Rates, Predictors and Variability of Interhospital Transfers: A National Evaluation

Abstract

Importance: Interhospital transfer (IHT) remains a largely unstudied process of care.

Objective: To determine the nationwide frequency of, patient and hospital-level predictors of, and hospital variability in IHT.

Design: Cross-sectional study.

Setting: Centers for Medicare and Medicaid 2013 100% Master Beneficiary Summary and Inpatient claims files merged with 2013 American Hospital Association data.

Patients: Beneficiaries ≥65 years and older enrolled in Medicare A and B, with an acute care hospitalization claim in 2013.

Exposures: Patient and hospital characteristics of transferred and nontransferred patients.

Measurements: Frequency of interhospital transfers (IHT); adjusted odds of transfer of each patient and each hospital characteristic; and variability in hospital transfer rates.

Results: Of 6.6 million eligible beneficiaries with an acute care hospitalization, 101,507 (1.5%) underwent IHT. Selected characteristics associated with greater adjusted odds of transfer included: patient age 74-85 years (odds ratio [OR], 2.38 compared with 65-74 years; 95% confidence intervals [CI], 2.33-2.43); nonblack race (OR, 1.17; 95% CI, 1.13-1.20); higher comorbidity (OR, 1.37; 95% CI, 1.36-1.37); lower diagnosis-related group-weight (OR, 2.02; 95% CI, 1.95-2.09); fewer recent hospitalizations (OR, 1.87; 95% CI, 1.79-1.95); and hospitalization in the Northeast (OR, 1.40; 95% CI, 1.27-1.55). Higher case mix index of the hospital was associated with a lower adjusted odds of transfer (OR, 0.36; 95% CI, 0.30-0.45). Variability in hospital transfer rates remained significant after adjustment for patient and hospital characteristics (variance 0.28, P = 0.01).

Conclusions: In this nationally representative evaluation, we found that a sizable number of patients undergo IHT. We identified both expected and unexpected patient and hospital-level predictors of IHT, as well as unexplained variability in hospital transfer rates, suggesting lack of standardization of this complex care transition. Our study highlights further investigative avenues to help guide best practices in IHT. Journal of Hospital Medicine 2017;12:435-442.

FIG. 1.. Cohort selection.^a

^aCenters for Medicare and Medicaid Services 2013 100% Master Beneficiary Summary and inpatient claims files.

FIG. 2.. Distribution of transfer rates across hospitals.

NOTE: All models are centered at the median transfer rate of 1.79%. Shaded boxes encompass transfer rates from the 25th percentile (Q3) for each model. Whiskers encompass transfer rates from the 2.5th percentile to the 97.5th percentile. For example, the null model demonstrates that half of all hospitals have transfer rates between Q1 = 0.83% and Q3 = 3.80%; after equalizing the patient characteristics, the interquartile ranges expands to Q1 = 0.78% to Q3 = 4.06%; however, measured hospital characteristics explain most of this variability, reducing the interquartile range to Q1 = 1.26% to Q3 = 2.54%.