Prognosis II Flashcards Preview

Research Methods > Prognosis II > Flashcards

Flashcards in Prognosis II Deck (21)
Loading flashcards...
1
Q

prognostic factors

A

general term that describes characteristics predictive of any type of future outcome is a prognostic factor

2
Q

risk factors

A

predictors of future adverse events usually are referred to as risk factors

target disorder/diease not developed. influence development of target disorder/disease

risk factor– we dont have illness yet ex smoking to lead to cancer

prognosis– have illness

3
Q

challenge of prognosis- 3

A

THE OUTCOME that are possible

THE LIKELIHOOD that the outcome will occur

THE TIME FRAME

4
Q

prognosis factor- 3 evidence based principles

A

evidence about the accuracy of the prognostic factor valid (quality assessment)?

does the prognostic factor generate important info about patient outcome (probability and magnitude considerations)?

can I apply the prognostic factor to a specific patient (clinical use)?

5
Q

quality assessment-evidence about prognostic factor valid? 4 questions

A
  1. well defined sample of patients? -inception cohort
  2. acceptable follow up rates? -complete and long enough to capture the outcome of interest
  3. utilize objective outcome criteria? -blind fashion
  4. subgroups appropriately considered?
6
Q

Prognosis important? 3

A

evidence about the prognostic factor

  1. probability of outcomes
  2. magnitude of association with factor
  3. precision of estimates
7
Q

Prognosis statistical analysis- linear regression

A

linear regression:

prognostic factor- cont (0-100 scale), categorical (pos or neg)

outcome-continuous (0-100 scale)

8
Q

linear regression prognostic factors

A

crude= one considered at a time (univariate or simple regression)

adjusted= all considered at same time (multivariate or multiple regression)

r2-estimate of total variance explained

9
Q

prognosis regression quantities

A

outcome (dep variable)- total variance (ex R square) explained in outcome

prognostic factors (indep variable)- relative contribution compared to another (standardized coefficients) and absolute contribution to prediction equation (unstandardized coefficients)

10
Q

prognosis statistical analysis-logistic regression

A

prognostic factors- continuous (0-100 scale), categorical (pos or neg)

outcome-categorical (pos or neg)

11
Q

Prognosis estimate- relative risk (RR)

A

the gold standard

cohort study design (prospective), exposure status is known, all subjects followed forward, outcome status is measured at later time, comparing probabilities

time consuming, expensive, and gold standard for cohort

RR= (A/A=B)/ (C/C+D)

incidence of outcome in exposed divided by the incidence of outcome in non exposed

12
Q

Interpretation of RR

A

RR- starts with prognostic factor (exposure), follows forward outcome, comparing probabilityof events

therefore, RR estimates probability of disease given presence of prognostic factor

13
Q

Interpretation of RR statistic

A

if RR > 1.0 increased probability of outcome among those with prognostic factor (exposed subjects)

if RR < 1.0 decreased probability of outcome among those exposed with prognostic factor (exposed subjects)

if RR= 1.0 probability of outcome is similar among all subjects

14
Q

Odds ratio

A

odds of exposure among individuals with outcome divided by the odds of exposure among individuals without outcome

OR= (AxD)/(BxC)

15
Q

prognosis estimate-OR

A

the “alternative”

doesn’t have to be a prospective study-

classic design-case sontrol (retrospective) subjects with and without disease are enrolled

compare the odds of disease in the two groups

16
Q

Case control study- recall bias

A

systematic error due to the differences in accuracy or completeness of recall to memory of past events of experiences

17
Q

Interpretation of OR

A

stats with outcome

follow backward to prognostic factor

comparing odds of events

therefore, OR estimates odds of prognostic factor given outcome

18
Q

Interpretation of OR statistic

A

if OR >1.0 increased odds of prognostic factor (exposure) among those with outcome

if OR <1.0 decreased odds of prognostic factor (exposure) among those with outcome

if OR= 1.0 odds of prognostic factor (exposure) are similar among all subjects. Nothing happens at 1

19
Q

OR and RR

A

often interpreted the same statistic

RR has probability built in, OR does not

20
Q

OR=RR estimate most likely when…

A

case control study-factors selected indep of exposure status

cohort tudy- overall incidence of the disease is low

21
Q

Interval estimates of OR/RR

A

post estimate of RR/OR-magnitude

interval estimate of RR/OR-precision, 95% confidence interval

range of values that the true OR or RR will lie 95% of the time

if interval estimate of RR/OR does not contain 1.0 then significant and more precise