PHEBP: Observation Studies in Epidemiology

Question 1

Define what is meant by exposure, risk factor and outcome in observational epidemiology

Answer 1

1) Exposure:

• In epidemiology, an exposure refers to contact with a potential risk factor or determinant that may influence the occurrence of disease
• This could be a specific behaviour (like smoking), a demographic characteristic (like age), a biological variable (like blood pressure), an environmental agent (like air pollution), a medical intervention (like a drug or vaccine), or other factors

2) Risk Factor:

• A risk factor is any attribute, characteristic or exposure of an individual that increases the likelihood of developing a disease or injury
• Risk factors are often revealed through observational studies where researchers find associations between these factors and certain health outcomes
• It’s important to note that identifying something as a risk factor does not imply a cause-effect relationship; it merely indicates an association

3) Outcome:

• An outcome in epidemiology is the health state or event that the study is designed to measure or observe in a population
• This could be the occurrence of a specific disease (like diabetes), an injury, a disability, or even death
• The outcome is essentially what the researcher is trying to predict or explain based on exposure to a certain risk factor

Question 2

Describe the basic design of cohort and case-control studies, recalling their main advantages and disadvantages​

Answer 2

Cohort Studies:

Cohort studies are a type of observational study where a group of individuals (the cohort) is followed over time

These individuals are typically free of the outcome of interest at the start of the study and are divided into groups based on their exposure to a certain risk factor

Advantages:

1. They can confirm multiple outcomes from a single exposure
2. They provide a natural way to calculate incidence and relative risk
3. They can be used to study rare exposures since the study population can be selected based on the exposure status
4. Temporality is clear - exposure is measured before the outcome occurs, which can support causal inference

Disadvantages:

1. They can be expensive and time-consuming, especially for outcomes that take a long time to develop
2. They may suffer from loss to follow-up, which can introduce bias
3. They may not be suitable for studying rare outcomes because large sample sizes and long follow-up periods are often needed

Case-Control Studies:

In case-control studies, individuals are selected based on their outcome status: cases (those with the outcome) and controls (those without the outcome)

The exposure status is then compared between these two groups

Advantages:

1. They are efficient for studying rare outcomes since cases are specifically sought out
2. They are quicker and less expensive than cohort studies as they do not require long periods of follow-up
3. They can evaluate multiple exposures for a single outcome

Disadvantages:

1. They are prone to recall bias as exposure data is often collected retrospectively
2. Selection of a suitable control group can be difficult
3. They do not provide incidence rates or relative risks directly (though odds ratios can be calculated and are often used as an approximation of relative risk)
4. Temporality can be difficult to establish - it may be unclear whether the exposure occurred before or after the outcome

Question 3

How would you Estimate a relative risk and interpret it along with the associated 95% confidence interval​

Answer 3

In a cohort study, relative risk can be estimated by:

RR = [disease (+) in exposed / the whole exposed group] / [disease (+) in unexposed / the whole unexposed group]

This ratio compares the probability of the disease in the exposed group to the probability of the disease in the unexposed group

Interpreting Relative Risk:

1. RR = 1: The risk of the event is the same in both groups. The exposure does not affect the risk
2. RR > 1: The risk of the event is higher in the exposed group. The exposure may be a risk factor for the event
3. RR < 1: The risk of the event is lower in the exposed group. The exposure may be a protective factor against the event

Interpreting the 95% CI for RR:

The 95% CI gives a range of values that likely contains the true RR in the population

1. If the CI includes 1, the RR is not statistically significant at the 0.05 level, meaning any observed difference could be due to chance
2. If the CI includes 1, the RR is not statistically significant at the 0.05 level, meaning any observed difference could be due to chance
3. If the entire CI is below 1, the risk in the exposed group is significantly lower

The width of the CI also gives an idea about the precision of the RR estimate - a narrow CI suggests a more precise estimate, while a wide CI suggests less precision and more uncertainty

Question 4

How would you Construct a 2x2 table of exposure by outcome and use it to estimate and interpret an odds ratio

Answer 4

An odds ratio (OR) is a statistic that quantifies the strength of the association between two binary data values

In the context of epidemiology, it’s often used to compare the odds of an outcome occurring in an exposed group to the odds of the outcome occurring in a non-exposed group

Odds of the outcome in the exposed group (disease/none) / odds of the outcome in the non-exposed group (disease/none)

OR = (a/b) / (c/d)

Interpreting an odds ratio:

• OR = 1: There’s no association between the exposure and the outcome. The odds of the outcome are the same in both the exposed and non-exposed groups
• OR > 1: The odds of the outcome are higher in the exposed group compared to the non-exposed group. The exposure may be associated with a higher risk of the outcome
• OR < 1: The odds of the outcome are lower in the exposed group compared to the non-exposed group. The exposure may be associated with a lower risk of the outcome

Just like with relative risk, you would also typically calculate a 95% confidence interval for the odds ratio. If the CI includes 1, the OR is not statistically significant at the 0.05 level. If the entire CI is above 1, it suggests a significant association with increased risk. If the entire CI is below 1, it suggests a significant association with decreased risk

Question 5

Explain what is meant by confounding, and how it might influence findings on the causal pathway​

Answer 5

Confounding is a distortion of the association between an exposure and an outcome that occurs when the study groups differ with respect to other factors that influence the outcome

In other words, a confounder is a variable that is associated with both the exposure and the outcome, but is not on the causal pathway between the exposure and the outcome

For example, let’s say we’re studying the association between alcohol consumption (exposure) and heart disease (outcome). Age could be a confounding factor, as older individuals are more likely to both consume alcohol and have heart disease. If we don’t adjust for age, we may mistakenly attribute the higher rates of heart disease to alcohol consumption, when they may be partly or wholly due to the higher age of the drinkers.

Influence on the Causal Pathway:

Confounding can distort our understanding of the causal pathway between exposure and outcome in two ways:

1) Confounding can create a spurious association:

• If the exposure is not truly associated with the outcome, but is associated with another factor (the confounder) that is associated with the outcome, it may appear as though there is an association between the exposure and outcome

2) Confounding can mask a true association or exaggerate it:

• If the exposure is truly associated with the outcome, a confounder can distort the size of the observed association
• If the confounder is positively associated with both the exposure and the outcome, it can make the observed association stronger than the true association (overestimation)
• If the confounder is negatively associated with either the exposure or the outcome, it can make the observed association weaker than the true association (underestimation)

Dealing with Confounding:

1. Design phase: Matching, restriction, and randomization can be used to ensure that the study groups are comparable with respect to potential confounders.
2. Analysis phase: Statistical techniques like stratification, standardization, and multivariable analysis can be used to adjust for confounders.
3. Counterfactual framework: This approach defines a confounder as a factor for which the counterfactual outcomes depend on its levels, given the levels of exposure. It supports techniques like propensity score matching, instrumental variable analysis, and g-methods (like g-computation, the g-formula, or the parametric g-formula)