Numeracy: disclaimer

1
Numeracy disclaimer

• This is hard for many medical personnel
• You can probably spelll beter than I can

2
Why understand measures of efficacy?

• Go tohttp//medinformatics.uthscsa.edu/ms3/rx

3
Feedback

• Current results

4
http//utxr.org
5
Example 1

• Bad outcome
• Drug 6 / 100
• Placebo 10 / 100
• What is the percent benefit?
• ___ (ARR)
• ___ (RRR)
• (there are two possible answers)
• What is the NNT ___ ?

6
Example 2

• Change which team member has the pencil/paper
• In a study of preventing death from a disease
• Experimental event rate (EER) 20
• Control event rate (CER) 30

7
Example 2

• Use http//clinical.uthscsa.edu/calc
• In a study of preventing death from a disease
• Experimental event rate (EER) 20
• Control event rate (CER) 30

8
Explanation
9
Explanation
10
Interpreting NNTs

• The NNS (number need to screen) for women aged
  40-50 with MMG
• 1200 according to the USPSTF (PMID 12204020)
• 2500 according to the NCI (http//www.cancer.gov/c
  ancertopics/pdq/screening/breast/HealthProfessiona
  l/page2)
• What is the NNS for prostate cancer according to
  PMID 10989402?

11
ARRs and NNTs

• 3 limitations

12
Calculating NNTs

• Can you calculate a NNT from the information in
  this NEJM abstract
• PMID 18836213
• How about using mortality data from the full text
  of the article?

13
What about this

• Bad outcome
• Drug 6 / 100
• Placebo control 10 / 100
• What is the percent benefit?
• ARR is 4
• RRR is 40
• NNT 25
• If a doctor uses this drug in patients that
  without rx have a 1 /100 risk of bad outcome.
  What is the
• ARR
• RRR
• NNT

14
Shared decision making and the law

• A residency program in Virginia lost a
  malpractice case when a resident physician used
  shared decision making, the patient declined PSA
  testing, and the patient later developed prostate
  cancer http//jama.ama-assn.org/cgi/content/full/2
  91/1/15.
• This famous case was determined by a locality
  rule. In Texas we do not use a locality rule
  (http//jama.ama-assn.org/cgi/content/full/297/23/
  2633).

15
Calculators

• Online calculator
• http//clinical.uthscsa.edu
• (click Calculators')
• Center for Evidence-based medicine
• (click 'EBM tool box')
• Palm
• UTHSCSA - Appraising applying medical evidence
• Use www.StyleTap.com to place on PocketPC
• Toronto
• Palm http//www.cebm.utoronto.ca/palm/ebmcalc
• EBM Tables (has many NNTs) http//www.cebm.utoront
  o.ca/palm/nnt/
• Pocket PC
• Toronto
• PocketPC http//www.cebm.utoronto.ca/pocketpc/ebm
  cal

16
End
17
Relative risk reduction

• Bad outcome
• Drug 6 / 100 EER 6
• Placebo 10 / 100 CER 10
• Relative risk reduction
• (EER - CER) / CER
• (10 6) / 10
• 4 / 10 40

Return
18
Absolute risk reduction

• Bad outcome
• Drug 6 / 100 EER 6
• Placebo 10 / 100 CER 10
• Absolute risk reduction
• CER - CER
• 10 6 4

Return
19
Number needed to treat (NNT)

• Bad outcome
• Drug 6 / 100 EER 6
• Placebo 10 / 100 CER 10
• NNT 100 / ARR
• 100 / 4
• 25 patients must be treated for one to be
  benefited by the drug

Return