Joel Shu, Vice President, Clinical Transformation & Population Health, Catholic Health Services of Long Island
Prior to today there were many good doctors and hospital systems. The problem was, since so many of us were documenting on paper, it was impossible to tell who the good and bad doctors were statistically. It was equally difficult to tell if programs and interventions were meaningfully improving patient care. It is hard to improve what you cannot measure. In 2018, the health care industry is a lot closer to being able to capture, document, aggregate, report, and most importantly take action on large sets of data, although much more progress needs to be made.
Since electronic medical records (EMR) became widespread, the onus of capturing and documenting has been placed on the health care provider. Many people think that having computers must make it faster to document. The reality is most physician practices would tell you they experience a 15-25 percent drop in their operational efficiency (e.g., how many patients seen per day) when they implement the EMR for up to a year.
“It is generally true that the younger generation of physicians has had an easier time adapting to computerized records”
Several different factors contribute to the drop in operational efficiency. One is the generational gap in using computers. It is generally true that the younger generation of physicians has had an easier time adapting to computerized records. The second is that EMRs are only beginning to have user interfaces that are user-friendly and come close to following previous workflow. Over the last 10 years, healthcare providers often have to bend their workflow to fit the EMR, as opposed to the EMR making what they already do easier. The third is that many, many more things are now required to be documented on a consistent basis. This can be good or bad – sometimes, it is unnecessary or redundant, much of the time it is a burden to the healthcare provider, but most of the time it is important to have that documentation for large-scale study and action. For example, your great doctor in 1970 may have ordered the right preventative screening exam for you, potentially saving you from developing a future problem - but without discrete documentation it’s hard for anybody but your doctor to know. No one else would know if your doctor discussed basics like smoking cessation and fall prevention, talked to you about advanced care planning, etc. Most people would say that they know health care quality when they see it, but capturing it and reporting it is done by proxy – that is, quality measures. These proxies are not exact, but in order to improve them, we need large scale study.
Aggregating data is perhaps the biggest current challenge in population health. Physician practices in a same healthcare system still utilize potentially hundreds of different EMRs. Despite some recent consolidation in the hospital EMR industry, hospitals are also on a multitude of different EMRs; the ones that have the same maker may be on different instances. Because EMRs were created in a competitive landscape, they do not have natural interfaces to each other. For a health care system, this leads to collecting results across 100 different systems to be intelligently reportable. Each of the 100 EMRs requires a two-week interface creation that may not be cost effective to complete ($15,000). Speeding up these EMR-to-EMR connections is a major opportunity for improvement.
Once all this data for a patient population is tied together, the next piece is the business intelligencesideof data management– namely, reporting and takingaction. Reporting is by no means easy, but it’s probably the most straightforward part of the chain. Many different systems can generate reports. If all else fails it could be done to some degree in Excel, provided there is good clean data in well-categorized fields. The data here is the raw ingredient. The better the data, the better the potential output. Better reporting systems make output easier, less time consuming, timelier, and more actionable.
Finally we arrive at the last part of the chain – what to do with the data. The concepts of population health are simple: get people to their doctors outside the hospital, proactively screen for conditions, control things like blood pressure and blood sugar, keep them from developing worse conditions that increase suffering and cost exponentially more. Just like a machine, earlier preventative maintenance avoids costly catastrophic events down the road. Unlike a machine, people are far more complex and may not see the direct link between what they eat, drink, and smoke to long-term disease processes. They may not have the knowledge or the access to proper care. The healthcare payment system is only beginning to reward preventative care.
Still, in 2018 the health care industry is at long last approaching the ability to monitor quality outcomes on a large scale, and devise the plans and programs to affect them cost-effectively. Despite all the challenges associated with EMRs, collecting that data is the only way to get there.