Skip to content
EqasOnlineAcademySign in

The exam

Sample questions

32 practice questions covering all six domains, with answers and explanations. None of them appear in the live exam, which also includes choose-two and type-a-number questions.

D1PT fundamentals and requirements

  1. D1.1Medical laboratories usually call proficiency testing…

    1. AMeasurement uncertainty
    2. BInternal quality control (IQC)
    3. CExternal quality assessment (EQA)
    4. DMethod verification
    Show the answer

    C. EQA is the medical-laboratory term. An EQA programme often adds education and advice to the scoring.

  2. D1.2Under ISO/IEC 17025:2017, taking part in PT supports which requirement?

    1. AControl of documents (clause 8.3)
    2. BExternally provided products and services (clause 6.6)
    3. CManagement reviews (clause 8.9)
    4. DEnsuring the validity of results (clause 7.7)
    Show the answer

    D. Clause 7.7.2 asks laboratories to monitor performance by comparison with other laboratories, including PT.

  3. D1.3Which of these breaks the rules of fair PT participation?

    1. ATesting the item with your routine method
    2. BDiscussing your results with another participant before the closing date
    3. CReporting in the units the provider asks for
    4. DFollowing the provider’s reconstitution instructions
    Show the answer

    B. Comparing notes makes the comparison meaningless. Results can be discussed after the closing date.

  4. D1.4What does the provider compare each participant’s result with?

    1. AThe participant’s internal QC mean
    2. BAn assigned value
    3. CThe manufacturer’s reference interval
    4. DThe same lab’s result from the previous round
    Show the answer

    B. Each result is compared with the assigned value, using criteria such as σpt set by the scheme.

  5. D1.5Which document sets the policy accreditation bodies follow on PT participation?

    1. AILAC P9
    2. BISO/IEC 17020
    3. CISO 13528
    4. DISO 9001
    Show the answer

    A. ILAC P9 is ILAC’s policy on participation in proficiency testing and other interlaboratory comparisons.

D2Designing and running a round

  1. D2.1For a homogeneity check, ISO 13528 typically calls for…

    1. AAt least 10 units chosen at random, each tested in duplicate
    2. B2 units, each tested once
    3. C1 unit tested 10 times
    4. DTesting every unit produced
    Show the answer

    A. Annex B of ISO 13528 describes testing g ≥ 10 randomly chosen units in duplicate.

  2. D2.2Between-sample SD ss = 0.07 and σpt = 0.20. Does the homogeneity check pass?

    1. AIt can’t be judged without participant results
    2. BYes: 0.07 is less than σpt
    3. CNo: 0.07 is more than 0.3 × 0.20 = 0.06
    4. DYes: 0.07 is less than 2 × σpt
    Show the answer

    C. The usual criterion is ss ≤ 0.3 σpt. Here 0.07 > 0.06, so the check fails.

  3. D2.3Mean before dispatch 10.42, mean after the round 10.30, σpt = 0.50. Are the items stable?

    1. AYes: the difference 0.12 is within 0.3 × 0.50 = 0.15
    2. BNo: the difference 0.12 is more than 0.10
    3. CYes: the difference is within 2 × σpt
    4. DNo: any change means the items are unstable
    Show the answer

    A. The stability criterion is a difference of no more than 0.3 σpt: 0.12 ≤ 0.15.

  4. D2.4Why do PT reports identify participants by code rather than by name?

    1. ATo keep each participant’s identity confidential
    2. BSo participants can compare notes
    3. CBecause codes are quicker to sort
    4. DBecause ISO 13528 allows numbers only
    Show the answer

    A. Confidentiality of participants’ identities and results is a basic requirement for PT providers under ISO/IEC 17043.

  5. D2.5What should the provider document before the round starts?

    1. AEach participant’s expected result
    2. BHow the assigned value and σpt will be set and how results will be scored
    3. CThe participants’ internal QC limits
    4. DWhich labs are expected to fail
    Show the answer

    B. Deciding the evaluation in advance stops the statistics being tuned after the results arrive.

D3Assigned value, σpt and Algorithm A

  1. D3.1In Algorithm A, results are adjusted to lie within…

    1. Ax* ± 3 s*
    2. Bthe 5th to 95th percentile
    3. Cx* ± 2 s*
    4. Dx* ± 1.5 s*
    Show the answer

    D. Each iteration replaces values outside x* ± 1.5 s* with the nearest limit, then recalculates x* and s*.

  2. D3.2Algorithm A’s starting value of s* is…

    1. Athe ordinary standard deviation
    2. B1.483 × the median absolute deviation from the median
    3. Cthe range divided by 4
    4. Dσpt
    Show the answer

    B. Multiplying the median absolute deviation by 1.483 gives an estimate of the standard deviation for normally distributed data.

  3. D3.3Which assigned value depends least on the participants’ own results?

    1. AThe median of your method group
    2. BThe robust mean of all results
    3. CA value from a known formulation
    4. DThe mean of the first ten results received
    Show the answer

    C. Formulation, certified reference values and reference laboratories are independent of the participants. Consensus values are not.

  4. D3.4A consensus from p = 25 results has s* = 0.50. What is u(xpt)?

    1. A0.020
    2. B0.100
    3. C0.625
    4. D0.125
    Show the answer

    D. u(xpt) = 1.25 × s* / √p = 1.25 × 0.50 / 5 = 0.125.

  5. D3.5What is the main benefit of a σpt fixed in advance from fitness for purpose?

    1. AIt protects against outliers
    2. BIt removes the need for an assigned value
    3. CIt always gives smaller z scores
    4. DScores mean the same thing from round to round
    Show the answer

    D. If σpt came from each round’s own spread, the standard of performance would move with the participants.

  6. D3.6u(xpt) = 0.09 and σpt = 0.25. Which score should be used?

    1. AEn, because the value is a consensus
    2. Bz, because u(xpt) is less than σpt
    3. CNo score can be given
    4. Dz′, because u(xpt) is more than 0.3 σpt = 0.075
    Show the answer

    D. When u(xpt) > 0.3 σpt the uncertainty is not negligible, and z′ includes it.

D4Performance scores

  1. D4.1xpt = 12.0 %, σpt = 0.40 %, your result 10.6 %. Your performance?

    1. AUnsatisfactory, z = −3.5
    2. BUnsatisfactory, z = −1.4
    3. CQuestionable, z = −3.5
    4. DSatisfactory, z = −1.4
    Show the answer

    A. z = (10.6 − 12.0) / 0.40 = −3.5, and |z| ≥ 3.0 is an action signal.

  2. D4.2If σpt matches the spread of competent labs, roughly what share of their results fall outside |z| ≤ 2?

    1. ANone
    2. BAbout 5 %
    3. CAbout 0.3 %
    4. DAbout 32 %
    Show the answer

    B. About 95.4 % of a normal distribution lies within ±2, so about 1 in 20 good results is questionable by chance.

  3. D4.3En = 1.4. Your performance?

    1. ASatisfactory: |En| < 3
    2. BSatisfactory: |En| < 2
    3. CUnsatisfactory: |En| > 1
    4. DQuestionable
    Show the answer

    C. En uses expanded uncertainties, so the limit is 1, not 2 or 3.

  4. D4.4Which quantities go into En?

    1. AThe expanded uncertainties of your result and of the assigned value
    2. BYour repeatability standard deviation
    3. COnly the standard uncertainty of the assigned value
    4. Dσpt and u(xpt)
    Show the answer

    A. En = (x − xpt) / √(U²(x) + U²(xpt)).

  5. D4.5z = +2.0 exactly. Your performance?

    1. ANot defined
    2. BQuestionable
    3. CSatisfactory
    4. DUnsatisfactory
    Show the answer

    C. Satisfactory is |z| ≤ 2.0, so exactly 2.0 is still satisfactory. Questionable starts above 2.0.

  6. D4.6A qualitative scheme expects “Salmonella detected”. You report “not detected”. The result is…

    1. ANot acceptable: it does not match the expected result
    2. BNot scorable
    3. CSatisfactory if your method is less sensitive
    4. DQuestionable
    Show the answer

    A. Qualitative results are judged against the expected answer. A false negative is a wrong answer.

D5Reports and trends

  1. D5.1Before investigating an unsatisfactory score, what should you check on the report first?

    1. AThat the result, units and sample match what you submitted
    2. BThe provider’s logo and address
    3. CHow many pages the report has
    4. DWhich other labs were unsatisfactory
    Show the answer

    A. Transcription, unit and sample-swap errors are quick to check and common. Reports also never name other labs.

  2. D5.2A report says the assigned value is the robust mean of n = 46 results. What does n tell you?

    1. AThe number of decimal places used
    2. BYour rank among the participants
    3. CHow many rounds you have taken part in
    4. DHow many results the consensus was based on
    Show the answer

    D. n shows how much data supports the consensus. A small n means a less certain assigned value.

  3. D5.3Which pattern deserves a look even though every score is satisfactory?

    1. AA single score of +1.0
    2. BEight scores in a row above zero
    3. CAll scores within ±0.5
    4. DScores alternating around zero
    Show the answer

    B. A long run on one side of zero suggests a consistent bias, even before any score fails.

  4. D5.4On a Youden plot your point is far from the diagonal: sample A high, sample B low. The most likely cause?

    1. ANothing; this is ideal
    2. BRandom error, or a problem with one of the samples
    3. CWrong units on both samples
    4. DA calibration bias
    Show the answer

    B. Systematic error moves both samples the same way, along the diagonal. Opposite directions point to random error.

  5. D5.5Your method group’s median differs from the all-method assigned value. What should you check?

    1. ASwitch methods straight away
    2. BNothing: method groups are for information only
    3. CAsk for your method group to be removed
    4. DWhich value your score is based on, and whether the difference is a known method bias
    Show the answer

    D. Scores may be based on all methods or on a method group. Know which, and whether the method is known to read high or low.

D6Investigation and corrective action

  1. D6.1You reported 5.71 but your laboratory system shows 5.17. Which cause category is this?

    1. AMethod
    2. BPT item
    3. CClerical (transcription)
    4. DEquipment
    Show the answer

    C. Swapped digits are a transcription error. Check the clerical trail first.

  2. D6.2A PT item that should arrive chilled arrives warm. What should you do?

    1. ARecord it and contact the provider before testing
    2. BChill it and test it without recording anything
    3. CTest it and say nothing
    4. DReport the result you expect
    Show the answer

    A. The provider can replace the item or allow for it, and a note made at receipt helps any later investigation.

  3. D6.3The root cause was a faulty calibrator lot. Which is a proper corrective action?

    1. AAsk the provider to widen σpt
    2. BEdit the PT result you submitted
    3. CSkip the next round
    4. DReplace the lot, recalibrate and confirm with QC and retained samples
    Show the answer

    D. Fix the cause, then show with evidence that the fix worked.

  4. D6.4Why review satisfactory PT results too?

    1. AAssessors only read satisfactory results
    2. BThere is no need
    3. CA trend in satisfactory scores can reveal a developing bias
    4. DTo improve the score afterwards
    Show the answer

    C. Trends show up before failures, and reviewing every round costs less than investigating a failure.

  5. D6.5You believe the provider’s assigned value is wrong. Best next step?

    1. AIgnore the round
    2. BRecalculate your own score and file that instead
    3. CContact the provider with your evidence
    4. DShare other participants’ results to prove it
    Show the answer

    C. Providers have a process for complaints and appeals. An evaluation error is possible and should be raised.

Take the examPractise z scores