Carolinas HealthCare System is using big data to yield big insights. By transforming the way it interprets data, it hopes to benefit patients and the community as well as the health care system.
In an initiative to use data such as clinical and financial information to predict local health needs and identify opportunities to improve care, they hope to mine data to identify challenges, define problems, and target solutions, in an effort to become an industry leader in advanced analytics and business intelligence.
To further that initiative, CHS created the Dickson Advanced Analytics Group in January, consolidating nearly 90 employees with specific data expertise into one department. The analytic group’s efforts are currently focused on the CHS’s metro hospitals. Carolinas HealthCare owns, leases or manages 33 hospitals in the Carolinas.
Through this initiative, CHS hopes to meld clinical data with financial information to focus on service lines that can be improved, and to help the health care system to be more assertive in adopting new strategies or adding services, rather than waiting for problems to arise.
Dickson Advanced Analytics (DA2)
“We see advanced analytics as a key strategy to enhance the value of our care,” says Allen Naidoo, Ph.D., vice president of operations at Dickson Advanced Analytics Group, or DA2, the preferred acronym for the CHS’s new big data analysis center. “This type of approach enables Carolinas HealthCare to be more assertive.”
He cites the additional example of the CHS app alerting users to the less busy of the system’s 21 urgent care locations. This simple, timesaving and possibly lifesaving application is but one of DA2’s early results.
“There are a lot of sophisticated algorithms involved in the wait time application,” says Michael Dulin, M.D., medical officer for analytics and outcome research at DA2.
Located in the former AAA Building on East Morehead, DA2 has only been in operation since February 2012. It is the successor organization to the R. Stuart Dickson Institute for Health Studies, an interdisciplinary and collaborative program of applied research and public health studies.
Naidoo, is responsible for the integrity of the massive data collected at DA2. Trained as a biostatistician, Naidoo likes the emphasis on statistics and analytics. Michael Dulin, who has a private practice in family medicine in addition to his work at CHS, insures the accuracy of the center’s clinical data.
Naidoo, age 48, came to CHS in May after a career in health insurance. “I’m beyond the term ‘big data,’” he says. “It is a thing of the past. I like to talk about big insights from data. Companies can have big data, but lack the computing and people power to do something with it.” At DA2, Naidoo and his team of specialists are taking big data to the user level. That often means the physician, nurse or nurse’s aide working at the patient’s bedside.
Best Practice Protocols
One of the first assignments given DA2 was to reduce the hospital readmission rate for elderly patients by 20 percent by the end of 2013. Using data collected from 11 home health agencies and statistical tools that isolate primary and secondary predictors, DA2 discovered the most salient and preventable determinant of readmission.
“The number one predictor was whether patients were adhering to their oral medications,” he says. “Those who had their medications carefully monitored by their home health care worker had the lowest readmission rates. Those on their own with a medication list and little else had the highest.”
Based on these findings, Naidoo and his team developed best practice protocols for all 11 home health agencies. When workers spent more time on oral medication instruction, the result was 200 fewer unnecessary readmissions, a six percent reduction. That Big Insight saved CHS over $2 million and helped improve the quality of life for 200 elderly patients who would prefer to stay at home.
Admittedly, big data is relatively new for CHS. As little as three or four years ago, medical data were handwritten on paper charts. Today, that’s all changed. Transcription software has made electronic medical records more accessible, whichs make diagnosis, treatment and outcome data more accessible and easier to analyze.
DA2 integrates vital clinical data with billing information, data from dozens of federally-mandated registries with prescription data and community data like address and English proficiency with psycho-social data such as patient satisfaction, age and sex. Thanks to Medicare and Social Security, physicians and patients have unique identifying numbers that are also entered into the mix. As Naidoo points out, that’s a lot of big data begging for big insights.
The statistical software DA2 uses to analyze their data was developed at North Carolina State University. Software Analysis Systems or SAS (pronounced sass), located in Cary, N.C., originally developed analytics programs for agriculture research. Today SAS is a major player in the business intelligence market and hospitals are some of its best clients.
The heart of DA2 is the Enterprise Data Warehouse, a concept developed in the late 1980s by IBM. A data warehouse gives DA2 a centralized, consistent and unified view across the entire hospital. To date, DA2 has a 1.5 petabytes of data in their warehouse. From this massive storehouse managers produce white papers, graphs and dashboard reports for physicians and administrators.
But that’s not its real strength. That comes from queries and forecasting—asking it intriguing health and business questions and having it build alternative, futuristic care models.
The hospital’s polychronic patients are a good example of the potential of the CHS data warehouse. Polychronics are typically elderly people with multiple chronic illnesses such as diabetes, congestive heart failure, coronary artery disease and asthma. Some are dually enrolled in Medicare and Medicaid—they are elderly, poor and chronically ill. Because of their multiple needs, polychronics consume approximately 80 percent of the hospital’s health care dollars and resources.
“These are patients having a tough time controlling their diseases,” says Dulin. On the plus side, they are a small group, less than 20 percent of all patients. In its data warehouse, DA2 has information on over 60,000 diabetics and 20,000 asthmatics.
By querying the data warehouse, DA2 staff can isolate the polychronics and build model treatment scenarios that lower the cost of their care while at the same time personalizing it. One innovative model mobilizes a corps of human resources and recruits them for the treatment team.
CHS staff coordinates and monitors the work of nurses embedded at the YMCA, home health agencies, hospice workers, families, friends and pharmacies. These medical and non-medical groups support the care plan, provide advance warnings of patient improvement or deterioration, keep the physician manager in the loop and help lower costs. They also provide warm fuzzies like encouragement, hugs and friendship. The medical staff provides all of the treatment oversight and care decisions.
Wait, there’s more...Using the DA2 data warehouse, CHS now sends lists to area physicians of their high risk diabetic patients. The list notes those that smoke, have elevated blood pressure and need their medications refilled.
“As a doctor, I look at the list and have my nurse call the 20 or so patients, ask them to come in for an appointment, get their medications refilled and remind them to quit smoking,” says Dulin. “We are using the DA2 data proactively to help patients stay healthy.”
Limits, Challenges and Concerns
All this emphasis on complication prevention, early action, proactive care and readmission reduction improves quality of life for patients, but not necessarily for the hospital.
“We are penalized to some degree for working on these quality initiatives,” says Dulin. “When you keep people healthy, you keep them out of the system and hurt the hospital’s revenues. We are not recognized for that under our current reimbursement system.”
As DA2 operation officer, Naidoo is troubled by personnel shortages. “It is a challenge to find qualified people to assume the key roles at DA2. Statisticians are not available; epidemiologists, not available; health economists, not available; biostatisticians, not available,” he says.
Academic doctorates are not staying at universities to teach, but Naidoo says, “They are being picked up by industry. With professors in short supply, the United States is not producing enough graduates in math and the computational sciences.”
In a 2011 report, McKinsey Global Institute, the business and economic research arm of McKinsey and Company, also focused on the talent shortages in big data. “The United States alone faces a shortage of 140,000 to 190,000 people with deep analytical skills as well as 1.5 million managers and analysts to analyze big data and make decisions based on their findings.” No wonder DA2 finds so many talent sectors unavailable.
Other issues concern the completeness of the data warehouse. “We aggregate data, but not all systems are involved,” says Naidoo. “There is a ton of data sitting out there untapped.”
In their 2011 report, McKinsey Global Institute raised the issue of data discarded by hospitals. They estimated that the United States health care industry discards 90 percent of all the data they generate. That’s a figure Dulin finds surprising.
“We don’t discard a lot,” counters Dulin. “We hold on to videos and imaging data like MRIs and CT scans. It might be that 90 percent of our data were not actively utilized in some type of report or study, but it’s not discarded.”
The Future of Big Data
In the next few years, big data will help lead the charge on greater transparency and reduced health care waste.
Americans have had access to their credit reports for years—why not their medical records since many are electronic files?
“Physicians are a little scared of that type of transparency, but I think it will be good for us,” says Dulin. In as little as a year to 18 months, Dulin envisions patients logging on to a secure Internet portal to view their most recent lab reports and physician notes. He predicts that transparency will result in greater patient autonomy, fewer data errors, enhanced doctor-patient communication and more shared decision-making.
A study published in September by the Institute of Medicine estimates that the United States health care system wastes $750 billion each year. Thirty cents of every health care dollar goes to unnecessary services, inefficient delivery of care, excessive administrative costs, inflated prices, prevention failures and fraud.
Both Naidoo and Dulin agree, many of these waste categories—especially unnecessary services and inefficient delivery of care—could be reduced if hospitals put their big data to work finding more best practices.