Statistics Flashcards
Which is the most appropriate analysis to establish the risk-benefit profile of new drug ??
Intention to Treat (ITT)
- includes all pts. post-randomization
- Provides true estimate of risk-benefit analysis
Draw back of Completer analysis ??
- It discounts pts. who have dropped out due to S/E
What is Pearson’s correlation coefficient ??
It is the covariance of the 2 variables divided by the product of their Std. deviation.
- Has a value b/w 1 & -1
- Indicates if it has a +ve / -ve relation b/w the 2 variables
- > 0.5 = strong correlation
What is Mann-Whitney test ??
- Non-Parametric test
- Compares similarities of 2 different population
What is Fisher’s exact test ??
Is used to analyse contingency tables (frequency distribution of multiple variables)
What is Chi Squared test ??
- Used to detect if there is a significant difference b/w Observed & Expected frequencies of a particular result in a set of data
What is ANOVA ??
- Compares differences b/w Means of groups of data
- eg.- the results from the same experiment performed 3 different times
What type of a study is it called when
- Investigator assigns exposure ??
- Investigator does’t assign exposure ??
- Experimental Study
- Observational Study
In an Observational study, what is it called when
- Comparison grp. is used ??
- Comparison grp. not used ??
- Analytical Study
- Descriptive Study
Name the types of Analytical study based on
- If Exposure is the start point ??
- If Outcome is the star point ??
- If both Exposure & Outcome are considered as start point together with Measurements ??
- Cohort Study (Observational & Prospective)
- Case-Control Study (Observational & Retrospective)
- Cross-sectional Study (provides a snapshot, sometimes called Prevalence studies)
Name the type of Descriptive based on
- If Reporting is done
- If Measurements are taken with Considering both Exposure & Outcome as start point ??
- Case report/ Case series/ Case study
- Cross-Sectional studies
What are the types of Experimental study based on
- If Randomization done ??
- If Randomization is not done ??
- Randomized Control Trial
- Non-Randomized Control Trial
What are the levels of evidence from Best to the least
1a : Evidence from Meta-analysis of RCT
1b : Evidence from at least 1 RCT
2a : Evidence from at least one well designed Non-RCT
2b : Evidence from at least 1 well designed Experimental study
3 : Evidence from Case, Correlation & Comparitive studies
4 : Evidence from Panel of Experts
What are the grades of Evidence ??
Grade A : Evidence from at least 1 RCT (ie. 1a or 1b)
Grade B : Evidence from Non-RCT (ie. 2a, 2b, 3)
Grade C : Evidence from a Panel of Expert (ie. 4)
When is a New drug is assumed to have an equivalent effect to the existing Rx ??
Equivalence margin is defined as
- [- delta to + delta]
If the Confidence Interval of the difference b/w the 2 drugs lies within this margin,
When is a drug called Non-inferior to the existing Rx. ?
Lower CI needs to lie within the equivalence margin (ie. - delta)
- Small sample sizes are needed for these trials
- Once it is shown to be non-inferior, large studies may be performed to show Superiority
What is the necessary condition for a new drug to be considered for Rx ??
Ideally it should show Superiority over the existing drugs
- BUT even if the new drug is Equivalent or even Non-Inferior, then they may compete on price or convenience
Define the following statistical terms
- Mean
- Median
- Mode
- Range
- Average of series of observed values
- Middle value when series of observed values are placed in an order
- The value that occurs most frequently within a dataset
- Largest value - Smallest value
Name the types of Data
Nominal
Ordinal
Discrete
Continuous
Binomial
Interval
Define the following
- Nominal data ??
- Ordinal data ??
- Discrete data ??
- Observed value can be put into set categories which have no particular order or hierarchy. You can count but NOT order/ measure it eg. Birthplace
- Observed values can be put into set categories which themselves can be ordered. eg- NYHA classification
- Observed values are confined to certain values, usually a finite no. of whole no. eg.- No. of Asthma exacerbations in a year
Define the following
- Continuous data ??
- Binomial data ??
- Interval data ??
- Data can take any value with certain range. eg.- Weight
- Data can take 1 or 2 values. eg- M/F
- Measurement where the difference b/w 2 values is meaningful, such that equal differences b/w values correspond to real differences b/w the quantities that the scale measure. eg- Temperature
What is Variance ??
It is a measure of spread of scores away from the mean
Variance = Square of SD
What is Pre-test probability ??
The proportion of people with the target disorder in the population at risk at a specific time (Point prevalence) or time interval (Period prevalence)
eg.- Prevalence of RA in the UK is 1%
What is Post-test probability ??
The proportion of pts. with that particular test result who have the target disorder
Post-test P = Post-test Odds/ (1+ Pre-test Probability)
What is Pre-test odds ??
The odds that the pt. has the target disorder before the test is carried out
Pre-test odds = Pre-test P/ (1- Pre-test P)
What is Post test odds ??
The odds that the pt. has the target disorder after the test is carried out
Post-test odds = Pre-test odds * Likelihood ratio
What is the formula for Likelihood ratio ??
Likelihood ratio for a (+)ve test result = Sensitivity/ (1 - Specificity)
What is Absolute risk ??
It is the likelihood or probability of an event or outcome occuring
Risk = Event Rate = No. of event/ Grp. total
What is Relative Risk or Relative Risk Ratio ??
It is the Ratio of Risk in the Experimental grp (Experimental Event Rate, EER) to risk in the Control grp (Control Event Rate, CER)
What is EER & CER ??
EER : Rate at which events occur in the experimental group
CER : Rate at which events occur in the control group
What is the significance of
- Risk ratio > 1 ??
- Risk ratio < 1 ??
- Risk ratio = 1
- Rate of an event is increased compared to control group. It is therefore appropriate to calculate the Relative Risk Increase if required
- The Rate of an event is decreased compared to control group. So, Relative Risk Reduction should be calculated
- Risk is same in both Event group & control group
When & how is RRR & RRI calculated ??
RRR is calculated when Risk Ratio is < 1
RRR = 1 - RR = 1 - ART/ARC
RR is Relative Risk
ART is Attributable Risk of Rx
ARC is Attributable Risk of Control
RRI = (ECR - CER)/ CER
What is NNT ??
The number of pts. who need to be treated for 1 patient to benefit
LOWER No. = Better Rx.
NNT = 1/ ARR
ARR = [c/(c + d)] - [a/ (a + b)]
What is NNH ??
The number of pts. who need to be exposed to a risk factor for 1 pt. to be hARmed.
HIGHER No. = Safer exposure
NNH = 1/ AR
AR = [a/(a + b)] - [c/(c + d)]
What is Hazard Ratio ??
A measure of how quickly or slowly an event of interest (eg.-disease onset or death) occurs in 2 groups
- Commonly used in Survival analysis, especially when studying Time-to-event data
Mention a few applications of Hazard ratio
Used in Cohort studies & Clinical trials where the timing of events is crucial & participants are followed until the event occurs
Difference b/w Hazard ratio & Relative risk ??
Both are very similar but HR is used when the time to risk occurrence is significant
HR: Emphasizes on TIMING of events
RR: Emphasizes OCCURRENCE of events
What is HR formulae & its Interpretation ??
HR = HR in Exposed Grp/ HR in Control Grp.
HR = 1; indicates NO difference in hazard rates in b/w 2 groups
HR > 1; a higher hazard (event occurring more rapidly) in exposed grp.,
HR < 1; Lower hazard
What is Odds & Odds ratio ??
Odds are a ratio of the no. of people who incur a particular outcome to the no. of people who do not incur the outcome
Odds ratio is the Ratio of the odds of a particular outcome with experimental Rx & that of control
How is Odds ratio interpreted ??
Typically used in Case-Control Study
Represents the Odds of exposure among cases (a/c) vs odds of exposure among controls (b/d)
OR = 1; Odds of exposure is = in cases & controls
OR > 1; Odds of exposure are greater in cases
OR < 1; Odds of exposure are greater in control
What is the difference b/w Odds & Probability ??
Probability is the fraction of times you’d expect to see an event in many trials
When expressed as a single no., it is always b/w 0 & 1. If we take the eg. of rolling dice
- Probability of rolling a 6 in 1/6 =0.16
- The odds of rolling a 6 is 1/5 = 0.2
What is Null & Alternate Hypothesis ??
It is a hypothesis which proposes that no statistical significance exists in the set of given observation
Null Hypothesis (H0) states that 2 Rx. are equally effective (& is hence negatively phrased) & is assumed to be true unless the evidence is strong enough to reject it
A significant test uses the sample data to assess how likely the null hypothesis is to be correct
Alternate Hypothesis (HI) is the opposite of null hypothesis, ie., There is a difference b/w the 2 Rx.
- If the evidence is strong enough we reject H0 in favour of H1, otherwise we can only say that there is only insufficient evidence to reject H0
What is p value ??
It is the probability of obtaining a result by chance at least as extreme as the one that was actually observed, assuming that the (H0) is true
- It is therefore = to the chance of making a Type I error
What are the types of error while testing Null Hypothesis ??
Type 1 : The H0 is rejected when it is true ie. Showing a difference b/w 2 groups when it does’t exist, a False Positive
- This is determined against a preset significance level (termed Alpha). As the significance level is determined in advance, the chance of making Type 1 error is not affected by sample size
Type 2 : The H0 is accepted when it is false ie. Faiing to spot a difference when one really exists, False (-)ve
- The probability of making a Type 2 error is termed Beta.
- It is determined by both sample size & Alpha
What is the Power of a study ??
The power of a study is the probability of (correctly) rejecting the H0 when it is false, ie., the probability of detecting a statistically significant difference
- Power =1-Probability of Type 2 error
- Power can be increased by increasing the sample size
Name the significant tests
The type of significant test used depends on whether the data is
- Parametric: Something that can be measured, usually normally distributed
- Non Parametric
Name the Parametric tests
Student’s T-test: Paired or Unpaired
- Paired data: Refers to data obtained from a single group of pt. eg.- Measurement before & after an intervention
- Unpaired data: Comes from 2 different groups of pts. eg.- Comparing response to different interventions in 2 groups
Pearson’s product-moment coefficient
- Correlation
Name the Non-Parametric tests
MANN-WHITNEY U Test
- Compares ordinal, interval, or Ratio scales of unpaired data
WILCOXON SIGNED-RANK Test
- Compares 2 sets of observations on a single sample; eg.- a ‘before’ & ‘after’ test on the same population following an intervention
CHI-SQUARED Test
- used to Compare proportions/ percentages eg.- compares the %age of pts. who improved following 2 different interventions
SPEARMAN, KENDALL RANK
- Correlation
Define Normal Distribution & its Properties
aka Gaussian distribution or Bell- Shaped distribution
- Symmetrical: Mean=Mode=Median
- 68.3% values lie within 1 SD of Mean
- 95.4% values lie within 2 SD of Mean
- 99.7% values lie within 3 SD of Mean
Within 1.96 SD of mean lie 95% of the Sample value
Define Confidence Interval ??
Range of values within which the true effect of intervention is likely to lie
- Specified Probability is called Confidence Level
- End points of CI is called C Limit
- 95% CI = range of mean [-1.96 SD to +1.96 SD] ie. If a repeat sample of 100 observations are taken from the same grp., 95 of them would be expected to lie in that range
What is Confidence Level ??
The LIKELIHOOD of true effect lying within the CI is determined by CL
What is
- SD ??
- Variance ??
SD is the measure of how much dispersion exists from the mean
- SD = Square root (Variance)
Variance = [SD] squared
Property of Skewed distribution
Normal G distribution: Mean = Median = Mode
(+)vely Skewed : Mean > Median > Mode
(-)vely Skewed : Mean < Median < Mode
What is Intention-to-Treat ??
ITT analysis is a statistical method used in clinical trials to evaluate Efficacy & Safety of a Rx. intervention & Randomization is maintained.
- Done to AVOID the effects of cross-over & drop-out, which may affect the randomization to the Rx. group
- Participants are analysed according to their original assigned Rx grp., regardless of whether they complete the Rx as intended or if they dropped out or deviated from the protocol
How are Prevalence & Incidence related in cases of
- Acute diseases ??
- Chronic diseases ??
- Prevalence & Incidence are similar; for conditions such as Common cold, the Incidence may be greater than the Prevalence
- Prevalence»_space;»> Incidence
Formulae that relates Prevalence & Incidence ??
P = I * Duration of condition
What is Confounding ??
Factors related to BOTH Exposure & Outcome (but NOT on causal path) distorts effect of outcome or leads to spurious results
Methods to reduce Confounding ??
- CROSSOVER studies (with subjects as their own control)
- MATCHING (pts. with similar characteristics in both Rx & control groups
- ANALYTIC Techniques (eg. regression analysis when confounding variables are known & measured)
What is Effect modification ??
Exposure leads to different outcomes in subgroups stratified by factor
- TRUE association exists
How to mitigate Effect modification bias ??
Stratified analysis (eg. after testing for interaction b/w OCP & Smoking, analyze risk among smokers & Non-smokers
What is a Funnel Plot ??
Primarily used to demonstrate the existence of Publication bias in meta-analysis
- They are usually drawn with Rx effects on the horizontal axis & study size on vertical axis
How is a Funnel Plot interpreted ??
Symmetrical, inverted funnel shape indicates that Publication bias is unlikely
Asymmetrical funnel: indicates a relationship b/w Rx. effects & study size
This indicates either publication bias or a systemic difference b/w smaller & larger studies (‘Small study effects’)
What are the phases of a Clinical trial ??
PHASE 1 : Determines pharmaco-kinetics & Pharmacodynamics & S/E prior to larger studies
PHASE 2 : Assess Efficacy + Dosage
Involves small no. of pts. affected by particular disease
- 2a : assesses optimal dosing
- 2b : assesses Efficacy
PHASE 3 : Assess Effectiveness
- Involves 100 - 1000’s of people, often as part of RCT; comparing new Rx with established Rx.
PHASE 4 : Postmarketing Surveillance
- Monitors long-term effectiveness & S/E
Name in which phases of Clinical trials are the following done
- Study conducted on Healthy volunteers ??
- Conducted on small no. of pts. with the particular disease being studied
- RCT ??
- Monitor long-term effectiveness & S/E ??
- Phase 1
- Phase 2
- Phase 3
- Phase 4
Define Reliability & Validity
Reliability is used in statistics to imply CONSISTENCY of a measure
Validity is determined by whether a test ACCURATELY measures what it is supposed to measure
A measurement can be Valid but not reliable or Reliable but not valid
What is Correlation & Linear Regression ??
Both the terms are related but not synonymous
Correlation is used to test for ASSOCIATION b/w variables (eg. whether salary & IQ are related)
Once Correlation b/w 2 variables has been shown, Regression can be used to PREDICT values of other dependent variables from independent variables
