Background:

Patients frequently admitted to the hospital use a disproportionate amount of healthcare resources. However, scant information exists about the clinical and social characteristics of this population, as well as the quality of their transitions of care. Qualitative and quantitative descriptions of these patients may aid in developing strategies for reducing future readmissions.

Methods:

We obtained administrative data on patients admitted in FY2011 to a 600–bed academic medical center participating in Project BOOST. We identified patients with an index admission on the medical service and >6 readmissions that year, randomly selecting a subgroup of patients for chart review. We examined demographic, clinical, and social data as well as process measures associated with discharge quality during their five most recent hospital stays. To quantify resource utilization, we obtained claims data looking at length of stay, ED visits, and ICD9 codes. Where possible, we compared measures to our overall inpatient medical population.

Results:

Of 3,900 patients admitted in FY2011, 39 patients (1%) were readmitted >6 times, accounting for 304/2275 (13%) of all readmissions. Eighty–seven percent of these readmissions were to the medical service. Of 30 patients selected for chart review, one was excluded because most readmissions were to a nonmedical service. The majority were housed, and many were discharged home with services (Table 1). Psychiatric illness, alcohol, and drug abuse were prevalent. Twenty–eight percent of patients had a terminal diagnosis, though only one was discharged on hospice. BOOST interventions (timely PCP appointments at discharge and follow–up phone calls) were successfully completed in a large proportion of these patients, but were lower than for the total population (53% and 40% vs 80% and 60%, respectively). Based on claims data, this population had 372 ED visits and were admitted 61% of the time. Thirty percent of all hospital stays were <48 h. These patients had a shorter average length of stay (5 vs 5.5), a lower case mix index (1 vs 1.5), and a higher government payer mix (89% vs 77%) compared to our general population. The top five principal diagnoses were sickle cell pain crisis, pneumonia, COPD, renal failure, and persistent vomiting, which differed significantly from our general population.

Conclusions:

We used administrative data and chart review to identify characteristics of a frequently readmitted population. They suffered from advanced and complex medical problems and social needs and had multiple points of contact for possible interventions. Given their complexity, a high–intensity multidisciplinary intervention may be necessary to reduce future readmissions of this patient population.

Table 1Chart Review of Frequently Readmitted Patients