With the Centers for Medicare and Medicaid Services pegging reimbursement to certain quality benchmarks, hospitals are making efforts to improve.

Nowhere is this more visible than in the 30-day readmission rates, the poster child for this effort. Increasingly, hospitals and health systems are using data to improve their readmission rates, though knowing what to do with that data certainly helps.

Financially, there’s a lot riding on these efforts. According to Loopback Analytics, hospitals were penalized about $108 million more in 2017 than in the previous year for readmissions alone.

“CMS has set this thing up as a horse race,” said Loopback CEO Neil Smiley. “The reason these readmission penalties keep growing is because you’re never going to be done with the thing. You’re compared against your peers, so the better they get, the better you have to get.”

What helps is to have a plan in place for all that data, and the first step is simple: identify relevant information. 

Health systems are sitting on a large amount of data that’s collected primarily for billing purposes and, with that, they can perform risk stratification. Some of the more common risk algorithms don’t leverage all of the necessary clinical markets, but others do — integrating across a variety of clinical platforms, and effectively serving as a data custodian.

Still, there’s no one-size-fits-all approach. Each health system is going to have its own anomalies for how it captures data. The key is to build a risk prediction model that leverages the data you’ve got.

“That’s usually where people stop,” Smiley said. “Then they’ll allocate a disproportionate share of their resources to the most expensive patients. It’s better than doing nothing, but there’s a fit aspect. They may have medication vulnerability, transportation issues and so forth. You have to identify that they’re high-risk, but how they’re high-risk, and match them to an appropriate intervention.”

That bleeds into an oft-overlooked part of the process: Matching the data. Simply put, you have to identify the fit, pinpointing what exactly makes the patient at-risk — medication adherence, for example.

“Once you apply an intervention to the population, it’s (about) measuring the efficacy,” said Smiley. “They’re going to have a highly varied response — some are highly responsive, some are for whatever reason not responsive. You need to use the data and put it into an eligibility model. If you do this right, you set yourself up for continuous learning and you stop applying resources to patients who are high-risk and unresponsive. You realize they need a different form of help.”

Mary Jane Konstantin, RN, senior vice president and general manager of Carepoint for AxisPoint Health, agrees that reducing readmissions starts with identifying those at risk — and for some of the tricker cases, a more personal touch may be warranted.

“We want to understand patients’ living situation as quickly as we can,” said Konstantin. “We can do that before they’re discharged, or within the first 48 hours. A home visit is great; we can see what their living situation is like, and identify family members who will participate in the support.”

Of prime importance is making sure the patient has an appointment with their primary care provider or another treating provider, and to work with them to ensure they have the ways and means to keep that appointment. For someone with a low income, that may be something as simple as getting a ride.

“Transportation can be an issue,” said Konstantin. “If they don’t have a means to keep that appointment, they probably won’t.” 

Missed appointments equal missed treatments, which link to a higher probability of a 30-day readmission.

That’s an opportunity that may emerge but it requires proper engagement with the data, and according to Smiley, that means an integrated data approach in which all of the pieces are connected to each other. But evaluating what comes of that is key.

“There’s a strong challenge in dealing with selection bias,” said Smiley. “When you go to enroll patients in a program, the way you’re doing that is going around the hospital and saying, ‘You look sick,’ ‘You need a coach,’ etc. But you don’t know. How do you isolate the effect of that? The other challenge you have is that you have a lot of initiatives, all rolling along at the same time.”

A good approach, he said, is to invest in data that identifies a risk-adjusted comparison group that hasn’t received the intervention — which allows hospitals to identify which interventions are working, and which are just adding cost.

Ultimately, it’s a strategy that will remain relevant for as long as these penalties exist. And that could be a long time.

“As more and more hospitals are getting their game on and using data to drive down readmissions, it forces those who think they’re fairly comfortable into being penalized again,” Smiley said. “You’ve got to keep moving, you’ve got to keep getting better. If you’re not in the penalty zone today, you may be there tomorrow.”

Poor medication adherence following hospitalization costs the U.S. healthcare system roughly $100 billion annually, according to a New England Journal of Medicine study, and is the most significant cause of readmissions. The job of improving medication adherence can be daunting because there are so many potential points of failure. The top three reasons for medication adherence failure: 1. Medications never get to the patient 2. Medications are not taken correctly 3. Medications are not refilled

Copyright 2019 Loopback Analytics, LLC | Privacy Policy