|
On 17th October,
at the Thames Valley Cytology Society meeting held at Chase Farm
Hospital, Linda Grosvenor from the Birmingham Cytology Training
School gave an interesting talk on how not to misuse statistical
analysis in relation to cytology screening using the previous speaker's
methods.
Following on
from Dr. Mohammed Mohammed's lecture about the misuse of
statistics Linda had performed a statistical analysis of her screeners'
work. She showed a graph of high grade reporting profile data for
screeners from her laboratory for 2001/02.
She had looked
at why variations in screener sensitivities occur, ie variation
outside normal limits. One screener had a much higher pick-up rate
compared to the others. When this was investigated it was seen that
this screener was screening the pre-screened abnormals from the
outsourced workload. Another screener had 3 high grades from one
tray in one day so this also gave a high pick-up rate.Low grade
reporting profiles were also looked at. Borderlines and milds were
being sent to the pathologist for a time without checking and during
this period a higher than normal rate of low grades was reported.
During the time when the reporting rates rose it was noticed that
a new pathologist had been appointed.
After looking
at the figures and reasons that might have affected reporting rates,
every abnormal referral was seen by a checker before the pathologist.
Multihead microscope sessions with the pathologist where there was
disagreement were set up. Smears were reported on concensus. This
resulted in a drop of borderlines reported by their laboratory.The
percentage of low grades from two screeners were just outside the
guidelines. When these cases were looked at one screener had had
time off sick following a car accident and the other had been on
long-term sick leave.
Finally, upheavals
in private lives and at work can cause lapses in concentration,
which could account for misses of small amounts of abnormal cells,
very small abnormal cells or very pale dyskaryosis. As a result
of this Linda recommended keeping a diary of events within the laboratory
which might help to explain a variation in performance of reporting
abnormals and to use the reporting profiles as a tool to improve
quality.
|