Assessment of performance
Deviation Index (DI) = (Result - Mean) / Standard Deviation (SD)
In the vast majority of cases a DI limit of + 2.0 is set as the limit of acceptable performance within groups for each parameter. For haemoglobin, normal INR and APTT ratio, D-dimer, low white cell (<1.0x10 9 /l) and platelet counts (<10x10 9 /l) where results are very tight, we have a limit of + 3.0, which seems more appropriate. For all other parameters the DI limits are 2.0. On no occasion would the limit exceed 3.0.
Hazardous performance would be designated as any DI exceeding + 4.0 (5.0 for normal PT, normal PTT/ratio & Haemoglobin) where this would affect clinical interpretation or therapeutic intervention. Where no underlying reason for hazardous performance can be identified (obvious reasons could be incorrect reconstitution or incorrect sample used for specific test in multiple sample dispatches) the participant is contacted ahead of a repeat sample or samples being dispatched. This enables investigation into hazardousperformance to start as soon as possible.
In all other cases, participants are issued a repeat sample or samples either during the results return period when their return is out with consensus or at the time of reporting when a result has been truncated.
Unsatisfactory Performance – Action Criteria
Participants are contacted by mail or email following two unsatisfactory performances in a given parameter. This could be within the same trial; an example of this is Factor VIII where two samples are tested at the same time. Alternatively as with for example, fibrinogen, unsatisfactory performance in two individual trials out of four, i.e. intervention before a fifth trial and possible third unsatisfactory performance. Non return without explanation is assessed as an unsatisfactory performance.
The Scheme, where possible, prompts investigation into likely problem areas and the centre is asked to report any corrective action(s) taken. Advice is given when a participant contacts the Scheme for assistance following an unsatisfactory performance. A participant can request samples with consensus mean values to verify any correctiveaction taken and report results to the Scheme to review.
We would like to emphasize that participants are notified of unsatisfactory performancesas soon as possible so that any problem may be swiftly addressed. In this way, we have found persistent unsatisfactory performance can be avoided.