M. Farrington et al., 15 PERCENT OF MICROBIOLOGY REPORTS ARE WRONG - FURTHER EXPERIENCE WITH AN INTERNAL QUALITY ASSESSMENT AND AUDIT SCHEME, The Journal of hospital infection, 30, 1995, pp. 364-371
Infection control teams critically depend on the quality of the diagno
stic microbiology laboratories with which they are associated. Interna
l quality assessment (IQA) by specimen resubmission measures the consi
stency of laboratory performance and can be adapted for medical audit,
but few laboratories appear to include these techniques as part of th
eir quality control procedures. We established an IQA scheme in our ba
cteriology laboratory in May 1989, and the mean discrepancy rate for t
he first two years was nearly 15%. During the next two years covered b
y this report, the scheme was extended to include audit of the consist
ency of medical microbiologists' comments on reports, and a standardiz
ed scoring scheme was introduced. Results from the scheme are collated
and analysed thrice yearly, and laboratory techniques altered and ext
ra staff training planned to reduce discrepancies. Four thousand four
hundred and fourteen specimens (1.4% workload) were resubmitted, and 3
25 (7.4%) gave discrepant results. During the two years, the overall d
iscrepancy rate fell consistently from 9.4 to 3.2%. Fifty-three discre
pancies (1.2%) were in microscopy, and 243 (5.5%) were between culture
results. Substantial decreases in technical discrepancies followed tr
aining initiatives in sputum bacteriology, and training in anaerobic r
ecognition and improved illumination in the anaerobic cabinet. Problem
s of consistent recognition of staphylococci and of urine isolates in
mixed culture, and of faecal microscopy have proved difficult to eradi
cate. There was a 20.3% discrepancy rate between medical comments, but
only 0.4% would have been likely to result in different patient outco
mes. The cost of running the scheme was pound 4474 per annum, equating
to a surcharge of pound 0.058 to each of our routine diagnostic speci
mens. Discrepancies were therefore reduced by about two-thirds at a co
st of pound 1.88 per discrepancy prevented; we believe this offers goo
d value for money.