Use of failure-to-rescue (FTR) as an indicator of hospital quality has increased over the past decade, but recent authors have used different sets of complications and deaths to define this measure. This study examines the reliability and validity of different FTR measures currently in use.Research Design:
We studied 3 definitions: (1) “original” FTR (using all deaths); (2) FTR-N, a “nursing sensitive” definition that uses only specific complications and deaths; and (3) FTR-A [another restricted definition of FTR used by Agency for Healthcare Research and Quality (AHRQ) for analyzing Healthcare Cost and Utilization Project (HCUP) data]. Each FTR measure was applied to 403,679 general surgical patients across 1567 hospitals reported in 1999–2000 Medicare MEDPAR data.Results:
Although FTR used all deaths, FTR-N and FTR-A definitions omitted 49% and 42% of deaths, respectively. Reliability was better for FTR than FTR-A or FTR-N (ρ = 0.32 vs. 0.18 vs. 0.18, respectively).Validity:
Hospitals ranked by adjusted mortality were highly correlated with FTR (Kendall’s τ = 0.83) and less correlated with FTR-A (τ = 0.43) and FTR-N (τ = 0.41). Adjusting for patient characteristics, all FTR measures showed strong associations with bed-to-nurse ratio, nursing mix, teaching status, and hospital size; however, hospital “high technology” was not as well associated with FTR-N.Conclusions:
For general surgery, more limited definitions used by FTR-N and FTR-A omit over 40% of deaths, display less reliability, and may have more questionable validity than the original FTR measure. We encourage analysts to use the original FTR definition that uses all deaths when analyzing hospital quality of care.