Skip to main content

Table 3 True and expected number of positive outcomes, risks, and risk ratio

From: Use of diagnostic likelihood ratio of outcome to evaluate misclassification bias in the planning of database studies

Group

Sample size

Number of true positives

Expected number of positive outcomes identified by the algorithma

Control

NCON

XCON

ECON = XCONSe + (NCON − XCON)(1 − Sp)

Test

NTES

XTES

ETES = XTESSe + (NTES − XTES)(1 − Sp)

Group

True risk

Expected risk based on the algorithm

Control

RCON = XCON/NCON

ECON/NCON = RCONSe + (1 − RCON)(1 − Sp)

Test

RTES = XTES/NTES

ETES/NTES = RTESSe + (1 − RTES)(1 − Sp)

Relative measure

True relative measure

Expected relative measure

Risk ratio (RR)

RRTRUE = RTES/RCON

RREXP = (ETES/NTES)/(ECON/NCON)

  1. Numbers of positive outcomes are those expected under the non-differential misclassification assumption. The numbers include both true- and false-positives based on the algorithm.
  2. aSe = sensitivity, Sp = specificity