home email us! sindicaci;ón

Don’t Look for Best Practices in Health Management

by Scott MacStravic

There is a strange disconnect in the arena of “health management” (HM) as opposed to sickness treatment.  In the latter, there is a constant search for and attempt to make universal whatever is found through evidence-based medicine (EBM) research to be the best practice for diagnosing, treating, and where necessary rehabilitating patients suffering from every identified sickness condition.  This is accompanied by efforts to minimize, if not eliminate “random variations” among providers’ approaches to patients, so that every one adheres to EBM guidelines and care is standardized for all.

This sickness care effort has had only modest success, as individual physicians rely on their own personal experience and specialty-specific preferences when treating individual patients, and as patients express their own preferences.  The Dartmouth Atlas Project annually reports on the wide variations in care across states and metropolitan areas in the way patients with the same conditions are treated, noting how much of it is unnecessary and wasteful. (www.dartmouthatlas.org)

In health management, variations are even wider, as HM provider organizations and individual physicians choose their own preferred approach to promoting health, reducing risks and managing diseases in order to reduce the incidence and prevalence of sickness and its consequences to health care and other insurance costs, as well as workforce costs and performance.  Government studies keep analyzing HM performance as if these widely varying approaches were the same “treatment”, and, hardly surprising, discover that their success varies at least as much as their approach.

When wiser heads prevail, there is a recognition that the first challenge in HM is to identify which approaches work, based on EBM-style analyses of the currently widely varying methods employed by HM providers.  Instead of damning disease management, for example, which gets most of the attention from the federal government, as “unproven”, it would be better if reports identified the specific interventions that are working, praising and paying them for their success, and promoting more widespread adoption of the approaches that are proven.

On the other hand, there is an underlying logic in HM that would be severely criticized in sickness care – there is simply no one best practice for any HM challenge.  There are two major reasons why this is the case:

1. customization based on individual “patient” characteristics and preferences works better for individuals than does “one-size-fits-all” standardization across populations; and
2. customization based on risk/reward potential delivers better financial results and returns than does standardization to one best practice

The chief aim of virtually all HM interventions (they are not “treatments”, because the patient is the main solution rather than the provider, since individuals’ “adherence” behaviors determine success) is to achieve positive financial results.  This means that the costs of each intervention have to be affordable relative to the financial gains in enough cases to make sure that success is achieved across populations.  And since the risk-reward potential of individuals varies widely, the best way to ensure the greatest net financial gains is to vary the costs of interventions to the risk/reward potential of individuals, or at least different risk/reward segments.

Fortunately, the science of predictive modeling, to identify the risk/reward potential of individuals and segments is far-advanced, though hardly perfect.  A major goal in the health risk assessments (HRAs) that are part of every HM intervention is to determine risk/reward potential, and predictive modeling systems, such as those used by MedAI, Inc. in Orlando, Florida (www.medai.com), and HealthMedia, Inc. on Ann Arbor, Michigan (www.healthmedia.com), simultaneously assess health risks and financial potential of individuals.

This potential reflects the combination of individuals’ present and predicted costs and their probability of having those costs reduced through HM interventions.  If predictive modeling were perfect, there could be a financial value potential calculated for each, reflecting dollar gain potential times the probability of achieving it.  This dollar value could then be used to set a limit on HM intervention costs for each person, given whatever ROI ratio is desired across the population being analyzed.

In the meantime, broader brush approaches are being used, such as stratifying populations into risk/reward categories, with different levels of intensity, frequency of contact, and kinds of interventions for individuals based on which category they fall into.  The most common approach seems to be using three categories, reflecting low, medium, and high risk/reward pot