Epidemiology Flashcards
Define ‘routine data’
Data that are routinely collected and recorded in an ongoing systematic way, often for administrative purposes
What are the advantages & disadvantages of using routine data?
- Advantages
- Relatively cheap
- Already collected/available
- Standardised collection procedures
- Disadvantages
- May not answer the question
- Variable quality
- Not every case captured
What are the coding systems for diagnoses and procedures in hospital?
- Diagnoses- ICD10
- Procedures - OPCS4
What are the advantages to using hospital episode statistics for study?
- Advantages
- Comprehensive
- Unbiased
- Based on case notes
- Disadvantages
- Data may not be complete
- In hospital death only
What are the top 3 causes of UK mortality?
- Heart and circulatory disorders 33%
- IHD 23%
- Stroke 10%
- Cancer 30%
- Trachea, bronchus, lung
- Breast, prostate
- Colorectum
- Respiratory disorders 13%
- Pneumonia
- COPD
What are the top 5 causes of mortality worldwide?
- IHD
- Stroke and other cerebrovascular disease
- LRTI
- COPD
- Diarrhoeal diseases
Define ‘clinical governance’
A transparent system for the continual maintenance and improvement of healthcare standards both on an individual and organisational level within the NHS
What are the elements of clinical governance?
- Education and training
- Clinical audit
- Clinical effectiveness
- Research and development
- Openness
- Risk management
What does the Care Quality Commission do?
- Independently inspect healthcare services against standards set by the DoH
- Investigate serious failures in healthcare services
- Publish regular ratings of NHS trusts
Define ‘clinical audit’
A quality improvement process that seeks to improve patient care and outcomes through the systematic review of care against explicit criteria and the implementation of change
How common is overweight/obesity in the UK and what risks does this confer?
- 20% obese, 50% females overweight
- Nurses’ health study
- overweight -> 3 years less life
- Obese -> 7 years less
- Active lifestlye (30 min walk per day) prevents 30% obesity and 45% new DM
How common are sexual health issues in the UK?
- 10% 16-24 year olds have ≥1 STI
- Increased in urban areas and amongst blacks and minorities
What is the recommended alcohol limit and how many people have alcohol issues in the UK?
14 units/week (1 unit = 10ml EtOH)
5% men and 2% women report problems
How common are mental health problems in the UK?
15% lifetime risk (most commonly depression and anxiety disorders)
What is the hierarchy of evidence?
- Systematic review and meta analysis
- RCTs
- Cohort or case control studies
- Descriptive studies
- Case reports
What is the point of descriptive studies?
Can generate hypotheses and provide frequency data (incidence/point prevalence); can’t determine causation
What are some examples of descriptive studies?
- Cross sectional surveys/census
- Ecological studies
What are case control studies?
Retrospective study of exposure in a case group (with the disease) and a control group (without). The proportion of exposed in each group generates an odds ratio
What are the pros and cons of case control studies?
- Pros:
- Quick and cheap
- Well suited for diseases with long latent periods
- Good to evaluate rare diseases
- Can examine multiple aetiological factors for a single disease
- Disadvantages
- Inefficient for evaluating rare exposures
- Cannot calculate incidence rates
- Temporal relationship between exposrue and disease can be difficult to establish
- Recall bias
- Selection bias
What is a cohort study?
Prospective study comparing development of disease in exposed and non exposed groups. Incidence of disease in each group -> relative risk
What are the pros and cons of cohort studies?
- Pros
- Good for evaluating rare exposures
- Can examine multiple effects of a single exposure
- Can elucidate temporal relationship
- Direct incidence calculation
- Cons
- Inefficient for evaluation of rare diseases
- Expensive and time consuming
- Loss to follow up affects results
What is an RCT?
A planned experiment designed to assess the efficacy of an intervention. Randomisation decreases selection bias and blinding decreases measurement bias
What are the pros and cons of RCTs?
- Pros
- Most reliable demonstration of causality
- Cons
- Non compliance
- Loss to follow up
- Validity depends on quality of study
- Ethical issues
- Selection criteria may limit generalisability
How do you interpret a forest plot?
- Square = OR
- size = size of study
- Line = 95% CI of OR
- Diamond = combined odds ratio
- Width = 95% CI
What are the 5 steps of EBM?
- Question
- Search
- Appraise
- Apply
- Evaluate
What are the Bradford Hill causation criteria?
- Strength of association
- Consistency
- Specificity
- Temporal relationship
- Dose response/biological gradient
- Plausibility
- Coherence
- Experiemntal evidence
- Analogy: if a similar association can be demonstrated in another setting
Define ‘bias’
Inaccurate data due to systematic error in selection, measurement or analysis
What is selection bias?
Systematic difference between the characteristics of those selected for a study and those who were not or differences between study groups
What is measurement bias?
When measurements or classifications of disease or exposure are inaccurate e.g. inaccuracies of instruments, expectations of observers or participants
What is analysis bias?
Error caused by participants being lost to followup or switching treatment groups.
What is confounding and what are some common confounders?
Error in the interpretation of an accurate measurement - a confounder is any factor which is prognostically linked to the outcome and is unevenly distributed between study groups. Not a confounder if it lies on the causal pathway between the variables of interest.
Commonly: age, sex, SES
How can you reduce confounding?
- Randomisation
- Stratification
- Standardisation e.g. SMR
- Ratio of observed:expected deaths
- Expected deaths derived from larger population
- Regression
What is the odds ratio?
=odds of exposure in cases/odds of exposure in controls
OR = ~RR if disease is rare (<10%)
What is the relative risk?
Increase/decrease in probability of a disease given a particular risk factor
=incidence in exposed/incidence in unexposed
What is the attributable risk?
The measure of excess risk due to a factor
What are the modified Wilson criteria?
For determining whether a screening programme is worthwhile:
- Disease
- Important health problem
- Well recognised pre clinical stage
- Natural history well understood
- Test
- Valid and reliable
- Simple and cheap
- Safe and acceptable
- Diagnosis and treatment
- Adequate facilities
- Acceptable treatment
- Early intervention is of more benefit than later treatment
- Cost effective
What is the difference between sensitivity, specificity, PPV and NPV?
- Sensitivity: detect true positives
- Specificity: exclude true negatives
- PPV: how likely to have disease if positive
- NPV: how likely to not have disease if negative
- PPV and NPV depend on disease frequency
What are some biases affecting the effectiveness of a screening programme?
- Selection bias: healthiest come for screening
- Lead time bias: disease diagnosed earlier -> increased survival time
- Length time bias: outcome appears better because more disease with good outcome detected
What is primordial prevention?
Prevention of factors promoting the emergence of risk factors: lifestyle, behaviours, exposure patterns
What is primary prevention?
Prevention of disease onset e.g. vaccination
What is secondary prevention?
Halting progression of established disease by early detection followed by prompt, effective treatment
What is tertiary prevention?
Rehabilitation of people with established disease to minimise residual disability and complications
What is the prevention paradox?
Many people exposed to a small risk may generate more disease than a few exposed to a large risk. So when many people receive a small benefit the total benefit may be large. However, individual inconvenience may be high to the many while benefit is only to a few.
What is likelihood ratio?
Measure of how much a test alters your probability of the disease.
LR = p [test result if disease present]/ p [test result if disease absent]
LR for positive result = sensitivity/(1-specificity)
LR for negative test result = (1-sensitivity)/specificity
Post test odds = pre test odds x likelihood ratio
What are the reasons to terminate a study early?
Safety concerns, overwhelming benefit, futility
