2.1 - 3: Epidemiology intro, diagnostic testing and screening Flashcards
Define epidemiology:
- The study of the distribution and determinants of disease in human populations
- Studies of distribution are largely descriptive (e.g. geography, time, age, gender, social class, ethnicity, occupation)
Define determinants:
- The factors that precipitate disease (aetiological or causal agents)
- e.g. biological (cholesterol), environmental (pollutants), social/behavioural (smoking)
Types of population:
- Target population: Population you are drawing inferences from
- Study population: Population you are collecting data from
Sources of data in epidemiological investigation:
- Routinely collected data: vital registrations like birth/death/infectious disease data, hospital databases
- Purposely collected data: Surveys, recruitment and follow-up
3 Key limitations of routinely collected data:
- Coverage: lack or variance in reporting of certain diseases, patchy sickness certification, cases who did not present to hospital
- Accuracy: Incorrect diagnoses of cause of death/illness
- Availability: Data may be barred by GDPR etc -> Research Ethic Committee approval can be obtained with justification
Mortality and morbidity:
- Mortality: Death due to disease in questions
- Morbidity: Being ill with disease in question
Issue of disease severity and reporting:
- ‘Disease iceberg’
- Less severe: more cases but less well recorded so unnoticed
Incidence and prevalence:
- Incidence: Number of new cases of the disease within a specified period of time
- Prevalence: Number of existing cases of a disease at a particular point in time -> strongly impacted by duration of illness
- Both are typically measured on a relative scale (e.g. incidence is measured in person-years)
- Prevalence measures are susceptible to survival bias
CI:
Full name, notation (not confidence intervals)
- Cumulative incidence risk
- Number of people who get disease during a period / number of people free of disease at start point
- Denoted H(t)
Crude mortality rate:
- Number of deaths in a specified period of time, divided by average population at risk during that period multiplied by length of study period
- Often not that informative without considering confounders -> Standardised rate required
Sensitivity and specificity: (Definitions and notation)
- Sensitivity = probability of diagnosing a true case as diseased
- Specificity = probability of diagnosing a truly non-disease person as non-diseased
- D and D-bar: Diseased status
- T and T-bar: Positive and negative test results
- FC 11
- Values are typically represented as a percentage
PPV and NPV:
- Positive predictive value: proportion of persons who are in fact diseased among those who test positive
- Negative predictive value: Proportion of persons who are in fact non-diseased among those who test negative
Balancing sensitivity and specificity via cutoff point:
- Receiver operating curve (ROC)
- y: Specificity, x: one - sensitivity
- Typically choose top-left-most point on curve
PPV expressed using Bayes thm:
- FC 14
Likelihood ratio of a positive test result:
- LR = sensitivity / (1-specificity)
Weighting for spec and sens.:
- M = w x sensitivity = (1-w) x specificity
- w can be maximised with respect to study or population criterion
- Study: w = within-sample disease prevalence
- Population: Population disease prevalence
Principle and aim of population screening for disease:
- Testing populations at risk who are asymptomatic for a disease
- Aiming to detect disease early -> better prognosis
- Generally cheap procedures followed up with more specific ones
- Often based on bio-markers from blood or urine samples (e.g. PSA, carcinoembryonic antigen)
Common screening programmes for cancer:
- Pap smear (cervical)
- Mammography ( breast)
- Colonoscopy (bowel/colorectal)
- Faecal occult blood test (bowel/colorectal)
- Derma check (melanoma)
Common screening programmes for foetal abnormalities:
- Alpha-fetoprotein
- Blood tests
- Ultrasound
Lead- and length-time and selection bias:
- Lead-time bias: Survival time since diagnosis is longer with screening -> need to compare mortality in screened / non-screened groups
- Length-time bias: Less severe cancers may be screen detected which may not be otherwise implicated prior to death so benefits seem greater
- Selection bias: Sub-groups may be more likely to attend for screening such as those with a family-history