Professional Documents
Culture Documents
Diagnosis of Ambulatory Community-Acquired Pneumonia
Diagnosis of Ambulatory Community-Acquired Pneumonia
- 20% pneumonia
- Also sounds like URTI, LRTI
- CXR
o Cons: Inconvenience, cost, small amount of exposure to radiation
o Pros: Rule out/in pneumonia, if pneumonia is caught then he won’t get worse
- Whether CXR given or not depends on values and preferences
Patient 2
- Upon Px there is no air entry
- 85% pneumonia
- CXR
o Management will differ
o If you don’t do CXR and give antibiotics (b/c high probability), can develop
resistance, there are downsides to antibiotics
o Severity of disease
o A Hx of CXR follow a patient over time, monitor progress
Patients: adult, >21 yo, acute febrile illness, <1 week duration (patient report, neighborhood
clinic, ER temperature), at least ¼ complaints (cough, coryza – runny nose, sore throat or
hoarseness), pregnant women and HIV status excluded
Validity:
- patient population that was chosen is too ‘broad’ if every patient w/ just hoarseness
and fever the other day is put into the study, the population leans more towards
‘normal’ rather than ‘abnormal’
- this makes your study look better physicians will most likely diagnose patients w/out
pneumonia as not having pneumonia (b/c it is obvious in the choice of population) and
vice versa
- the 2 normal distributions will be plotted as further apart from each other
- this will overestimate the power of the test
- could have included some SOB in criteria
- independent comparison
o people doing the test were unaware of results of GS and vv
o separate respirologist and radiologist doing CXR and they were blinded
- risk of bias (from 0-10) is estimated to be 5-6
o met criteria of independent and blind study but not really w/ appropriate
population
Pretest Probability
Posttest Probability
Sensitivity
- Portion of people that are correctly identified in those who are target positive
- 14/19 = 74%
Specificity
- portion of people that are correctly identified in those who are target negative
- 194/231 = 84%
Conclusions
- physicians are good at ruling out but not ruling in
- when are they right/wrong
Nature of Problems
Validity
- the use of correct methods to find the truth
- risk of bias
- Give skewed results, misled
o Wrong methods used Bias (systematic deviation from the truth)
o Interpret accurate results in a misleading way
o Random error
Particularly if sample size is small
- Bias
o systematic deviation from the truth
Results
- random error
Applicability
- inappropriate patient population (symptoms, etc.)
- inappropriate outcome measures find the truth about something else
Target condition
- the truth
- ie. pneumonia or no pneumonia
- participants are target positive or target negative
- should enroll people w/ diagnostic uncertainty
- use test in people who are going to use the test
Criteria for validity of diagnostic test study
- Appropriate population people with diagnostic uncertainty
- independent and blind test – gold standard
Criteria for validity of treatment study
- randomization
-
** Find out post test probability for patients 1 and 2 would differ from 20% and 80% if their CXR
were positive or negative look into likelihood ratios
**