- 1 Suppression Guidelines for File Transfers
- 2 Survival (or Failed to Survive) Curves
- 3 Minimums and Medians
Suppression Guidelines for File Transfers
Please double check your output counts. The purpose of checking your counts is to ensure that no protected health information (PHI – for more on PHI see HHS HIPAA website) is transferred off the system and that the material being moved has no cells with counts of beneficiaries or events less than 11 or rates/proportions/percentages that be used to derive a count of beneficiaries or events less than 11. The full CMS data suppression policy can be found here. Note, cells with a value of "0" are acceptable. Pay attention to your derived suppression!!
Below are a couple of examples of derived counts of beneficiaries or events:
Examples of Derived Counts
Example 1 – Count Derived from Total
Consider the following table:
|Diagnosis||Count of beneficiaries|
In the table the count for diagnosis C has been suppressed. But we know that the counts for A, B, and C must add up to 60 (the total). It is then possible to determine that count for diagnosis C is 5.
Possible solutions to this include not releasing the total, suppressing the next smallest count, or grouping the diagnoses together.
Example 2 – Two-way Table of beneficiaries
In this example, the issue is similar to example 1. The suppressed count can again be derived from the other information in the table. Unlike example 1, it is not sufficient to simply suppress one other cell.
In the table below, the user has tried to fix the issue by suppressing the count of women with diagnosis C, but it is still possible to derive the counts.
The best solution is to suppress one other count from the column for men.
In general, with a two-way table, for a given suppressed cell, you must suppress one other cell in the row, and one other cell in the same column. But other solutions include grouping the data differently, or not releasing the totals.
Example 3 – Counts Derived from Percentages of beneficiaries
In the example above, the user has correctly suppressed the counts, but it is still possible to derive them using the associated percentages. The percentages associated with any suppressed counts must also be suppressed.
Survival (or Failed to Survive) Curves
- If you are requesting release of a survival curve (only the curve itself) from the DAC servers, we require >11 at the beginning, end, and to have failed overall.
- If any other data is accompanying the curve, you may not include the number remaining in your graph or any tables. If you are planning to release the starting number of beneficiaries, it must be substantially large enough that the end result is 11 or greater. In that case, you will not be allowed to release the starting number of beneficiaries.
Minimums and Medians
- Per ResDAC, we must suppress any statistic that is less than 11, which includes minimums and medians. As in all cases, this is specific to beneficiaries AND events.