A multi-center task force is evaluating the diagnostic performance of a novel quantitative… — USMLE Step 3

A multi-center task force is evaluating the diagnostic performance of a novel quantitative serum biomarker, matrix metalloproteinase-9 (MMP-9), for the early detection of severe joint destruction in patients with rheumatoid arthritis.

Extensive population studies reveal the following about the biomarker's distribution:
- In healthy patients without joint destruction, MMP-9 levels are normally distributed with a mean of $40 \text{ ng/mL}$ and a standard deviation of $5 \text{ ng/mL}$ .
- In patients with confirmed severe joint destruction, MMP-9 levels are normally distributed with a mean of $60 \text{ ng/mL}$ and a standard deviation of $5 \text{ ng/mL}$ .

The current diagnostic threshold for a 'positive' screening result is set at $\geq 50 \text{ ng/mL}$ . However, rheumatologists have noted an unacceptably high number of missed diagnoses (false negatives) using this threshold. To improve the test's sensitivity, the task force officially recommends lowering the cutoff threshold to $\geq 45 \text{ ng/mL}$ .

Assuming the 68-95-99.7 (empirical) rule applies, by what absolute percentage (percentage points) will the test's specificity decrease as a result of lowering the cutoff from $50 \text{ ng/mL}$ to $45 \text{ ng/mL}$ ?

Answer: 13.5 %

Answer

13.5

The absolute decrease in specificity is mathematically equivalent to the absolute increase in the false positive rate (FPR). Specificity represents the proportion of healthy individuals who correctly test negative (values below the cutoff). The FPR represents healthy individuals who incorrectly test positive (values above the cutoff). At the original cutoff of

50 \text{ ng/mL}

(+2 SD from the healthy mean), the empirical rule states 2.5% of the healthy population falls in this upper tail, making the initial FPR 2.5%. When the cutoff is lowered to

45 \text{ ng/mL}

(+1 SD), 16% of the healthy population falls in the upper tail, making the new FPR 16%. The absolute increase in false positives is

16\% - 2.5\% = 13.5\%

. Consequently, the test's specificity drops by exactly 13.5 percentage points.

Step-by-Step Solution

Identify the population distribution needed for specificity calculations.

Specificity and false positive rates are derived exclusively from the healthy population (mean

40 \text{ ng/mL}

, standard deviation

5 \text{ ng/mL}

Specificity is the probability of a negative test in patients who do not have the disease.

Calculate the False Positive Rate (FPR) for the original cutoff of

50 \text{ ng/mL}

The FPR is 2.5%.

The cutoff

50 \text{ ng/mL}

is exactly 2 standard deviations above the mean of 40. Using the 68-95-99.7 rule, 95% of the area is within

\pm 2 \text{ SD}

. The remaining 5% is split between two tails, so the area above

+2 \text{ SD}

is 2.5%.

Calculate the False Positive Rate (FPR) for the new cutoff of

45 \text{ ng/mL}

The FPR is 16%.

The cutoff

45 \text{ ng/mL}

is exactly 1 standard deviation above the mean of 40. Using the empirical rule, 68% of the area is within

\pm 1 \text{ SD}

. The remaining 32% is split between two tails, so the area above

+1 \text{ SD}

is 16%.

Calculate the absolute decrease in specificity.

13.5 percentage points.

Specificity is

100\% - \text{FPR}

. The old specificity was

100\% - 2.5\% = 97.5\%

. The new specificity is

100\% - 16\% = 84\%

. The absolute decrease is

97.5\% - 84\% = 13.5\%

Key Concept

The quantitative relationship between shifting diagnostic cutoff thresholds, false positive rates (FPR), and specificity within normally distributed populations.

Question

Answer

Step-by-Step Solution

Key Concept

Common Mistakes

Topics

More questions from this topic