The McKinsey Study and the U.S. Health Care System: Now for Some Good News…
As I said in my recent post on the U.S. health care system, the U.S. cannot continue at its current spending rate. Certainly, the McKinsey Study (and many other publications) makes this very clear. Now, you and I may have very different opinions on how this can be fixed depending on our political leanings, etc., but ignoring this problem is not going to make it “go away.” Lucky for us, McKinsey takes a look at how big data and analytics can alleviate health care costs in some very promising areas. Whether you share my views on health care reform or not, it’s clear that we need to figure out how to align health care policy and regulations with economic incentives designed to move the industry towards a more collaborative, data sharing approach. To me, this is the “real” health care debate!
If you are a card carrying member of the big data community but don’t know much about the state of health care data, you may be surprised at just how antiquated data collection and usage is. If you have been steeped in the health care industry (and not just in the health care reform debate), this will be old news: health care data is not systematically collected, stored, and used. It is the only trillion dollar-plus industry in the U.S. without a modern information technology infrastructure. Look at is this way: while medical technology has advanced, much of the infrastructure that supports it is paper-based. For example, most of our medical records are stored in files (and I mean physical files) at various hospitals and doctors’ offices. Whenever you see a new doctor or go to a new clinic or hospital, chances are you fill out the same forms (again and again) documenting your medical history, risk factors, allergies, etc.
However, the times they are a’changin. In 2009, the federal government (via the American Recovery and Reinvestment Act) set aside approximately $19 billion to encourage widespread adoption of EMRs (Electronic Medical Records). Hospitals, large clinics, and other health care providers are spending billions of dollars on transactional systems like Hospital Information Systems (HIS) and EMR systems. Why the push? To streamline operations, introduce efficiencies, and provide quality care at lower costs while at the same time, ensuring that privacy (via HIPAA and HITECH) is preserved. However, there is still much to do. According to the McKinsey study, health care data can be divided into four main information silos:
- Clinical data from providers—30% of which is offline.
- Claims and cost data which are not collected in a standardized format and are usually stored in legacy IT systems (often incompatible with other systems).
- Pharmaceutical and medical product R&D data which is considered the most advanced in terms of digitization and use.
- Patient behavior and sentiment data which could be used to look at how lifestyle could influence treatment as well as how to promote wellness initiatives.
Of course, one of the fundamental challenges our health care industry is facing is how to incentivize the sharing of data between organizations while overcoming the technological barriers that prevent data from being shared. Beyond that, McKinsey looks at ways in which the analysis of big data sets can be applied (they call them “levers”) to realize savings or create value in five key areas: clinical operations, payment and pricing, research and development, new business models, and public health.
Clinical Operations. Essentially, this area is focused on how clinical care is provided. Certainly, there is an enormous amount of data collected from patient records (that’s why EMRs are so important), operations, and performance. If this data could be aggregated, analyzed, and anonymized, it could be used to predict which treatments work best for patients, help clinics make “better decisions” about care, streamline remote patient monitoring, and identify patients’ future health care problems (in order to address underlying issues that could prevent the problem from occurring). For example:
“In one particularly powerful study conducted at a pediatric critical care unit in a major U.S. metropolitan area, a clinical decision support system tool cut adverse drug reactions and events by 40 percent in just two months.”
Payment and Pricing. This is the area we are all most familiar with as it involves health care payment and pricing. Automated payment systems that include fraud detection (see my post on how the credit card industry does this so well and Terence’s post on “Fraud Detection by the Numbers”) could help to reduce health care costs as well as streamline payment processing. For example, fraudulent Medicare claims were estimated to be about $47 billion (out of $440 billion) in 2009. Perhaps a bit more controversial (but for those of us who have spent considerable out-of-pocket money on prescriptions, worth considering) is the sharing of risk for new drugs where pharmaceutical and medical product (PMP) companies agree to share part of the costs:
“Several pharmaceutical pricing pilot programs based on Health Economics and Outcomes Research are in place, primarily in Europe. Novartis, for example, agreed with German health insurers to cover costs in excess of €315 million ($468) million per year for Lucentis, its drug for treating age-related macular degeneration.”
Research and Development. The area with the most data would certainly benefit from aggregation and analysis. Predictive modeling can be applied to predict clinical outcomes, determine optimal trial designs, and analyze trial data so that resources can be better allocated as well as identify those products that have suboptimal outcomes earlier in the process, and analyze disease patterns and trends (like a far more scientific, algorithmically based Google flu trends). It could also have a profound impact on a relatively new concept, personalized medicine:
“Impressive initial successes have been reported, particularly in the early detection of breast cancer, in prenatal gene testing, and with dosage testing in the treatment of leukemia and colorectal cancers. We estimate that the potential for cost savings by reducing the prescription of drugs to which individual patients do not respond could be 30 to 70 percent in some cases.”
New Business Models. Once health care data is digitized and “released” from its silos, there is certainly an opportunity for new businesses. For example, there will be an ongoing need for aggregated data sets that cross the silos or go beyond them in order to explore the “unknown” (not just finding needles in haystacks but finding more haystacks and unknown needles). Online platform and communities like PatientsLikeMe.com, Sermo.com, and Participatorymedicine.org, generate valuable data that could be included with other data sets.
Public Health. Of course, it almost goes without saying that aggregated health care data can be used to identify and track infectious diseases to prevent large scale outbreaks. This also would help to make the U.S. better prepared for possible outbreaks; managing laboratory capacity to deal with all aspects of care (for example, increased production of flu vaccinations).
So, what are we talking about in terms of savings or value? Well, beyond the improvement of the quality of care we all receive (how do you put a price on “better” care?), McKinsey estimates $300 billion in value per year with two-thirds of it coming in the form of health care expense reductions (this represents about 8% of the 2010 health care spend). For those of you who think this is just a “drop in the bucket,” consider this: McKinsey is only estimating direct savings or value from big data and analytics initiatives. Tertiary benefits, such as the impact of a healthier population on future costs or how the prevention of full scale outbreaks could reduce future health care spend (in terms of clinical visits, hospital stays, vaccination production, etc.), have not been factored in but will certainly deliver value.
What do I know for sure? As the financial services and retail industries (the big data and analytics application experts) have already demonstrated, once large data sets are available for analysis there are waves of efficiencies that result. Funnily enough, each one is usually introduced by the simple question: what if (see my post on exploratory analysis)? Yes, our health care system is antiquated and in trouble, especially when you compare the cost of care with the quality of care. But, to borrow (liberally) from The Six Million Dollar Man (a television series from the mid-1970’s and yes, I am old enough to remember it!): we now have the technology to rebuild the health care system. The only thing stopping us is, well, us (and this includes our politicians, current health care programs like Medicare and Medicaid, the health care industry, and of course, the lobby groups that support the various interests). Isn’t it time to stop arguing and start fixing?
One more thing: If you are non-profit working on ways to fix this problem and need help with the big data side of the equation, please give us a call. We want to do our part in helping to fix it!