Biostats Flashcards
What is the mode?
data point that occurs most frequently
What is the mean?
average of all the data points
What is the median?
halfway between highest and lowest value
What is the range?
distance between highest and lowest value
Describe a right skewed distribution and relation to mean, median & mode
Right skewed = tail is to the right, peak to the left
from tail –> peak: mean, median, mode
Define incidence
the rate at which new disease occur
Define prevalence
total number of cases in a population at any given time
Define Precision
data points around 1 point = points in a cluster. Doesnt mean they have to be right, just clustered
Define Accuracy
= validity. Are the points in the correct spot? *gold standard tests
Define Reliability in terms of tests
can a test outcome be reproduced
What % is 1 SD? 2SD? 3SD?
1SD = 68% 2SD = 13.5% more (95% total) 3SD = 2.5% more (99.7% total)
If 40,000 students take step3 each year. The mean score is 222 with a SD of 16. How many students scored above 254? explain your answer.
1st how many SD above the mean is 254? (222-254)/SD(16) = 2
2nd what do they want? they want those that scored ABOVE 254 aka those that are 3SD area. 3SD = 2.5% at each tail. we want those at the tippy tippy top. ==> 2.5% of 40,000 = 1,000
What is the Standard Error of the Mean(SEM)?
a measure of how tightly grouped data is
What does a small Standard Error of the Mean(SEM) tell you?
the data is tightly grouped = more precise.
What is a Z-score?
tells you how many SD up or down from the mean the data is.
What is the Z-score: Average 240, SD 18. Zscore for 254?
254-240 = 14. 14/18(SD) = 0.78.
Zscore = 0.78
What is a confidence interval?
tells you how precise a group of data is. tells you where 95% of the data is located
What does it mean if a confidence interval contains 1?
CI contains 1 = data is not reliable, way too spread out = useless =(
What is a correlation Coefficient(r)? What does a +r mean? -r? what if r=0?
R = relation between data R = + positive correlation R = - negative correlation R = 0 no correlation
When would you want to use a T-test?
when comparing 2 groups of data
When would you want to use ANOVA test?
when comparing 3+ groups of data
When would you want to use Chi-Squared test?
when comparing non-numerical data; are these groups related?
What is a Randomized Controlled Trial?
RCT = prospective trial
*avoids many forms of bias
What is a Cohort Study?
Cohort = follow the fish! ~ observes prospectively over time what happens to a group of patients with certain exposures or underlying illnesses. *used to assess risk
What is Relative Risk? Which study type is used to calculate this?
RR = risk of disease based on who was exposed in the past.
*cohort studies are used to calculate this
What is a Case Control Study? Whats the bias?
CCS = retrospective studies looking for the odds of a previous exposure on the development of a rare disease manifestation. (taking ppl with disease and looking for exposures)
Bias = subject to recall bias
What is an Odds Ratio?
OR = those who have a disease and looking at the chance of exposure.
==> (case exposured/cases not exposed)/(control exposed/controls not exposed)
What is Berkson Bias? How can you solve this?
hospitalized patient are used as trial subjects instead of the general population resulting in different results from the general population.
solution: random selection of trail subjects
What is the Hawthorn Effect? How can you combat this?
those being studied know they are being studied and change their behavior.
solution: using a placebo and blinding both the investigator and the participants.
What is lead time bias?
early detection is confused with increased survival
What does it mean if a P = 0.05?
there is a 95% chance the ALTERNATE hypothesis is true(reject null)
What does it mean if a P <0.05
Null hypothesis is rejected
What does it mean if a P > 0.05
accept null
What is a type 1 error?
false positive
What is a type 2 error?
false negative
What is sensitivity? How do you calculate this?
the likelihood a test will detect all the ppl with the disease.
TP/(TP + FN) = sensitivity
What is specificity? How do you calculate this?
the likelihhood that ppl without a disease are correctly identified as disease negative
TN/(TN+FP) = specificity
What is NPV? How do you calculate this?
If the test is negative, what is the likelihood you really dnt have the disease?
TN/(TN+FN) = NPV
What is PPV? How do you calculate this?
If the test is positive what is the likelihood you really do have the disease?
TP/(TP+FP) = PPV
What is Absolute Risk Reduction(ARR)?
% decrease in the risk of death or disease from a treatment compared to 100% of ppl in a population
ex: 3% no tx vs 2 % with tx arr=1%
What is Number Needed to Treat(NNT)? How do you calculate this?
NNT = 1/ARR ~number of ppl you have to treat to save 1 life
What is the Number Needed to Harm(NNH)? How do you calculate this?
NNH = number of ppl you have to treat to harm 1 person.
NNH = 1/Absolute Risk
Absolute Risk = incidence w/exposure - incidence without exposure
How do you calculate Absolute Risk?
Incidence with exposure - incidence without exposure
