Information on QA Percentages & the QA Algorithm

Below is a brief outline on the way the QA percentage algorithm works in epaPRO.

Setting the standard QA beneath 100% will have a varying effect depending on other QA percentages.

QA Rules

• Void type result - never QA

• Manual flag set against the result - always QA

• Component rate of 0 - never QA

• Otherwise it will follow the order below

There is a hierarchical order in which the QAs are checked:

1. Assessment Component

2. EPA

3. Provider

4. Standard

5. Specialism

6. Assessment Method (if enabled)
   
   

• If the Assessment Component is set to Never/0* then it overrides all the other QA’s - they will not be checked, even if they contain a percentage.

• If the Assessment Component % is set to 0, then it will prioritise this over the other general rules (Standard, Specialism, EPA, Provider, Method).

• If the Assessment Component % is set to a non-zero value, then the engine will look at all the general rules equally (standard, specialism, EPA, Provider, Method) and will include Assessment Component as an equal check.

• If any of the rules are set to 100% then it will force a QA, this is because as the algorithm checks each one to see if the target has been met and if it comes to one set to 100%, it will say the target has not been reached and that it needs QA.

• If the % is set to either a 0% or a 100%, the randomness mentioned below is ignored.

Finally - it is worth noting that if the assessment component QA is above 0, or it is blank and any of the other QA percentages are above 0, then it is also possible for there to be a random override that causes the QA to be carried out. If this happens, it can be seen under QA reason on the QA Audit report.

An example:

Assessment component is set to: Default/Blank
EPA is set to: 25%
Standard is set to: 50%

• This result will be QA'ed if the EPA’s QA rate is less than 25% across the previous 100 results they added for that standard (the 100 results need to be within the last 3 months).

• If the above passes, the result will still be QA'ed if less than 50% of the results for that standard have been QA'ed over the previous 100 results/3 months.

• The result will not be QA'ed if both criteria are met.

• QA may also be triggered by random chance - if any of the checks have 100% or 0% then randomisation will be disabled, even if one of the other rules are not 100% or 0%.

  For example:

  • Component - 50%

  • EPA - 25%

  • Provider - 0% ← Disables random

  • Standard - 15%

  • Specialism - 20%

  In the example above, the QA engine will verify the above percentages are met, but due to the provider having 0% it will disable the random chance for that result.

The binary options for QA setting against assessment components have been replaced with a QA minimum percentage, allowing for increased customisation.

The new QA minimum percentage works in the following manner:

• 0% will always prevent the element from being QA’d, with the other checks (such as provider/EPA etc.) not taken into consideration.

• An empty/blank value will ignore the element check and use the other checks (such as provider/EPA etc.) to determine whether the element should be QA’d.

• Any other value will work using the percentage & previous QA history to determine the QA status, as per functionality of the other existing rules.

Given the functionality detailed above, the existing QA settings have been converted into the following percentages, to help maintain existing functionality:

• “Always” becomes “100%”

• “Never” becomes “0%”

• “Default” becomes empty/blank