The Centers for Medicare & Medicaid Services (CMS), the nation's largest health care payer, has been giving the squint-eye to some types of hospital readmissions for years—and oncologists are wondering when the government's gaze will turn their way.
“I think it's coming,” said Carl Schmidt, MD, a surgical oncologist at the James Cancer Hospital at the Ohio State University Wexner Medical Center.
He is one of many researchers around the country who are trying to understand hospital readmissions for cancer patients: How many readmissions are too many? Should readmissions be a quality of care metric? Are some readmissions avoidable? If so, which ones—and how?
They are responding to the so-called value movement in health care, in which government and private insurers wish to reward physicians and hospitals that provide high-quality, low-cost care and penalize those that do not. The value movement, still in its infancy, requires a reconsideration of what good care means.
“If you went back 15 or 20 years ago—maybe even 10 years ago—and asked most physicians about readmissions, a lot of people would tell you that, in fact, readmitting your own patients to care for them after certain problems related to treatment or operation or disease is how you provide great care,” Schmidt says.
Fast forward to 2016, when 2,665 hospitals—nearly three-quarters of all the hospitals in the country—are seeing their Medicare pay docked because CMS says they had too many readmissions within 30 days after patients were discharged.
To date, CMS is only looking at readmissions for a handful of diagnoses, none of which are cancer-related. And that's the way it should stay, says Richard Bold, MD, a surgical oncologist at the University of California-Davis Health System.
He and several colleagues reviewed a large database of cancer patients and found that most hospital readmissions are not preventable (Ann Surg. 2014;260:583-591).
“So you shouldn't penalize people for what is really good care,” Bold said. “On the clinical side, we are pushing back and saying these [penalties for readmissions] weren't really thought out.”
CMS' View on Readmissions
CMS cast its eye on hospital readmissions more than a decade ago when it recognized that many Medicare patients with certain diagnoses frequently bounced between home and hospital. The agency thought some of these readmissions were preventable if the hospital helped patients successfully transition to home after a discharge.
CMS was right. The national readmission rate—instances in which a patient, regardless of diagnosis, returns to a hospital within 30 days after discharge—fell to 17.5 percent in 2013 after many years in the 19 percent range (American Hospital Association Trendwatch, 2015).
The government thinks hospitals can do better. In 2013, CMS' Hospital Readmissions Reduction Program began reducing Medicare payments to hospitals deemed to have “excess” readmissions for patients being treated for heart attack, congestive heart failure, and pneumonia. In the years since, the penalty has increased to a maximum of 3 percent of a hospital's total Medicare pay and other diagnoses—chronic obstructive pulmonary disease and hip/knee replacements—have been added to the watch list.
CMS measures excess readmissions by dividing a hospital's number of 30-day readmissions for those five conditions by the number that would be “expected,” based on an average hospital with similar patients. A ratio greater than one indicates excess readmissions.
More than 3,400 hospitals are subject to the government's readmissions reduction program, but certain types of hospitals, including comprehensive cancer centers, pediatric hospitals, and psychiatric hospitals are not.
Despite that, Joanna-Grace Manzano, MD, a hospitalist at The University of Texas MD Anderson Cancer Center, thinks all types of cancer care providers should monitor their readmission rates because they might be an important quality measure. She and her colleagues recently analyzed data for a subset of MD Anderson patients to find a baseline readmission rate and are exploring various interventions that might reduce it.
“After we have completed our investigation we will have a better understanding of whether this is a good metric for our population and, if it is, what should be taken into consideration in terms of risk adjustment,” she said. “We have to be better informed.”
What Does Good Look Like?
Because they have been studied for years, readmissions for congestive heart failure and many other conditions are well-understood. Health systems can easily benchmark their performance against their peers, and interventions to reduce potentially preventable readmissions have been developed, tested and spread throughout the industry.
Not so for cancer care. Although some institutional consortiums share data that allow members to compare their readmission statistics for cancer patients, most provider organizations have no way of knowing how their performance stacks up. Indeed, most have never analyzed their own readmission data for cancer patients.
Manzano wants to find out whether some readmissions can be prevented, and she needed baseline data to start answering that question. Analysis of 884 hospitalist discharges on MD Anderson's general internal medicine hospitalist service during a six-month period of 2012 found a 22.6 percent unplanned 30-day readmission rate. The study included only readmissions to MD Anderson; the researchers did not know if patients were readmitted to another hospital (JOP2015;11(5):410-415 DOI:10.1200/JOP.2014.003087).
She and her colleagues believe some readmissions can be avoided, although which ones—and how—is not yet clear.
“There really is no standardized definition of what a preventable readmission is in the cancer context,” she said. “Until we have an evidence-based definition, the validity of readmissions as a measure of quality care for our cancer patients will always be challenged.”
She points out that many readmissions are related to the nature of the cancer itself, disease progression, or treatment side effects. In her study, the most common reasons for unplanned readmissions were complications related to metastatic disease such as malignant ascites; intestinal obstruction; gastrointestinal hemorrhage; septicemia; biliary tract disease; acute renal failure; complications with device; and urinary tract infection.
“For most of these, management in the hospital setting is appropriate considering that many of our cancer patients are elderly and have several comorbidities,” Manzano said.
To explore effective interventions, the hospitalist service is borrowing ideas that have helped reduce readmissions for other types of patients. Those include:
- Interdisciplinary rounds, in which physicians, advanced practice providers, pharmacists, nutritionists, therapists, nurses, social workers, and case managers try to identify and proactively address problems that patients may encounter after discharge;
- Medication reconciliation for patients at risk of drug-related adverse events or poor compliance with prescriptions; and
- Post-discharge phone calls.
The service is considering a post-discharge clinic for patients believed to be at high risk for readmission.
“We are trying to attack this at several different levels,” Manzano said. “Whether any of this is effective [in reducing readmissions] remains to be seen. It's certainly worth trying because whatever intervention we put in place—care coordination, early follow-up, a phone call within 72 hours—is good patient care.”
Measuring readmission rates for cancer patients is complicated by the fact that, unlike congestive heart failure and some other medical conditions, cancer is so many different diseases.
Alberto Montero, MD, Quality Improvement Officer at Cleveland Clinic Cancer Center, has led a team that has reviewed data for the cancer center's palliative medicine and general medical oncology inpatient services over a 16-month period in 2013 and 2014 and calculated a 30-day readmission rate of 25.7 percent.
“Within our own data, there's a lot of variability,” Montero said. “You would expect that a normal readmission rate will depend on the kind of cancer, even what stage the patient is and what treatment they are receiving.”
That said, Cleveland Clinic found that its overall rate could be reduced. As part of a system-wide focus on reducing readmissions, the cancer center introduced three interventions in April 2014—provider education; nurse phone calls to patients within 48 hours after discharge; and follow-up appointments within five days—on the two services and saw the readmissions rate fall to 20.4 percent over the next eight months.
Nurses in the outpatient center called patients to check their status, make sure they were taking medications appropriately, address patients' questions, and verify that a follow-up appointment was scheduled. Compliance with the call-backs and the early follow-up has been high, but the readmission rate nonetheless drifted a bit upward in late 2015, Montero said.
They have been searching for possible explanations, such as increased outside referrals of heavily pretreated patients and a higher comorbidity index. Meanwhile, he and his colleagues are looking for a way to pre-empt emergency department visits that inevitably lead to inpatient admission.
“We are exploring the idea of having nurse practitioners make home visits to see if we can intervene so the patient can be cared for at home and not have to come back to the hospital,” he said. “We are also looking at social factors because there's a lot of literature that shows people who don't have social support are predicted readmissions, independent of the underlying medical condition.”
Predicting Readmission Risks
Looking at Medicare patients only, analysis of inpatient admissions—including surgical and medical patients—at OSU's James Cancer Hospital for a six-month period in 2011 and 2012 revealed an unplanned readmission rate of 11 percent. Schmidt and his colleagues at James also track readmissions by service and by diagnosis, where considerable variation is seen (JOP 2015;11:410-415).
“I try not to automatically come to a conclusion that a number is too high or just right or low, and therefore laudable, because we're still learning,” he said. “The number means much less to me than coming up with better systems and providing better care.”
To that end, his research team did chart reviews to discover whether some readmissions could have been prevented. That is an imperfect process, Schmidt said, because it required the reviewers to make judgments based on the documentation in a patient's chart, which may fail to capture details that informed decision-making at the time.
The reviewers determined that 21 percent of readmissions were potentially preventable, and the most common causes were errors in discharge planning, inadequate treatment of infection present at the index admission and missed diagnosis.
“In fact, most readmissions looked unavoidable,” Schmidt said. “They were caused by new complications related to existing disease or treatment that weren't present at the time of index discharge.”
He believes every institution should track its readmission rates for specific subsets of patients—for example, those with a certain diagnosis or who meet certain criteria—and work to reduce that rate. “I'm absolutely sure there are some readmissions we can prevent by providing better care,” he said. “We need to figure out what those are and do it.”
The James' researchers have developed a model to use medical record data to predict patients at high risk of readmission so the medical team can proactively address the problems that might trigger readmission. The model is undergoing validation before it is published.
Schmidt encourages provider organizations to share their readmission data so that providers, payers, and patients can eventually understand what reasonable readmission rates are and the causes—appropriate and inappropriate—for variance.
Groups of providers “should work hard to establish the benchmarks themselves and not leave it to insurers or any other entity that may have motivation for setting benchmarks that are not patient-focused or are dangerous or do not reflect the reality of patient care,” he said.
Readmissions as Quality Measure
After reviewing the data of more than 2.5 million patients hospitalized for cancer, Bold, the UC-Davis researcher, believes the 30-day readmission rate is not an appropriate quality measure for use in payment policy.
He and his colleagues analyzed data about cancer patients admitted to 240 hospitals—academic medical centers, including National Cancer Institute-designated comprehensive cancer centers, and their affiliated hospitals—that belong to the University Health System Consortium (UHC) during a 44-month period ending in September 2013.
Their findings included:
- NCI-designated cancer centers and hospitals that serve a high volume of cancer patients had higher rates of readmissions than other hospitals.
- Factors associated with readmission include discharge from a medical service (rather than surgical service); emergency surgery as the reason for the initial admission; and complex procedures during the initial admission. None of these are modifiable.
However, Bold thinks readmissions can be prevented in some situations. Data from his own institution, University of California-Davis Medical Center, showed that cancer patients hospitalized between January 2010 and September 2013 had a seven-day readmission rate of 3.4 percent; a 14-day rate of 5.8 percent; and a 30-day rate of 9.4 percent.
The most common readmission diagnoses—at seven days and 30 days—were infections, nausea/vomiting/dehydration, and thromboembolism. The medical center's efforts to predict which patients might develop an infection that required a readmission have been unsuccessful. But Bold thinks improved care transitions, including post-discharge phone calls and early follow-up appointments for patients at risk of nausea, vomiting, and dehydration are preventing some readmissions.
“What we have learned here in the last couple of years is that there are some areas in which we can move the needle, but that there are some that we can't,” he said.
He encourages all hospitals to start tracking their readmission rates, knowing that the rates will differ significantly based on patient mix, institutional size, and other factors.
“I don't think comparison without adjusting for some of those factors is relevant,” he said. “But we are probably going to get to a point of doing risk adjustment based on some of the factors that we do know influence readmission so that hospitals will have a way to benchmark themselves.”