176 ◾ John T. Hansmann
data elements originate. Included in the data denitions is the timeframe (annual, year to date
[YTD], quarterly) of the data to provide a consistent comparison.
Comparison peer groups are typically decided by factors such as size of operation (e.g., bed
size, number of patient days, number of procedures), patient type/mix, community populations,
and/or geographic distributions. In many cases the data need to be normalized for wage dier-
ences (wage index) or patient mix (using case mix index [CMI]) or severity adjusted to include
patient complexity in the comparisons. Additional operational characteristic details such as unit
design (circle, square, X, H, I), unit storage (centralized vs. bedside), and available technology
(automatic dispensing cabinet) rene the comparison even more. e point of the comparison is
to select other hospitals that are as similar in operations as possible to which you can compare your
hospital/department. e more similar the operations, the more any variances may be attributable
to actionable dierences.
Step 2: Variance Analysis: Identify Performance Variance
and Potential Improvement Opportunities
e second step for a benchmarking/comparison analysis initiative is to analyze the comparison
data and determine if a potential improvement or savings opportunity exists. is variance analysis
calculates the dierence between how one organization performs compared to another. e vari-
ance or potential savings opportunity is calculated by taking the dierence in operating perfor-
mance (e.g., productive hours per unit of service) between the two organizations and multiplying
by the volume of the focused organization. is eectively calculates the focused organization’s
cost of operations using the comparison organization’s cost structure and the focused organiza-
tion’s volume. In others words, it is measuring how much can be saved if the focused organization
could operate at the other organization’s cost structure.
Most comparisons use either quartile or percentile grouping methodology. Quartile and per-
centile groupings organize the comparison data in rank order from best to worst to identify a
quartile or percentile for each hospital. In a quartile ranking system, the best or top quartile is the
1
st
quartile, the 2
nd
quartile is the next best, and so on. In a percentile ranking, the data are rank
ordered from best to worst, with the percentile rank identifying the percent of hospitals that are
worse on the list. For example, if a hospital is at the 75
th
percentile, its performance is better than
75% of the hospitals on the list.
Normally a high-level target is established to perform at the 50
th
percentile, 67
th
percentile, or
75
th
percentile levels. is is equivalent to the best and second-best quartiles. e methodology
ultimately identies a specic department or hospital as the target hospital. e target hospital is
used to calculate the variance or potential savings opportunity. It is important that the comparison
target is an actual hospital that is really performing at the identied level versus an average or other
mathematical variance that is unattainable.
Many options exist to estimate an improvement goal for the department. e ultimate goal is
to achieve the same or better performance as the target department, but in some cases that may not
be realistic in the short term. Practice and experience show that in a short time period (dened as
one year), it is rarely feasible to make changes greater than 20% of the total expenditure. Another
concept that has been used to set the goal is to improve one spot in the rankings. is is especially
useful in the case where the ultimate target may be a real stretch goal for the department. A third
concept to set a goal creates a potential savings range, and is sometimes described using terms such
as conservative and aggressive targets. However it is done, the point is to identify a hospital that is