This population-based study examines socioeconomic and clinical factors associated with scheduled and unscheduled readmissions after discharge among older patients with diabetes.
Published Online: October 15, 2010
Hongsoo Kim, PhD, MPH; Joseph S. Ross, MD, MHS; Gail D. Melkus, EdD, C-NP; Zhonglin Zhao, MD, MPH; and Kenneth Boockvar, MD, MS
Objectives: To describe rates of scheduled and unscheduled readmissions among midlife and older patients with diabetes and to examine associated socioeconomic and clinical factors.
Study Design: Population-based data set study.
Methods: Using the 2006 California State Inpatient Dataset, we identified 124,967 patients 50 years or older with diabetes who were discharged from acute care hospitals between April and September 2006 and examined readmissions in the 3 months following their index hospitalizations.
Results: About 26.3% of patients were readmitted within the 3-month period following their index hospitalizations, 87.2% of which were unscheduled readmissions. Patients with unscheduled readmissions were more likely to have a higher comorbidity burden, be members of racial/ethnic minority groups with public insurance, and live in lower-income neighborhoods. Having a history of hospitalization in the 3 months preceding the index hospitalization was also a strong predictor of unscheduled readmissions. Almost one-fifth of unscheduled readmissions (constituting approximately 27,500 inpatient days and costing almost $72.7 million) were potentially preventable based on definitions of Prevention Quality Indicators by the Agency for Healthcare Research and Quality. Scheduled readmissions were less likely to occur among patients 80 years or older, the uninsured, and those with an unscheduled index hospitalization.
Conclusions: The predictors of scheduled and unscheduled readmissions are different. Transition care to prevent unscheduled readmissions in acutely ill patients with diabetes may help reduce rates, improving care. Further studies are needed on potential disparities in scheduled readmissions.
(Am J Manag Care. 2010;16(10):760-767)
This study used a population-based data set to examine socioeconomic and clinical factors associated with scheduled and unscheduled readmissions after hospital discharge among older patients with diabetes.
- A substantial percentage (12.8%) of readmissions were scheduled at least 24 hours in advance.
- The causes and predictors of scheduled and unscheduled readmissions were distinct, suggesting that scheduled readmissions should be separated from unscheduled readmissions in studies of readmission.
- The lower likelihood of scheduled readmissions among older and uninsured patients implies less access to advanced coordinated healthcare services for these groups.
Diabetes mellitus affects about 23.6 million individuals in the United States and is the seventh leading cause of death.1 Because of the complexity of the disease and its management, patients with diabetes are more likely to use healthcare services than the general population,1 with a total direct medical cost of about $116 billion and a mean cost that is 2.3 times higher than that of individuals without diabetes.1,2 Moreover, almost 6.3 million hospital stays in 2004 were by patients with diabetes as a principal or coexisting condition, costing almost $57.8 billion, or about 20% of total hospital costs, while hospitalizations principally for diabetes cost about $3.9 billion.2
Many of these hospital stays included readmissions of the same patients, and hospital readmission has been a major quality concern in diabetes care. However, there are few population-based studies on readmission among patients with diabetes and its risk factors. Jiang et al3 reported that more than 30% of older patients with diabetes were readmitted during the same year as their index hospitalization in 5 states in 1999. Readmissions accounted for 55.2% of total hospital stays, and the mean total hospital cost for patients with readmissions was 2.5 times higher than that for patients without readmissions. They also reported a high risk of readmission among racial/ethnic minority groups and variations in readmission rates by insurance type.4,5 Robbins et al6 and Robbins and Webb7,8 observed racial/ethnic disparities among hospital readmissions in Pennsylvania hospitals between 1994 and 2001 and noted the positive effect of public health clinics in reducing hospital readmissions among low-income patients with diabetes.
A gap in the literature on hospital readmissions among patients with diabetes has been the lack of differentiation between scheduled and unscheduled readmissions.3-5 Readmissions are often assumed to be unscheduled, but this may not be true, especially for patients with existing complex conditions such as diabetes. The need for attention to the type of readmission has been raised,9 but few studies6,7 have explicitly focused on unscheduled readmissions only. No study seems to have assessed what the reasons for and the factors associated with scheduled readmissions are, and whether they are similar to those associated with unscheduled readmissions.
Another gap in the literature is that studies have often focused on readmissions within 1 month of hospital discharge6,7,10 or on readmissions among Medicare patients.10,11 This short observation window is because hospital readmission has been widely studied in terms of its relationship to hospital performance.7,12 However, this may be too narrow an approach to evaluate readmission risk from the perspective of patients, for whom avoiding a later readmission is as important as avoiding an early readmission. Investigations focusing on patient risk for readmission have often used a 3-month window following hospital discharge,5 which we adopted.
The objectives of this study were to examine and compare the demographic, socioeconomic, and clinical factors associated with scheduled and unscheduled readmissions in midlife and older patients with diabetes using a population-based data set from California. We also examined common reasons for scheduled and unscheduled readmissions and estimated the cost and extent of potentially preventable readmissions using Prevention Quality Indicators (PQIs) of the Agency for Healthcare Research and Quality (AHRQ).13
We analyzed the 2006 California State Inpatient Dataset (SID), developed as part of the Healthcare Cost and Utilization Project (HCUP) sponsored by the AHRQ.14 The publicly available California SID includes hospital discharge abstract data (such as demographic characteristics, clinical information, resource use, and payers) from short-term, nonfederal, general, and specialty hospitals. California was selected for the analysis because it has a large and ethnically/racially diverse population15 and provides in its SID an encrypted unique patient identifier (UPI) that is used to link hospital readmissions to a distinct patient. The California Office of Statewide Health Planning and Development (OSHPD)16 routinely monitors the quality of hospital inpatient data and corrects or edits the data if necessary. They also ask reporting institutions to make corrections when their reports do not meet the error tolerance level (now 2%) that the state has established.
Our sample included patients 50 years or older with diabetes as the principal or a secondary diagnosis, identified using the AHRQ’s Clinical Classifications Software (CCS 45 [diabetes mellitus without complications] and CCS 50 [diabetes mellitus with complications]),17 all of whom were admitted to California hospitals at least once between April and September 2006. We defined the first admission as the index hospitalization and identified readmissions within 3 months of the index hospitalization using the encrypted UPI. As did Jencks et al,10 we counted no more than 1 readmission for each patient and focused on the characteristics of the first readmission (ie, scheduled or unscheduled) in the analysis. Similar to the approach by Jiang et al,3 we excluded patients with a missing UPI, patients with the same UPI but with an inconsistent age (differing by >1 year) or sex, and patients with a missing admission or discharge month.
The outcome variables of this study were scheduled and unscheduled readmissions. We used the SCHED variable in the California SID,18 which refers to whether an admission is scheduled at least 24 hours in advance. Therefore, we defined unscheduled readmissions as readmissions that were not scheduled at least 24 hours in advance and scheduled readmissions as those that were scheduled at least 24 hours in advance. The SCHED variable is derived from the type of admission in the California hospital inpatient data report collected by the California OSHPD, from which the California SID is developed for the HCUP. The California OSHPD16 provides detailed guidelines for each variable in the report, including the type of admission, and conducts multistep validation checks and edits.
We examined patient sociodemographic and clinical characteristics that were potentially associated with readmission variables, representing the 3 major categories of the Behavioral Model of Health Services Use.19 This model posits that potential and realized health services use is determined by interactions among predisposing, enabling, and need characteristics of individuals and the healthcare systems in the communities where they reside. Predisposing factors are individual characteristics existing before illness that determine the likelihood of seeking care and are measured by age, sex, and race/ethnicity. Enabling factors provide the means to use health services and include individual-level and community-level factors measured by primary payer type, resident location, and neighborhood median income. The California SID does not provide the patient’s exact income level, so we used as a proxy variable the median household income for the patient’s zip code in 2006. Perceived health status and evaluated health status determine the need for health services measured by comorbidities, prior hospitalization history, and variables related to index hospitalizations. The severity of conditions was measured by the number of chronic conditions in the 18 body systems using the Chronic Condition Indicator developed by the AHRQ.20 We counted only 1 chronic condition per body system, as did Wolff et al,21 although some patients might have had more than 1.
For data analysis, we first computed unadjusted rates of scheduled and unscheduled readmissions within 3 months of the index hospitalization by patient demographic, socioeconomic, and clinical characteristics. Second, logistic regression models were developed to examine the predictors of scheduled and unscheduled hospital readmissions. The reference group for odds ratios was patients with no readmissions in the 0 to 3 months following their index hospitalizations. Third, we compared common clinical conditions between the scheduled and unscheduled readmission groups using the CCS for the International Classification of Diseases, Ninth Revision, Clinical
Modification developed by the AHRQ.17 The CCS is a uniform and standardized coding system for aggregating more than 14,000 diagnosis codes into smaller, mutually exclusive, clinically meaningful categories. We ran a single-level CCS algorithm against our data, which categorized all the diagnoses into about 300 clinical conditions, among which we selected and compared the top 15 conditions for the scheduled and unscheduled readmission groups.
PDF is available on the last page.