Predictive modeling is not a new concept. It is relatively new in the context of being more widely available for practical application for employer use. morgan/sigma partners has followed its development and application for our client’s use. Our experience with the positive and creative application of predictive modeling is about as long as it has been around for client use.
Predictive modeling is usually found in the same sentence or paragraph with disease management. Predictive modeling provides the identification of employees that benefit from the tools of disease management; it’s a very symbiotic relationship.
Predictive modeling is a fantastic instrument for employers who are partially self funded for design, budget and reserving issues.
Predictive modeling is a process by which a model is created to try and best predict the probability of an outcome. Our concern is medical claims modeling applied to individuals within a group as well as the group itself. Focusing on the group, it is possible to predict the aggregate claims for the group with a great deal of accuracy (see slide 15 in presentation below)for a future 12 month period as well as compare the group to national norms and create “if/then” scenarios. It is also possible to generate data It is also possible to provide workforce modeling, workforce health profiles and potential claims savings by medical category.
Predictive modeling when applied to individuals can identify individual employees within the group and place them into several health risk categories for intervention or prevention. This tool also allows for identification of employees where gaps in care, based on national standards, appear. A very important aspect of predictive modeling can forecast emergency room and inpatient stays by individual employee.
An employer can have a “scorecard” for each physician treating employees within the group summarizing past costs, forecasted costs, care gaps, prescribing patterns and other data for assessing physician performance.
Predictive modeling is an extremely useful tool for employers who:
-
Are partially self funded and want to be fully aware of how to budget their plan or design their plan for a maximum balance between costs and plan design (benefits delivered)
-
Are interested in an assessment of the physicians that are used by employees and dependents of the medical plan(s)
-
Are interested in identifying individuals within the employer’s health plan who have chronic illnesses and would like to provide that employee with an intervention, on a voluntary basis, to improve that employee’s health
-
Would like to take advantage of the benefits of a Health Risk Assessment program but find they do not want to be intrusive, yet still would like to identity their high risk employees to help plan and budget their health plans and would be in a position to better negotiate with their stop-loss carrier (reinsurer).
-
Are in the process of initiating a Health Risk Assessment program but do not wish to wait until the assessments are complete and the overall program has been implemented to get actionable results.
-
Would like to design plans to drive their employees to the most cost effective physicians.
-
Would like to provide their employees with information on physicians with a good track record of care, by national standards and cost.
Predictive modeling can be done to accomplish a one time project or can be on-going, producing monthly and quarterly analysis.
A predictive model for a given employer will be done using 12-24 months of medical and prescription drug claims (drug card, outpatient) and or by the data gathered and entered from health risk assessments completed by the employee. The effectiveness of the predictive model for an employer can be greatly enhanced by the inclusion of biometric data of each employee.
Below you will find useful information on predictive modeling. One is a Word document synopsis of predictive modeling for employer use provided by a very good predictive modeling vendor and partner. The other is a Power Point presentation created by the same vendor with the need for the predictive modeling process. We hope they are beneficial.
