Readmission rates

The US Centers for Medicare and Medicaid Services initiated public reporting of 30 day readmission rates for patients with acute myocardial infarction, pneumonia, and heart failure in 2009,1 and financial penalties began to be imposed on underperforming hospitals in 2012. Since then hospitals and health services researchers have been searching for ways to reduce hospital readmissions while trying to understand the causes and importance of readmissions. The ultimate objective is to improve care, but the path to that goal is not clear. Two linked studies help point us in the right direction.2 3

The literature on readmission rates has three general themes. The first consists of challenges to the suitability of readmission rates as quality measures,4 including their potential to decrease access to care,5 and a vexing concern regarding a possible inverse correlation between 30 day readmission for heart failure and mortality rates.6 7 The second includes studies of various interventions to reduce readmission rates,8 such as risk prediction tools.9 Studies of these efforts report mixed results, and comprehensive programs seem to be more effective than focused interventions.8 10 The two linked studies contribute to the third theme—investigations into the drivers of readmissions.

Dharmarajan and colleagues (doi:10.1136/bmj.f6571) analyzed the inpatient claims of all US beneficiaries of Medicare readmitted within 30 days of an index hospital admission for heart failure, myocardial infarction, or pneumonia during 2007 to 2009. They found that high performing hospitals (those with lowest readmission rates) had fewer readmissions across all three diagnoses.

Donze and colleagues (doi:10.1136/bmj.f7171) analyzed data from 10 731 patients discharged from a single US hospital during 2009 to 2010, 2398 of whom were readmitted within 30 days to one of four hospitals within the same network. They determined that five of the most common primary readmission diagnoses were related to one of seven comorbidities. Furthermore, in patients readmitted with a diagnosis related to a comorbidity, most had a different primary diagnosis at the time of their index admission. In other words, the patient’s comorbidities are as likely to be the cause of readmission as the principal diagnosis at the time of the first admission.

What lessons can we learn from the observations in these two studies? The first confirms the findings of others: patients are readmitted and not diagnoses.11 Interventions to reduce readmissions that focus solely on the principal diagnosis at the time of the index admission are unlikely to achieve optimal results.

The second lesson is similar and is equally intuitive: hospitals that are good at reducing readmissions for one diagnosis are good at reducing readmissions for all diagnoses. There is apparently something in the way these institutions provide care that is not disease specific and that results in favorable outcomes across conditions. This finding is reminiscent of a qualitative study by Curry and colleagues,12 which found that hospitals with low 30 day mortality rates after myocardial infarction were differentiated from poorly performing hospitals by an organizational culture that supported efforts to improve care across the hospital and not by specific interventions. It is therefore possible that excellent performance on 30 day readmission rates could also be driven by organizational characteristics.

Through these efforts a clearer picture is beginning to emerge of underlying causes of high readmission rates and the approaches to improving not only this metric but also measures such as mortality and quality of life that give a more holistic view of care. Nevertheless, further research is needed in several other areas. Firstly, we need an in-depth understanding of the specific interventions that high performing hospitals use to reduce readmissions and insights into organizational characteristics that may be the true drivers of success. Secondly, the causes of readmission—including the contributions of inadequate treatment, poor systems for handing over care between incoming and outgoing staff, socioeconomic factors, and progression of disease—need to be understood more fully. Thirdly, we also need a better understanding of the impact that various “actors”—including hospital physicians, outpatient providers, hospitals, payers, and the patients themselves—exert on readmissions. In addition, studies should determine the true value of the various risk prediction tools for reducing readmissions in a broad array of hospitals. What are the crucial elements in these tools, and how can we implement them more effectively?

Finally, the use of readmission rates as accountability measures for large numbers of hospitals raises further questions that deserve prompt attention. Are hospitals gaming the system by artificial maneuvers, such as placing patients in “observation status” to make the numbers look better? And the most fundamental question of all: what is the balance of benefits and harms associated with using readmission rates in incentive programs for providers? The proposed benefits of transparency may have unintended consequences, such as distracting providers from other quality improvement efforts and decreasing access to care for the sickest patients. The possible inverse relation between 30 day readmission and mortality rates for patients with heart failure has yet to be tested in the setting of the strong financial incentives now placed on all US hospitals.

Despite intense interest among health services researchers, our understanding of the causes of readmissions, effective interventions to avoid them, and the full impact of using readmission rates as measures in a hospital incentive program remains rudimentary. We need additional insights of the type provided by Dharmarajan and colleagues and Donze and colleagues to provide the right answers for our patients.