The problem.
When Mark Hulse, current vice president and CIO, joined H. Lee Moffitt Cancer Center & Research Institute in Tampa, Fla., around 2010, the medical center already had a working data warehouse. However, the data warehouse was not meeting the research and patient outcome needs of Moffitt. Moffitt's approach to personalized medicine needed an updated data warehouse.
"Our approach to treating cancer, or any disease, involves diagnosing the disease and then following the set treatment. However, some patients respond well to certain treatments and others do not, even if they both receive identical treatments," says Mr. Hulse. "In order to understand why the treatment does not work for one patient, you need to understand the genetic and molecular level of the disease and the patient." Mr. Hulse knew a data warehouse that could gather, store and manage a vast amount of data — information about the specific disease, the individuals' demographics and past treatments, would improve Moffitt's ability to offer research genetic and molecular levels of diseases and thus, improve overall research capabilities.
The solution.
Moffitt created the Health & Research Informatics platform to aid the gathering of both clinical and molecular data for the Total Cancer Care Protocol. Phase 1 of Moffitt's platform included only Moffitt patient data while the next phase, set to go live this spring, will include data from a consortium of 18 hospitals across ten states. The Health & Research Informatics platform allows a researcher submit a data request for a research protocol or study. The researcher can enter a patient profile into the data warehouse from their desktop and find patients that meet their requirements. The system can follow a variety of specific guidelines such as the disease type, disease staging and various treatment approaches. "Finding a patient sample used to take months and now it only takes a few minutes or so," says Mr. Hulse.
To build a successful data warehouse, Mr. Hulse recommends the following six best practices:
1. Begin with the end in mind. Before developing data warehouse architecture or hiring IT consultants, it is important to define the needs of the data warehouse. The first time Moffitt made a data warehouse, it was not all-inclusive and could not support the needs of all stakeholders meaning researchers, clinicians, administrators and patients, says Mr. Hulse. The second time Moffitt included stakeholder consideration and identified the tools needed to accomplish stakeholder goals. "[Upfront planning] is essential to avoid wasting time and money, says Mr. Hulse.
2. Consult the stakeholders. The input of stakeholders will help develop a comprehensive understanding of expectations of the data warehouses' use and impact so that an ideal and functional structure can be selected. Working with all the stakeholders is important. If you do not recognize who will use the data in the data warehouse then you may face hardships later when the data warehouse does not fulfill expectations. Moffitt had all the right people at the table so that the design of the data warehouse would support expectations. "It is imperative that you know what the administration, clinicians and researchers will be looking for and how they will use the data," says Mr. Hulse.
3. Use stakeholder input to inform case examples. Clinician, researcher and administrator input is most beneficial when the input is utilized throughout the data warehouse planning process. Moffitt developed several use case stories based on how the Health & Research Informatics platform should meet stakeholder needs. For example, the case studies were similar to scenarios detailing the disease in a research trial, the researcher's hypothesis, the search process and the function of the platform in the process. Moffitt then used the case stories to guide the design of the data warehouse. "It is not so much about that data from the start, but more about the architecture of the warehouse," says Mr. Hulse. "The end design is what will support the data for the long haul." The case studies helped Moffitt to picture the ideal data architecture.
4. Define a solid data governance process. Data governance does not have to be a separate option from a data warehouse. "Data governance should be the process of decision making that gets wrapped around the data warehouse architecture," says Mr. Hulse. "Agile methodology helps you narrow your focus so data requests move more quickly." Agile data governance applies learning-capable software to access multiple data locations and then organize and manage data centrally. When building a data warehouse, run-ins with data discrepancies are likely. "The reality is that certain types of data exist across areas. A patient's demographics could be captured at scheduling or registration. Systems do not always match up," says Mr. Hulse. When systems do not match it can cause data quality issues in a warehouse, Mr. Hulse believes it is important to trust the data and use a data governance system to help ensure data quality.
5. Hire a CHIO. Given information is a critical asset to any organization, another suggestion for addressing data governance is to establish an office that is charged with developing data standards and ensuring data quality. Moffitt hired a chief health information officer to oversee their data architecture and work with IT, as well as key stakeholders, to determine key elements over time. "Our chief health information officer is a critical contributor to the design of our data warehouse and oversees how data is managed and used across the organization," says Mr. Hulse. "She also oversees the development of standard data definitions, capture of metadata, which is information about the data such as which source system it came from, and data quality issues."
6. Consider multiple vendors. Moffitt had unique needs for their data warehouse so they developed a Request for Concept (RFC) that was sent out to several potential solution partners. When there are many data warehouse business intelligence vendors to choose from an RFC process allows the vendor to pitch how they would design and implement the data warehouse based on how it is intended to be used. "A request for concept allows you to evaluate vendors and their approach to your needs," says Mr. Hulse. "We selected a several vendors to work together in partnership to guarantee our data warehouse’s success."
The Health & Research Informatics platform is an example of an informatics platform utilizing date warehouse and business intelligence tools along with agile data governance. Moffitt's warehouse is successful because of the upfront planning and effort put into the project. Although Moffitt is an example of a specific data warehouse goal, cancer and chronic disease research, the implementation process could be applied to other hospitals based on the amount of data stored and managed. Other hospitals in need of data warehouses could easily utilize the above best practices to build a successful and usable data warehouse.
Related Articles on HIT:
5 Differences Between Static Data Warehouse Design & Agile Data Governance
Spurring Innovation in Healthcare Delivery: 5 Best Practices of Health System Leaders
The Future of Healthcare: 9 Capabilities for Post-Reform Success
When Mark Hulse, current vice president and CIO, joined H. Lee Moffitt Cancer Center & Research Institute in Tampa, Fla., around 2010, the medical center already had a working data warehouse. However, the data warehouse was not meeting the research and patient outcome needs of Moffitt. Moffitt's approach to personalized medicine needed an updated data warehouse.
"Our approach to treating cancer, or any disease, involves diagnosing the disease and then following the set treatment. However, some patients respond well to certain treatments and others do not, even if they both receive identical treatments," says Mr. Hulse. "In order to understand why the treatment does not work for one patient, you need to understand the genetic and molecular level of the disease and the patient." Mr. Hulse knew a data warehouse that could gather, store and manage a vast amount of data — information about the specific disease, the individuals' demographics and past treatments, would improve Moffitt's ability to offer research genetic and molecular levels of diseases and thus, improve overall research capabilities.
The solution.
Moffitt created the Health & Research Informatics platform to aid the gathering of both clinical and molecular data for the Total Cancer Care Protocol. Phase 1 of Moffitt's platform included only Moffitt patient data while the next phase, set to go live this spring, will include data from a consortium of 18 hospitals across ten states. The Health & Research Informatics platform allows a researcher submit a data request for a research protocol or study. The researcher can enter a patient profile into the data warehouse from their desktop and find patients that meet their requirements. The system can follow a variety of specific guidelines such as the disease type, disease staging and various treatment approaches. "Finding a patient sample used to take months and now it only takes a few minutes or so," says Mr. Hulse.
To build a successful data warehouse, Mr. Hulse recommends the following six best practices:
1. Begin with the end in mind. Before developing data warehouse architecture or hiring IT consultants, it is important to define the needs of the data warehouse. The first time Moffitt made a data warehouse, it was not all-inclusive and could not support the needs of all stakeholders meaning researchers, clinicians, administrators and patients, says Mr. Hulse. The second time Moffitt included stakeholder consideration and identified the tools needed to accomplish stakeholder goals. "[Upfront planning] is essential to avoid wasting time and money, says Mr. Hulse.
2. Consult the stakeholders. The input of stakeholders will help develop a comprehensive understanding of expectations of the data warehouses' use and impact so that an ideal and functional structure can be selected. Working with all the stakeholders is important. If you do not recognize who will use the data in the data warehouse then you may face hardships later when the data warehouse does not fulfill expectations. Moffitt had all the right people at the table so that the design of the data warehouse would support expectations. "It is imperative that you know what the administration, clinicians and researchers will be looking for and how they will use the data," says Mr. Hulse.
3. Use stakeholder input to inform case examples. Clinician, researcher and administrator input is most beneficial when the input is utilized throughout the data warehouse planning process. Moffitt developed several use case stories based on how the Health & Research Informatics platform should meet stakeholder needs. For example, the case studies were similar to scenarios detailing the disease in a research trial, the researcher's hypothesis, the search process and the function of the platform in the process. Moffitt then used the case stories to guide the design of the data warehouse. "It is not so much about that data from the start, but more about the architecture of the warehouse," says Mr. Hulse. "The end design is what will support the data for the long haul." The case studies helped Moffitt to picture the ideal data architecture.
4. Define a solid data governance process. Data governance does not have to be a separate option from a data warehouse. "Data governance should be the process of decision making that gets wrapped around the data warehouse architecture," says Mr. Hulse. "Agile methodology helps you narrow your focus so data requests move more quickly." Agile data governance applies learning-capable software to access multiple data locations and then organize and manage data centrally. When building a data warehouse, run-ins with data discrepancies are likely. "The reality is that certain types of data exist across areas. A patient's demographics could be captured at scheduling or registration. Systems do not always match up," says Mr. Hulse. When systems do not match it can cause data quality issues in a warehouse, Mr. Hulse believes it is important to trust the data and use a data governance system to help ensure data quality.
5. Hire a CHIO. Given information is a critical asset to any organization, another suggestion for addressing data governance is to establish an office that is charged with developing data standards and ensuring data quality. Moffitt hired a chief health information officer to oversee their data architecture and work with IT, as well as key stakeholders, to determine key elements over time. "Our chief health information officer is a critical contributor to the design of our data warehouse and oversees how data is managed and used across the organization," says Mr. Hulse. "She also oversees the development of standard data definitions, capture of metadata, which is information about the data such as which source system it came from, and data quality issues."
6. Consider multiple vendors. Moffitt had unique needs for their data warehouse so they developed a Request for Concept (RFC) that was sent out to several potential solution partners. When there are many data warehouse business intelligence vendors to choose from an RFC process allows the vendor to pitch how they would design and implement the data warehouse based on how it is intended to be used. "A request for concept allows you to evaluate vendors and their approach to your needs," says Mr. Hulse. "We selected a several vendors to work together in partnership to guarantee our data warehouse’s success."
The Health & Research Informatics platform is an example of an informatics platform utilizing date warehouse and business intelligence tools along with agile data governance. Moffitt's warehouse is successful because of the upfront planning and effort put into the project. Although Moffitt is an example of a specific data warehouse goal, cancer and chronic disease research, the implementation process could be applied to other hospitals based on the amount of data stored and managed. Other hospitals in need of data warehouses could easily utilize the above best practices to build a successful and usable data warehouse.
Related Articles on HIT:
5 Differences Between Static Data Warehouse Design & Agile Data Governance
Spurring Innovation in Healthcare Delivery: 5 Best Practices of Health System Leaders
The Future of Healthcare: 9 Capabilities for Post-Reform Success