Got asked to post this from section leaders today because the method varies from the answer key (which uses the ROC table).
Thanks to my study group for pointing out how to do this (Chi, Adrian, Lekha).
========================================================================================
g) Calculate the Net Benefit of a hospitalization strategy using the ABCD2 cutoff in (a). Is it higher or lower than the NB of the “Treat All” strategy? [2]
Net Benefit = (Patients Treated Appropriately – C/B × Patients Treated Unnecessarily)/(All Patients)
Had C/B from an earlier subquestion but here it for completion sake:From Vignette: "Remember that an unnecessary admission of a TIA patient who doesn’t have a stroke in the next 2 days is 1/25 as bad as failing to admit someone who does have a stroke in the next 2 days. "
C = cost of regret associated with treating someone without
disease (admitting 25 patients for 2 days)
B = cost of failing to treat someone with disease (discharging someone with a
stroke in the next 2 days)
25C = B or C = B/25 depending on how you read into it
C/B = C/25C = 1/25 = 0.04
=============================
|
Score |
D+ |
D- |
% of TIA Patients |
|
6-7 |
1.70% |
19.30% |
21% |
|
4-5 |
1.85% |
43.16% |
45% |
|
0-3 |
0.34% |
33.66% |
34% |
|
|
3.89% |
96.11% |
100.00% |
Cut off was for a score >4 so we are interested in the D+ and D- in those corresponding rows (bolded above)
Net Benefit = (Patients Treated Appropriately – C/B × Patients Treated Unnecessarily)/(All Patients)
· C/B = .04 (from above)
· Patients treated appropriately = 1.85% + 1.70% = 3.55% or .0355
· Patients treated unnecessarily = 43.16+19.3 = 62.46% or .06246
=============================
Net Benefit = (Patients Treated Appropriately – C/B × Patients Treated Unnecessarily)/(All Patients)
Net benefit = (.0355 - .04 * .6246) / 1
Net benefit = (.0355 - .0249) / 1
Net benefit =.0106