Basic Research

1
Basic Research

• Designs
• Statistics

2
Study Designs
Case Report
Case Series
Cross Sectional Study
Case-Control Study
Cohort Study
Randomized controlled trial
3
Case report
New, rare, how is it special
Case Series
Completed History, PE, Lab, X-ray Investigation
Review literature
How I do it
4
Cross Sectional Study
5
Case-Control Design
6
Case-Control Study
7
Cohort Study
Reference Population
Study Population
Non-Randomized Selection
Exposed
Not Exposed
Disease
No Disease
Disease
No Disease
8
(No Transcript)
9

• - ?????????????????????????
  ????????????????????
• ???????????????????????????
  ???????????????????
• - ?????????????????? ?????????????????????????
  ??????????? ??????????????
• ?????????????????????????????
  ?????????????????????
• - ??????????????????????? ???????????? , ????
• ?????? ??????????? ?????????????????????
  ????
• ??? ???????????????????????? ???????????
  ???????
• ????????????
• - ????? HIV positive 400 ??
  ??????????????????????????????????????????????????
  ????????????

10
Others

• Quasi-experiment (non-randomized)
• Time series
• Measurement
• Validity (Diagnostic test)
• Reliability (Kappa,Pearson,Correlation..)
• Survey Research
• Systemic Review/Meta-analysis

11
Category of evidence

• Ia meta-analysis of RCT
• Ib at least one RCT
• IIa at least one controlled study
• IIb at least one quasi-experiment
• III non-experiment
• IV expert report/series

12
Survey Research Component

• 1. Sampling
• Non-probability
• convenient, snowball, quota, focus
• Probability
• simple random
• systemic
• stratified
• cluster/multistage
• mixed

13
2. Question design

• clear, coding, checking, field test, scale etc.
• not good good
• 10 9 8 7 6 5 4 3 2 1
  0 VAS
• Excellent Very good Good Fair Poor
  evaluate
• Always Usually Sometimes Rarely
  Never freq.
• Very satisfied Generally s. Somewhat
  dissatisfied Very d.
• Completely agree Generally a. Generally disagree
  Complete d.
• see extra page

14
3. Mode of data collection

• Advantages-Disadvantages
• telephone, self-administration, mail internet,
  interview (personal, group)
• Bias
• Selection not represent
• Information recall, interview, nonrespose
• Assessment intention to find, blind

15
Bias

• Information bias
• recall , interview , classification ,
  measurement , non response.
• Selection bias
• not represent (hospital , community) , not be
  comparable
• Assessment bias
• Intention to find , blind

16
Confounding (another type of bias)

• The association between an exposure (risk factor)
  and outcome (disease) is due to the presence of a
  third factor related to both exposure and disease
• a third variable distorts the association
  between an exposure and an outcome

17
Example

• taking vitamin C are less likely to have high BP
• ( taking vit. C are more likely to exercise ,
  drink less , less BW )
• taking coffee are more likely to have CA lung
• ( taking coffee are more likely to smoke
  cigarette )
• Solving stratified , adjusted , multivariate
  regression
• randomize , restricted , match population

18
Suggestion

• How important
• Design correspond with Question
• Clarify of subjects, method and results
• Possible bias and chance

19
One sample
Proportion
Means
Unknown Var
Known Var Z test
npggt5 Z test
Nlt200 t-test
Ngt200 Z, t-test
npglt5 binomial
20
Two samples (means)
Dependent Pair t-test
Independent
Unknown Var
Known Var Z test
Non-equa Var t-test Estimate df
Equa Var t-test df (n1n2)-2
21

• Three samples comparing means
• ANOVA
• Relationship chi-square test

22
Terms

• Ratio quotient of two number A B
• Proportion fractional component A/(A B)
• Rate the ratio of changing from one time to
  another time
• ex 100 lung cancer cases per 10,000 pop per yr
• Prevalence proportion of case within a
  population
• Incidence occurrence of new case within a
  population

23
Risk Ratio
Disease
Non Disease
Incidence rate of disease
Total
a a b
Exposed
a
b
a b
c c d
c d
c
d
Non Exposed
a a b c c d
RR incidence in exposed
incidence in non exposed

24
Odds Ratio

Disease
Non - Disease
Exposed
a
b
Non - Exposed
d
c
Odds a 1 - a
a c b d
OR Odds of being expose in disease
Odds of being expose in non disease
ad bc


a b c d
OR Odds of disease develop in expose
Odds of disease develop in non
expose
ad bc


25
Questions to ask when reading a journal article
1. What is the research question? 2. What
is the study type? 3. What are the outcome
factors? 4. How are the outcome factors
measured ? 5. What are the study factors
(exposures)? 6. How are the study factors
measured? 7. What is the reference
population? 8. How was the study population
selected from the reference popn? 9. In a
experimental study (trial) how were participants
assigned? (In a case-control study, are
the controls appropriate?) 10. Are statistical
tests performed? 11. Is the study ethical? 12.
What conclusions did the authors reach?