Karteikarten
HRSS THEORY EOS
Karten in diesem Stapel (58)
- 1What are Cross-sectional study design features?- Also known as a cross-sectional analysis, transversal study, or pervalance study
- Is observational
- Is descriptive
- Collects data from a population at one specific point in time
- Groups determined by existing differences, not random allocation - 2Advantages of cross-sectional study design?1. Snapshot of a population at one point in time
2. Can draw inferences from existing relationships or differences
3. Can use large numbers of subjects
4. Relatively inexpensive
5. Can generate odds ratio, absolute risk, relative risk an prevalence - 3Disadvantages of cross-sectional study design?1. Results are static. No indication of a sequence of events or historical or temporal contexts
2. Does not randomly sample
3. Cannot establish cause and effect relationships - 4Ethical, Cultural, Practical considerations must promote?o Aims of research (knowledge, truth, and avoidance of error)
o Values that are essential to collaborative work (trust, accountability, mutual respect, and fairness)
o Public support for research
o Moral and social values (social responsibility, human rights, animal welfare, compliance with the law, and public health and safety) - 5CASP Appraisal Guidelines - 4 braod issues to be considered when appraising the report of a descriptive/cross-sectional study and thinking about the associated statistical analyses?- Are the results of the study valid?
- Is the data analysis reasonable?
- What are the results?
- Will the results help locally? - 6Background - the 5 P'sPopulation (who is the focus of this paper?)
Problem (which specific issue is the paper addressing?)
Prevalence (how common is it/how many people are impacted?)
Positive & negative clinical implications (why is this review and/or project important?)
Proposal (what do you propose about this situation?) - 7Pearson's product-moment correlation analyses were...Conducted on the continuous outcome variables to examine their inter-relationships
Further can be conducted to examine whether the continuous outcome variables were associated with the set of independent variables - 8Pearson’s correlation coefficient?Measures linear
relationship between two variables with ρ=0 suggesting ‘no
linear’ relationship - 9Aspects of Regression Modelling?To investigate whether an association exists between the variables of interest
• To measure the strength (as well as direction) of an association between the variables
• To study the form of relationships: - 10What examination (regression) for a continuous outcome and a categorical outcome?For a continuous outcome, relationships can be examined by linear or non-linear regression models
For a categorical outcome, logistic regression is usually used to examine possible relationships - 11Positive and Negative AssociationA positive slope implies upward slopping line and
a positive association
A negative slope implies downward slopping line and a
negative association - 12Linear regression data considerations - The response or outcome variable must be? - The independent variables can be? in linear regression, what do we test?o The response or outcome variable [DV] must be continuous (eg. Weight, balance measure)
o The independent variables can be categorical or continuous, or a combination of both
We test the null hypothesis of no relationship between the DV and the IV. - 13Assumptions of linear regression?- The relationship between DV and IVs is linear
- The observations are independent and randomly selected
- Homogeneity of variances – constant variance
- The residuals (differences between observed and predicted observations) are independent and normally distributed
- The effects are additive
- Absence of outliers and multi-collinearity - 14What are the data exploration steps prior to examining the data for relationships?Descriptive statistics of all variables
Distribution of outcome variables using histogram, QQ plot, normality tests
Scatterplot to examine linearity
Collinearity diagnostics
Appropriate transformation for normality if an outcome variable is not normally distributed
Use median, range, inter-quartile range to summarise non-normal data - 15What is multi-collinearity? In presence of multi-collinearity, regression models...In regression analysis, multi-collinearity refers to IV's that are correlated with other IV's.
may not give valid estimates of the individual predictors. - 16What is a variance inflation factor?measure of how much the variance of the estimated regression coefficient is “inflated” by the existence of correlation among the IVs in the model
o VIF = 1 -> no correlation among the predictors
o VIF > 4 warrants further investigation
o VIF > 10 are signs of serious multicollinearity - 17Constant variance?- If a plot of the fitted values against the residuals scattered randomly around zero, this is an evidence of homoscedasticity
- In addition, we can use statistical tests to examine constant variance (homoscedasticity) assumption
- An insignificant p-value (P > 0.05) of the homoscedasticity test supports the constant variance assumption. - 18If the outcome data are not suitable for any transformation to make them normal and/or sample is small, alternative options to explore relationships include?Spearman Rank-correlation coefficients
Quantilte regression - 19What descriptive statistics would you use to describe a skewed ariable?Median, range, inter-quartile range
- 20Observational studies? - Observed in? - groups of subjects that are compared are? - subjects may be? - there is no?- Subjects are observed in their natural state
- The groups of subjects that are compared are self-selected eg. Manual workers vs. non-manual workers or subjects with and without disease
- Subjects may be measured and tested (eg. Disease status ascertained) but there is no intervention or treatment (eg. Patients not allocated to different exercise programs, or to new drug or placebo) - 21Cohort Longitudinal studies - Population of subjects is identified by?a common link (eg. Living in the same geographical area, working in the same environment, attending the same clinic, diagnosed with the same condition / disease eg. Brain injury).
- 22Cohort studies consider factors not?under the control of the researcher
- 23Strengths & Weaknesses of Cohort studiesStrengths:
- We can find out incidence rate and risk.
- More than one disease related to single exposure
- Can establish cause-effect
- Good when exposure is rare
- Minimises selection and information bias
Weaknesses:
- Losses to follow-up
- Often requires large sample
- Ineffective for rare diseases
- Long time to complete
- Expensive
- Ethical issues - 24Chi-square test? Assumption?Can be used when you want to see if there is a significant relationship between two categorical variables.
Assumption: Expected frequency in each cell >/=5 - 25Fisher's exact testWhen you want to conduct a chi-square test but one or more of your cells has an expected frequency less than 5.
- 26Independent Samples t-test? What about ANOVA?used when you want to compare the means of a normally distributed continuous variables for TWO groups
ANOVA when comparing more than two groups - 27Odds RatOiomeasure of strength of association between an exposure and an outcome.
- 28OR = 1? OR > 1? OR < 1? CI of an OR contains 1?Exposure does not affect odds of outcome
Exposure is associated with higher odds of outcome
Exposure is associated with lower odds of outcome
Relationship is likely to be insignificant. - 29Tests eg. chi-square tests, t-tests and odds ratios) cannot?adjust the associations for the potential effects of other covariates
This is where logistic regression comes in! - 30Logistic regression is... addresses the same research questions that multiple regression does but...a regression with an outcome variable that is categorical and independent variables that can be a mix of continuous and/or categorical values
WITH no distributional assumptions - 31Logistic regression identifies...factors that determine whether an individual is likely to benefit from a certain type of rehab program
- 32LR assumptions- Ratio of cases to variables – using categorical variables requires that there are enough responses in every given category
o Linearity in tlf there are too many cells with no responses, parameter estimates and standard errors may be extremely large.
o This can also make maximum likelihood estimation (MLE) of parameters of the model impossible
- Linearity in the logit – the regression equation should have a linear relationship with the logit form of the outcome
- Absence of multicollinearity and outliers
- Independence of residuals - 33Logistic Regression models for different types of outcomes?Dichotomous outcome - Binary logistic regression
Polychotomous outcome - Multimodal logistic regression
Ordered outcome - Ordinal logistic regression - 34Influential cases?can be identified by plotting the standardised residuals against the index participants
- influential case suggests further examination - 35Goodness of Fit?Hosmer-Lemeshow goodness of fit test can be used to examine whether the estimated LR model fits the sample data.
- P-value 0.05 suggests a good fit. - 36Receiver Operating Characteristics- ROC (receiver operating characteristics) curve can be used to measure predictive accuracy of the fitted model
- 37Area = Predictive power?- Area = 0.5 -> no predictive power
- Area = 1 -> perfect predictive power - 38What is RCT?- Individuals are allocated at random to receive one of a number of interventions (minimum 2)
o All participants have equal chance of allocation to each intervention
o NOT determined by research
o NOT predictable
- Experimental
- Often comparative studies to answer the question as to whether an intervention is effective compared to something else - 39Historical controls? (RCT)- Compare results of new treatment on new patients (tested prospectively) with records of previous results of a historical group (historical controls) who received the standard treatment
- 40Historical controls - What are differences between old and new patients?* Change in patient population over time (?healthier or less healthy)
* Change in diagnostic criteria over time
* Changes in international coding systems
* Overall standards of patient management improve over time
* Differences in data quality between groups - 41Non-randomised concurrent control?- There are two groups each one receiving a different treatment roughly at the same time. Less costly option to RCTs, relatively simple
We could end up with groups that are not comparable.
Subconscious bias in allocation to either group.
No control of potential confounding factors. - 42Quasi-randomised design- Allocation to intervention is not truly random. For example, alternative allocation, allocation by order of enrolment etc, day of enrolment
(Subject to contamination by confounding variables)
(Without proper randomisation, statistical tests can be meaningless as unexpected factors might affect results) - 43Purpose of Random Allocation?- Gives “gold standard” evidence
- Eliminates bias in treatment assignment
- Covariates are equally distributed across groups at baseline – unbiased distribution of confounders
- Experimental and control groups are treated exactly the same, except for the intervention received
- Facilitates blinding of treatments from investigators, participants, assessors
- Allows researchers to make causal inferences
o Randomisation ensures any differences between groups is due to chance
- BUT
o Expensive: time and money
o Ethical considerations
o Hawthorne effect - 44Types of RCTs - Parallel trials?o Researcher (i) selects sample from population; (ii) measures baseline variable; (iii) randomises participants; (iv) applies intervention; (v) measures outcome variables and analyses results
- 45Types of RCTs - Crossover RCTo Researcher (i) selects sample from population, (ii) measures baseline variables; (iii) randomises participants; (iv) applies intervention; (v) measures outcome variables; (vi) has a washout period to reduce carryover effect; (vii) applies intervention to previous control group; (viii) measures outcome variables
- 46Types of RCTs: - Factorial RCTo Researcher (i) selects sample from population; (ii) measures baseline variables; (iii) randomly assigns 2 active interventions and controls to 4 groups (iv) applies intervention (v) measures outcome variables.
- 47RCTs provide causal inferences about the effect of the intervention if?- If we randomise sufficient participants and avoid selection and performance bias
- 48Integrity of RCT can be undermined by:o Inadequate generation of randomisation sequence
o Inadequate concealment of allocation
Possible selection bias if participants or researchers can predict which treatment participant is receiving.
o Inadequate blinding/masking
Potential performance bias if the intervention researchers provided to intervention and control group is not equal
Potential detection bias if outcome assessors know what treatment the participant has received
o Excluding participants or significant attrition
Attrition bias if there are systematic differences between groups in withdrawals and incomplete data
o Analysing participants in the wrong group
o Selective reporting of outcomes eg. Only significant findings
Reporting bias if there are systematic differences between reported and unreported findings - 49Ethical considerations- Social and clinical value
- Scientific validity
- Fair subject selection
- Favourable risk-benefit ratio
- Independent review
- Informed consent
- Respect for potential and enrolled subjects - 50Sample size vs. effect size- Sample size is inversely associated with effect size
o Effect size = 0.5 -> n = 102
o Effect size = 0.2 -> n = 620 - 51What is Intention To Treat Analysis?- An intention to treat analysis compares treatment groups as originally allocated (as per random allocation), irrespective of whether patients received or adhered to their treatment protocol.
ITT promotes external validity as it is a pragmatic approach that aims to evaluate the effectiveness of an intervention in routine practice - 52What is Per Protocol Analysis?- Per protocol analysis compares treatment groups as originally allocated but includes only those patients who completed the treatment protocol, which compromises the internal validity of the findings.
- 53Longitudinal data structureObservations are collected from each individual on multiple occasions or at multiple time points.
Each subject has multiple observations.
Observations from each subject are clustered within the subject.
Not necessarily require the same number of observations on each subject - 54Data Analysis in repeated measuresRepeated measures of analysis of variance
Longitudinal data analysis (mixed effects modelling, marginal modelling) - 55Features of RMANOVA?o Complete case analysis
o Assume everyone is measured at the same time and at equally spaced time intervals
o Require restrictive assumptions about the correlation structure
o Does not provide parameter estimates (just p-values)
o Cannot handle time-dependent covariates (predictors measured over time) - 56Longitudinal Data Anaylsis helps us to...o Assess changes in a response variable over time
o Measure temporal patterns of response to treatment
o Identify factors that influence changes
o Include time-varying predictors in the model
o Investigate causality
o Better handling of missing data (all available data - 57Generalised Estimating Equations?- Generalised Estimating Equations (GEE) is an extension of the general linear model (GLM) of statistical regression for modelling clustered or correlated data
- GEE offers robust estimates of standard errors to allow for clustering of observations
- GEE produces consistent estimates of regression coefficients and their standard errors
- GEE can deal with both normal and non-normal outcome data
- GEE is useful when the aim is to investigate differences in population averaged responses - 58LDA reduces...- In LDA, the unexplained variation is divided into components, making the ultimate error variance smaller
- LDA reduces unexplained variability in the responses, which provides better estimates of the effects
Informationen
- Beschreibung
- Karteikarten am HRSS THEORY EOS, erstellt von Sam Kim am 11/11/2017.
- Erstellt von
- Sam Kim · Nov 2017
- Aktualisiert
- 16 Nov 2017
- Aufrufe
- 1
- Bewertung