The Heisenberg Uncertainty Principle in PHI Evaluation

When I began my career in proactive health management 20 years ago, while VP for strategy and marketing at a multi-hospital system in Denver, I approached the Coors Brewing Company as an example of employers interested in the idea, at least in worksite wellness. I learned that while Coors had a strong interest and major investment in wellness, it evaluated its investments entirely in terms of health insurance expenditure reductions that resulted.

I felt then, and have become increasingly convinced since, that evaluating investments in workforce health should include attention to a far-ranging set of impacts that they can have on overall business performance. The only real limit to such impacts is that they must logically come from changes in human capital asset value and overall performance, as opposed to new technology, for example. But expanding the scope of evaluations brings with it an unavoidable complication, an uncertainty about which and how much impacts should be attributed to PHI investments, as opposed to any number of other causes.

In effect, there is a “Heisenberg Uncertainty Principle” at work. The original principle applied to quantum mechanics, noting that it is impossible to simultaneously determine both the position and momentum of matter. The more precisely that one can locate the position of something, the less precisely one can be about its velocity, and vice versa. (www.wikipedia.org/wiki/Uncertainty_principle)
A similar principle applies to evaluation: the more broadly we look at dimensions of value that PHI interventions could have affected, the less certain we will be about whether changes detected were caused by such interventions.

Given this reality, there are two main “solutions” to the problem: 1) compete with other causes, particularly when there are other functions, departments, investments, interventions, etc. that are likely to or have claimed credit for changes noted; or 2) cooperate with them to reach a consensus on sharing the credit. There is a third option, of course, namely to take a “scientific” approach to discovering the truth about causation and credit – but that is unlikely to be acceptable in most situations.

While government insurers have taken a scientific approach to evaluating Medicare and Medicaid interventions into beneficiary health, they are the exception. Employers normally avoid such approaches for good reasons. Random or matched pair assignment to “intervention” vs. “control” populations is both too much trouble/cost and sure to be unpopular among employees. In one of the few examples where this was tried, a state education system randomly assigned school districts to intervention vs. control conditions, but as soon as it realized how much money it was saving through the intervention, it assigned all the control populations to the intervention condition. Science could not trump economics.

An alternative to rigorous scientific study might well be one of measuring the full "value chain" of effects that PHI requires to achieve its results. If it can be shown that specific interventions are linked to specific changes in participants' knowledge and attitudes about healthy behaviors, particular changes in related health behaviors, in health status indicators, then in productivity, performance, and value -- this would make a stronger case for a causal relationship than citing value indicators alone. Arguably, this is not "proof", but at least it is "circumstantial evidence".

Competing with other causes is sure to be contentious, since there are likely to be a number of other “silos” that could claim credit for improvements in workers’ productivity, performance, and value to the employer. And the broader the range of value dimensions and specific performance indicators involved, the more causes will have grounds to join the fray.

Consider the following list of value impacts proposed for evaluating the workforce training and development function: increased sales/revenue growth; decreased labor costs; improved customer satisfaction, quality, on-time delivery, productivity, cycle time, employee satisfaction/morale; reduced waste, worker injuries, turnover. [D. Brown “The True Value of Learning” Skillsoft.com 2008] Every one of these dimensions should at least be considered as possible effects of PHI interventions as well.

Human Resources, compensation practices (e.g. pay-for-performance), benefits other than health, and a wide range of morale boosting policies or processes may well deserve and demand credit for improvements in the value that employees deliver. A host of other investments (e.g. capital equipment, information technology, supplies, work environment, etc.) may also insist on their share of credit. Competing with them on a “political” rather than scientific basis could create far more harm than good. PHI may or may not have high enough credibility to get a fair share of the credit, and lose out in competing for future investment. On average, well-designed evaluations of PHI have shown ROI levels in the 2:1 to 5:1 range, which should allow plenty of room for sharing credit.

Cooperation, e.g. by either sharing equally in the credit or negotiating something other than equal shares, may work better, as long as agreement/consensus can be achieved amicably, quickly, and credibly. Of course, if the total value improvement noted does not more than equal the total costs of interventions competing for credit, cooperation may break down. Naturally, CEO and CFO support will be essential for such cooperation to work. Supporting this possibility is the likelihood that there will be plenty of credit to share, particularly in the long haul, as examples of specific interventions have already demonstrated.

Initiating a pay-for-performance system, replacing a uniform hourly wage system for windshield installers at Safelite, for example, resulted in an immediate 44% improvement in productivity, at only a 10% increased cost in wages. Empowering corporate office employees to set their own hours and choose their own place to work improved productivity by 35% and cut annual turnover from 16.7% to zero. Neither example provided any information on any other interventions that might have claimed credit for some of the value added, but had any done so, there would clearly have been plenty of room to share the credit.

There is actually another source of "credit competition" that may have to be considered -- public, community, and governmental interventions that may be affecting the health of entire populations, in addition to those members of the population that are employees or dependents thereof. IBM, for example, has recently initiated a major research effort to identify factors that affect health, as the basis for investments by anyone interested in population health. ["IBM Launches Research Effort to Build 360 Degree View of Factors Affecting Human Health" IBMMay 6, 2010 (www-03.ibm.com)]

Fortunately, while investments by sponsors other than the employer may well deserve a share of the credit, the IBM project should also identify which are the most cost-effective interventions in population health. Armed with such information, PHI should be able to compete effectively with any other sources of impact. Broad-based interventions are already increasing with recent health reform, and these investments may also deserve some credit. They will probably focus mainly on reducing sickcare costs, however, and thereby leave plenty of room for PHI investments more focused on workforce productivity, performance, and value.

The majority of employers still seem to rely on their faith in the clear connection between employee health and productivity, performance, and value when making PHI investment decisions. On the other hand, far better decisions, particularly about which among competing human capital interventions, as well as which health management problems and solutions to invest in, can be made if there is more and better data on the full range of value impacts such interventions might have. It may be true that “When ignorance is bliss, ‘tis folly to be wise”, but data, analysis, and understanding of causes and effects are likely to be a far better basis for making investment decisions. And cooperation seems likely to be both more welcome and advantageous than competing for credit when effects are known.