Objectives. To evaluate the hospital multistay rate to determine if it has
the attributes necessary for a performance indicator that can be applied to
administrative databases.
Data Sources/Study Setting. The fiscal year 1994 Veterans Affairs Patient T
reatment File (PTF), which contains discharge data on all VA inpatients.
Study Design. Using a retrospective study design, we assessed cross-hospita
l variation in (a) the multistay rate and (b) the standardized multistay ra
tio. A hospital's multistay rate is the observed average number of hospital
izations for patients with one or more hospital stays. A hospital's standar
dized multistay ratio is the ratio of the geometric mean of the observed nu
mber of hospitalizations per patient to the geometric mean of the expected
number of hospitalizations per patient, conditional on the types of patient
s admitted to that hospital.
Data Collection/Extraction Methods. Discharge data were extracted for the 1
35,434 VA patients who had one or more admissions in one of seven disease g
roups.
Principal Findings. We found that 17.3 percent (28,300) of the admissions i
n the seven disease categories were readmissions. The average number of sta
ys per person (multistay rate) for an average of seven months of follow-up
ranged from 1.15 to 1.4.5 across the disease categories. The maximum standa
rdized multistay ratio ranged from 1.12 to 1.39.
Conclusions. This study has shown that the hospital multistay rate offers s
ufficient ease of measurement, frequency, and variation to potentially serv
e as a performance indicator.