Across the country, practices are creating billions of clinical notes every year. These are increasingly stored electronically using Electronic Health Records software.
Together these records add up to mind-boggling amount of data. And with great data, comes great possibility. Big health data has the potential to improve public health, improve the quality of care, and make hospitals better. But great data also comes with great responsibility. The data contained in EHRs is sensitive, and incredibly valuable to hackers and well as hospitals.
Either way, the data is only growing.
Healthcare reform is one of the biggest contributors to the explosion in health care data, specifically, requirements that providers invest heavily in information technology. The EHRs designed to digitize patient records for data capture and sharing enable healthcare organizations to meet requirements of Meaningful Use. EHRs are supposed to improve the quality of care and reduce cost by making patient medical history more accessible, reducing errors through computerized order entry, and increasing charge capture.
So how’s that working out? Let’s investigate.
Big data can improve public health
Automating and routinizing data capture makes gathering information cheap. And that leads to insights.
For instance, it turns out Ben Franklin was right, at least about the happy: people who go to bed later and get up later are less happy than people who go to bed and get up earlier. Widespread tracking also allowed researchers to discover that South Dakota residents eat more fruits and veggies than Californians. This is the kind of information big data yields. Researchers at Stanford, such as Associate Professor of Medicine and Genetics Euan Ashley, MD, PhD, are sharing health data from medical apps like MyHeart Counts and finding fascinating new connections.
The big health questions that big data can answer include which factors actually lead to good health. If we can identify the behaviors, risk factors, and early indicators of disease, we can prevent it more effectively.
Consider that at any one time, an estimated 26% of Americans age 18 and older are living with a mental health disorder and fully 46% of Americans will have a mental health disorder over the course of their lifetime. Serious mental illness has many co-occurring conditions, which reduce life expectancy by 8-17 years on average and big data can reveal the behaviors associated with developing mental disorders, thus identifying them early and saving lives.
Another possibility for big data in healthcare is increasing medication adherence, a $300 billion problem and one of the leading causes for hospital readmission. We can identify which patients need more help taking their meds on time. “Maybe he or she suffers from depression or knows English only as a secondary language,” said SyTrue CEO Kyle Silvestro. “These are the critical factors, and the information is already there, but we need the ability to select and use it easily, to manage our populations correctly.”
Big data can make hospitals safer
The limits of self-reported data are well-documented. Not shockingly, when caregivers are asked to document harm they cause, they only identify a small percentage of total harms. This makes figuring out where errors are most likely to occur harder. Big data to the rescue. When we know which events or lab measurements are often associated with harm, we know where to apply greater scrutiny. And it works.
Researchers created a “trigger tool,” to find instances of harm across 600 medical charts from six U.S. children’s hospitals. It flagged the medical chart for detailed exploration if a trigger occurred.
They found that almost 25% of patients included in the chart review had experienced at least one harm, and that 45% of these harms were probably preventable.
This flagging can be done automatically and incredibly quickly. Watson can read through many thousands of medical documents per second, searching for clues, correlations and insights. It can help train medical students, and doctors. IBM is developed a software program called Watson Paths, allows a doctor to see the underlying evidence Watson uses and inference paths Watson takes to make a medical recommendation, and display the information visually.
Clinical settings such as oncology are even starting to deploy AI to offer diagnostic and treatment recommendations.
However, we’ve got to be real about where we are right now. So far, the systems have not been smart or fast enough to really help doctors in day-to-day practice.
Danny Hillis, an artificial intelligence expert, explained to the New York Times what is required to fully utilize technology like Watson Paths, “The key thing that will make it work and make it acceptable to society is storytelling.” Not so much literal storytelling, but more an understandable audit trail that explains how an automated decision was made. “How does it relate to us?” Mr. Hillis said. “How much of this decision is the machine and how much is human?”
But perhaps the biggest hurdle to fully utilizing all the amazing data that’s being collected is that right now so much of it is not structured.
According to SyTrue, the cross-industry consensus is that approximately 80% of all healthcare data remains unstructured. This estimate is based in part on the structuring required to meet coding and reporting requirements. “Clinical notes have, so far, been excluded from analytics data flows because they are not discrete and the clinical richness they provide remains lost. If anything, the EMR/EHR– penetration has added digital waste and administrative costs.”
To fully realize the public health potential of big health data, the technology to analyze the data will need to progress a little more. It’s also imperative that the data isn’t just gathered, but structured.
Big data means big breaches
According to Healthcare Dive:
- A data breach now costs organizations an average total of $3.8 million, up 23% from 2013.
- The worldwide per capita cost of a stolen record in the healthcare industry is $363, making it the most expensive industry for a breach.
- 49% of breaches in the U.S. are the result of criminal attack, rather than human error or a system glitch.
And no one is sure what the regulation around health data requires. GovHealthIT warns that mining individually-identifiable health information could constitute a breach of patient privacy if the analysis falls outside of the scope of HIPAA. “It is not clear whether using patient data to improve products, as opposed to health outcomes, is allowed under this law. And an even more concerning scenario could take shape if health information were combined with other personal, non-medical data for the purposes of user profiling.”
So how do you mediate the cost of a breach? One study indicated that an incident response team can create over $12 in per-capita savings. In addition, medical data security best practices such as encryption, employee training, BCM involvement, a CISO appointment, board-level involvement, and insurance protection can all make a data breach less likely.
EHR implementation is leading to record-breaking amount of medical data. If properly structured and analyzed with machine help, this health care data has the potential to drastically improve public health and make hospitals safer.
However, patient data is sensitive, and valuable to hackers. It’s essential hospitals and clinics take care to protect this information.
What is your practice doing with big data? How can you better use the healthcare big data you have? Let us know in the comments!
Images by Abby Kahler
Looking for Electronic Medical Records software? Check out Capterra's list of the best Electronic Medical Records software solutions.