Transcript An Amateur's View of MQSA

An Amateur's View of MQSA
Victor E. Anderson, C.H.P.
Radiologic Health Branch
California Department of Health Services
The Circle Problem
• Inspectors are required by eye to determine
if a part is “out of round”, e.g., not a circle.
• Not a hard process, unless you define
“roundness.”
Mammography
• Similar problem:
– Is it a cancer or ?
• Some Factors
–
–
–
–
Quality of image
Skill of interpreter
Size of tumor mass
Physical/Psychological
Problem
• Screening test.
– Not expected to detect every tumor
• When does the interpreter and system fail?
– Easy answer: does it meet MQSA Standards?
• What about false negatives and positives?
Facility Failure
• How bad is bad?
• Or “How many cancers went undetected
that should have been seen?”
• Two interrelated areas:
– Physical factors (“Physics”)
– Ability of the Interpreter
Review Actions
• How far back in time does the problem go?
– Previous inspections
– Clinical Image Reviews
• MQSA records
– Image Quality
– Interpreter skill (Additional Mammography
Review)
How Many?
• Typically 30 additional cases.
• Breast Cancer Rate
– Various with age
– About 0.008 on the average per screening
• Out of 1,000 cases, eight may have
detectable cancer!
Rates and Sampling
•
•
•
•
Mammography facilities see lots of patients.
As much as 20 per day per machine.
About 100 per week per machine.
A ten machine facility could find about
eight cases per week.
• What is bad?
False Negative Rates
False Negative Rate
Reference
Total Population
As a % of
Total
Population
As a % of the Total
Number of Women
Found to have Cancer
1
27,305
0.1831%
15.4%
2
752,081
0.025%
5.1%
3
389,533
0.081%
19.1%
0.04%
10%
Weighted Averages
False Negatives
• Indications are that
– For every eight
– About two will be missed.
• Given a poorly performing facility
– How many films to review?
– How bad is bad?
How Far Back?
• Ideally no further than last inspection.
• Indicators
–
–
–
–
–
Daily checks
Physics reports
Referrals
CIR
AMR
One Testing Solution
• Make a time estimate.
• Using average rates, determine number of
cases seen.
• Total number of films.
• How big a sample?
Sample Size
• Mil spec tables or AQL
– Period of potential sub standard performance is
10 weeks.
– Four machine facility.
– 4,000 patients.
– AQL table indicates 200 samples with an
acceptance level of one false negative.
– And a rejection level of two false negatives.
Issues
• Cost of sample size.
• What is at stake?
• Rational assurance that proper screening
occurred.
• Follow up
Conclusions
• MQSA provides good QA/QC
• Need Sampling scheme to follow up.
• Provide rational basis for extent of a
problem with respect to time.
• How many patients to notify?
• Aid in corrective actions.
References
• False-negative breast screening assessment: what
lessons can we learn?, Burrell HC, Evans AJ, Wilson
AR, Pinder SE., Clin Radiol 2001 May;56(5):385-8
• Comparison of full-field digital mammography with
screen-film mammography for cancer detection: results
of 4,945 paired examinations., Lewin JM, Hendrick
RE, D'Orsi CJ, Isaacs PK, Moss LJ, Karellas A, Sisney
GA, Kuni CC, Cutter GR., Radiology 2001
Mar;218(3):873-80
• The evaluation of false negative mammography from
malignant and benign breast lesions., Wang J, Shih TT,
Hsu JC, Li YW., Clin Imaging 2000 Mar-Apr;24(2):96103