September 16, 2017
Defining Population Medicine and Health Analytics is akin to the proverbial story of the elephant described by the four wise but blind men. One ‘saw’ it as a huge pillar while he felt one of its legs. Another described it as a huge fan while touching its ear. The third one described it as a long tapering pipe while sensing the trunk. While the fourth described it as a rope with frayed ends while perceiving the tail. All were right, of course, from their own perspectives. However, each was wrong if we look at the big picture.
Population Health Analysis may mean studying and analyzing the whole population of the community. On the other hand, it may mean the population that is being taken care of by a specific group of providers. Alternatively, it may selectively reference a particular sub-group of the population segregated by age, gender, medical condition or some other identifying feature. In our case, we have approached the whole population of patients being taken care of by our medical group, comprising of employed and affiliated providers. We have then widened our scope in certain areas to reach out to the whole community, of which we have been and continue to be an integral part for more than a couple of decades.
When we look at the whole panel of patients being taken care of by our Independent Providers’ Association and our ACO, we have found that the most reasonable approach is to risk-stratify them as Very high-, High-, Moderate-, and Low-Risk, based on their diseases, prior history and experience. This risk-stratification is best done in a payer-blind manner, in a uniform approach established by our medical systems and processes.
Using simple data systems and reviews, with our preliminary focus being on real-time and accurate data, we found several streams of information which help with our risk-stratification. These data sets are from clinical, financial or operational inputs, from both internal and external sources. Clinical data can be referral-based by providers, information gleaned by our case managers or care coordination center, our electronic medical records, patient requests, census data, reports from Health Maintenance Organizations or Center for Medicare and Medicaid Services. The financial information comes from claims reports, pharmacy use or our own claims platform. The operational information comes from non-clinical personnel, advanced data reviews or any other touch-point.
The clinical touch-points are office settings, hospitals, skilled nursing facilities (SNFs), assisted living facilities (ALFs), acute rehabs, home health care services, etc. We took all of the above and created systems that reacted to and became proactive based upon what we were able to see in real-time. A team of our case managers who followed the patients actively in hospitals to the SNFs to homes or ALFs maintained continuity of care. A utilization management system also monitored the referrals and their medical necessity according to ODAG rules (Organization Determination Appeals and Grievances), and locally and nationally covered determinations. Pooling all this data in an enterprise data warehouse and tracking it across the spectrum of various points of services ascertained comprehensive health care to our patients.
In the hospitals, we have closely followed Geometric Length of Stay and number of hours in observation status along with the medical necessity of each admission. We also tracked re-admits and did root cause analyses on each one. We were able to reduce re-admits from SNFs from 30% to nearly 10%, GLOS from 120% to approximately 100-105% and observation hours to less than 24 on average. A more advanced analysis of this data should be able to predictively tell us which patients are at a higher risk for further admits and catastrophic illnesses.
We also instituted measures geared towards the whole community, making our approach truly population-based. Our providers started giving lectures on good health behaviors, personal hygiene, wholesome nutrition, exercise programs and integral approaches to health care in our community. We videotaped them to be made available online to anyone interested globally, free-of-cost.
As we continued to apply population analysis to our panel of patients, we realized that our patients can be categorized into four (4) broad categories: low-, moderate-, high-, and very high risk. The low-risk patients usually have one or two medical conditions, such as hypertension or high cholesterol or they may have no medical concerns at all. Such patients do well with annual wellness visits, routine preventive screening, exercise programs, dietary and nutrition counselling and can be managed with few or no medications.
The moderate-risk patients need a patient-centered medical home (PCMH) setting with good coordination and access to care. A robust patient education program with a coordinated approach is useful for them along with a more intense nutritional counselling, exercise program and medication management program. These patients usually have one or two serious medical condition with possible complications, such as Diabetes Mellitus with vascular manifestations, Liver Cirrhosis , Emphysema, Congestive Heart Failure, Coronary Artery Disease and Cerebro-vascular Accidents, Severe Drug Dependence or Psychiatric Conditions, etc., but are usually well-managed and controlled. As one climbs up the pyramid of complexity, the need for case-management and team-based care models increases.
The high-risk patients have multiple serious medical conditions affecting several organ-systems and are on a plethora of medications. This is where our interventions become intense and multi-dimensional. This is where case and care management play a larger role as does our trained staff in the call center. There is much improved and frequent communication and it is a team approach. Patients are seen in the office more frequently. Should the patient end up hospitalized, then the hospitalist steps in and our on-site care managers then fill in the gaps so that the primary care doctors are not left in the dark. The same is true for when the patient needs a Skilled Nursing Facility or an acute rehab. We have SNFists and on-site care managers who provide the same service here as in the hospital.
And then we come to the very apex of the patient-care population, the very high risk patients. These patients are so fragile that they require hand-holding at an unprecedented level and that which is not usually seen in most practices. For these patients, we have provided the soft touch of personal phone calls from the call center to listen to them and to provide that constant attention and communication that these patients and the families of patients like this need. This is also a more intensive case management model where case managers create and coordinate care plans acting as a go-between from provider to patient and back again. These patients may even be seen in high risk clinics in the office if necessary or we may utilize home health as an adjunct to everything else. We have found that a comprehensive approach across the whole continuum of care ensuring patient safety, engagement, compliance and quality is able to improve health care and reduce complications. We were able to improve our Medicare Risk-Adjustment scores compliantly while improving our STAR ratings and HEDIS scores, CAHPS and HOS scores while reducing cost of care using evidence-based medicine and standards of care.
As we have improved our information systems utilizing an enterprise data warehouse, pooling our data, slicing and dicing it, growing in our ability to manage Big Data and Small Data, and sharing our knowledge across our various teams as a Knowledge-Based Organization, we were able to achieve significant savings while improving patient satisfaction and outcomes in a ‘meaningful’ way.
Our uniqueness of approach depended on using our care managers on the ground, in the facilities, and in constant communication with patients, families and providers, along with our hospitalist and SNFist programs along with home care visits. We did not use data in an isolated manner but as a means to improve our clinical practices, operational expertise and technical competence. We used our learning management systems to become a Learning Organization while approaching Population Health in the truest sense, i.e., seeing the whole community as the population that needs to be impacted. Admittedly, this is barely the beginning for us. But, we believe, this is the right direction, solution and approach to the vast range of possible health care interactions that is Population Medicine and Analytics.