#4: Quality Control

Proposed by

lemoene

Proposal type

Functionality

Assigned to release

• Bika LIMS 2.0

State

being-discussed

Definitions

Many of the concepts here are demonstrated
in the QC part of the Bika Fireassay model

Standards

A Standard is a sample of which the chemical composition is known and kept in the system

In the terminology used here, such a Standard can be used for QC purposes as
        Control samples
        Calibration samples

Sometimes the lab will make up it's own Standard in a container and repeatedly use some of it for analyses. Other times it will be ordered from a Standard Supplier and be delivered in a batch of say 24 ampoules, and 1 ampoule is used per analyses

The above explanation suggests that Standards AND Standard Batches in the system be treated as a Sample on which successive Analysis Requests are performed

Standards have expiry dates where after the system does not allow analyses on them

Standard Suppliers are maintained in the system per standard Bika 'organisation and contact' model with the addition of a tab listing all the Standards supplied to the lab. From this listing, links lead to the labs performance stats and graphs on that Standard

All Standards in the system can be listed off a main menu item too, leading similarly to their performance graphs

Duplicates (Checks)

Worksheets often include part of the same sample twice and the system compares their results - the exact same result must be returned for the 2 analyses in the perfect lab. Should they differ more that a specified %, the labmanager is alerted and a 'reject or retest' workflow entered

Duplicates don't have to be Standards. Most or the time labs uses any arbitrary sample on the worksheet or instrument tray

Blanks

An empty sample container is put through the analysis proses - result must be 0. Above specified % error, alarm is raised and reject or retest' workflow entered

Calibrations

Some standards are run through the analytical instrument before, and sometimes also after, the routine analyses on a tray/worksheet with the purpose of calibrating the instrument

In this design, calibration results are captured but not used as of yet

Not in the scope: In phase 2, these calibrations will be used to plot calibration graphs and if there was sufficient drift between the 1st and last analyses on the worksheet, correction factors calculated and applied

Specifications

Out of range limits are maintained to measure QC results against and labmanagers alerted when exceeded

Standards have these values stored on their individual records

Fault tolerances for Blanks and Duplicates are set per sample type and analyses in the LIMS set-up

Motivation

Required by Bika 2 sponsor

Proposal

To refine the design for Bika's QC presented here

Deliverables

Development and Implementation arising from the functional description in this document

Some of the concepts here are demonstrated
in the QC part of the Bika Fireassay model

QC Workflow

The laboratory includes a duplicates and control analyses on each worksheet. Should the results returned for it fall outside the allowable tolerance specified

• the result is displayed against red background

• all analyses on the same worksheet are flagged for rejection or re testing and the lab manager alerted

Reporting

QC results are looked up from a special page 'Control Analyses'

When clicking through to a control analyses, the user sees a page with the analysis' data summarised and graphed. Out of range values are displayed against shaded background

The control analysis result is also statistically compared with previous results for the Standard presented on Performance and Distribution graphs

The default period for statistical analysis is the previous 6 months' data. The user may specify any valid date range on the page and reload it - handy for looking at older analyses

These graphs and tables are also available, per sample type and analysis, from a QC reports menu

The following graphs are available

Trend - showing performance over time. true value, min & max measured and a trend plotted through the data. An example for an standard with true value of 3.00 g/t:

Distribution - shows true value, the lab's mean and normal distribution around it

An example for a result of 11.20 %/vol, lab mean of 11.15 and true value of 11.00. The standard deviation is also shown

Representations of graphs here are 'hand drawn' 
To see Bika's production quality graphs
please visit the Bika Interlab page

Duplicates - Relative % error

The % error on duplicates in the results range tested. An example