Epidemiology 1: Fundamentals of Epidemiology: Research Concepts and Designs PDF

Epidemiology 1: Fundamentals of epidemiology: Research concepts and designs Tom McAdams 1  Overview of Epidemiologic Research Populations & Samples Outlin Measures of occurrence and e effect Study Designs Systematic Error (Bias)...

Epidemiology 1: Fundamentals of epidemiology: Research concepts and designs Tom McAdams 1  Overview of Epidemiologic Research Populations & Samples Outlin Measures of occurrence and e effect Study Designs Systematic Error (Bias) 2 Overview of Epidemiolog ical Research 3 Epidemiology: the study of the distribution and determinants of disease in populations 4  To describe adverse health outcomes When? Where? Among whom? To explain patterns of disease and their causes Temporal, sub-populations, disparities Goals of epidemiologi Types Exploratory (Descriptive) cal research Predictive o Diagnostic: Identify those with a disease o Prognostic: Identify those likely to get disease Causal (Identify causes of disease) o Long-term/overall goal is to prevent disease 5 A brief history of epidemiology 1850’s: John Snow & the Broad Street Cholera Outbreak 6 7 8 A brief history Risk Psychiatric factors Chronic and Infections (behaviour disease Psychosoci al and al Epi structural) In the 20th century, Epidemiology increasingly shifted its focus onto chronic diseases and environmental risk factors, e.g. socioeconomic factors, lifestyles, and health behaviours e.g. Brompton Hospital case-control study of smoking and cancer risk (Doll & Hill, 1950) This new focus eventually led to the emergence of Psychiatric and Psychosocial Epidemiology 9 Psychiatric & Psychosocial Epi Psychiatric Epidemiology: Prevalence, distribution, and causes of mental disorders at the population level E.g. What are the causes of depression in children and young people? Psychosocial Epidemiology: Psychological and social determinants of health and disease and underlying mechanisms E.g. Is childhood trauma related to an increased prevalence of physical and mental disorders in adulthood? 10 Types of Epidemiology By By By Exposure Outcome Method psychosocial epi psychiatric epi descriptive epi clinical epi cancer epi environmental CVD epi life course epi epi infectious social epi disease epi epidemiology of analytic epi genetic epi aging 11  Research Process Observati on Inductiv e Theory Hypothesi s/ Research Question Deducti Empirical data collection ve testing analysis Results & Inference 12 Populatio ns & Samples 14  Definitions Target Population: t he entire population, or group, that a Sample: researcher is Group of interested in people researching selected from and the population understanding of interest and about which the researcher wants to make inferences 15  Who you want to make conclusions Target Population about Target Population Cohort Eligible for study Sampling Frame Analytic Cohort Study Sample Included in study 16 Types of Sampling: Probability Any method of sampling that utilises some form of random selection, so that each member of the population has a known likelihood of being included in the sample 17 Types of Sampling: non- Probability Does not involve random selection Non-probability sampling methods: Quota: Sample selected, not randomly, but so that specified groups will appear in numbers proportional to their size in the target population Self-selecting/voluntary response: Sample selected on the basis of members’ own intention to participate in the study Convenience/opportunity: Sample selected because they are easily available for testing 18  Measures of Occurrence and Contrast 19 Prevalence The proportion of a population that currently has the outcome (disease) Cross-sectional (a snapshot in time) Usually a single moment Can be over a window of time (e.g. 2016) Numerator = current cases Denominator = total population Measures of prevalence: prevalence count, prevalence proportion, prevalence odds 20 Incidence the frequency at which new cases occur among those currently at risk of the outcome At risk = (usually) the healthy subgroup; those “eligible” for disease requires the prospective monitoring of a disease-free subgroup to determine how quickly they get the disease Numerator = new cases within defined period of time Denominator = population that is “healthy” (eligible to be cases), monitored over time Measures of incidence: incidence count, incidence proportion, incidence odds, incidence rate 21  Measures of Disease Contrast Prevalence/Risk Occurrence Count difference (RD) Prevalence/Risk ratio Incidence proportion (risk) or (RR) prevalence Odds ratio (OR) proportion Incidence rate Odds difference (IRD) Incidence rate Incidence rate ratio Survival (IRR) Time-to-event Number needed to treat 22 Measures of Occurrence – Risk (Incidence Proportion) 23 Measures of Occurrence – Risk (New Illness) Time = Disease 24 (D+) Measures of Occurrence – Risk Incidence proportion (risk) = Disease 25 (D+) Measures of Occurrence – Risk Incidence proportion (risk) = Disease (D+) 26 Measures of Occurrence – Risk Incidence proportion 5 (risk) No recurrent events/only counts first outcome per person 20.25 or 25% = Disease 27 (D+) Measures of Occurrence – Risk (Incidence Proportion) Assumptions The only way to interpret a risk is to know the length of Closed population the time period over which the risk applies No subject has disease at start Must have specified follow-up time Subjects followed for entire period (no loss to follow-up) No competing risks 28 Measures of Occurrence – Incidence Rate 10 people 10 months Time 3 = Disease 29 (D+) Measures of Disease – Incidence Rate 10 people Incidence Rate 10 months Time 3 = Disease 30 (D+) Measures of Disease – Incidence Rate 10 people 10 months Incidence Rate 3 Time 3 10 0 0.03 events per unit Can change scale (e.g., 3 events per 100 time) or units of time (e.g., months to years) for interpretability = Disease (D+) Measures of Disease – Incidence Rate 10 people 10 months Incidence Rate Time 3 3 96 4 0.031 events per unit of time NB: deaths/drop outs will impact total "units" = Disease (D+) Measures of Disease – Incidence Rate Assumptions Closed population No subject has disease at start Must have specified follow-up time All subjects followed for entire time (no loss to follow-up) No competing risks 33 Measures of Contrast (Effect) Measures of contrast allow us to compare prevalence/incidence of disease/disorder/condition measured in one group vs another Can have more than two groups, but can only compare two groups at a time Groups are typically framed as ‘exposed’ (E+) and ‘unexposed’ (E-) Choosing an appropriate comparison group 44 for an exposure is critical for study design Measures of Effect Risk/prevalence differences, ratios, and odds ratios compare proportions (incidence or prevalence) Unexposed Exposed 45 Measures of Effect E+ E- D+ 5 2 7 D- 15 18 33 20 20 40 46 Measures of Effect Incidence/ Prevalence Exposed (E+) Unexposed (E-) E+ E- D+ A B D- C D Odds Exposed (E+) Unexposed (E-) 47 Measures of Effect Incidence Exposed (E+) Unexposed (E-) E+ E- D+ A B D- C D Odds Exposed (E+) Unexposed (E-) 48 Measures of Effect Incidence Exposed (E+) Unexposed (E-) E+ E- D+ 5 2 7 D- 15 18 33 Odds Exposed (E+) Unexposed (E-) 20 20 40 49 Measures of Effect Risk Difference E+ E- D+ A B Risk Ratio D- C D 50 Measures of Effect Risk Difference E+ E- D+ 5 2 7 Risk Ratio D- 15 18 33 20 20 40 51 Measures of Effect Odds Ratio E+ E- D+ 5 2 7 D- 15 18 33 20 20 40 **If disease incidence is low (