Presenting accuracy data as natural frequencies rather than probabilities improves post-test probability estimation and speed of calculations¶

Claim · All claims
Evidence · 1 study · review r3 · q2
Currency · newest study 2015 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q2 i? Five studies found post-test probability estimation was more accurate with natural frequencies than probabilities. → Whiting Penny F 2015 q2 i? Two studies found health professionals spent an average of 25% more time assessing probability-format scenarios than natural-frequency scenarios. → Whiting Penny F 2015

Evidence¶

Whiting Penny F 2015¶

Whiting Penny F, Davenport Clare, Jameson Catherine, Burke Margaret, Sterne Jonathan A C, Hyde Chris, Ben-Shlomo Yoav. (2015). How well do health professionals interpret diagnostic information? A systematic review. BMJ Open. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4521525/

q2 · i? · review · r3

Narrative synthesis of five studies including three RCTs comparing natural frequency and probability formats. The review reports estimation was more accurate with natural frequencies, and two studies found professionals spent an average of 25% more time assessing probability-format scenarios.

"Five studies (3 randomised controlled trials (RCTs), 1 two-group study, and 1 single-group study) found that post-test probability estimation was more accurate when accuracy data were presented as natural frequencies19 26 31 32 than as probabilities"

Discussion¶

Related Claims¶

• Evidence is mixed on whether likelihood ratio formats yield more accurate post-test probability estimates than sensitivity and specificity — a narrower finding that bears on this claim
• Natural frequencies combined with graphical aids produced the highest proportion of correct post-test probability estimates in one RCT — a narrower finding that bears on this claim
• Post-test probability is more poorly estimated for negative than positive test results in some studies — related
• Health professionals' estimation of post-test probability is generally poor with a tendency to overestimation — related