Observed and expected outcomes in transfer and nontransfer patients with a hip fracture.

OBJECTIVES

Hospitals and providers that accept transfer patients risk lower ratings on publically reported quality measures that are inadequately adjusted for infirmity and complexity. We compared the outcomes of transferred patients and nontransferred patients after treatment of a hip fracture and sought to determine if expected outcomes based on an expansion of All Patient Refined-Diagnosis Related Groups (APR-DRGs) norms are accurately adjusted for transfer status.

DESIGN

Retrospective cohort study.

SETTING

Tertiary care referral center.

PATIENTS

Four hundred six consecutive patients 65 years and older who received operative treatment of an acute hip fracture.

INTERVENTION

Patients who were transferred from another acute care hospital or a skilled nursing facility before treatment were classified as transfer patients (n = 123), and all other patients were nontransfer patients (n = 283).

MAIN OUTCOME

Groups were compared with respect to in-hospital mortality, length of stay (LOS), excess days over the geometric mean length of stay (GMLOS), and readmission rate as well as expected length of stay (Exp LOS) and expected mortality (Exp Mort) based on APR-DRG norms and additional adjustment for transfer status.

RESULTS

Transfer patients had significantly greater LOS (10.2 vs 9.6 days; P < 0.05), Exp LOS (9.7 vs 7.7 days; P < 0.001), Exp Mort (0.07 vs 0.03; P = 0.004), and excess days over the GMLOS (4.1 vs 3.3 days; P = 0.025) than nontransfer patients, near-significant greater in-hospital mortality (9.8 vs 4.9%; P = 0.069), and similar readmission rates. The differences in LOS and Exp LOS were nonsignificant in both transfer (P = 0.49) and nontransfer patients (P = 0.10).

CONCLUSIONS

Patients 65 years and older transferred to a tertiary care facility for treatment of an acute hip fracture have worse outcome than nontransfer patients. Unadjusted data such as in-hospital mortality may be misleading, but risk adjustment using the APR-DRG methodology and additional correction for transfer status may provide meaningful benchmarks.