Title | EPID exam notes TAKE IN |
---|---|
Author | Cecilia Nguyen |
Course | Epidemiology and Biostatistics 481 |
Institution | Curtin University |
Pages | 1 |
File Size | 126.2 KB |
File Type | |
Total Downloads | 47 |
Total Views | 146 |
Download EPID exam notes TAKE IN PDF
Epidemiology: study of distribution and determinants of disease (health related states/events) in specified populations and the application pf this study to control health problems 3 Common Comparisons: 1) Comparing diseased w/ not diseased (comparing cases w/ controls -> case control studies) 2) Comparing not diseased w/ risk factor to fully not diseased (smokers vs non-smokers) 3) Experiment to manipulate an exposure whether it results in different outcomes for experiment and control group (experimental/intervention studies) Study designs asses value of evidence Evidence Based Health Care (5 Steps): -Uncertainty to answerable q’s - Find best available evidence - Appraise evidence (validity, reliability, relevance + applicability) - Apply evidence in practice - Evaluate Def: combined research evidence, clinical experience + patient preference Standard error is std of sampling distribution Small skewness std error: skewness similar across few samples in populatn Large skewness std error: skewness varies significantly from few samples in populatn
same for kurtosis std error
Graphs for CATEGORICAL data: bar, pie Graphs for CONTINUOUS data: histogram, box plot, polygon
stem + leaf plot, frequency
Prevalence: proportion of current cases in populatn at given time = # cases in population at given time Incidence: new cases during specific time period divide by total Total population that of same disease-free personattime observation in population at risk time = # of new events Total event-free person-time of observatn in pop at risk
Conceive, design, conduct, analyse and use study Observational designs: Intervention/experi - Randomised design - Cross-sectional - Non-randomised or Quasi- Ecological/correlational experimental design - Cohort - Case control
Inferential statistics: 95% CI, p value, pearson’s correlation coefficient ‘r’, r2, relative risk, odds ratio E.g. Total of 225 people w/ lung cancer (cases) and 833 people w/out lung cancer (controls) selected. Data showed 176 cases and 85 controls reported exposed to asbestos in past. Calculate odds ratio. Draw 2x2 table. Cases
Controls
Exposed
176
85
Nonexposed
49
749
225
833
Odds cases = exposed cases / non-exposed cases = 176 / 49 Odds controls= exposed controls / non-exposed cont. = 85 / 748 Odds ratio = odds cases / odds controls = (176/49) / (85/748) = 31.61 Interpretation: those w/ lung cancer were 31.61 times more likely to be exposed to asbestos than d= .20 small those w/out. IF calculated odds of EXPOSURE than d= .50 med same result but asbestos instead. d= .80 large
Type 1 error reject Ho when shouldn’t have; depends on level of significance (set at 5% or 1/20) False (+) = correct null hypothesis rejected Type 2 error accept Ho when shouldn’t have; depends on power of study (set at 80%) Higher power= lower chance making type 2 error False (-) = null hypothesis incorrect but not rejected Reliability/precision= lack random error Validity/accuracy= lack systematic error or bias...