Errata

Page 38. Derivation of No Treat-Test Threshold Probability with a costly dichotomous test.  It says p[+|D-]B when it should say p[+|D+]B.  It should read:

$$\frac{(p[+|D-])C + T}{(p[+|D+])B + (p[+|D-])C} = \text{No Treat-Test Threshold Probability}$$

Page 59, Box 3.2.  In the table below the figure, first column, “BNF” should read “BNP.”

Page 321, table at the bottom for answer to problem 2.5b.
Multiple entries in the table are off a bit, and in some cases internally inconsistent, such as the entry “153 − 123 = 31” in the RST-Positive Low risk cell.  A corrected table is pasted below.

The calculations as well as the results are shown in the following table.

Population Overall risk by pedigree analysis  
High risk Low risk Total  
RST Positive 153 × 80%= 122 153 − 122= 31 2464 × 6.2%=153 PPV= 80%
Negative 2311 − 2126= 185 2311 × 92%= 2126 2464 × 93.8%= 2311 NPV= 92%
Total 122 + 185= 307 31 + 2126= 2157 2464  
  Sensitivity=
122/307= 39.8%
Specificity=
2126/2157= 98.6%
   

Page 322, left column, answer to part 2.5c, 2nd paragraph reads, “Similarly, to get the posttest odds above which we would treat without testing…”
This should read, “Similarly, to get the pretest odds above which we would treat without testing …”

A little lower down, same mistake.  It reads, “So the posttest probability = 4.22/5.22…” and should read, “So the pretest probability = 4.22/5.22 …”

Thank you to Songphol Tungjitviboonkun (Epi 204 student in 2024) for spotting this error!

Page 339, right column, bottom, answer to part f, there are some percent signs that should not be there, and the number 0.9611 is erroneously written as 0.0911.  This part of the answer should read:
NB(Treat All) = .0389 – (1/25).9611=0.000456, about 0.05%.

Page 340, left column, answer to part g, replace “1/100th” with “1/4th (1.05%/3.89%)” as follows:

This is higher than the treat all strategy, but it’s still only about 1/4th (1.05%/3.89%) as good as being able to admit someone destined to get a stroke without having to admit anyone unnecessarily.