Department of Family Medicine
EBM Biostatistics and Prevention
| 1. |
|
| 2. | Measures of "spread" of scores:
Standard Error of the Mean (SEM) = the standard deviation of a number of sample means and is related to the SD as follows: SEM = SD (square root of variance) 95% Confidence Interval is two times the SEM. The interpretation of the 95% CI is that the true population mean is likely to be within the interval and we are 95% confident. |
| 3. | Null hypothesis = the hypothesis of no effect |
| 4. | p value = the probability that the observed result assuming the null hypothesis is true. |
| 5. | alpha = the level
set by the investigator at the onset of the study which will
determine "statistical significance." Totally arbitrary
and usually 0.05. 0.05 means that if the p value (calculated from your
data) is less than 0.05, you are willing to reject the null hypothesis (no difference). Note: you can never prove without a doubt that there is a difference. |
| 6. | Statistical power = probability of finding a difference if a difference really exists. Is dependent upon difference to be detected, sample size, alpha, and variability of responses. |
| 7. | Type I error = the
null hypothesis is rejected when it should not be. (Finding a difference
when one does not exist.) Governed by alpha (0.5). |
| 8. | Type II error = accepting the null hypothesis when it is false (finding no difference when there is a difference). Governed by statistical power. |
| 9. | Statistical TESTS Two types of variables: dichotomous (sex, race, etc.) and continuous (age, BP, etc.) t-test = one continuous and one dichotomous example: difference in mean birth weights between males and females. chi-square = two dichotomous variables example: deaths from bike accidents depending on helmet use between two time intervals. correlation coefficient = two continuous variables examples: average daily fat intake and incidence of breast cancer. meta-analysis = a statistical method used to combine results of several studies into a single study. |
Prevention
1. Primary prevention = prevention of disease occurrence2. Secondary prevention = early detection and screening
3. Tertiary prevention = prevention of morbidity and mortality
Secondary Prevention (Screening)
sensitivity = given disease, how many have a + test
specificity = given no disease, how many have a - result
Sensitivity/specificity are test characteristics. They depend upon the chosen cut-off points and do NOT depend of the prevalence of disease.
predictive value + = given a + test, how many have disease
predictive value - = given a - test, how many do not have disease.
PREDICTIVE VALUE DEPENDS UPON THE INCIDENCE/PREVALENCE OF THE DISEASE. IF LOW INCIDENCE, LOW PV+ (means most who test + will really not have disease).
Receiver Operating Characteristic (ROC) Curve:
Graphically looks at the true positive rate (sensitivity) to the false positive rate (1- specificity) at varying cut-off points.
Uses:
1. Finding best cut-off point
2. Comparing two tests
Point or curve closest to the upper left corner is best.