The Value of Decreasing Health Cost Volatility

Marc Herant, PhD, MD, and Alex J. Brown, MEng, MBA

The Health Coaching in Primary Care randomized control study^{1} showed a clear effect on leading clinical variables (eg, glycated hemoglobin control for diabetes), However, a subsequent analysis, presented in *The American Journal of Managed Care* article, “Costs for a Health Coaching Intervention for Chronic Care Management,^{”2} did not show a statistically significant impact on average health cost per patient within the same year that the coaching was delivered. Two factors likely contributed to the lack of a statistically significant signal: 1) a population with a standard deviation (SD) of cost so much larger than the mean; and 2) a 1-year observation window too short to realize the full benefits of the intervention. It would be interesting to compare the mean year 2 and year 3 healthcare costs for both analysis populations as a follow-up investigation, even with no additional coaching.

We were struck by how dramatically the SD of individual cost changed between the control group ($12,844) and the arm receiving the health coaching intervention ($6552). Without further knowledge of the underlying cost distribution, we cannot say for certain whether this change is statistically significant, but we can say that the magnitude is meaningful from an economic perspective. Financial firms routinely trade off average rate of return against volatility, and a reduction in health cost variability is valuable to a payer in terms of reducing reserve requirements and/or reinsurance needs. This will be a significant effect for small risk pools. As shown in the**Table**, we worked out the value for a group of 100 individuals with an average cost of $3500 per year, which turned out to be about 4% of cost (comparing the value of 1-year options to hedge against above-average costs).

We note that this analysis makes certain normality assumptions about the characteristics of the volatility that may not hold in situations where power laws generate significant long-tail risk, as is possibly the case here. However, in such circumstances, the option value of reduced volatility will be even larger and the Black-Scholes value^{3} should represent a lower bound.

Interventions such as health coaching decrease variability in cost through 2 mechanisms: 1) ensuring that all individuals get routine care results in an increase in costs in the left side of the cost distribution, and 2) preventing significant medical issues from snowballing into full-blown medical disasters results in a tightening of the right side of the cost distribution. Even if this does not change the average cost, decreased cost volatility should be considered valuable.

We were struck by how dramatically the SD of individual cost changed between the control group ($12,844) and the arm receiving the health coaching intervention ($6552). Without further knowledge of the underlying cost distribution, we cannot say for certain whether this change is statistically significant, but we can say that the magnitude is meaningful from an economic perspective. Financial firms routinely trade off average rate of return against volatility, and a reduction in health cost variability is valuable to a payer in terms of reducing reserve requirements and/or reinsurance needs. This will be a significant effect for small risk pools. As shown in the

We note that this analysis makes certain normality assumptions about the characteristics of the volatility that may not hold in situations where power laws generate significant long-tail risk, as is possibly the case here. However, in such circumstances, the option value of reduced volatility will be even larger and the Black-Scholes value

Interventions such as health coaching decrease variability in cost through 2 mechanisms: 1) ensuring that all individuals get routine care results in an increase in costs in the left side of the cost distribution, and 2) preventing significant medical issues from snowballing into full-blown medical disasters results in a tightening of the right side of the cost distribution. Even if this does not change the average cost, decreased cost volatility should be considered valuable.