Research Design, Biostatistics, and Literature Evaluation

is around 89% (Answers A, B, and C are incorrect). The

PPV is calculated by dividing 189/289, and the NPV is

calculated by dividing 100/112.

Answer: D

When evaluating this study, it is important to consider

that the outcome related to a biomarker is simply a

surrogate as opposed to a true clinical outcome. This

could be problematic or lack validity because other ste-

roids have previously been studied for outcomes such

as mechanical ventilator days and mortality (Answer

D is correct). The study may not have met the criteria

for the original power analysis that was defined by the

authors, but it did achieve power to find a statistically

significant difference according to the outcome (Answer

A is incorrect). Although the sample size is small, it may

represent a reasonable population with internal valid-

ity to use the information in a limited way, particularly

at the institution(s) where it was performed (Answer B

is incorrect). The power analysis was created around a

continuous outcome, but it is reasonable to present the

data in two different ways (in this case, patients who did

and did not achieve a CRP decrease of at least 10 mg/L)

as long as the reader understands that the study was not

powered around this analysis (Answer C is incorrect).

Answer: A

The NNT is calculated by 1/ARR. The absolute risk

reduction is 3%, or 0.03. The NNT is 1/0.03 = 34 (round-

ing up to the nearest whole number) (Answer A is correct;

Answers B, C, and D are incorrect).