Quality Assurance & Quality Control

What amendment is responsible for QA and QC laws

CLIA'88

QC refers mainly to what phase

analytical

Quality Assurance is a continuous improvement process that main 2 purposes is to decrease what?

decrease active errors (obvious). decrease latent errors

Latent errors are errors of what type

organization and design of the lab, overall system errors

quality assurance refers to which phases of the lab

pre-analytical, analytical , post-analytical

What is important in both QA and QC

documentation

QC documents

precision within a run or batch

QC detects

variations in precision (analytic errors)

QC prevents

reporting or erroneous patient values (ensures accuracy)

QC helps with

predicting the future and correcting it in advance

Normal distribution

Gausian (Bell) Curve. The distribution of values of the normal population. What reference range is based on.

Frequency

number of samples with a particular value

average/mean

add all the samples and divide by number of samples

median

middle value

mode

occurs most frequent

Range

interval with upper and lower limit.

Reference range/Normal range

range that the normal population falls within+/_2SD.

QC range

range that QC falls within +/_2SD.

Standard deviation

How much the numbers deviate from the mean

confidence limits

range +/-2SD for reference or QC

QC is in

within confidence limits

QC is out

not in confidence limits.

wider bell curve

greater SD, deviates more, less precise

Skinner bell curve

decrease SD, deviates less

When does a new QC range need to be established

new lot #, change methodology or instrument

How many QC level for a particular test and what are they?

at least 3=low, normal, high

How often is QC run

at least run 2 levels per shift (8hrs.) (CLIA'88) (mostly for chemistry)

How often is QC evaluated

every time you run QC done by MLT, MT, MLS. weekly and monthly done by MT, MLS.

What is the minimum number of data points to determine QC range

20

Who determines QC range

MLT, MT, MLS

Standard deviation measures the degree of...?

imprecision

+/-1SD includes what % of values

68%

+/-2SD includes what % of values

95%

+/-3SD includes what % of values

99.7%

How many days do you run QC serum alone

20-30 days

Levey-Jennings chart has how many charts?

One chart for each level of QC sera for each test (3 at least)

Random errors

no trend or means of predicting

Random errors examples

pipetting inconsistencies, improper mixing of sample, voltage or temperature fluctuations

What to do in case of random errors

re-assay (run again)- results will usually fall within limits

How often will QC be out from inherent variability

5% (document)

Systematic Error will be seen as

a trend (drifting) or shift (abrupt change) in the data

Systematic error affects

accuracy-values consistently too high or too low.

Examples of Systematic errors

poor calibration, reagent deterioration, bad reagent lot #

Westgard Rules applies to?

when 2 levels of control are used

Runs

any time you run QC (may run 2 levels or more within a run)

Across runs

comparing more than one day

Within run

values within same run. compare multiple levels

across levels

comparing different levels within the same run

Which Westgard Rules are Random errors

1-2s, 1-3s, R-4s

Which Westgard Rules are Systematic errors

2-2s, 4-1s,10-x

Which Westgard Rules get Warning

1-2s (possibly 10-x depending on the lab)

Which Westgard Rules get Reject

1-3s, 2-2s, R-4s, 4-1s, (possibly 10-x depending on the lab)

1-2s

one control value exceeds the +/- 2SD (other levels are ok).=random error, warning

1-3s

one control value exceeds the +/-3SD (other levels are ok). = random error, reject

2-2s

2 consecutive values exceed the same + or - 2SD 1)across runs same QC level outside the same 2 SD. OR 2) Within run- each QC level is outside same +or - 2 SD (one each).=systematic error, reject.

R-4s

"Across the range" 1)range between levels within a run exceeds 4 SD (one level +2SD and the other -2SD). Only to be used within a run, NOT across runs. =random error, reject

4-1s

1)four consecutive values exceed the same + or -1SD. OR 2) Same side exceeds + or - 1 SD across control levels 2 of each for a total of 4). =systematic error, reject

10-x

ten consecutive values on the same side of the mean. 1)same control OR 2)across control levels (5 each). = systematic error, warning or reject

Delta check

result for patient today compared to result for patient's past (within certain time frame) is very different. Instrument will notify you with this symbol. Happens everyday.

Proficiency testing is done by

tech with testing reagents sent by CAP

Evaluation of new methods includes

[precision studies, linearity studies =done by tech] [specificity and sensitivity studies=done by manufacturer]

Accuracy studies done by the manufacturer include

interference studies and recovery studies (add known amount of pure substance to serum pool and calculate how much is detected by a particular method)

Who does comparison of method, new vs. old studies

tech

coefficient of variation (CV)

SD expressed as a percentage. Allows comparison of 2 or more methods/instruments that have dissimilar SD ranges. CV = (SD/mean)*100. More reliable when comparing precision at different concentration levels (or units).