Readmissions: How can you keep patients from bouncing back?

Readmissions: How can you keep patients from bouncing back?

Researchers offer clues on how to prevent readmissions

September 2016
SHARE

Published in the September 2016 issue of Today’s Hospitalist

IF YOU’RE STRUGGLING to find the right mix of tools and strategies to stem readmissions, you’re not alone. Hospitals facing steeper reimbursement penalties are anxious to take action, and they are looking to researchers to identify the factors that lead to readmissions.

Study efforts, however, are having varying degrees of success. “The story behind readmissions is more subtle than patients just coming back to the hospital,” says Amanda Mixon, MD, MS, MSPH, an investigator in Nashville’s Vanderbilt Center for Health Services Research and an assistant professor of medicine in Vanderbilt’s section of hospital medicine.

In a recent study for which she was lead author, for instance, Dr. Mixon hoped to find a connection between readmission risk and reports by patients about how prepared they felt to go home at discharge. Unfortunately, her study didn’t show consistent, strong correlations between patient perspectives about discharge and readmissions, compared to administrative data.

“The one aspect most commonly associated with risky early discharge was poor communication.” 

cover-readmission-auerbach~ Andrew D. Auerbach, MD, MPH, University of California, San Francisco

Research has found, however, that some 30-day readmissions can be prevented, but data show that those make up a minority of patients who bounceback. That leaves hospitals wondering what approaches they should take.

Should hospitals rely on classification systems and 
prediction tools to identify those patients? What’s the
 best combination of tools to use, and the most cost-effective? Can you draw lessons from multi-site studies, or do you need to drill down into your own hospital’s data?

And even when a tool shows promise in identifying patients at high-risk of being readmitted, it may still not be ready for prime-time implementation. Amid both encouraging and disappointing surprises, the good news is that evidence is gathering to help hospitals get a handle on which readmissions they may be able to prevent.

Diving into avoidable readmissions
If you have a fixed budget to put a lid on readmissions, what should you do first?

Start by bringing together an interdisciplinary team, says Andrew D. Auerbach, MD, MPH, professor of medicine at the University of California, San Francisco (UCSF), and director of UCSF’s hospital medicine research division. Dr. Auerbach was the lead author of a study published in the April 2016 issue of JAMA Internal Medicine that paved the way to practical solutions by taking what he called “a deep clinical dive” into readmission data.

Researchers looked at 1,000 general medicine patients readmitted within 30 days to 12 academic medical centers throughout the U.S. The authors first surveyed patients and physicians, then used two-physician case review to determine which factors contributed to those readmissions and which might be preventable.

“The story behind readmissions is more subtle than patients just coming back to the hospital.”

cover-readmission-mixon~Amanda Mixon, MD, MS, MSPH, Vanderbilt Center for Health Services Research, Nashville

 

Among the findings: Just about one in four (27%) readmissions was potentially preventable. Further, about half of those readmissions were “thought to represent gaps in care during the initial inpatient stay,” the authors wrote.

The following factors were most commonly associated with potentially preventable readmissions:

  • emergency department decision-making (readmitting a patient who didn’t need to be hospitalized);
  • failure to report important information to outpatient health care professionals;
    premature discharge from the index hospitalization; and
  • lack of discussion about care goals among patients with serious illness.

Dr. Auerbach, who is also editor in chief of the Journal of Hospital Medicine, says he was surprised that premature discharge was such a big factor.

“We can’t not discharge,” he says, “and we can’t extend length of stay unnecessarily, so we have to think more about what may enhance patient readiness to go home.”

That analysis leads to one conclusion, he adds: “The one aspect most commonly associated with risky early discharge was poor communication.” Suboptimal communication also extends post-discharge with outpatient physicians and ED doctors when patients present to the ED again.

Dr. Auerbach suggests bringing together hospitalists, nurses, case managers, outpatient providers, home health clinicians and others. Have that team track how often specific readmission factors come up over time, then prioritize solutions to address them.

“We hope that will reduce readmissions without breaking the bank,” Dr. Auerbach says. He’s now working to develop a data platform to help other hospitals look at their readmission data and develop improvement projects.

“We were surprised at how many readmissions in our study were deemed not preventable.”

cover-readmssion-janjigian~ Michael Janjigian, MD, Bellevue Hospital, New York

 

He also hopes such data will prompt hospitals to seek broad-based solutions. The problem with ED physicians readmitting patients unnecessarily, for example, is a “limitation of the health system itself,” not of emergency medicine, he wrote in the study. The “fix” might be improving staffing in local clinics later in the day if that’s when readmissions from the ED occur.

Dr. Auerbach also notes that the study was not meant to provide detailed solutions, but to lay the groundwork to help hospitals look for answers.

“There’s no penicillin for any of these problems,” he says. “Think of what you have on hand and what can help patients manage better at home.”

Classification effort can pay off
While Dr. Auerbach and his colleagues looked at readmissions across 12 sites, one New York academic center did a taxonomy of its own seven-day readmissions. Any hospital can use the five-category algorithm that center used to develop targeted interventions.

That’s according to hospitalist Michael Janjigian, MD, director of inpatient general medicine at New York’s Bellevue Hospital and assistant professor of medicine at the NYU School of Medicine. His study at the 800-bed Bellevue Hospital Center found that 46% of readmissions were due to unpredictable or unpreventable complications like disease progression.

Readmissions due to patient behavior (substance abuse, nonadherence) made up another 19%. Discharge-process deficiencies accounted for about 17%, readmissions related to leaving against medical advice were 12%, and unnecessary readmissions were 7%. Across those categories, researchers determined that 24% of readmissions were deemed preventable.

Here’s how the classification system works:

      • Was the readmission medically necessary?
      • Did the patient leave against medical advice during the first admission?
      • Was the readmission due to a deficiency in the index discharge process?
      • Was the readmission caused by substance use or nonadherence related to the first admission?
      • Was the readmission related to a complication of the primary disease or treatment, or to an unrelated condition that could not have been predicted or prevented?

Consider an oncology patient being treated for a new diagnosis of colon cancer; he is discharged to an oncology clinic with appropriate follow-up but then readmitted with worsening abdominal pain. The readmission was medically necessary, the initial discharge process was appropriate, and the patient was not using drugs or alcohol. As a result, the readmission would be categorized as class 5: an unavoidable, unpredictable disease progression.But say the patient came to the ED with nausea that could have been better controlled if the ED had called the oncologist. That readmission would then be a class 1: It wasn’t medically necessary, and it could have been prevented through better communication.

Such a labor-intensive classification should be done at year’s end so you have enough charts to choose from, then periodically to monitor trends, Dr. Janjigian says. He studied 400 charts, but says fewer would work for a smaller institution.

He recommends giving the job to senior clinicians— his study used hospitalists—who better understand how hospitals work and have more patient care experience. Here’s his example of how that advantage might play out with a patient discharged after a heart failure exacerbation who comes back with shortness of breath and lower-extremity edema.

A junior faculty member may think the readmission from the ED was appropriate. But a senior clinician would understand that the patient could have instead been discharged from the ED to a heart failure clinic. The patient could also have benefited from an assigned case manager who would make sure the patient received the right medications and appointments.

Once your results are in, says Dr. Janjigian, look for areas where low-cost, low-resource interventions can succeed. For example, Bellevue did improve access to its heart failure clinic so the ED would feel more comfortable not readmitting such patients.

Despite that success, “we were surprised at how many readmissions in our study were deemed not preventable,” Dr. Janjigian says. “It’s frustrating when you realize how much of this is out of your control.”

Would a prediction tool work?
Reliable. Valid. Generalizable. Finding a tool that meets those criteria to predict potentially avoidable 30-day hospital readmissions just got easier. But it may not be quite ready for prime time, according to the author of a recent study on the HOSPITAL score.

The encouraging news is that the study found the
 HOSPITAL score generalizable across more than
117,000 patients in nine large hospitals in four differ
ent countries, giving a boost to the usability of the straightforward prediction model.

“Studies show that intervention are not always effective in reducing readmission probability because they’re not targeted to high-risk patients,” says Jacques D. Donze, MD, MSc, associate physician at the Bern University Hospital in Switzerland and lead author of the study published in the April issue of JAMA Internal Medicine. “HOSPITAL can make us more effective in targeting those patients.”

Seven variables are scored according to the HOSPITAL acronym:

  • H: last available Hemoglobin before discharge (positive if <12 g/dL)
  • O: discharge from an Oncology service
  • S: last available Sodium level before discharge (positive if <135 mEq/L)
  • P: any Procedure performed during the hospitalization
  • Index admission Type (emergent or urgent, not elective)
  • A: number of Admissions in the previous 12 months
  • L: Length of stay (positive if > 5 days)

The score can be automatically calculated through the EHR or at the bedside for smaller hospitals before discharge, Dr. Donze says. That way, case managers can put transitional care interventions in effect to try to prevent readmission. He does note, however, that the LOS and lab-value predictors aren’t known until discharge, so those don’t help identify high-risk patients early.

Additionally, Dr. Donze explains, the variables are predictors and not necessarily modifiable risk factors. For example, a patient with a low hemoglobin at discharge doesn’t mean that level has to increase to reduce readmission risk. It just means that general interventions—not necessarily related to a specific predictor—are called for.

So does using the score reduce readmissions? Dr. Donze is planning a study in which he’ll identify patients at high risk of readmission, then randomly divide them into two groups. One will receive usual care, while the other will have additional interventions, such as medication reconciliation before discharge and two follow-up phone calls. He wants to see if the combined identification and interventions can reduce readmission risk.

“It’s the next logical step,” Dr. Donze says. “Hospitals now are just providing interventions to all their patients without knowing if they will work or not.”

Do patient self-reports at discharge matter?
When Vanderbilt’s Dr. Mixon looked for a relationship between how well-prepared patients said they felt at discharge and how often they were readmitted, she found that self-reports didn’t trump other types of tools. But that study, which was published online in February by the Journal of Hospital Medicine, shed some light on identifying potential bouncebacks.

The study tested two patient-reported preparedness measures: the 11-item brief-PREPARED (B-PREPARED) tool and the three-item Care Transitions Measure-3 (CTM-3).

B-PREPARED measures how patients feel at discharge on such metrics as self-care and medication information and community services needed. The CTM-3 uses a four-point scale to ask patients to rate how well their preferences for transitional needs had been considered, as well as their understanding of post-discharge self-management and medications.

B-PREPARED showed promise: A four-point increase in that score was associated with a 16% decrease in 30-day and 90-day readmissions or death. The CTM-3, however, did not predict either. And researchers found that the predictive value of both paled in comparison to the LACE index, a measurement of length of stay, acuity, comorbidity and emergency department use.

“I’m an advocate for the patient, what he thinks and fears about going home,” says Dr. Mixon. “Yet our results don’t show that patient-reported outcomes really added anything outside of administrative data.” She also “didn’t anticipate that the LACE index would be so powerful.”

That raises the question of why use self-reported measures, especially if the results come in after discharge. “To be proactive, we want to try to give information to people before they go home,” such as setting up home services, she says.

For now, Vanderbilt is using its own 10-measure in-house predictive model, dubbed “Cornelius,” that considers clinical variables, including the number of hospitalizations in the last six months.

However, Dr. Mixon admits she doesn’t know how useful that model is. Even though it generates a score in the medical record, not all physicians use it—or even know what it means, she says. The health system is considering ways to better use the tool, perhaps by incorporating its results into clinical huddles. But she wonders how it would compare to LACE.

“We haven’t used LACE,” she says, “but it’s caught our eye.”

Paula S. Katz is a freelance health care writer based in Vernon Hills, Ill.

Factoring in socioeconomics

IT’S THE READMISSIONS ELEPHANT in the room. Dinging hospitals for 30-day readmissions without considering socioeconomic factors may unfairly penalize some hospitals, says John Martin, PhD, MPH, executive director of Premier Research Institute in Charlotte, N.C., an analytics arm of Premier Inc., a health care improvement company.

His study of data from 15 million discharges found that clinical factors such as heart failure (20%), COPD (198%), renal failure (17%) and septicemia (17%) are the strongest predictors of readmissions. But so too are race, income, distance traveled and payer status, he wrote in an article in the April 29156 Journal of healthcare Quality.

Because the Centers for Medicare and Medicaid Services (CMS) doesn’t account for such factors, hospitals that serve a disproportionate share of vulnerable populations will appear to do worse in terms of readmissions.

“We weren’t surprised by the results because we know socioeconomic factors are associated with outcomes,” Dr. Martin says. “We know hospitals dealing with riskier patients have the potential to be penalized.”

Hospitals can’t just change the socioeconomics of the community they serve. “That’s the point of the article,” he notes. “We want to be sure there’s no concern for these hospitals to bring in higher-risk patients. They should just focus on treatment, as would any other hospital.”

The CMS has heard that concern. Last year, it published “A Guide to Preventing Readmissions Among Racially and Ethnically Diverse Medicare Beneficiaries.” And its 2016 fiscal year final report notes that the agency is working with eh national Quality Forum to test risk adjustment for such sociodemographic factors as income and occupation as well as race, ethnicity and primary language. That trial is supposed to last two years.

Who’s at high risk of readmission?

AN INTERNATIONAL GROUP of researchers validated what they call the HOSPITAL score, a tool designed to help predict which patients are at high risk of potentially preventable readmissions. The acronym stands for several clinical variables and patient characteristics. Here’s how those variables played out in terms of percentages of patients with potentially avoidable 30-day readmissions:

  • Urgent or emergent index admission: 83%
  • 1 procedure in index admission: 78%
  • Low hemoglobin at discharge (<12 g/dL): 71%
  • Index admission LOS > 5 days: 54%
  • No. of hospital admissions in past year:
    <1: 43%
    <2-5: 40%
    <5: 18%
  • Low sodium at discharge (<135 mEq/L): 18%
  • Discharge from an oncology division: 7%

Source: JAMA Internal Medicine

 

NO COMMENTS