3M Commentary

Incorporating socioeconomic factors into payment adjustment – you get what you pay for but what do you want?

February 12, 2014 / By Norbert Goldfield, MD, Richard Fuller, MS

When talking with practitioners at the sharp end of health care there is concern about the unintended results of policies that create penalties or rewards to encourage change. Take, for example, the familiar Hospital Readmissions Reduction Program, a major cost containment initiative for CMS. There is argument over the structure of the penalties, the lack of a positive financial reward, accuracy of the risk-adjustment, what constitutes a readmission – but there is little argument that it has grabbed the attention of the hospital community and has elicited a response over and above terse denunciation of the policy.

In some ways the structure of the payment incentive works. It makes a single entity accountable; the penalty is sizeable enough to warrant action, but not too big to cause instant extinction for a failing provider. There are, of course, numerous failings which many in the industry have highlighted. In particular, criticism has focused on the absence of adjustment for socioeconomic status (SES). Study after study has shown that the likelihood of readmission is linked not just with SES factors associated with an individual, but with the percentage composition of patients with low SES (Fuller, Atkinson, McCullough, & Hughes, 2013; Gu et al., 2014). Hospitals serving a large volume of low income patients have worse patient outcomes than predicted by individual patient poverty. Such observations are inextricably linked with measures of racial disparities in health outcomes.

Policy therefore needs to balance conflicting objectives. First, it has to fairly account for the additional resource burden of low income patients – to not do so jeopardizes both access to and quality of care. Second, it needs to encourage improvement in outcomes quality for all hospitals, which requires equitable treatment for hospitals regardless of local income disparities. This is essential if we seek to reduce the correlation between race, income, and disparities in care outcomes. Such an effort is eminently doable and can be operationalized in a variety of ways.

As discussed in previous blogs, in conjunction with making any high level adjustment for SES, such as one based upon DSH status, we should identify those characteristics associated with groups of individuals likely to be in lower SES that are not directly dependent upon income status. These characteristics range from complex clinical factors such as severe mental disability, substance abuse, high severity of illness, and children with complex illnesses, to social factors such as homelessness or foster care placement. Separately adjusting for these factors permits more accuracy in identifying a more general SES adjustment. Calculation of an SES adjustment should require identification of SES variables associated with individuals, rather than the facilities that treat them, to control for any potential relationship between hospital outcomes quality and local income levels. The result would deliver a more balanced patient based SES adjustment that accounts for correlation between low income areas and institutionalized lower quality of care.

Given the likelihood that the worst performing facilities are in the poorest areas, even with an SES adjustment for individuals described above, it is important to emphasize that they are likely to have the thinnest margins and least resources to redirect towards “dealing” with a new policy. To restate the earlier concern, the policy is going to get in the way in an already challenging environment. Penalties may have the unwanted result of putting too much pressure onto hospitals struggling in poorer neighborhoods, which may inadvertently reduce access to care and worsen quality.

This returns us to the all important question of the policy objective. If hospitals are to be encouraged to improve the health outcomes of the communities they serve, then there must not only be fairness in measurement but recognition of the differing levels of work required to overcome local challenges. A blind penalty that ignores this dynamic does not empower change—it incentivizes the change but does not make the incentive actionable as resources required for action are not available. Conversely, preserving the status quo, crossing ones fingers and hoping for change will not get us there either.

One way to bridge the divide is to transform penalties into performance grants. A penalty can be booked against an underperforming facility in a given year with notice that the performance deficit will incur an actual monetary transfer after a pre-defined period elapses. Allowing the penalty to be offset by future year improvement permits the hospital to invest in quality and avoid penalties or take no action and incur penalties. There are multiple variations on this approach. There is no guarantee that this approach would elicit the desired outcome of quality improvement but the concerns of safety net hospitals and low income communities are addressed in a significantly more thoughtful and, dare I say, fairer manner.

Payment incentives are powerful enough to drive any change – intended or unintended. You have to make sure you target what you want without burning down the house to get it. In the next blog we will continue discussing the important topic of incorporating SES into payment incentives and will comment on the MedPAC perspective.

Richard Fuller, MS, is an economist with 3M Clinical and Economic Research.

Fuller, R. L., Atkinson, G., McCullough, E. C., & Hughes, J. S. (2013). Hospital readmission rates: the impacts of age, payer, and mental health diagnoses. The Journal of ambulatory care management, 36(2), 147–55. doi:10.1097/JAC.0b013e3182866c1c

Gu, Q., Koenig, L., Faerberg, J., Steinberg, C. R., Vaz, C., & Wheatley, M. P.  (2014). The Medicare Hospital Readmissions Reduction Program: Potential Unintended Consequences for Hospitals Serving Vulnerable Populations. Health services research. doi:10.1111/1475-6773.12150