Instructions for using the matrix

April 2000 Table of Contents

The practice evaluation matrix used by Exempla Medical Group is available here for downloading. Although the group’s personal data has been altered in this version, the matrix itself is intact, allowing users to view the formulas and understand how the matrix works.

Notes:

Familiarity and competence with Microsoft Excel is necessary to use the matrix effectively.
The matrix was designed specifically for Exempla Medical Group and may not be easily transferable to other physician groups without some significant modifications.
Benchmark data used in the original evaluation matrix was taken from the Medical Group Management Association (MGMA) “Physician Compensation and Production Survey” from 1997 using 1996 data. More current data is available from the MGMA.

Functionally there are three components to the matrix:

1. Input.

Practices applying the matrix to their own situations would need to input into the following items:

Practice names and/or codes: Actual names of the physician practices, or codes that represent the physician practices, must be input to identify each line.
FTE (full-time equivalent) providers by specialty: These values are critical to calculating a weighted benchmark if the practice is a multispecialty primary care practice. The existing matrix has two sections for input, for use with different variables. This would not be necessary in all cases.
Data associated with the evaluation variables (e.g. encounters, revenue, expenses).
Weighting values: Based on the variables selected, weights are assigned to reflect the relative value or importance of each variable relative to the others. The weights we chose are shown in the sample matrix.

2. Raw score calculation.

Raw scores for each variable are calculated based on the actual value’s deviation from the benchmark or mean for that variable. To achieve a balanced distribution of raw scores (i.e., to prevent all practices from falling into one category), it may be necessary to adjust the amount of deviation from the benchmark that is necessary to trigger a higher or lower score. For example, a raw score of 5 may be assigned to values that fall within 25 percent of the mean or total values that fall within 50 percent of the mean, etc.

We used raw scores of 0, 5 and 10. These values were selected arbitrarily to achieve some level of differentiation in addition to the weighting. Any set of incremental values could be used.

3. Weighted score calculation.

Weighted scores are simply calculated by multiplying the raw score by the weighting value for each variable. The total score is the sum of all weighted scores.

For the original use of this matrix, it was necessary to calculate a total possible score and evaluate each practice with respect to how they scored relative to the total possible. This was due to the fact that many practices were not evaluated on some criteria, including marketing and physical plant.

Ranking was based on the percentage of the total possible score. Rankings were performed both overall and by geographic location. We also included rankings from the prior year to attempt to identify trends in performance relative to the other practices in the group.

Finally, it should be noted that any quantitative variables may be used in this matrix. In actuality, there were many iterations of this matrix, some having only four variables.