MLT GlobeLaboratory Operations: Quality control & QA – page 2
54 Laboratory Operations MCQs on Quality control & QA with answers and explanations.
Two consecutive results for the same control level both exceed the +2 SD limit. Which rule is violated and what error does it suggest?
Two consecutive values beyond the same 2 SD limit violate 2-2s, which detects systematic error (bias). R-4s requires values on opposite sides of the mean.
In one run, the normal control is +2.3 SD and the abnormal control is −2.2 SD. Which rule is violated?
The range between two controls in the same run exceeds 4 SD (one above +2 SD, one below −2 SD), violating R-4s, a random error rule. Neither value exceeds 3 SD.
On a Levey-Jennings chart, control values rise steadily over eight days in one direction. The most likely cause is:
A trend is a gradual, progressive change in one direction, typical of slowly deteriorating reagents, lamps or filters. Single events such as a bubble cause isolated outliers, not trends.
Immediately after a new reagent lot is placed in use, all control values abruptly move to a new level above the mean and stay there. This pattern is a:
A sudden, sustained change in the level of control values is a shift, typically from a new reagent lot, recalibration or instrument change. A trend changes gradually over time.
In an interlaboratory comparison, your lab mean is 105 mg/dL, the peer group mean is 100 mg/dL and the peer SD is 2.5 mg/dL. The standard deviation index (SDI) is:
SDI = (lab mean − peer mean) ÷ peer SD = (105 − 100) ÷ 2.5 = +2.0. A positive value means the lab reads higher than peers; |SDI| ≥ 2 suggests bias needing review.
A technologist is unsure of a proficiency testing result and wants to send the PT sample to a reference lab to confirm it before reporting. The correct action is to:
Regulations (e.g. CLIA) require PT samples to be tested in the same way as patient samples. Referring PT samples to another lab or discussing results with other labs before submission is prohibited.
A test is evaluated in 200 subjects: true positives 45, false positives 15, false negatives 5, true negatives 135. The positive predictive value is:
PPV = TP ÷ (TP + FP) = 45 ÷ 60 = 75%. 90% is the sensitivity (45 ÷ 50) and 90% is also the specificity (135 ÷ 150); 96% is the NPV.
A clinician wants a test to confirm (rule in) a disease with the fewest false-positive results. The most important test characteristic is high:
High specificity means few false positives, so a positive result reliably rules in disease. High sensitivity (few false negatives) is best for ruling out disease.
According to CLSI EP28, the minimum number of healthy reference individuals recommended to establish a reference interval by the nonparametric method is:
CLSI EP28-A3c recommends at least 120 reference individuals per partition for the nonparametric method. Twenty samples are used only to verify (transfer) an existing interval.
During method validation, running the same sample 20 times over several days mainly estimates:
A replication experiment measures the spread of repeated results, i.e., random error expressed as SD or CV. Systematic errors are assessed by comparison-of-methods, recovery and interference studies.
A comparison of a new method (y) with the current method (x) gives the regression equation y = 1.00x + 8 mg/dL. This shows:
A y-intercept different from zero with a slope of 1.00 means the new method reads a fixed amount higher at all levels: constant error. A slope different from 1.00 would indicate proportional error.
The lowest analyte concentration that can be reliably distinguished from the limit of blank is the:
Per CLSI EP17, LoD is the lowest concentration reliably distinguished from the LoB. LoQ is the lowest concentration measured with acceptable total error; LoB is the highest result expected from a blank.
A patient's MCV is 72 fL today but was 95 fL two days ago, with no transfusion. The QA tool that flags this is the:
A delta check compares a result with the same patient's previous result; a large unexplained change (MCV is very stable) suggests a specimen mix-up or error. Westgard rules evaluate control data, not patient results.
During method evaluation, a sample is split and a potential interfering substance is added to one portion. This interference experiment mainly estimates:
The interference experiment measures the difference caused by an added substance, which is usually a constant systematic error. Proportional error is estimated by the recovery experiment.
A test is evaluated in 300 people: true positives 80, false negatives 20, true negatives 180, false positives 20. What is the negative predictive value?
NPV = TN / (TN + FN) = 180 / (180 + 20) = 0.90, or 90%. Sensitivity here is 80/(80+20) = 80%, which is a common confusion.
When a new lot of control material is introduced, the laboratory's own mean and SD should ideally be established from at least:
CLSI C24 recommends at least 20 results from 20 separate runs or days to capture day-to-day variation; ranges from only a few runs are unreliable. Values can be refined as more data accumulate.
Under CLIA regulations, calibration verification must be performed at least every:
CLIA requires calibration verification at least every 6 months, and also after major maintenance, reagent lot changes (if needed) or QC problems. Weekly or monthly checks are not the regulatory minimum.
Under CLIA, for most quantitative chemistry tests without an approved alternative QC plan, the minimum control testing is:
The default CLIA requirement for quantitative tests is at least two levels of control material each day of patient testing. Weekly or monthly controls do not meet this minimum.
On a Levey-Jennings chart, control values become widely scattered above and below the mean without a trend. Which cause is most likely?
Increased scatter reflects random error, such as intermittent bubbles in the pipetting system. Lamp deterioration or reagent evaporation usually causes trends, and a calibrator change causes a shift.
Two tests for the same disease are compared by receiver operating characteristic (ROC) curves. Test X has an area under the curve of 0.92 and test Y of 0.65. This means:
The area under the ROC curve summarises overall discrimination; 1.0 is perfect and 0.5 is no better than chance. A higher AUC does not guarantee better sensitivity or specificity at every single cut-off.