7 Factors Limiting Benefits of Big Data Healthcare
Many experts have touted the use of Big Data in healthcare as one of the major hopes to improve care delivery and reduce costs. However, challenges remain, and a variety of them need to be resolved before Big Data can make an impact. A report from the Harvard School of Public Health, Mathematica and the University of Michigan took an in-depth look at the limitations of Big Data as part of a report on the state of information technology. (Photo: Fotolia)
1. Data Protection
A major concern with Big Data is data security, and who has the authority to de-identify and use the information. HIPAA has guidelines that regulate sharing personal health information across covered entities. However, the guidelines will need additional refinement to take into account changes in the nature of healthcare data, the volume of information and the types of available information. Even with de-identified data, there’s the potential to identify patients, especially when data is associated with a narrowly defined population. (Photo: Fotolia)
2. Data Ownership
Another concern is data ownership. When patient information was only on paper, patients had to provide written authorization for paper-based information to be shared with other entities. “However, with the advent of electronic health records, patient-generated information and even non-structured data available through social media, determining data ownership has become less clear,” the report said. (Photo: Fotolia)
3. Data Access and Availability
Barriers to accessing and sharing data, across health professionals and health systems, remain a significant challenge. Providers have a vested interest in blocking access to data and not sharing it. Healthcare organizations believe that data sharing might cause them to miss chances to make their own discoveries with the information, “and, thus, an opportunity to leverage their health information, whether for financial gain or personal recognition,” report authors note. “They also fear that shared data may compromise their market share, thus reducing overall revenue.” (Photo: Fotolia)
4. Data Quality
Researchers have continued to question the reliability of claims data as well as the usability of electronic health records data for predictive analytics. That’s because several factors can compromise data quality, including human error, variability in data capture and storage, and inconsistency or evolution of terminology and practice standards, the report says. In addition, Big Data analyses often use secondary data that’s being used “to answer questions for which the data was not initially intended,” the authors contend. (Photo: Fotolia)
5. Data Validity
Varied, imperfect approaches for de-identifying and aggregating health information gets in the way of effective analytics in healthcare, the report said. These factors complicate efforts to combine data across systems, while increasing the likelihood that errors will be introduced, thus impacting the validity of the final data set. The work required to minimize inaccuracies and integrate disparate sets of information wastes time and resources. (Photo: Fotolia)
6. Data Analytics Tools
The report notes a 2013 study that found, “Without standardized algorithms and analyses to support the systematic analysis of Big Data, researchers are likely to generate spurious results or chase patterns that are not real.” (Photo: Fotolia)
7. Data Analysis Strategies
A major limitation to the potential of Big Data in healthcare is the lack of strategic approaches for data analysis. Using large data sets to identify “random findings is both flawed and lazy,” the report suggests, adding that only a multidisciplinary team can unlock lessons from the data by using both inductive and deductive reasoning, “leveraging a combination of analytical and clinical skills.” (Photo: Fotolia)
How to Best Use Big Data in Healthcare
The report concludes that organizations can best accomplish the effective integration and utilization of Big Data through coordinated efforts by individuals who have different skill sets. In particular, this team needs a subject matter expert, an individual with formal training in research, a computer scientist and a system administrator. The report, “Health Information Technology in the United States, 2015,” which the Robert Wood Johnson Foundation supported, is available here. (Photo: Fotolia)