Vitreos Publication

Why Most Care Management Programs Fail to Deliver Results

By Kirit Pandit, MS, MBA

It is now fairly common knowledge that Care Management (CM) programs have had mixed success in reducing the Per Member Per Month (PMPM) cost for a population. There are many publications that site case studies and compile savings and ROI numbers for care management programs across the country in the last 5 years. The results are all over the place. These research publications conclude most CM programs that are successful are those that are highly integrated, high touch programs.

However, these studies mostly ignore the other important question – are the CMs going after the right cohort of population? Our recent studies have indicated that most CM programs are not picking the right candidates for appropriate care management programs.

VitreosHealth® (formerly PSCI) did a recent study with a Medical Home population of about 11,000. We used EMR data for calculating clinical state-of-health (SOH) risk scores and claims data for calculating utilization (PMPM costs). PMPM cost included both acute, ambulatory, post rehab, and skilled nursing facility.


Fig 1 illustrates the framework we used to analyze the “At-Risk” population. We segment the population on the basis of clinical risk score and PMPM cost. The clinical risk score is a composite of the individual disease risk scores and is calculated from EMR (clinical) data that includes vitals and lab results.

The top right quadrant (“Critical”) is the cohort of high cost, high clinical risk score patients. These patients are clinically risky based on the current state-of-health and are also high utilizers today and account for about 50% of the total population spend. The lower right quadrant represents the cohort (“High Utilizers”) that are high utilizers today even though they are relatively at lower clinical risk based on their state-of-health analysis using EMR data. Typically, they are emergency room (ER) and medication ‘abusers’ and are either hypochondriacs, and/or may have socio-economic and access-to-care problems. Both these segments are typically identified through claims analysis in most population and disease management programs and become ‘high risk candidates’ for care management programs.


However, there is a far more important category of patients which is the upper left (“Hidden Opportunity”). This cohort comprises of members that are clinically at higher risk today based on EMR data analysis, but have historically not been high utilizers, hence are not identified by claims based risk scores that are biased towards historical utilization costs. In most cases, they account for only 10% of the total spent and have very low PMPM costs, so most of these members are ignored by CM programs.

However, through repeated ACO case studies, we have found that within 12-18 months, 15 - 20% of the “Hidden Opportunity” members transition to the ‘Critical’ category if they are ignored by care management programs. This is illustrated in Fig 2. Once they move to the right, they account for anywhere from 40-50% of the spend of the “Critical category” the following year. This means anywhere from a quarter to half the spend associated with the “Critical” category comes from these new patients that did not exist at the beginning of the year in the “Critical” category.

Yet, the “Hidden Opportunity” category is largely ignored. Why? One reason is that most care management programs are driven by claims data analysis which cannot identify this “Hidden Opportunity” population. However, predictive clinical risk scores that use both the harvest EMR data along with claims data can easily identify this hidden opportunity cohort. In addition to using EMR data, these 10-15% of the hidden opportunity cohort that are the future ‘liabilities’ can be identified through a multidimensional risk model which combines this clinical risk with other risk factors such as compliance risk, socio-economic risk, access-to-care risk and mental well-being risk.


VitreosHealth® has been able to identify this population consistently in retrospective analysis. Once these are identified, published studies have proven that a high-touch, integrated CM program can successfully reduce the PMPM by 20-25% and potentially avoid the movement of this cohort to the catastrophic “Critical” segment. These new and unique findings we believe represents a paradigm shift in care management based on the potential impact id deciding the success of a population management program.


Fig 3 shows that an ideal Care Management program is one which

Prevents the 10-20% of the hidden opportunity category from becoming “critical”. Ensure the high clinical risk patients “P1” do not move to the right “critical” category.
Makes sure the “critical” population’s health and acuity remains in check and reduce their utilization through effective case management and care coordination and move this population to the left.
Identifies the causal factors for the “High Utilizers” (socio-economic, access-to-care, mental well-being) to design tailor-made care management programs address their unique mental and social well-being needs.
Identify future high-risk patients early in the disease cycle (pre-diabetic, obesity, hypertension, anxiety, etc.) from the current “Relatively Healthy” cohort and continue to keep them healthy through fitness, wellness programs and disease counseling.
It is important to note that traditional claims based analysis can only provide a partial picture, since they lack clinical records such as vitals, lab results, family history, etc. which can be used in disease models to predict more accurate and segment the population more precisely. A combination of clinical, claims and demographic data and a multi-dimensional risk model can segment the population more accurately and provide the correct candidates to put into a CM program.

Kirit Pandit, MS., MBA, is President of, VitreosHealth®, based in Plano, Texas, selected as the Most Promising 2014 HealthTech Company at the Health Tech Conference in Mountain View, CA. He can be reached at This email address is being protected from spambots. You need JavaScript enabled to view it..