research results to patients, not through the cycle of peer-reviewed publication if that cycle needs to be short-circuited. In order to short-circuit that cycle responsibly, unlike the recreational genomics companies, we've created something called an ICOB. We call this whole project the Informed Cohort, because this cohort finds out what's going on, participates, and benefits. The ICOB is the Informed Cohort Oversight Board, which includes ethicists, lawyers, scientists, and risk-communication experts, and we figure out what is actionable from the research, what needs to be returned to patients, what should be returned to patients, who will benefit, what the consent looks like, and what the message looks like. And this is hard, but we actually are doing this. So, again, ethics, risk communication, personalization, and modeling individual preference; this is very difficult.
We're at an interesting moment in genomic research and research around understanding the interaction with environment, health. The commodity is the full DNA sequence of the individual. I mean, I wonder what Watson and Crick would have thought if they knew that we could do the full DNA sequence. Okay? But actually getting the phenotype, like, I've had asthma for four years. But that's the part we're struggling with now. Very interesting moment in time, but we've got to catch that part up, okay, to the basic science technologies, and then we're really gonna have some very interesting synergies.
I will end with this slide, where I just thought about what are some particular challenges that the CDC might want to address. One is promoting adoption of these things. And that's both at the societal level -- we need to get the data flowing out of electronic health-record systems into personal health-record systems as well, and between electronic health-record systems. Les is very, very involved in this effort. And the CDC is leading there. Consumer adoption, also. And that's gonna really relate to trust. Consumers are gonna adopt these things when, A - they're useful, when they do something that the consumer wants. They're gonna be pulled into consumers' homes, not pushed into consumers' homes. Okay? With EMRs right now, we have a model where we're gonna pay docs to push these technologies into their offices as opposed to the other technologies that docs and hospitals are buying and pulling into their office. I think we really need to think about pushing technology on people versus making something that they can't -- that's just flying off the shelves. And that really has to do with making superb products. And successful businesses understand this. We've got to imagine what the killer apps are for public health and ones that actually engage consumers. We've got to understand getting bandwidth on the consumer-engagement channel, because there will be paid advertising on that channel. We've got to define new models for surveillance, and think about, I think, this consented model. And then, I think we may have the opportunity, and, you know, I leave this, I want this to be a discussion with people who know more about these surveys than I do, but I think that we may be able to reimagine the national surveys in terms of returning results to participants and longitudinalizing the phenotype in the way that I described.
So I'm going to leave it right there, and there may be questions. There may be questions from remote locations, or people may just want to go out and enjoy the overwhelming heat.
[ Laughter ]
[Question] Where are we?
[Dr. Mandl] We're very early.
[Question] Is there any data?
[Dr. Mandl] Well, you know, those data are hard to come by, but if I had to guess, I would guess that there are less than 100,000 people with accounts across Google, Microsoft, and Dossia, if I had to guess. And of those, my guess is that most people are not actively using them yet. And you know what, I think that's perfectly fine, because, you know, the reason that Google, Microsoft, and Dossia are the organizations doing this is because they have very deep pockets. The data have to be given to patients for free. They can wait a little while. Okay? So the data are sitting there, and what's happening is the ecosystem is developing. At some point, there'll be some critical mass of useful applications that connect to these things, such that people will want them. Okay? If there was nothing that you could download to your iPhone, people wouldn't particularly, if you didn't have iTunes, okay, initially, right, and then all this other stuff, right, people would not have flocked to that platform. Once you've got the applications, you will see people pulling that technology into their lives, and I think we, as we go, we just have to make sure -- one role for the CDC, I think, is to think about what the health impacts are of the different models and the different policies. Scott?
[Scott McNabb] Thank you. I’m Scott McNabb with NCPHI. Thank you very much for this presentation; it was really terrific. And it's wonderful to see some of the population-health benefits from a personally controlled health record. So, I was thinking about the public health workflow. Another set of stakeholders here is the county-state health department, epidemiologist, the state – the public health nurses, et cetera, which are sort of, one could envision a similar platform for public health purposes, which would follow down traditional reporting requirements that are made by the states for, you know, the 77 reportable conditions, so to speak. And I'm wondering -- the question I have, actually, is do you -- would you envision some similar platform for a public health nurse? I carry a G1 Google phone, by the way, rather than an iPhone. It would be an open-source type of approach.
[Dr. Mandl] Yeah, sure. Yeah, so, absolutely. And, you know, I didn't mention, in our immunization project, we engage the school nurse as one user of that immunization registry, for example. And the WIC office, as well. So, I think that, again, we can think about what that -- what the architecture looks like. It might be that we have these -- we have a platform, and the public health is just one of those – public health applications are just one of those squares on the iPhone. Okay? Or there could be a separate public health network that somehow interacts with the traditional healthcare system network. But, you know, thinking about public health -- so, reporting, you know, the reporting comes from -- the reporting does ultimately come from an individual's data.
So, where are those data living? If the data were living in a platform that was entirely based on patient health records, then the data would naturally be reported from there. If the data are coming from institutional records, then the application would sit on the institutional platform.
So, let's just talk about -- let's forget personal health records for a second and think about that from a non-personal health record perspective. I've got, let's say, the e-clinical work system in my office practice, and I want to report out public health. If you had a platform with some standards, okay, or hopefully not too many, hopefully not 50 platforms, but maybe like with the personal health records, maybe there are two, three, four. Okay? Then CDC could help define what the application needed to be. So let's think -- swine flu. Right? We weren't really -- we didn't really have a swine flu workflow before a couple of months ago, right? We could produce an application like that that's substitutable, comes, sits on the platform, and adds functionality. Now, this is very similar to what Les has already envisioned with the grid. Okay? It is a very similar concept, because you've got data flowing and you have applications that can connect to the grid, and the question is, with the investment right now in EMRs, and possibly PHRs -- I mean, that comes in there, too; it comes in the conversation in interesting ways. I don't fully understand. There may be people in the room who understand the relationship that's going on between EHRs and PHRs behind closed doors. I have some sense. But I think the question is, with that emerging story, where does the public health app fit in, and under what circumstances? And I think the jury is out. But this is the time, these next six months, would be a very good time to be thinking about that.
[Question] Very nice presentation, Ken. I'd like to take a step back and think about the traditional challenges in public health, which frequently targ -- the diseases look for vulnerabilities in the public, which tends to frequently be lower socioeconomic populations, and so at one level, obviously, you're talking about kind of, Boston, very highly wired, and then we kind of got Mississippi and other places at the other extreme, and inner cities, et cetera. What's the experience in terms of technology adaptation in other domains outside of health? Are there ways where we could somehow get back into those lessons? So, for example, would we need to provide, let's say, free software to inner city practices and physicians, the storefront docs, et cetera, in order to allow us to have those kind of entrees using this model?
[Dr. Mandl] Yes, I believe that the model that’s emerging under the ONC spending plan is to pay -- is to do exactly that, to pay docs to provide some free systems and to pay docs to put those systems in. Okay? So they're very, very low starting costs. And what I would argue for, as we go forward with that plan, I think that plan, I think that's kind of a done deal. That's what's happening. I would argue that we really think about making sure that these systems have some kind of programming interface so that if we want to drop in new functionality that is not created necessarily by the original vendor, but that might be created by a really hot, small start-up that just -- they just get it, they just made something that everybody wants, that just because you've taken advantage of the free start-up package for your EHR, you're not now locked out of an innovation that happens one week later. Okay?
There are ways that companies who want to can get that innovation into your practice. Okay? And the costs will be lower, I think, if there's competition for your desktop, and subsidy is going to, I think, probably be very important to reduce disparities among different types of practices. I would see that as a very likely policy decision.
[Question] Oh, push that? So, among those 100,000 people are my graduate students at Georgia Tech, who were required as part of a course I teach to create a Google health record. I have yet to have a graduate student who didn't find that intimidating. And it seems to me, some of the same issues that plague EMRs also plague personal health records -- complex data set, lots of unfamiliar terms, unless you're a healthcare professional. Just interested in your comments on that and where you think things are going to help with that.
[Dr. Mandl] I think we have to be very careful with what we share with patients and make sure that it's really in a usable format. I think that the recreational genomics is one example of giving non-actionable data back to patients, or consumers. You know, NaviGenics, 23andMe, I don't know what people are supposed to do with that, or whether the data have any value. Most of the data are in fact predictive of what they say they're predictive of. Similarly, ICD codes caused a bit of a flap. People may or may not have seen -- it made the news when a patient at Beth Israel Deaconess who was a patient, a very articulate patient advocate called E-patient Dave took out a Google health record and found inconsistencies and inaccuracies in the ICD-9 codes that he was -- that were being displayed to him. These are diagnostic codes that were being displayed to him in his health record, kind of in parallel with the clinical data. And anyone who's ever billed for healthcare or looked at billing codes knows that those are not directly translatable into clinical variables. And within a day or two, Beth Israel Deaconess had to pull the ICD codes out of the feed into the records.
So there was an example of probably putting in something that was -- it might be useful for a machine, so a decision-support engine might really want to have access to those codes and be able to process them in interesting ways with artificial intelligence, and it could discount codes that it knew -- but a person just reading through these may be misled and not particularly benefited by looking through these codes as if they're clinical data. So, I think we have a lot to learn about presentation of the information and what's useful. I think largely, the raw data for most patients, unless you have a biomarker for your disease that you're following, that you need to follow -- hemoglobin A1C is of course the greatest example in the world of this. If you've got a biomarker that if you follow it, you're going to be healthier, and you act on it, that's great. But most diseases aren't like that, and I think it's gonna be a processed intelligence that's coming from these data that's gonna help guide people, and not so much just flipping through and scrolling through the raw data on the machine, which is really where those records are right now. Yeah?
[Question] Has there been any thought to allowing financial information to go into these records? I mean, one of the headaches that patients have is dealing with all of the bills from all the different places. And you talked about Quicken, and so I wonder if you've thought about doing that.
[Dr. Mandl] Well, Quicken has thought about it. Their products in the health-record space, and they came to our meetings, too, but they were very clear that they're gonna be much smarter than everyone else and they're gonna go -- you know Sutton's Law, right? They're gonna go where the money is, and they're gonna help people manage their finances across the healthcare systems. So you should take a look at what some of the Intuit products are doing. So, yes, very hard for people, and we'll see what happens under healthcare reform, if it gets more difficult for individuals to understand, or less. It's not clear.
[Question] In your presentation earlier, you implied, in addition to technology, there is a significant gap, absence of public policy around personal health records, and having it naturally adapt is a huge catalyst to adoption. Traditionally, policy set by large corporations, whether financial or technology, are not always in synergy to best public interest. Do you see, in current efforts, like IHE efforts, they don't focus on public policy at all. Do you see, in the future, a separate initiative just targeting public policy, setting a public policy where all the interested groups are brought in and maybe down the road, having an agent, such as Consumer Financial Protection Agency, similar to that, created specifically for this purpose.
Dr. Mandl] What did you say about consumer -- I missed the part you said about consumer finance something?
[Question] Currently, a Consumer Financial Protection Agency is being created. Something similar for the personal health records.
[Dr. Mandl] Yeah, so, I think there are emerging efforts to do that. There's -- look at using the federal certification process for EHRs, for personal health records. There are advocacy groups, many of which are very privacy-oriented that are looking at this. It will be very important to have some consumer advice available. What we're working on in our system -- one thing that's really -- if you want to think about this in a semi-technical way, that layer, the interface between all the data and all the programs that are dropping down, those little squares on your iPhone, that's where all the magic's gonna happen in terms of whether you're being really protected in your interests, because that is gonna either transparently make it very clear what that application is gonna do, like that application is gonna suck all the data into it and then send it to the drug company. Okay? That might be okay. But it should be very, very clear what's happening. Okay? And there may also be policies and guidance that we can have around what those applications should do. Like, perhaps it should be stated that they should use the minimum amount of data that they can, and store the minimum amount of data that they can. Because once the data leaks into these applications, you don't have all those nice personal controls that you have that are built around the secure repository.
So this is a real issue, and one that it is extremely hard to explain to a consumer, and I think that we -- it's, we're gonna be obligated to learn to educate people so that they can understand something about this, but also to just make it safe enough that you don't have to understand too much to use these things very safely. But I am sure there will be some authoritative voices around consumer protection as these things get used more.
[Question] Hi. I'm actually a guest here from Georgia Tech, but I wanted to follow up a little bit on your thoughts on education. And considering the questions about how hard it is for patients sometimes to use their personalized patient records, do you think some thought needs to be put into some kind of policies about including appropriate levels of education and decision support together in applications that actually allow people to view their medical records? Or whether just sharing information, just data, clinical data, with them might be harmful rather than helpful in some cases?
Dr. Mandl] So, whether there should be policies about that? I think probably only at a very broad level. I mean, I think what we need to do – I mean, what's gonna happen, I think, is that useful applications will be developed that people will want to use. There may need to be some -- there may be ways to, in some way, certify certain applications as sort of safe. It may be through a real certification process, or it may be through a professional organization that just says, "This thing seems like a really good thing you might want to use." Or maybe the American Academy of Pediatrics says that the CDC gives really good advice about immunizations. You know, you -- "this is where you should go." And a hospital that's giving out, or a primary care practice that's giving out records to patients could pre-load them with some decision support that it thinks is good.
We don't have to make it so you have to navigate the whole world for every consumer. But it's gonna -- I think there's got to be some innovation, some competition among applications, emerging, good applications, and then some way to make sure that there's a lot of transparency, and then ultimately, some regulation about practices that would be abhorrent in some way.
[Question]Yes, this is a little off-point, but it just, I think it follows the other question. Now, I've really been taken by how popular, at least, in my son's circles and the whole part of Houston he lives in, and we actually bought one, too -- a Wii Fit -- if you're familiar with that. And it's so popular, you know, it's hard to find them in stores sometimes; they keep selling out. For those who don't know, it's a way of getting physical activity through either doing the physical activities themselves or doing games, like bowling, and so forth. It seems to me there's an opportunity there, because it asks questions about your age and weighs you and does all that stuff. There's opportunities there to ask some questions about if you're eligible or whether you've had a colonoscopy or a pneumococcal vaccine or flu vaccine or whatever. It also seems like that with the -- with drugs, there are opportunities to ask questions of patients, for example, "the health department offers TB drugs free. Would you like to get your free TB drugs from the health department?" So, I guess the general question is, you know, how receptive do you think people who are producing various applications would be to talking with us about embedding, you know, public health messages in various systems?
[Dr. Mandl] You know, it's a great idea. I've had some exposure to conversations like that through my work on the advisory committee, where big companies like Target and Walmart are very interested in understanding what they can do for the public health. But there does seem to be, there is a line that gets drawn at some point. So, I – you know, in other words, I don’t think -- so Target would seem like they probably wouldn't be flashing information up about colonoscopies in their aisles, because that kind of doesn't go with their branding and the dog with the -- but they would help promote things like flashlights for kids going out for Halloween.
So... but, you know, that's the business world. And I think there's two approaches to that. One is to continually educate them and to remind them that their employees, for whom they're paying for healthcare, are also customers and that we need to do something at the national level. And the other is to be creative and to understand at least within that channel what messages the CDC can work out to get into. And I know the CDC is very engaged in placement in television programs about public health messages. So I think you're – I mean I think it's very important to start to embed these messages in different applications and in different media.
[Announcer]For the most accurate health information, visit www.cdc.gov or call 1-800-CDC-INFO, 24/7.