I recently attended the HIMSS Big Data and Healthcare Analytics Forum in San Francisco. The event was a great opportunity to learn from peers at other healthcare organizations who presented their data and analytics work. A plethora of vendors who specialize in healthcare analytics pitching solutions around the hottest new technology developments, such as natural language processing, cognitive processing, machine learning, the Internet of Things, and the smart use of big data? These vendors talked about their intriguing developments and the promise of revolutionary capabilities. Yet, my experience at the event reminded me that there is a lot of non-technical work needed before these new technologies can reach their full potential.
Analytics growing pains
In their data maturity models, most healthcare organizations I heard from are just getting past basic reporting and descriptive analytics. This means they may still be comparing data in Excel spreadsheets, dealing with departmentally siloed data marts and manually pulling reports from disparate systems, often with considerable delays. If these organizations implemented more advanced technology like what the vendors pitched, the investment would be costly and would probably yield minimal benefit.
Though current market systems and applications can drastically contribute to more advanced analytics and improvements to value-based care, one of the biggest roadblocks to healthcare data analytics resides in what needs to be developed outside of technology. To mature their analytics programs, organizations need to look at their culture, processes and workflows around collecting and consuming data before digging in on the technology.
In healthcare facilities, clinicians often question data quality, have minimal access to the data they need or don’t have the governance in place to be given the access to the data they need confidently. These are major roadblocks.
If clinicians don’t trust the data, they won’t use or value it. Similarly, if they can’t access the data in near real-time, care will not improve, and they will feel the hours they spend documenting are only being used to meet Meaningful Use measures and not contribute to patient care. In short, clinicians need to believe data is accurate and accessible, and there needs to be processes in place to leverage the data well and securely.
A basic analytics approach for real results
In contrast to implementing many of the new technologies for complex and advanced functions, I learned a lot of organizations at the event are focusing on developing the right culture and processes and have found ways to leverage more basic analytics to deliver value. For example, one speaker at the Big Data and Healthcare Analytics Forum spoke about how UCLA used analytics to improve scheduling. To improve patient wait times, UCLA created a dashboard to clearly see which clinics have a 30-day wait period and which have same-day wait periods. The dashboard allowed them to identify poor patient experience and then allowed them to work with those providers to reduce wait times and improve satisfaction.
Another speaker at the forum discussed how her organization focused on addressing data quality problems before using that data for more advanced analytics. At the end of an analysis, administrators found that a lot of patient notes were being copied and pasted by providers from one patient to the next. Clinicians often face long, 12+ hour shifts where they must balance their time for patients and proper documentation to ensure quality care. Many clinicians will find themselves waiting until the end of their shift to catch up on their notes from the day, so we can see where the copy and paste technique seems more efficient. To address this issue, the organization used basic natural language processing to find which providers were copying and pasting and not documenting properly. This was done to start improving the data one care provider at a time in order to create a culture where users trusted the data.
Helping hospitals manage data
Recently, I worked with a client to help them implement a population health management solution and risk stratify their patient population. For this initiative, the health system wanted to take into account socioeconomic data points in order to assess which patients would benefit from outreach programs the best.
Though natural language processing can help sift through free notes to see if someone drives or is a heavy smoker, the organization did not want to unnecessarily manage or invest in extraneous technology when they could focus on processes instead. As a solution, my team knew we could achieve the same benefits by simply changing workflows. We worked with the clinicians to have them document patient information in discrete data fields instead of free text notes, which allowed for improved capturing of data within the system with minimal additional work required by the care team.
Takeaways
After listening to many of the speakers’ experiences around healthcare data analytics, I walked away thinking most organizations are still in the early stages of their data maturity model and that all the hype out there is not necessarily reflective of the current healthcare market.
Healthcare is still developing and must take a holistic approach to analytics-based improvements to care. This means many advancements can be made by working with healthcare’s many stakeholders to better the collection and consumption of data. By improving culture, processes and workflows and investing time in these to build a stronger foundation for success, organizations can pave the way for the more advanced analytics capabilities and the benefits we are being promised.
The growing pains are inevitable and necessary, but the future and capabilities of big data and analytics in healthcare are exciting. By continuing to share ideas and learn from each other at events like the Big Data and Healthcare Analytics Forum, healthcare will make significant strides towards what matters the most – patient outcomes.