Intro bio stats (1) Flashcards
Preclinical research -
- Basic
- In labs on animals
- Focused on theory and understanding mechanisms of disease/therapies
Clinical research -
- Appled
- Human subjects
- Focused on testing theories that help find new and better ways to detect, diagnose, treat and prevent disease/develop therapies
Step 1 of the clinical research process?
Identify the research question
Identify the research question (step 1) -
- specify question to investigate
- most important and often most difficult because it controls the direction of all subsequent planning and analysis, design and statistics included
Step 2 of the clinical research process?
Design the study
Design the study (step 2) -
To plan methods of implementation to investigate the research question considering:
- Who will be studied
- How subjects will be chosen
- What to measure
- What statistical methods for data analysis
Step 3 of the clinical research process?
Implement the Study
Implement the Study (step 3) -
- To implement the plans designed in steps 1 and 2
2. To collect data - most time-consuming part of the research process
Step 4 of the clinical research process?
Analyze the Data
Analyze the Data (step 4)
- To analyze, interpret, and draw valid conclusions about the obtained data
- To apply statistical procedures to summarize and explore quantitative information in a meaningful way to address research question
Step 5 of the clinical research process?
Disseminate Findings
Disseminate Findings (step 5) -
- To fulfill a responsibility to share findings with the appropriate audience so others can apply the information to clinical practice or to further research
- To pull together of all the materials and apply them to a generalized/theoretical framework
3 types of research:
- Descriptive research
- Exploratory research
- Explanatory research
Descriptive Research -
- qualitative as well as quantitative
2. describe a group of individuals on a set of variables, to document their characteristics
Descriptive research uses what sources?
It uses questionnaires; interviews; direct observation; or the use of databases
Types of descriptive research:
Case report study - document unusual conditions/effect of innovative interventions
descriptive study
Exploratory research -
- observation WITHOUT intervention
2. Finds relationships between factors
Types of exploratory research:
- Cohort studies - prospective to determine status with respect to disease or outcome and their exposure to certain risk factors
- Case-control studies - compare patients with (cases) and without (controls) a disorder or outcome of interest (determines risk relationship between risk factor and disorder)
Explanatory research -
- Experimental with intervention
2. Establish cause and between interventions and outcomes
Type of explanatory research:
- RCT - comparison of an experiment intervention and a placebo
Random assignment
Controls for bias
What type of study is a description of interesting, new and unique cases to build a foundation for clinical science and as a means of sharing special information among researchers
case study
What type of study design provides an overall picture of the group’s characteristics
Descriptive study
cohort study -
- [cohort: a group of individuals who are followed together over time]
- to select a cohort who do not yet have the outcome of interest and follow them to see if they develop the disorder
case-control study -
- [cases: group of those with the disorder]
- [controls: comparison group without the disorder]
- to select cases and controls and look backward in time to determine if the groups differ with respect to their exposure histories
cross-sectional study -
study a cohort of subjects at one point in time and draw conclusions about a population
Longitudinal study -
to follow a cohort of subjects over multiple points in time performing repeated measurements
RCT -
- It assigns subjects randomly to at least two comparison groups
- It provides the strongest evidence for cause and effect relationships
- It is considered the gold standard of true experimental designs
Continuous data -
ex:
Data with numeric values
ex: Age, Weight, Height, A1c level, test score
Categorical data -
ex:
Data with categorical values
ex: Gender, Race, Exposure/Disease status
Normal distribution -
symmetrical around the mean (bell shaped)
Central limit theorem -
as the sample size tends to infinity, the sample mean is normally distributed
What % is between:
1SD
2SD
3SD
1SD - 68%
2SD - 95%
3SD - 99.7%
Bimodal distribution -
suggestive of 2 different populations
skewed to the right (right skewed) -
tail of distribution is longer on the right
mode
skewed to the left (left skewed) -
tail of distribution is longer on the left
mean
What are three measure of central tendency for continuous data?
- Mean (sum of values/total number of values)
- Median (value in the middle of ranked data)
- Mode (value that occurs most often
What are the measures of dispersion for continuous data?
- Variance
- SD
- Standard error of the mean
Graphic representation of continuous data?
- Histogram
- Box Plot
- Line graph
What are 2 descriptive stats for categorical data?
- Frequency
2. Proportion
Graphic representation of categorical data?
- Pie chart
2. Bar graph
What are stats used to make decision to answer the study question?
Inferential stats
What is the difference between the population group and the sample group?
- Population group: the complete collection to be studied
2. Sample group: a part of the population of interest selected for study
Is it more practical to test population group or sample group?
Sample group
What is the numerical property of population group?
Parameter (population mean, population proportion)
What is the numerical property of sample group?
statistic (sample mean, sample proportion)
Make an inference for a population group from a sample group using what?
- Statistical Hypothesis Testing (SHT)
2. Effect Size with Confidence Interval (CI)
Null (Ho) hypothesis -
hypothesis of no effect or no difference
Alternative (Hi) hypothesis -
hypothesis of some difference or effect
Effect -
difference between the population value and the null hyp value (mean diff between outcome for treatment group and control group)
Significance level (alpha) -
- standard defined by the probability of rejecting a true null hypothesis (false positive)
- .05 of .01
If you have a lower alpha, it requires (weaker/stronger) evidence for you to reject the null hyp?
Stronger
p-value -
quantifies how consistent your sample statistics are with the null hypothesis
Is the p-value (low/high) if sample results are consistent with a null hypothesis that is true
high
Is the p-value (low/high) if sample results are not consistent with a null hypothesis that is true
low
If the significance level = .05 and the p-value is
reject the null hypothesis and accept the alt hyp
If the significance level = .05 and the p-value is >.05, you would…
fail to reject the null hyp
Type 1 error -
false positive (null hypothesis is incorrectly rejected)
Type II error -
false negative ((null hypothesis is incorrectly not rejected)
Power -
true positive
- goes up when the sample size is larger with all other conditions kept
ex: H0: Drug A and B are equally effective H1: Drug B is more effective than A Your study uses a significance level of alpha = 0.05. The power of the test was 0.80.
If the alternative hypothesis is actually true, what is the probability that the study will show a significant difference in efficacy between the two drugs?
power = 0.8
EX:
The type I error of this study was 0.078. Which does the analysis represent for this study?
It represents the probability of incorrectly rejecting the null hypothesis
Confidence interval -
quantifies the uncertainty in the estimates
If your CI is narrower, what does that imply about your estimate?
narrower interval implies higher precision with less variability
If your CI is wider, what does that imply about your estimate?
wider interval implies lower precision with increased coverage
How do you make inference about population from the sample based on the CI?
- Check if the CI contains a value of no effect (a null value)
- It fails to reject H0 if the CI contains the null value
- It rejects H0 if the CI doesn’t contain the null value
What is the null value for the mean difference?
EX: The mean difference of SBP was 2 and its 95% CI was (0.5, 4.5). Is the mean difference of 2 significant?
0
Yes, mean diff of 2 is sig because null value of 0 not found in CI, reject null
What is the null value for the ratio?
Ex: The odds ratio of patients with and without breast cancer for having had diet with low rate of Vitamin A was 2 and its 95% CI was (0.5, 4.5). Is the odds ratio of 2 significant?
1
No, null value of 1 is found in the CI so fail to reject null
EXample: The systolic blood pressures (in mmHg) of 20 women between the ages of 20 and 35 were measured before and after the administration of a newly developed oral contraceptive. The mean change of systolic blood pressure was 2 and its 95% confidence interval was (0.5, 4.5). Which of the following is a CORRECT inference made based on the computed mean change of systolic blood pressure with its 95% confidence interval?
statistically significant, because it does not contain 0
Example: In a study of the epidemiology of breast cancer and the possible
involvement of dietary fats, the odds ratio of patients with vs. without breast cancer for having had diet with low rate of Vitamin D was 2 with the associated 95% CI of (0.5, 4.5). Which of the following is a CORRECT inference made?
not-significant because the associated CI overlaps 1, the null value of odds ratio
What test would you use to compare the means of two independent groups?
Student’s t-test
What test would you use to compare the means of three or more independent groups?
Analysis of variance (ANOVA)
What test would you use to compare the proportions/ratios between independent groups?
Chi-square test (X^2)
EX: what test would you use to compare the mean BP among those three different hospitals?
Analysis of variance (ANOVA)
EX: what test would you use to compare the percentage over those three different hospitals who have essential hypertension?
Chi-square test (X^2)
EX: what test would you use to compare the mean blood pressure (BP) between men and women?
Student’s t-test