Calculations And Correlation

Exam logistics you calculate against

The MLS (ASCP) Board of Certification examination is 100 multiple-choice questions delivered in 2 hours 30 minutes by computer adaptive testing (CAT). Scores run on a scale of 100-999 with a minimum passing score of 400, and candidates receive a preliminary pass/fail result immediately at the test center. Because CAT adjusts question difficulty to ability, there is no fixed number-correct or raw-percentage cutoff; do not equate 400 with 40%. Chemistry is one of the largest content areas (roughly 17-22% of items), so calculation fluency directly affects your score.

Spectrophotometry and Beer's law

Most quantitative chemistry rests on Beer's law (Beer-Lambert): A = a x b x c, where A is absorbance, a is the molar absorptivity, b is the path length (usually 1 cm), and c is concentration. Absorbance is directly proportional to concentration. If a standard of known concentration gives a known absorbance, find the unknown by ratio:

C(unknown) = C(standard) x [A(unknown) / A(standard)]

Worked example: a 100 mg/dL standard reads 0.250 A; the patient reads 0.500 A. C = 100 x (0.500/0.250) = 200 mg/dL. Note %T and absorbance are inversely related: A = 2 - log(%T), so 1% T = 2.0 A and 100% T = 0 A.

Dilutions and concentration math

• A 1:10 dilution = 1 part sample + 9 parts diluent; dilution factor = 10. Multiply the measured result by 10 to get the original concentration.
• Serial dilutions multiply: a 1:10 followed by another 1:10 yields a total 1:100 dilution.
• The dilution formula C1V1 = C2V2 solves how much stock to use.

Unit conversions appear too -- e.g., converting mg/dL to SI mmol/L using the analyte's molecular weight (glucose mg/dL x 0.0555 = mmol/L).

Method evaluation and result correlation

Diagnostic test performance: sensitivity = TP / (TP + FN) (ability to detect disease, few false negatives); specificity = TN / (TN + FP) (ability to exclude disease). High sensitivity is best for screening. Precision (reproducibility, expressed as %CV = SD/mean x 100) differs from accuracy (closeness to true value).

The correlation half of this skill is judgment, not arithmetic. Before reporting, ask whether the result is physiologically possible and whether it matches the clinical picture. Examples of failed checks: a glucose of 5 mg/dL with no symptoms (suspect glycolysis in an unseparated tube), a potassium of 8.0 with a hemolyzed specimen (pseudohyperkalemia), or a calcium that conflicts with the albumin. Delta checks flag a result that differs sharply from the patient's prior value, prompting verification before release.

Always integrate the analyte pattern -- a high AST/ALT with high bilirubin and obstruction-type ALP/GGT, for instance -- into one coherent interpretation rather than reading each value in isolation.

Quality control and Westgard rules

Chemistry calculations extend to QC. A Levey-Jennings chart plots control results against the mean +/- standard deviations. The Westgard multirules flag error: 1-3s (one control beyond 3 SD) and R-4s (two controls in a run differing by more than 4 SD) detect random error, while 2-2s, 4-1s, and 10x (ten consecutive on one side of the mean) detect systematic error or shift, and a steady drift across days is a trend that often signals a deteriorating reagent or calibration. A 1-2s warning alone (one control between 2 and 3 SD) is expected occasionally and does not by itself reject a run.

Recognizing which rule was violated -- and whether the error is random or systematic -- is exactly the procedural judgment the exam measures.

Standard curves and final result calculation

Many analytes are read from a calibration curve. A linear standard curve lets you interpolate, but a result above the highest standard is beyond the analytical measurement range and must be diluted and re-run, then multiplied by the dilution factor -- never extrapolated past linearity. Combine this with delta checks and physiologic plausibility before release.

The unifying message of the chemistry domain is that a number is only a result once you have confirmed the calculation, the dilution factor, the QC status, and the clinical correlation; reporting a value that fails any of these checks is the error pattern the MLS exam consistently penalizes.

Sensitivity, specificity, and predictive value in context

Go one step past the formulas. Predictive value depends on disease prevalence: even a highly specific test produces many false positives when prevalence is low, so positive predictive value falls in screening a healthy population. That is why confirmatory testing follows screening. A ROC curve plots sensitivity against (1 - specificity) across cutoffs, and the test with the larger area under the curve discriminates better. When an item asks which test to use for screening, favor high sensitivity (to avoid missing disease); when it asks which to use for confirmation, favor high specificity (to avoid false alarms).

Tie this to method validation: a new method is judged against a reference by linear regression and correlation, where the slope detects proportional (constant-ratio) error, the y-intercept detects constant error, and the correlation coefficient and standard error of the estimate quantify scatter. Knowing whether an error is constant, proportional, or random tells you whether to recalibrate, change reagent lots, or investigate imprecision -- the closing calculation-and-correlation skill the chemistry domain ties together.