By Melinda Young

EXECUTIVE SUMMARY

Decision support technology can help case managers improve transitions of care and more easily access patient information necessary for an optimal discharge.

  • The algorithm can collect data on patients’ functional status, cognition, caregiver status, and other important characteristics.
  • Advanced practice nurses helped develop the algorithm by studying cases to determine if they would refer these patients to post-acute services.
  • Case managers can partner with information technology staff to implement the electronic decision support tool.

Hospital case managers can leverage decision support technology to improve transitions of care and the discharge process.

Case managers know where to find information about patients’ comorbidities, caregiver support, admissions history, and other factors that affect discharge. But this information may be scattered throughout the patient’s chart, or is unknown, says Kathryn H. Bowles, PhD, RN, FAAN, FACMI, professor of nursing and van Ameringen chair in nursing excellence at the University of Pennsylvania School of Nursing. An algorithm that works within an electronic health record (EHR) can find the necessary information and allow case managers to access it quickly and efficiently.

For instance, the Discharge Referral Expert System for Care Transitions (DIRECT) algorithm can help case managers and clinicians regarding post-acute care referrals and level of care.1 “DIRECT puts it all in one place for them and gives them the gestalt of the patient,” Bowles says.

Case managers can get a feel for what is needed for an optimal discharge. But without a technological tool to provide the needed information in one place, case managers might not be able to synthesize the data as efficiently. With the support, they can quickly know the patient’s function has declined over the hospital stay, and they will know the caregiver status.

Decision Support

In research and development of DIRECT, the electronic record ran the algorithm 24 hours after patients were admitted to the hospital. Case managers could receive the decision support several times a day, if desired. It produces a spreadsheet that characterizes patients in terms of whether they will need post-acute care. It also recommends the level of care, home care, or facility.

“It gives case managers an early heads-up that this is a patient they should consider for post-acute services, based on the patient’s characteristics,” Bowles explains

These are the most common characteristics included in the algorithm:

  • Functional status;
  • Cognition;
  • Caregiver status;
  • Admissions history;
  • Comorbid conditions;
  • Access to the house and condition of the house.

The decision support tool makes a case manager’s job easier, but developing and integrating the technology is more challenging. Bowles worked on developing the tool for years and has researched the use of DIRECT and an earlier version of the algorithm.

Bowles and co-investigators built and validated an expert clinical decision support system for discharge referral decisions about post-acute care. They published a study that showed highly satisfactory predictive summary statistics on the algorithm. The authors suggested evidence-based decision-support tools for discharge planning can alleviate workloads for discharge planners.2

Researchers studied patients taking multiple medications and histories of previous conditions. Their likelihood of readmission was high. “You would expect that people who are at high risk of readmissions would come out of acute care and receive skilled nursing facility services, or something else to further their recovery,” she explains. “The majority of the patients went home to self-care.”

The investigators asked advanced practice nurses to study these cases and determine if they would refer these patients to post-acute services. The nurses said they would have referred all but a couple of the patients to post-acute care.

“We looked at those patients who didn’t get services, and 49% of them were readmitted within 12 weeks,” Bowles says.

Among patients who went home without services, those who were flagged as needing services were readmitted at a rate of 5% higher than those who were not flagged by the algorithm.3

Case management and care transition guidelines and best practice models also are useful tools. But these lack the dynamic flexibility of an electronic decision support tool that can be adapted to individual hospitals and units. “The elements within the algorithm need to be pulled from the databases of hospitals,” Bowles says.

The first-generation discharge algorithm that preceded DIRECT was built into software and used by dozens of hospitals, she adds.

Partner with IT

The integration process for this tool works more smoothly with a hospital champion. “Case managers are key people to say, ‘We really want this and want to have better outcomes for our patients,’” Bowles says.

Once hospitals decide to use a decision support tool, case managers can partner with information technology (IT) staff to decide what information to collect and how to collect it. Case managers explain the data they need to determine patients’ functional status, and IT professionals explain how the databases work and how things are collected in the electronic record.

One thing to keep in mind is that while the algorithm suggests which patients would benefit from certain post-acute services, physicians and patients might disagree and choose other options.

“The algorithm identifies people who needed services and didn’t get them, and people who did not need the services and did get them,” Bowles says. “We haven’t studied why this happened, but one reason why the algorithm identifies people who don’t get the services is because of patient refusals.”

For example, the authors of another study performed a chart review of patients who refused post-acute services. They found the most frequent reason for refusal listed by men was their spouse would be the caregiver.3

Other reasons for refusals of services include:

  • Patients do not want someone in their home.
  • Patients do not carry enough insurance coverage for the service or cannot afford the copay.

“We built our algorithm using expert clinicians, doctors, nurses, social workers, and physical therapists who evaluated case studies,” Bowles explains. “We had almost 1,500 case studies of actual, hospitalized patients, and we had them review these cases and tell us what they would do and whether they would refer them or not for post-acute services.”

The algorithm was built on clinical need and expert opinion, and did not consider barriers such as homebound status or insurance, Bowles says.

After integrating the decision support tool in the EHR, case managers need instruction on how to use the tool. Optimally, they were involved in the integration process and were familiar with and supportive of the tool.

“We educated case managers about the algorithm, how we developed it, how it worked,” Bowles explains. “We find out how they want to receive the information, which information they want, and where in their workflow they could review it.”

Then, the program sent case managers a spreadsheet, twice daily, of their patients and algorithm advice.

Investigators found the patients who refused the services were twice as likely to be readmitted within both 30 days and 60 days, compared with those who accepted the referral.3

The reports can be sent to case managers at any time and as many times a day as desired. Using the information, case managers can help patients and their families prepare for a post-acute referral by introducing the idea that the patient may be sent to a skilled nursing facility or need home health services after leaving the hospital, Bowles says.

REFERENCES

  1. Bowles KH, Ratcliff SJ, Holmes JH, et al. Using a decision support algorithm for referrals to post-acute care. J Am Med Dir Assoc 2019;20:408-413.
  2. Bowles KH, Ratcliffe SJ, Naylor MD, et al. Nurse generated HER data supports post-acute care referral decision making: Development and validation of a two-step algorithm. AMIA Annu Symp Proc 2018;2017:465-474.
  3. Keim SK, Bowles KH. Comparison of algorithm advice for post-acute care referral to usual clinical decision-making: Examination of 30-day acute healthcare utilization. AMIA Annu Symp Proc 2018;2017:1051-1059.