Many policymakers, health care and insurance executives, and consumer advocates argue that better integrating medical care with long-term supports and services (LTSS) can both save money and improve the well-being of frail elders and younger people with disabilities. It seems logical, and there is some tantalizing evidence that it is true.

Unfortunately, health care is filled with ideas that sound like they ought to succeed, but don’t. A new study by researchers at Johns Hopkins University and the Long-Term Quality Alliance aims to determine whether combining medical treatment with care management and social supports such as personal care aides and nutrition programs really can reduce avoidable high-cost medical care.

Unfortunately, the results of this careful study are…ambiguous. After looking at five fully-integrated health plans and comparing their results with those of a similar fee-for-service Medicare population, the authors concluded that the plans may be reducing overall costs. But they can’t prove it.

They said, “The directionality of the results is consistent and strong…and provides a view of the potential in integrated LTSS to achieve substantial health care savings.” Yet, they added, “It is not possible to prove that there was a causal relationship or to calculate medical savings.”

The problem is that potential may not be good enough to convince plan administrators to jump into integrated medical care and social supports with both feet.

The opportunity is undeniable. Anne Tumlinson, who worked on the LTQA team, estimates that Medicare spends far more on beneficiaries with functional limitations than it spends on those with similar health who do not have those impairments. For example, in 2015, Medicare spent an average of about $5,500 on beneficiaries with two or fewer chronic conditions. But it spent nearly $13,000 on similar beneficiaries who also suffered from some functional impairment, such as an inability to bathe without assistance.

Could managed care plans save even a fraction of those health care dollars by providing a suite of social supports to their members? Imagine a plan could reduce that $13,000 in health costs to $10,500 by preventing one fall that led to one one-day hospital admission. And imagine it could do so by spending $2,000 on supports such as home-delivered meals or installing grab bars in a member’s shower.  That would increase its margin on that patient by $500—money it could use to reduce premiums or add new benefits.

Last year, new legislation and regulations allowed Medicare Advantage plans to offer such supplemental benefits for the first time. But plans have been slow to participate, in part because they are uncertain whether providing non-medical services could save them money on the health side.

But some plans already have been doing this for members who receive both Medicare and Medicaid (the dual eligibles). The LTQA/Hopkins study looked at five of those plans and found some important hints about the potential for cost-savings from fully-integrated delivery models. For instance, the researchers concluded that members of three Senior Care Options (SCO) plans in Massachusetts had consistently lower hospitalization rates than similar fee-for-service Medicare beneficiaries.

Similarly, a New York Program of All-Inclusive Care for the Elderly (PACE) program was very successful at reducing emergency room visits, perhaps because its care teams could identify and treat problems before  sending a participant to the hospital. But overall results were more mixed for the non-SCO plans. One reason the SCOs may be more successful overall: They have been doing it longer than some other models.

And we know that some specific social supports do reduce costs and improve the well-being of older adults. Nutrition programs seem to accomplish both goals. So does Community Aging in Place–Advancing Better Living for Elders (CAPABLE). This initiative, for low-income older adults living at home with functional limitations, resulted in fewer hospitalizations, emergency room visits, or skilled nursing facility stays. In an initial demonstration, it saved Medicaid more than $20,000 per recipient over two years, net of costs, and now has been adopted in 22 cities and rural communities in 11 states.

Keep in mind that the LTQA/Hopkins study looked only at costs, and not at the quality of life of participants. But avoiding unnecessary hospitalizations probably is as good for beneficiaries as for the bottom line of the plans.

It also is important to remember that the plans themselves, and their actuaries, have access to detailed member-level data. So even if independent researchers can’t say for sure, the plans are learning whether full integrating services does reduce their overall costs.

So for now, we’ll just have to watch the market. If more plans jump into the LTSS world , that will be pretty strong evidence that they think adding social services to their list of benefits really is a winner.

Full disclosure: I served as an unpaid member of the technical advisory panel for this project.