AMCHP 2005 ANNUAL CONFERENCE
DELIVERING RESULTS, IMPROVING PREGNANCY & BIRTH
February 19-23, 2005
DAVID HOLLAR, JR: Thank you, Lori. And I thank all of you for being here today. The title of my presentation is Linking Public Health Databases, Developing Partnerships, Infrastructure and Testing. And this is kind of an overview of the work that we've been doing with our two grant funded projects. One from Health Resources Services Administration and the other one from the Agency For Healthcare Research and Quality.
Electronically linked health databases call for the opportunity to improve healthcare delivery, provide single secure website access to data for children and anyone with those health records, provide comprehensive medical and health‑related data. Reduce redundancies in the data and to reduce medical errors, both errors of commission and errors of omission. Provide long‑term tracking of children with various conditions to make sure they're receiving appropriate services, such as keeping up on their immunizations, for instance, and of course avoiding lost cases which is an important issue that's brought up by, for instance, Allen Hinman has brought up a number of very eloquent presentations, some very good data illustrating how important this is to avoid losing any children within the system.
A lot of these recommendations and arguments for electronic linked health databases have come from reports from the Institute of Medicine, as well as from other groups, including the Public Health Infomatics Institute, which has run projects such as All Kids Count, and the Connections Community of Practice, as well as groups such as the Public Health Data Standards Consortium, just to mention a few.
The Tennessee Genetics Plan has the goal of linking together electronic health records, principally newborn screening, hearing screening and vital records and integrating this with the health infrastructure. So to provide improved health services to children with special healthcare needs and particularly children with genetic or metabolic conditions.
This plan was developed by the initial applicants for the HRSA planning and then implementation grants to link together these records and to improve healthcare services.
The grant was awarded in April of 2003. Major goals of this HRSA funded project are to develop an electronic child health profile that links together newborn screening for genetic or metabolic conditions, hearing screening, birth certificate, vital records and other databases as they become available. And additionally to make this information available in a secure environment that's accessible to providers with appropriate permission from parents and/or guardians.
Here you see the overall scheme of our project. On the left we have major databases that we are looking at linking together to make a central data warehouse, which we call Ten CHP, or the Tennessee Child Health Profile, a company called Neometrics which many are familiar with, the Division of Matis, and they had a booth here at the conference. They link together newborn screening and hearing screening data that are made available at the State Laboratory in Nashville Tennessee, and the state contracts with them to do that.
In Tennessee, like in all of your states, there is, when a child is born, blood spots are collected on filter papers with demographic information that's provided and if an audiologist is available, they will conduct a hearing test at the birth hospital and all that information will be recorded on the filter papers and they're sent directly to the state lab.
The state lab runs the test, gets the results and puts them into electronic format, and includes linking the hearing records with the newborn screening results.
One of the initial successes of our project was establishing a virtual private network linkage or connection to transfer this data and make a replicate copy of it at the state health department located in downtown Nashville at the Cordell Hall building, and we established that connection, and our data manager who works at the university but is based at the health department updates those records as they come in on a daily basis over the VPN linkage and stores this replicated data on Oracle database servers behind the secure fire walls at the Tennessee Department of Health there the data manager is working at linking those data with birth certificate and vital statistics, particularly hospitalization records. And other data including case management data as they become available. And we're looking at other databases as well.
Together, these linked data comprise Tennessee Child Health Profile. We're almost completed with the secure web browser application, which will be utilizing in a pilot test that will be starting in April or May. There have been a few delays there. And I'll be talking about that as some of the issues that we face.
There's a secure parental permission form where the parent grants permission to designated providers to access health records. So here are the providers who will be involved in the pilot test. Namely physicians and genetic counselors and social workers at the University of Tennessee Developmental Center in Knoxville . Children's special services case managers for the east Tennessee region, which covers about 15 counties, and pediatric endocrinologist at East Tennessee Children's Hospital who work with special cases, particularly conditions such as hypothyroidism, for instance.
Any how, when a family comes in with a child for follow‑up visits, they will be read and told about the project and invited to sign the permission form and at that point, once they've given permission and that's been sent to the channels, the user will be able to access the records.
Our partners for the project, again, UT Department of Medical Genetics and Tennessee Department of Health. There are many divisions of the Department of Health that have worked very closely with us in getting the agreements together on linking the data and sharing the data, and in developing the project for the pilot test. Project development included internal review board approvals from the University of Tennessee Graduate School of Medicine and the Tennessee Department of Health, as well as a project approval from the Tennessee Office of Information Resources, which is in the Department of Finance and Administration. OIR has to approve any type of electronic data sharing protocols, any projects involving the sharing of confidential information. We might add Derrick Chapman is in the office and he helped us out considerably with the development of that project approval.
Our guidance comes from a state genetics coordinating committee, which consists of about 36 members from across the state, very broad based in terms of its composition, representing the key stakeholders in providing services for children with special healthcare needs. They include the three primary genetic centers in Memphis , Knoxville and Nashville . The Tennessee Department of Health, the Tennessee Department of Education, Early Intervention Services, as well as child development centers at UT Health Science Center in Memphis , the Bowling Center for Developmental Disabilities and the Vanderbilt University Center for Child Development. Also the Council On Developmental Disabilities, Tennessee Department of Mental Health and Developmental Disabilities, the Bureau of Ten Care, which is the state supported Medicaid program. The Tennessee Division of Mental Retardation Services, the major universities in the state, including the major health sciences centers, consumers, including parents of children with genetic or metabolic conditions. I might add also consumer advocacy groups including the March of Dimes and Family Voices. We recently added two state legislators to the committee.
Additionally, there are numerous very active subcommittees for the State Genetics Coordinating Committee that help in various aspects of the project. For example, there's a newborn screening committee and newborn hearing screening committee, newborn screening finance committee. Additionally, an ethics committee, which is headed by a pediatrician who is also a lawyer. Other activities including professional development and these groups work closely. They meet twice annually in Nashville to discuss issues related to the project and present project results.
In 2003, out of nearly 79,000 births, there were 125 confirmed cases of the following genetic or metabolic conditions. At that time there were six conditions being screened as part of the Universal Newborn Screening Protocol. In 2004, several additional conditions were added to the screening. These included homocystisurea, Maple syrup urine disease, and MCAD, and most importantly, due to advocacy from consumer groups, the legislature mandated tandem masspectrometry screening for amino acid organic acid and fatty acid oxidation disorders and about 50 of these conditions were screened beginning between April and July of last year. So currently Tennessee is screening approximately 60 genetic or metabolic conditions.
In 2004, there were 169 cases across these 60 conditions. That may increase by one or two, a few cases presumptive positives that were assessed in late December, so that might increase by a few. And I'll thank Jim Gibson at the state lab for getting me to figure on your handout the one error is right here, has a greater than 72,000. He sent me an e‑mail giving me the exact number. Almost 81,000 births. You can make that minor adjustment.
Major task in our project that I'll be discussing during the next few minutes including developing partnerships, sharing data, our major successes of which there have been several. Setbacks and challenges that we've encountered. Linking data, and our efforts to get this implemented or at least in the pilot test format within a few months.
In developing partnerships, the key issues here really boil down to one thing: One‑to‑one communication with the key holders of the data. And working very closely with them. You know we can't rely just on e‑mailing or telephone communications. We've got to get there and have the meetings. And we had countless meetings during the last year to establish the data sharing agreements and outline the efficacy of the project and how it has one overall goal, to benefit the children. Now, I don't have that listed here but really that's what this is all about, the benefit to children. So we had to identify the key holders of data because there were a lot of data silo issues. There still are. People don't want to let go of their data. They're worried about HIPAA regulations and the conserving the confidentiality of the data.
In approaching the key holders to data we identified liaison experts who could help us in getting to the right people to talk to and setting those meetings up and repeatedly meeting with them and explaining to them how this is going to benefit the children.
Communications with division directors to explain to them what we're doing. The Department of Health is a big place, and it's easy for multiple projects to get mixed up. So communication was critical. We worked through influential supporters in the department. While there were a lot of concerns, I would say the majority of the people in the department supported our project, and continued to do so.
We clearly outlined project benefits, identified and address partner needs and tried to address the things that can help their project such as we worked with the CDC early hearing detection program. We're interested in helping them with that and leveraging our project and help theirs.
And throughout it all, the important thing is to maintain intensity and momentum. There's certainly bad days when it doesn't look like anything is going to get accomplished, but we've been able to turn those things around by working with the key holders of the data.
In terms of sharing data, we've had to address data silo issues. As I said, that continues and will probably constantly be an issue throughout the project.
We demonstrated the project efficacy, and we've been able to cite other states such as Rhode Island, for instance, which has done so well with their Kids Net Program, and just trying to model ours after their program, and also other programs that are moving along in the forefront of this effort such as Oregon and Colorado, for instance.
Communicating with the data holders again and ensuring that we're going to be preserving the data in a secure fashion. The Ten CHP data warehouse is a read‑only format. So there's not going to be any modifications in the data. Everything that comes into the state system goes through the appropriate channels and the appropriate departments so the data can be stored, and then we look at linking it in the replicated format at the Tennessee Department of Health. And of course establishing accountability and putting in all the necessary safeguards for the data warehouse.
Our major successes have been achieving those data sharing agreements and working at linkages for three databases. We've had pretty much success with two of those databases. The third one, hospitalization database, we've still got to work on, because pretty much the databases that were given to us did not have many identifiers in them. So that's going to be a problem that we're going to have to work with, as we move along.
As I mentioned before we had approval from the Office of Information Resources for a project that, a pilot project that will last one year once we start it in April and May. We have two IRB approvals, including the appropriate authorization on the permission forms for the parents to sign for the case managers can access the data.
We have involvement of the case managers. They've completed a questionnaire outlining their particular data needs in the web browser applications so that we can put those things in as we finalize that. And last January our project officer for the HRSA grant, Debbie Windsor, suggested we go after an HRQ IT planning grant and it was a lot of competition for these grants. One figure it was quoted to me was up to a thousand applications and 38 were awarded, and we were one of the fortunate awardees. And so during the course of the at least since September we've been working on planning partnerships to expand our HRSA grant not to include just children with genetic or metabolic conditions, but to go into other conditions various types of developmental disabilities, to more comprehensively cover the many different types of special needs faced by children.
An additional benefit from this is that we discovered a group at Vanderbilt at the biomedical infomatics, Center for Better Health headed by Mark Frisk, called the Volunteer E‑health Initiative. This is also an HRQ funded project, I think something like $5 million put into the project. Additionally Governor Brettison has put in several million more dollars into this project, and they are using Vanderbilt's Star Chart Medical Record System, which was developed in‑house as the data infrastructure for linking health records for 12 regional hospitals in the Memphis area in a three‑county area surrounding Memphis . It's an ambitious project and they're going at it very aggressively. They have a great design for it in the way they're approaching it. And they're working with various hospitals there, such as Labonner and St. Judes to link together health records on everyone in that particular region. So we see this as a great opportunity to get not just the pilot test done for the HRSA project, and not just implementation for our art grant, we'll be going after an implementation grant in the limited competition this spring, but not just looking at those two projects but also working with the Vanderbilt project in three different sites across the state to get this thing closer to a statewide implementation. So it has really high potential if we can get all the teamwork together.
So we have statewide cooperation. Now, it is concentrated, our project, with the HRSA grant, with the Ten CHP project, it's based in Nashville, but we're going to be pilot testing it in the 15 county region surrounding Knoxville.
Our IT planning grant with ARC would also be pilot tested in this area. I'll describe that later. And, of course, the Vanderbilt project is over in Memphis in a three‑county area here. So concentrating on the three major metropolitan areas but expanding into the surrounding rural areas, which is very important. But ultimately we potentially can get the state linked together and the advantage that we have to Vanderbilt's project is that we're focusing on children with special healthcare needs, whereas they're looking at medical imaging and Ten Care records and things like that for everyone in the service area.
So we have that focus, which benefits them, and their infrastructure can benefit us. So we each have something to give to each other. Setbacks and challenges that we face include lots of personnel changes at the Tennessee Department of Health. There have been political changes there. And additionally, like most states, there have been budget cutbacks and you have departments where there are maybe three people doing the job of 16. And situations like that. And so there's been a turnover, and of course some highly qualified individuals have sought good opportunities elsewhere, but we continue to communicate with them. And additionally with their own project our data manager accepted a position with the state that was higher paying. So we had a little bit of a slow‑down on our project. But actually he ended up in a position that may help us. So that was good. And currently with our two grants we have two data managers on board now who are working at finalizing the Ten CHP project for the pilot test and also working on getting the infrastructure for our ARC grant.
Technical issues, of course, have been identifier issues that I'll get to shortly. Also, various aspects of the data formatting and getting those linked together. We don't see that as the major obstacle, those can be overcome with the time and the programming involved in getting those together. But throughout it all again it's the communication issue and the teamwork issue and the overall goal of making something that's going to be useful that's going to work and it's going to benefit children.
Linking data, we were looking at matching records between databases. There are serious identifier issues but we think we'll have a rough system working for the pilot test and we'll iron out details on that as the pilot test goes during the course of the year.
There are lots of data duplications and lots of other issues, all kinds of things that crop up in the data, such as like in one database there may be a name of a child with their natural parent but in other database it turns out they're adopted so we're going to have to screen that out so nobody finds out that type of confidential information.
In terms of looking at data linkages, this is just between the newborn screening and the birth certificate records. We've got records all the way back from current all the way back to 1991 on the newborn screening. For the birth records, I think we've got everything from 1990 through 2002 that has been cleaned by the state. Initially, when not many conditions were screened between 1990 and 1995, when we look at those data, there's not much in terms of matching. About 18%, when you're looking at all children in the state. However, with more conditions being added and looking at records and newborn screening, that are also present in birth certificate, we readily get 62% records matching.
Now, the other records are undoubtedly there, but they're things like name misspellings and maybe using the mother's maiden name and things like that, where we're missing them in this matching. So these are deterministic results in our matching, and we're working on probalistic approaches and combination of identifiers using systems such as Sound Dex which Colorado is using to look at misspellings in names. So we'll be using that as well as using fairly sophisticated but certainly workable algorithms such as what Sherry Spence is doing in Oregon in terms of probablistically identifying the records between different databases.
The best identifiers that we've found so far include the child's last name. The child's first name doesn't work because usually it's reported as boy or girl. Okay. Date of birth is fairly, is pretty accurate. As is ethnicity, gender, mother's last name and it improves when you add in maiden name, which is often reported as the last name. Mother's county of residence is much better than the birth county, as you would guess.
In terms of implementing the pilot project, the web browser development is close to being finished. One of our data managers specializes in that area and he's finishing that up. And that will be ready for training session for the case managers that will be held in March.
The case managers have completed a questionnaire that's helped in finalizing the details on the web browser application in terms of what fields they want to get to right away and again the training session set for March and pilot test we're looking at starting in April or May at the latest.
Our methods of evaluation, in looking at the data, include looking at percentage confirmed matches on subsequent passes in the probablistic approaches we'll be using. The percentage of live births screened and those adequately screened given that often in the newborn screening there are results that come in several times and we want to look at follow‑up records as well.
Timeliness of diagnoses and timeliness of treatments are issues that we want to look at, as well as long‑term assessment of infants with positive conditions for various genetic or metabolic conditions.
Additionally, percentage of false positives for a test. Percentage of infants with pre‑birth identified medical home, as well as percentage of public and private providers who are submitting data to the integrated database, to look at how well the hospitals both public and private are providing information into both birth certificate and newborn screening.
Also looking at user satisfaction with the Ten CHP database will be issuing surveys for the case managers as they use the database and they pinpoint problems, and of course we'll still be ironing out the identifier issues as that goes along. So we'll be looking at how well the records that are present in Ten CHP match up with the hard copy records that they and that we have at the UT Developmental Genetics Center. And certainly we'll be providing questionnaires to parents and guardians in terms of their satisfaction with services received.
And of course overall we're looking at, in this project, addressing specific goals for Healthy People 2010, particularly targeting chapter 6, disability and secondary conditions as well, and also looking at strategic goals for health improvement for children with special healthcare needs as identified by HRSA.
In terms of project expansion, as I mentioned, we have a second grant which is a one‑year planning grant that 38 groups across the country received and those 38 grantees will be competing in a limited competition for 14 implementation grants this spring.
With our IT planning grant, we're looking at developing a parallel database, which we call UT Chip, University Tennessee Child Health Profile, that isn't based at the health department in Nashville but is instead based at the UT Memorial Hospital in Knoxville . But we're pretty much making it parallel with what we're doing with the Ten CHP database. And what we're looking at is direct electronic transfer of newborn screening, hearing screening, other genetic and metabolic disorders and developmental disabilities identified by child development centers directly to the database that we're making in Knoxville . And this wasn't our original diagram. With our discovery of the Vanderbilt project, also funded by ARC, the volunteer E‑HEALTH Initiative, we're looking at data sharing with them and looking at the infrastructure they're using in terms of designing our own and sharing data between the two. So data will be flowing into our database as well as into the volunteer E‑health initiative data vaults, as they refer to the way that they're storing data, and other databases will be incorporated.
Just like with the Ten CHP database, we'll provide a secure web browser access with parental control over who accesses their child's data. And the primary users, when we get to implementation, and we're confident that we'll reach that point, not just one acute care hospital but several. Certainly UT Memorial Hospital in Knoxville will be one of the acute care hospitals but certainly Vanderbilt Hospital and also UT Health Sciences Center in Memphis and other hospitals involved in the E‑health initiative will be involved in this effort as well.
The three genetic centers, plus various satellite centers in the state, child development centers, particularly the Bowling Center for Developmental Disabilities in Memphis under Dr. Fred Palmer, and also the Center for Child Development at Vanderbilt under Dr. Robert McWilliam. As well as pediatricians in both urban and rural settings. We're working with pediatricians who will be users of this database. And I might point out, while the Ten CHP database is going to be in a read‑only format, UT CHP is going to be predominantly in read only format but we're looking at providing a capability of where various users can input data for tracking. Particularly, at Vanderbilt, at the Child Development Center there, Robin McWilliam had been working with Rue Simianson at North Carolina with field testing of the children's version of ICF. I don't know how many of you are familiar with this. The International Classification of Functioning Disability and Health. It's a neat instrument for identifying functioning of individuals and for tracking their development. And it complements ICD‑9.
So we're looking at that capability as a two‑way information reporting scheme, whereas the database will be in mostly in a read‑only format.
Our partners, as I've already outlined, include the UT Department of Medical Genetics and university HEALTH systems, which actually operates UT Memorial Hospital as a private entity separate from the university. And the Department of Information Services will house the UT CHP database. Vanderbilt University , the Department of Pediatrics, the genetic center there, and the Center of Child Development and the Department of Biomedical Infomatics which has the Volunteer E‑health, ARC funded grant, as well as the Center for Better Health. UT Health Sciences Center in Memphis including the Bowling Center for Developmental Disabilities and the Memphis Genetics Center .
So, once again, statewide cooperation, and with three projects, one in Knoxville, one in Nashville and the Memphis project with Vanderbilt, we're very positive that sharing data between these three projects we can get closer to statewide implementation of a child health profile.
Additional collaborations that have been very helpful include a regional grant to the Southeast Regional Genetics Group, headed by Genetic Centers at Emery University and Tulane University, and the purpose of this grant is to improve genetic services information and collaborations between genetic groups and the south eastern states, namely Louisiana, Alabama, Mississippi, Georgia, Florida, South Carolina, North Carolina, Tennessee and Kentucky.
Additionally, we are members of the Public Health Infomatics Institute Connections Community of Practice, which is a group of people groups across the country, various states, that are involved in these data linking projects, and those have been very fruitful collaborations, and this is very connected with their previous project which was the All Kids Count Project. Additionally, I've been involved with data standards, reports that have come out of the public health data standards consortium, which is a group of stakeholders across the country interest in establishing proper standards for data messaging and setting up electronic health records that will work and that will be useful for providers. And the Public Health Data Standards Consortium is based at John Hopkins University under Dr. Anna Ovalova. Additionally we've corresponded with other HRSA and ARC grantees as we've outlined plus other HRSA grantees, such in Oregon and Colorado , discussing the issues they faced in getting advice from them, and we've also networked with CDC Early Hearing Detection Intervention Projects in looking at their approaches to data linking.
Our long‑range goal is to link all of these databases together to provide a very comprehensive expanded child health profile for all children in the state, not just children with special healthcare needs.
Acknowledgments: Members of the State Genetics Coordinating Committee, staff at the Tennessee Departments of Health and Education Early Intervention Services, and, of course, our support, as I've outlined, comes from HRSA and AHRQ.
For more information, feel free to contact me. My business card is stapled to the handouts. And, of course, our principal investigator on the project and a fantastic person to work with, she's been doing this probably four decades now, Dr. Carmen B. Losio, who is the principal investigator on the project and her information is provided as well. Thank you.