Lab management, Statistics Flashcards
PPM
Must be performed by a provider. May have competency assessments but proficiency testing is not required.
DOT categories
A - Pathogens, high biological risk
B - Routine biological specimens, some risk
HIPAA exceptions
Can send information to insurance groups and other providers without express consent.
Inspections: JCO, CAP, COLA
JCO: q3yr hospital, q2yr labs.
CAP: q2yr, off-year self-inspection. Phase 2 worse than 1.
COLA: q2yr, voluntary. For smaller labs
Medical director requirements
Mod/High complex: Doctorate with 1+yr experience, or PhD with boarding.
Waived testing: None required.
Director can delegate roles. May cover max of 5 labs.
Competency testing
Annual (or q6mo if new hire)
Material retention
Most CP things - 2yrs
Cyto slides - 5yrs
AP materials - 10yrs
Forensics - Forever
Suggested cytogenetics and flow 10+ yrs.
Productivity
Tests (or product) per tech (or labor).
Higher with automated testing and SMALLER labs?
Lean
Toyota method
Identify and remove wasteful steps.
Use physical maps, spaghetti diagrams.
Three day stay
Outpatient lab services occurring up to 3 days prior to admission are billed to the admission
AHRQ
Agency that examines effectiveness of new treatments
Analytical sensitivity
Analytical specificity
A. sens: Ability to detect analyte at low concentrations
A. spec: Ability to detect analyte with interferants
Calibration verification vs assay calibration
Calibration verification: Test known concentrates throughout the RR.
Assay calibration: Adjust instrument output to match known concentrates.
Validating FDA-approved assays
Need accuracy, precision, RR. Confirmation of manufacturer RR (20-40 cohort OK)
Frequency of QC checking
q24hrs (exception: blood gas q8hr or per run)
CLIA certificates
Waiver - Most common. For labs that only do waived tests.
Registration - Allows for complex testing to proceed until inspection
FMLA
12 weeks
Unpaid, job-secure
no gender/sex difference
ISO Standard 15189:2007
International standard for medical lab quality and competence.
Voluntary. Has no legal authority.
Levey Jennings plot
Bland-altman plot
Levey-Jennings: Graphs Mean+/- SD over runs or time.
Bland-Altman: Compares % difference to average values
PT testing: Unsuccessful, unsatisfactoryu
Unsatisfactory: Failed to attain adequate result on an assay
Unsuccessful: Unsatisfactory result on 2+ consecutive attempts.
PT testing: Send-outs
NEVER DO IT, just specify that you would normally.
Root cause analysis
Factor mapping Tree diagrams Ishikawa fishbone Pareto analysis Fishbone analysis
Seeks to identify many factors in a blameless fashion. Results in an actionabl solution
Six Sigma
Aims to minimize product defects. Uses “Black belts” and “green belts”…
MBA rubbish.
Sec 1877 of SSA (42 USC 1395)
Stark low. No referrals for financial benefit (including to family members, etc).
Westgard rules
1(2S) - Warning 1(3S) 2(2S) 4(1S) R(4S) 10(x)
SD containers
1SD: 68%
2SD: 95%
3SD: 99%
Tests to compare datasets
If parametric, use student’s T-test.
If non-parametric, use Mann-Whitney.
How is efficiency calculated?
TP + TN / All outcomes
Reagent lease
Pay for reagents and depreciation of instrument. Manufacturer retains ownership of instrument.
Semi-variable costs
“Step variable”, increases with testing volume but not linearly. eg. Supervisors.
Turnover rate
departures / total staff in a given time period.
Address these with exit interviews.
DRGs
Only reimburse HOSPITALS for inpatient services (eg, Medicare part A).
Adjustments exist for geographic location, setting, hospital type, and patient demography?
SSTs
Separates cells and serum. Not good for drug or immunoassay testing (adsorption)
Effect of standing, tourniquets
Both: Third-spacing (hemoconcentration)
Tourniquets: Lactic acidosis, maybe hyperkalemia (fist clenching)
Optical interferants
Bilirubin: 340-500 nm
Heme: 412 nm (“Soret band”)
Lipids: Scatters, doesn’t absorb.
Likelihood ratio
Sens / 1 - spec
Refers to the odds of a positive result in a disease patient relative to non-diseased.
CV%
SD / Mean
How is post-test probability counted?
Multiply the pre-test ODDS RATIO by the LR. Then convert back to probability.
Eg: 25% population diseased, test with 5x LR
25%»_space; 33% pre-test odds ratio.
166% post-test odds ratio
166/266 or 62.4% post-test probability