While we can all see problems in the care of older adults all around us in both our personal and professional lives, figuring out how to deliver better care at a lower cost is not easy.
The John A. Hartford Foundation has been working on demonstration programs for many years and while some are great successes, it is not uncommon for even well-designed interventions to increase costs due to the added services being delivered and the discovery of unmet needs.
Even worse, it is also possible for an intervention designed with all the expertise and good will in the world to fail to change health outcomes or even patient satisfaction with care.
A frequent explanation of such perplexing results is that there was a failure of targeting.
Targeting is something that Goldilocks discovered in her interactions with the three bears—some things are too hard/hot/big and some things are too soft/cold/small and only a few things are just right. In geriatric care, it’s often the case that for a particular intervention, some patients will be too sick and some will be too well and only a narrow group will be “just right.”
For example, in the federal Medicare Health Support demonstration program aimed at providing care management to fee-for-service beneficiaries with heart failure or diabetes, one of the complaints of the disease management firms was that the patients were “too sick.” These firms, whose prior experience was largely with insured working populations, and who primarily relied on telephonic self-management support tools, felt that the needs of this population were too great. As a result, they were unable to improve outcomes or lower costs of care.
The opposite error is also possible. If relatively young Medicare beneficiaries without serious chronic illnesses had been selected, these interventions might still have failed, as there would be little room for improvement in the health of such a population and certainly few costs that could be avoided.
The needs of the target population (and therefore the opportunity for improvement) must be a “just right” match for one’s intervention. So, when an intervention fails to demonstrate the desired results overall, mysterious failures of “targeting” frequently get blamed. Model proponents often suggest that their program would have “worked” if better targeting had been used and sometimes look at the benefits of an intervention for particular subgroups to test this proposition.
Over the years, I have come to believe that this is largely wishful thinking. Targeting algorithms such as self-rated health (fair or poor), age (>=70?), PRA+ scores, Hierarchical Condition Categories, or counts of chronic illnesses or functional limitations, don’t seem to identify very different populations and the evidence is scarce that using one metric versus another or changing the cut points would produce different cost-effectiveness outcomes. Even when a demonstration is observed to work for some subgroup (but not overall), this “targeting” phenomenon must be prospectively replicated to ensure that it is not a statistical artifact.
If experts who have been trying to perfect targeting systems for more than 20 years still seem to be grasping at the targeting “straw” to explain when an innovation fails to work, perhaps it is just not the right question.
Rather than obsessing about correctly targeting a relatively inflexible service model to just the right beneficiaries, a better approach might be to tailor a more flexible service model to the current needs and responses of beneficiaries and modify the mix of services over time to help the person achieve some target state. Treating to target rather than targeting who to treat.
Treating to target is one of the key ingredients of the very successful and cost-effective IMPACT model of depression treatment in primary care. In the IMPACT model, patients are repeatedly assessed for their level of depression and response to treatment (or lack thereof). Patients generally begin on relatively low-cost/low-intensity courses of treatment, but if they fail to improve, the intensity of services can be stepped up to help get patients on the path to recovery. When people achieve the target outcomes, services can be tapered off.
The challenge would seem to be to adapt the IMPACT, treat-to-target approach from a single focus on depression (albeit in a very multi-comorbid population) to creating a more general target metric. One could perhaps use functional status as a target metric on the argument that everyone wants more function.
But perhaps more individualized, and therefore more motivating, would be to use the attainment of personal goals as the target metric. Our new grant to the National Committee on Quality Assurance seeks to develop feasible and meaningful personal goals of care as quality measures for complexly ill patients.
I would love to see an innovation that used such personal goals to establish “targets of care” and stepped up (or down) a package of interventions to help older adults get closer to their personal goals.