The Measurement Challenge
What has kept health productivity management (HPM) from flooding the market is the extreme difficulty of measuring its impact on total performance. It is known that healthy workers are significantly more productive than are unhealthy ones, but most employers are not sure how much more. Absences related to employee sickness are difficult to measure, since employees take “sick days for personal reasons, and personal time off for sickness reasons. Declines in productivity at work caused by “unhealth” has usually been found to be many times as expensive as absence alone, but identifying such declines as a major challenge.
Most of the research has been done on the negative performance impact that employee unhealth has. In an analysis conducted of Dow Chemical Co. employees, looking at the impact of ten categories of chronic conditions, for example, the total cost impact was gauged at $7204 times every employee Dow had. Of this total amount, only $1218, or barely over one-sixth of the total, reflected direct medical care costs for treating such sicknesses. Presenteeism, by contrast, accounted for the vast majority, roughly four-fifths of the total.
Gauging the total savings from HPM, however, is even more difficult than gauging the total costs of employee unhealth. The savings from employees who have an existing chronic condition will reflect the difference in their sickness care expenses and productivity losses when their conditions are controlled, perhaps even “reversed” (defined as requiring only lifestyle vs. medical management) as compared to when not controlled. This will normally represent a large share of the difference between employees with given conditions and those with none, but by no means all of that difference.
Once people have a chronic condition, they tend to cost more to employers, no matter how well managed their conditions are. Moreover, the majority of people who have one chronic condition have more than one, so while one may be under control, the other may not. And relapses into “out-of-control” status are common even after conditions become under control. And, of course, this creates the challenge of measuring something that does not happen, the situation that can be measured, compared to a situation based on predictions of what would have been the case if employees’ conditions were not managed.
This is nowhere near as simple as comparing “after” to “before” situations. Evaluation of disease management (DM) efforts, for example, often overstate savings by comparing costs for patients newly diagnosed with a given condition, when their health care and productivity costs are likely to be greatest, to the costs in the subsequent year, when “regression to the mean” tends to reduce costs below the first year level even if nothing is done to cause such reductions.
Moreover, there tends to be a “self-selection bias” affecting those who end up participating in a DM intervention. Those most motivated to improve their health and comply with recommendations for taking their medications and changing their lifestyles will be more likely to participate, and to achieve reductions in costs. If average costs across all employees with the same condition “before” are compared to average costs of only those who participated “after”, the differences will tend to overstate the impact of the intervention, by including the impact of self-selection.
Even tougher is the evaluation of disease risk management (DRM) efforts aimed at preventing the onset of chronic or acute conditions. All risk reduction efforts have the challenge of predicting what would have happened without such efforts, then the predictable costs of the “would have happened” situation, in order to measure the differences. While the reality of actual costs for employees identified as at risk can be calculated or estimated, the prediction of costs that will be compared to the actual will necessarily involve an “educated guess”, since prediction is always fraught with uncertainty.
Part 3 will appear tomorrow on HFMA Views. To read Part 1, click here.
