Critical numbers Flashcards
Define prevalence.
Number of existing cases in a defined population at a defined point in time, divided by the number of people in the population at that time.
Define risk
The number of new cases in a defined population at risk during a specified time people.
The defined population at risk does NOT include existing cases
What is the secondary attack rate?
Number of exposed people developing the disease within the range of the incubation period ÷ total number of people who are exposed/susceptible x 100
Define odds.
Number of new cases in a specified time period ÷ number who did not become a case during that time period
Define incidence rate.
Number of new cases in a specified time period ÷ total person-time at risk during that time period.
From what study type can you not calculate a risk ratio?
Case control studies
What is a p value?
A calculated probability of finding the observed or more extreme results when the null hypothesis of a study question is true. Provides an estimate of the probability that the results are due to chance.
What p value suggests statistical significance?
0.05
What is a null hypothesis?
A statement that proposes that no statistical significance exists in a set of given observations.
The null hypothesis is set up in opposition to an alternative hypothesis and attempts to show that no variation exists between variables, or that a single variable is no different than its mean.
What is the alternative hypothesis?
Opposite of the null hypothesis. Usually the hypothesis that you are trying to prove.
Name four types of linear regression
- Linear regression
- Logistic regression
- Poisson regression
- Cox regression
What do the four types of of regression estimate or model?
Linear regression = estimated mean differences between groups
Logistic regression = binary outcomes, models an odds ratio
Poisson regression = models rate ratios
Cox regression = models hazard ratios
What is regression analysis?
A powerful statistical method that allows you to examine the relationship between two or more variables of interest. Helps to understand how the typical value of the dependent variable changes when one of the independent variables is varied and the others remain fixed.
What would a relative risk of 1 suggest?
Incidence in the exposed group is the same as incidence in the non-exposed group.
No increased risk = no association
What would a relative risk of <1 suggest?
That incidence in the unexposed group is greater than incidence in the exposed group.
Decreased risk = negative association
What would a relative risk of >1 suggest?
Incidence in the exposed group is greater than incidence in the non-exposed group.
Increased risk = positive association
Define bias.
A systematic error in studies that lead to an error in conclusions or skewed results.
How do you eliminate/reduce bias?
Blinding
What is blinding?
When patients/subjects do not know which treatment they are receiving in order to ensure that the results are not affected by the placebo effect (power of suggestion).
What is double blinding?
Both the researchers and the subjects/patients do not know which treatment they are receiving.
What is simple random sampling?
A process in which each member of the population has an equal probability of being selected.
What is systematic sampling?
Members of the population are selected at equal intervals
What is stratified sampling?
The population is partitioned into strata (groups) and a sample is randomly selected within each strata.
What is cluster sampling?
The population are partitioned into clusters (groups) and a sample of clusters is selected. All people in the selected clusters are included in the sample.
What is absolute risk?
The number of people experiencing an event in relation to the population at risk.
What is relative risk?
The probability of an event occurring in the exposed group vs the probability of the event occurring in the non-exposed group.
What is a confidence interval?
A range of values that is likely to contain the population parameter of interest.
What would a 95% confidence interval suggest?
That you can be 95% certain that the interval contains the true mean of a population
Is a test significant if the 95% confidence interval contains zero?
No.
What is absolute risk reduction?
The amount by which a therapy reduces the risk of a bad outcome.
What is absolute risk increase?
The amount by which an exposure e.g. therapy increases the risk of a negative outcome.
How would you calculate absolute risk reduction?
Control event rate - experimental event rate
e. g. if a drug reduces the risk of a bad outcome from 50% to 30% the ARR is:
0. 5-0.3 = 0.2 (20%)
How would you calculate absolute risk increase?
Control event rate - experimental event rate
e.g. gastric band surgery caused an increase in mortality to 60% from 40%
ARI = 0.4 - 0.6 = 0.2 (20%)
What is number needed to treat (NNT)?
The number of patients who need to be treated to prevent one additional bad outcome/ for one patient to benefit
What is the number needed to harm (NNH)?
The number of patients that need to be treated (or exposed to a risk factor) in order for one patient to have a particular adverse effect.
What is the equation for NNT?
1/ARR
ARR = control event rate - experimental event rate
What is the equation for NNH?
1/ARI
ARI = control event rate - experimental event rate
What is an observational study?
Recording of outcomes without intervening in the care of the patient in any way other than what is routine clinical care.
What is an experimental study?
Recording outcomes following planned intervention in the care of patients.
What types of study would be classed as a DESCRIPTIVE observational study?
- Case report series
- Ecological studies
- Cross sectional study
What types of study would be classed as an ANALYTICAL observational study?
- Cohort study
- Case-control study
- Cross sectional study
Define ecological study.
Studies of risk modifying factors on health or other outcomes based on populations that are defined geographically or temporally.
Define case-control study.
A type of observational study in which two existing groups of people, differing in outcome, are identified and compared on the basis of some supposed causal attribute.
Often used to identify factors that may contribute to disease by comparing people who have the condition (cases) to those who do not (controls) but are otherwise similar.
Define cohort study.
A type of longitudinal study that sample a group of people who share a defining characteristic, performing a cross-section at intervals through time.
Define randomised control trial.
A study in which a number of similar people are randomly assigned to two or more groups to test a specific drug, treatment or other intervention. The experimental group receives the intervention that is being tested and the other group (control) receive and alternative treatment.
Define a cross sectional study.
A comparison of different population groups at a single point in time.
What should a good study have?
- Randomisation of participants to interventions
- Show causation rather than association
- Have outcome measures/results for at least 80% of the population
What is a confounder?
A variable that is associated with the exposure and the outcome of interest.
What criteria must a variable meet in order to be defined as a confounder?
- Associated with the outcome
- Associated with the exposure
- Is not an intervening variable between the exposure and the outcome (i.e. it must not be a consequence of the exposure).
What are the advantages/disadvantages associated with an ecological study?
Advantages:
- Uses routinely collected data
- Units of analysis are populations
- Few ethical issues
Disadvantages:
- No link between individual exposure and effect
- Bias
- Absence of records for individual attributes
What are the advantages/disadvantages associated with a case-control study?
Advantages:
- Number of subjects is kept to a minimum (good for rare diseases)
- Relatively inexpensive
- Results can be obtained quickly as there is no need to wait for the disease to develop
Disadvantages:
- Selection and information bias
- Retrospective - tendency for participants memory to be selective and records of past events may be incomplete
- Because it is retrospective it is difficult to determine whether an association is causal or not.
- Difficulty in choosing controls
- INCIDENCE of a disease within a population CANNOT BE CALCULATED
What are the advantages/disadvantages associated with a cohort study?
Advantages:
- Possible to distinguish antecedent causes from concurrent associated factors
- Incidence can be determined which allows attributable, absolute and relative risks to be calculated
- Less chance of bias since exposure is measured before the development of disease.
Disadvantages:
- Cannot be certain that exposures are causal
- Long periods of study and large populations = expensive
- Problems with follow-up especially if study is long
- Diagnosis of cases may change over the years as medical science becomes more advanced.
What are the advantages/disadvantages associated with a RCT study?
Advantages:
- Randomisation means that confounding factors (age, sex etc) are equally distributed = minimises confounding
- Minimes bias
Disadvantages
- Tend to be large and expensive trials - often multicentre
- Volunteer bias
- Ethical difficulties in withholding treatment from the control group or offering what is seen to be an inferior treatment
What are the advantages/disadvantages associated with a cross sectional study?
Advantages:
- Quick, cheap
- Rapid feedback of current events in the community
- Results used to generate hypotheses
Disadvantages:
- Could just be reporting medical oddity
- Prone to bias
- No time reference
What are the five most important aspects of evidence based medicine?
- Asking focused questions
- Finding evidence
- Critical appraisal
- Making a decision
- Evaluation
What is considered to be the gold standard of evidence based medicine?
Systematic reviews of randomised control trials.
Define systematic review.
Systematic methods are used to identify, select and analyse relevant research in order to answer a specific question.
What is critical appraisal and why is it important?
Critical appraisal is about assessing validity, reliability and applicability. It is important because it means you can provide your patients with the best possible evidence and information.
What are the advantages and disadvantages of systematic reviews.
Advantages: can prove causation, compares similar studies
Disadvantages: time consuming, expensive.
What are the three types of cohort studies, depending on when the data was collected?
- Concurrent - groups of exposed and unexposed individuals have been assembled but the disease has not occurred yet
- Retrospective - The investigation begins at a point in time after both exposure and outcome have already occurred
- Ambidirectional - data are collected both retrospectively and prospectively in the same cohort. This type is applicable for exposures having both short-term and long-term effects
What is standard deviation?
A measure of variability. When we calculate the standard deviation of a sample, we are using it as an estimate of the variability of the population from which the sample was drawn.
What is standard error?
Tells you how accurate the mean of any given sample from that population is likely to be compared to the true population mean
What is positive confounding?
When confounding exaggerates or over-estimates the true effect
What is negative confounding?
When confounding hides/under-estimates the true effect.
When does standard error increase?
When standard deviation, i.e. the variance of the population, increases.
When does standard error decrease?
Standard error decreases when sample size increases – as the sample size gets closer to the true size of the population, the sample means cluster more and more around the true population mean
EXAMPLE - CALCULATING SECONDARY ATTACK RATES
Consider an outbreak of the plague in which 18 people in 18 different households became ill. If the population of the community was 1000, then the overall attack rate would be 18/1000 x 100 = 1.8%
One incubation period later, 17 persons in the same households of the primary cases developed the plague. If the 18 households included 86 persons, calculate the secondary attack rate.
86-18 = 68 (17/68) x 100 = 25%
Give two examples of non-random sampling?
- Convenience sampling
2. Purposive sampling
What is nominal data?
A qualitative classification of data - a naming system.
What is ordinal data?
When data has numerical scores existing in order.
Likert Scale is a popular ordinal data example. For a question such as: “Please express the importance pricing has for you to purchase a product.”, a Likert Scale will have the following options which are coded to 1,2,3,4 and 5 (numbers). 1 is lesser than 2, which is lesser than 3, which is lesser than 4, which in turn is lesser than 5.
What is binary data?
Data that can only take one of 2 possible states. E.g. alive or dead.
What is discrete data?
Data with a finite number of values/ data that can only take certain values e.g. Number of people living in a house (you can’t have half a person!)
What is continuous data?
Data that can take any value e.g. height.
Define morbidity.
Suffering from a disease.
Define mortality.
The number of deaths per year for a specified disease.
Define clinical significance.
The results are significant enough to be worthwhile clinically. They will have a genuine effect on day to day life.
Define statistical significance
The results are used to accept or reject the null hypothesis. They are not necessarily clinically significant.
How is a 95% confidence interval calculated?
Sample statistic + or - 1.96 x standard error
How is a 90 % confidence interval calculated?
Sample statistic + or - 1.65 x standard error
How is a 99% confidence interval calculated?
Sample statistic + or - 2.58 x standard error
How is standard error calculated?
SD/square route of the number of people
How would a confidence interval change if a 99% level of confidence was used?
It would be wider and more CI from possible samples would contain the population incidence
How would a confidence interval change if a 90% level of confidence was used?
It would be narrower and fewer than 90% of the CI would contain the population proportion
If the confidence intervals do not overlap and do not contain the number 0, does this imply statistical significant?
Yes.
How do you calculate an incidence rate?
no. of new cases during time period/ total person time (size of population at risk x average duration of follow up)
What does ‘adjusted mean’?
Multi-variate analysis = observation and analysis of more than one statistical outcome variable at a time
What is the affect of adjustment on the odds ratio?
Adjusted OR will decrease if the covariates tend to increase the incidence of disease and will increase if the covariates decrease the incidence.
How would you work out incremental-cost effectiveness ratio (ICER)?
ICER = (cost 1 - cost 2) / (effectiveness 1 - effectiveness 2)
