Before replacing its potassium method, a laboratory measures 40 patient samples on both methods and draws the difference plot shown. Its acceptance limit for mean bias is ±0.10 mmol/L. The correct conclusion is:

A constant bias that exceeds the limit: not acceptable

A constant bias within the limit: acceptable

Random scatter only: acceptable

A proportional bias above the limit: not acceptable

Compared with the current method, the new creatinine method in the comparison plot shown has:

A proportional systematic error

Linearity samples for an AST method give these results (expected / measured, U/L): 100 / 102, 300 / 296, 600 / 610, 900 / 872, 1200 / 1015. The allowed deviation is 10%. Up to what activity is the measuring range verified?

900 U/L; dilute samples above it

600 U/L; dilute samples above it

1015 U/L; the top measured result

Before using an unmodified commercial method for patient testing, a laboratory accredited to ISO 15189 must at least:

Verify the manufacturer's performance claims

Fully validate it as an in-house method

Set its own reference interval from 120 people

Rely on the package insert without testing

Method validation, verification and reference intervals

9 DOH Abu Dhabi (HAAD) lab licensing practice questions on Method validation, verification and reference intervals, part of General Laboratory, Quality and Safety. The first 9 are below; the first 5 with answers and explanations (tap Show answer after you have chosen), the rest in Learn with a free account.

Question 1 · Hard
Before a new potassium method goes live, the laboratory sets the hemolysis index above which potassium will not be reported. A patient serum pool is split into five aliquots. Four receive hemolysate to give free hemoglobin of 0.5, 1.0, 2.0 and 4.0 g/L (50 to 400 mg/dL); the fifth receives the same volume of saline. Mean potassium of triplicates (mmol/L): saline control 4.00; 0.5 g/L, 4.15; 1.0 g/L, 4.32; 2.0 g/L, 4.61; 4.0 g/L, 5.24. The laboratory accepts an interference bias of up to 5% of the control value. What is the highest free hemoglobin tested at which potassium can still be reported?
1. A0.5 g/L
2. B2.0 g/L
3. C4.0 g/L
4. D1.0 g/L
Show answerHide answer
Answer: A · 0.5 g/L
In a paired-difference study, every aliquot is compared with the saline control from the same pool, so the bias is the difference from 4.00 mmol/L. The biases are +0.15 mmol/L (3.8%) at 0.5 g/L, +0.32 (8.0%) at 1.0 g/L, +0.61 (15%) at 2.0 g/L and +1.24 (31%) at 4.0 g/L, against an allowed 5% of 4.00 = 0.20 mmol/L. Only 0.5 g/L passes, so potassium is not reported, or is reported only with a comment as policy states, when the index is higher. Comparing each aliquot with the one before it, instead of with the control, wrongly passes 1.0 g/L. Potassium rises because red cells hold roughly 25 times as much potassium as plasma.
- B: The bias is 0.61 mmol/L (15%), three times the allowed limit.
- C: The bias is 1.24 mmol/L (31%), so grossly hemolyzed results would be released.
- D: The bias is 0.32 mmol/L (8%); this cut-off comes from comparing each aliquot with the previous one.
Question 2 · Hard
Before replacing its potassium method, a laboratory measures 40 patient samples on both methods and draws the difference plot shown. Its acceptance limit for mean bias is ±0.10 mmol/L. The correct conclusion is:
Figure: MedExams
1. AA constant bias within the limit: acceptable
2. BRandom scatter only: acceptable
3. CA proportional bias above the limit: not acceptable
4. DA constant bias that exceeds the limit: not acceptable
Show answerHide answer
Answer: D · A constant bias that exceeds the limit: not acceptable
The differences (new minus current) scatter around a mean of +0.20 mmol/L, and they stay the same size from low to high potassium, with no slope. A bias that is the same size across the range is a constant systematic error. At +0.20 mmol/L it is twice the acceptance limit of 0.10 mmol/L, so the new method cannot replace the current one until the cause, such as a calibration offset or a blanking problem, is found and corrected and the comparison repeated. A proportional bias would show differences that grow as the concentration rises.
- A: The mean difference of +0.20 mmol/L is twice the ±0.10 mmol/L limit.
- B: Every difference is above zero, so there is a clear bias, not just scatter.
- C: Proportional bias would make the differences rise with concentration; here they are level.
Question 3 · Medium
Compared with the current method, the new creatinine method in the comparison plot shown has:
Figure: MedExams
1. ANonlinearity at the high end
2. BRandom error from imprecision
3. CA proportional systematic error
4. DA constant systematic error
Show answerHide answer
Answer: C · A proportional systematic error
The points follow a straight line (r = 0.999) with a slope of 1.12 and an intercept close to zero, so the new method reads about 12% higher at every concentration. A difference that grows in proportion to concentration is a proportional systematic error, typically from calibration (for example a wrong calibrator value) or a matrix difference. A constant error would move the line up or down parallel to the line of identity, changing the intercept rather than the slope. Random error shows as scatter about the line, which is small here.
- A: The points stay on a straight line up to the highest concentration.
- B: The points lie close to the regression line (r = 0.999), so imprecision is small.
- D: A constant error changes the intercept; here the intercept is near zero and the slope is 1.12.
Question 4 · Hard
Linearity samples for an AST method give these results (expected / measured, U/L): 100 / 102, 300 / 296, 600 / 610, 900 / 872, 1200 / 1015. The allowed deviation is 10%. Up to what activity is the measuring range verified?
1. A900 U/L; dilute samples above it
2. B600 U/L; dilute samples above it
3. C1200 U/L; the method is linear
4. D1015 U/L; the top measured result
Show answerHide answer
Answer: A · 900 U/L; dilute samples above it
Each level's deviation is (measured - expected) / expected: +2.0%, -1.3%, +1.7%, -3.1% and -15.4%. Only the 1200 U/L level exceeds 10%, so the measuring range is verified up to 900 U/L, the highest level that passes. Samples above 900 U/L are diluted with a validated diluent and rerun. Falling recovery at the top of the range is typical of substrate depletion in kinetic enzyme methods.
- B: The 900 U/L level is only 3.1% low, so 600 U/L is too conservative.
- C: The 1200 U/L level reads 15.4% low, outside the 10% limit.
- D: 1015 U/L was measured at a level that failed, so it is not a verified limit.
Question 5 · Easy
Before using an unmodified commercial method for patient testing, a laboratory accredited to ISO 15189 must at least:
1. AVerify the manufacturer's performance claims
2. BFully validate it as an in-house method
3. CSet its own reference interval from 120 people
4. DRely on the package insert without testing
Show answerHide answer
Answer: A · Verify the manufacturer's performance claims
ISO 15189 separates verification, which confirms that the laboratory achieves the manufacturer's stated performance (for example precision and trueness) in its own hands, from validation, which is needed for laboratory-developed, modified or off-label methods. Verification uses smaller studies, such as the CLSI EP15 precision and trueness protocol. Using the package insert with no local data is not acceptable. A manufacturer's reference interval can usually be transferred and verified with about 20 local subjects rather than a full study.
- B: Full validation is required only for laboratory-developed or modified methods.
- C: A full 120-person reference study is not required for every new method.
- D: The laboratory must show with its own data that it meets the claims.
Question 6 · Hard
To adopt a manufacturer's reference interval, a laboratory tests 20 healthy local adults, as CLSI EP28 recommends. Three of the 20 results fall outside the interval. The laboratory should next:
1. ATest a second group of 20 healthy adults
2. BWiden the interval to include the three results
3. CAccept the interval, as 85% fall within it
4. DReject it and start a 120-person study at once
Answer and explanation in LearnFree account
Question 7 · Medium
In an interference study, hemolysate is added to one aliquot of a patient pool and an equal volume of saline to a second aliquot. The difference between the two results estimates:
1. AThe carryover between adjacent samples
2. BThe constant error caused by hemoglobin
3. CThe proportional error of the method
4. DThe random error of the method
Answer and explanation in LearnFree account
Question 8 · Medium
In a recovery study, serum is mixed 9:1 with water (baseline) and 9:1 with a standard that should add 5.0 mmol/L of glucose. The baseline reads 4.6 mmol/L and the spiked sample 9.4 mmol/L. What is the percent recovery?
1. A49%
2. B96%
3. C104%
4. D188%
Answer and explanation in LearnFree account
Question 9 · Hard
A new method is compared with the current one using patient samples and linear regression. The slope is 1.00 but the y-intercept is clearly different from zero. What kind of error does this show?
1. ARandom error only
2. BNo error
3. CConstant systematic error
4. DProportional systematic error
Answer and explanation in LearnFree account

More General Laboratory, Quality and Safety questions

Answer them one by one, keep your score and see your weak topics in Learn. Signed-out visitors try 25 per exam; sign in with Google to see them all.

Study General Laboratory, Quality and Safety

Method validation, verification and reference intervals

Before replacing its potassium method, a laboratory measures 40 patient samples on both methods and draws the difference plot shown. Its acceptance limit for mean bias is ±0.10 mmol/L. The correct conclusion is:

Compared with the current method, the new creatinine method in the comparison plot shown has:

Linearity samples for an AST method give these results (expected / measured, U/L): 100 / 102, 300 / 296, 600 / 610, 900 / 872, 1200 / 1015. The allowed deviation is 10%. Up to what activity is the measuring range verified?

Before using an unmodified commercial method for patient testing, a laboratory accredited to ISO 15189 must at least:

To adopt a manufacturer's reference interval, a laboratory tests 20 healthy local adults, as CLSI EP28 recommends. Three of the 20 results fall outside the interval. The laboratory should next:

In an interference study, hemolysate is added to one aliquot of a patient pool and an equal volume of saline to a second aliquot. The difference between the two results estimates:

In a recovery study, serum is mixed 9:1 with water (baseline) and 9:1 with a standard that should add 5.0 mmol/L of glucose. The baseline reads 4.6 mmol/L and the spiked sample 9.4 mmol/L. What is the percent recovery?

A new method is compared with the current one using patient samples and linear regression. The slope is 1.00 but the y-intercept is clearly different from zero. What kind of error does this show?