Cost Effectiveness Analysis in Healthcare Modeling

Healthcare dollars are becoming scarce. Doctors want to provide the best healthcare possible, many times using the latest techniques, procedures and medications. It is becoming ever more critical that the latest developments are actually effective in delivering the best healthcare outcomes at a realistic price.

But how does one evaluate a new procedure? In the past, financial analysts used return on investment (ROI) calculations. This analysis has worked well in determining investments in companies (e.g. buying stocks), but actually falls short when used to evaluate healthcare resources, whether their utilization or their deployment.

In healthcare, outcomes can become rapidly complex. A treatment with medication can lead to side effects or complications which can be very costly. Another treatment may be more effective that the other therapy, but cost more. How can we determine if such treatments are worth doing? Perhaps one has an idea of how much one is ultimately willing to pay for such a treatment. However, using cost-effectiveness analysis, by building decision-tree models, provides a responsible and quantifiable solution to these various problems.

A decision-tree model can be custom-built to the particulars of a specific problem and can compare various treatments to not only determine the least cost but simultaneously determine the most effective path. The measurement in CE analysis is cost (dollars) per QALY (Quality Adjusted Life year). A QALY of 1 indicates perfect health per one year period. Death has a QALY of O, and the sicker a patient is, the small the QALY fraction. So someone with a QALY of 0.7 is sicker than a patient with a QALY of 0.8. If two methods cost the same, but one gives the patient a better outcome (longer life or feeling better), the QALY will be closer to 1.0 than the inferior method, such that the CE Ratio is less. This makes it easer to compare apples to apples.

What if you want to compare methods or therapies that affect chronic diseases (CHF, COPD, etc) versus one time problems such as a cancer/tumor. In this case we use Markov Analysis. With this method, you may design a model that compares disease states or conditions, such as being in a hospital, using and emergency room or just routine doctors’ visits, with each assigned a specific probability. You assign different probabilities to each state based on the treatment. Than a computer model can predict over time what will happen to the patients, based on the type of treatment they are given.

The flexibility of this analysis lends itself to answer a wide variety of healthcare questions. It can be use in disease management programs, health and wellness programs, no equipment, such as MRI, PET, surgical robotic purchase, new treatment programs, new medications, surgical procedures, us of urgent care centers and emergency rooms, population health programs, smoking cessation programs, narcotic/pain management programs, utilization management programs and disease management programs.