Cohort studies

Cohort Studies. Mpho Tlali School of Public Health and Family Medicine mpho.tlali@uct.ac.za May 2022.

Lecture outline. Recap of classifications of study designs Features of cohort studies Measures of outcome and their interpretation in cohort studies The Framingham study – and example of a landmark cohort study for cardiovascular risk factors Strengths and weakness of cohort studies.

Cohort studies. You should be able to: Describe the key features of an analytic cohort study Outline the design considerations in a cohort study Interpret the results from a cohort study Describe the strengths and limitations.

Recap of study designs. Did the investigator control exposure?.

What is a cohort. Ancient term that was used to referred to a military unit in a Roman army Use has been extended to include: A group with common characteristic In epidemiology the group must be as similar as possible to the exposed group with respect to all factors except the exposure.

Descriptive cohort study. Descriptive cohort study Describes the occurrence of disease in a group of people No comparison group E.g. birth cohort, group of individuals with a particular diagnosis - natural history of disease (such as COVID-19).

Analytic cohort study. Key features A study in which a group of people ( a cohort ), free of disease at the start of the study are classified into subgroups according to exposure and are then followed up over a period of time to see how subsequent development of new cases of the disease differs between the subgroups Aim to examine a possible relationship between an exposure and an outcome.

Cohort study. Population. People without the disease.

Features in the design of cohort studies. At baseline All participants must be free of the disease. Must screen out all who have the disease so that we can be sure that the “exposure” came before the “disease” The comparison/unexposed group needs to be as similar as possible to the exposed group with respect to all factors except the exposure.

Features in the design of cohort studies. Follow-up of participants Follow-up period needs to be appropriate for outcome Ideally complete follow-up of all participants Loss to follow-up major source of bias in cohort studies If loss to follow-up large (30-40%) then raises doubts about validity of study results If loss to follow-up related to exposure/outcome or both this could results in a biased estimate of association between the exposure and outcome If loss to follow-up - try to ascertain differences between those lost to follow-up and those that remain in the study.

Measuring the outcome – 1. Cumulative incidence/Incidence proportion [risk] No. of new cases of disease during a specific period of time/No. of people at risk of the disease at the start of the period Expressed as number of cases per 1000 population.

Measuring the outcome – 2. Incidence rate No. of people who get the disease (new cases) in a specified period/ Sum of the length of time during which each person in the population is at risk For each individual the time at risk is that time during which the person remains disease free Expressed as number of new cases per total person-years of observation.

Example (person years calculation). •slK-uosEd 00b = IP.IOI pennooo eseesp ueL4M = X.

Measuring the association between the exposure and the outcome.

Relative risk. A ratio that compares the risk/rate of disease among the exposed to the risk/rate of disease among the unexposed Numerator: Incidence (CI or IP) in the exposed Denominator: Incidence (CI or IP) in the unexposed Gives an indication of individual risk Useful to clinicians when dealing with individual patients.

Attributable risk. Measure the difference in frequency of disease (risk) between the exposed and unexposed groups Gives an indication of the risk of disease in the exposed group that is attributable to the exposure, after taking into account underlying levels of disease Important public health implications - u seful measure of the extent of a health problem caused by a particular exposure/risk factor.

Example: calculating relative risk. Cohort study conducted to determine if there is an association between cigarette smoking and coronary heart disease (CHD) 288 smokers free of CHD at baseline were followed up, and 112 subsequently developed coronary heart disease 312 non smokers free of CHD at baseline followed up, and subsequently 88 developed CHD.

Example: calculating the relative risk. CHD No CHD Total Smokers 112 176 288 Non smokers 88 224 312.

Example: interpreting the relative risk. Relative risk = 1.38 The risk of developing CHD is 1.38 times higher for smokers compared to non-smokers Clinical setting –if you smoke you are 1.38 times more likely to get CHD compared to if you did not smoke.

Interpreting relative risks. RR < 1 RR = 1 RR > 1 Risk comparison between exposed and unexposed Risk for disease is lower in the exposed than the unexposed Risk of disease is equal for the exposed and unexposed Risk for disease is higher in the exposed than in the non exposed Exposure as a risk factor for the disease? Exposure reduces the risk of disease (protective) The exposure is not a risk factor Exposure increases the risk of disease (risk factor).

Risk and Precision. How precise is the Risk Estimate?  Confidence Intervals For example, RR = 1.38 (95% CI 1.2 - 6.3) means that a) the risk is increased b) you are 1.38 times more likely to have the outcome if you have the exposure c) you are 95% confident that the true RR lies between 1.2 and 6.3..

What is the implication of a 95% Confidence Interval?.

If a RR = 2.9 (95% CI 1.3 - 6.1) then. increased risk that is significant.

CHD No CHD Total Smokers 112 176 288 Non smokers 88 224 312.

Example: interpreting the attributable risk. The CHD risk difference between the group of smokers and the group of non-smokers is 0.107 Not all of the disease (CHD) is due to exposure (smoking). Note IR of CHD among non-smokers is 0.282 (not 0) Quantifies the “extra” risk presented by an exposure (smoking) – the risk that can be attributed to the exposure.

Example: interpreting the attributable risk. Gives an idea of the public health benefit if we could eliminate that particular risk factor. So in this study population if we could have prevented smoking we could have prevented 107 cases of CHD per 1000 individuals Useful measure of the extent of a health problem caused by a particular exposure – predicts the burden on the health services. AR of 0 means no difference so no additional risk caused by that exposure.

The Framingham Study. Introduced the notion of “risk factors” for CVD Framingham Heart study is synonymous with remarkable advances made in the prevention of heart disease Regarded as one of the most important epidemiological studies in medicine Example of an analytic cohort study.

Prior to the Framingham study. Widely believed that atherosclerosis was inevitable part of ageing BP was supposed to increase with age Notion that individuals could modify risk factors related to cardiac disease not standard medical practice Relationship between elevated cholesterol levels and cardiac disease not well understood Since 1930s epidemic of cardiac disease – WHY?.

The Framingham Study. A group of adults living in Framingham, Massachusetts, who were free of heart disease were recruited between 1948 - 1950. Group divided into exposure categories. Exposures of interest included : Smoking Alcohol use Obesity Hypertension High cholesterol levels Low levels of physical activity.

Framingham study. Participants examined every 2 years for these risk factors and for the outcome of interest - heart disease. Also hospital surveillance. Association between risk factors (exposure) and heart disease was assessed..

Framingham study: results. Major risk factors for CVD identified Better understanding of effect of lifestyle (smoking, exercise, diet) on CVD.

Cohort studies: Strengths. Confident that exposure precedes outcome (temporal relationship) – useful in determining causal relationship Allows for calculation of incidence of disease – so risk of disease can be measured directly Can assess if a single exposure has multiple outcomes Efficient for studying rare exposures.

Cohort studies: Limitations. Expensive Time-consuming Loss to follow-up Prone to information/detection bias – reporting or ascertainment of disease different in exposed and unexposed group Inefficient to study rare disease – because will need a very large sample size.

Cohort studies. You should be able to: Describe the key features of an analytic cohort study Outline the design features in a cohort study Interpret the results from a cohort study Describe the strengths and limitations.

General epidemiology. Measures of disease frequency Prevalence Cumulative incidence Incidence Measure of effect Relative and absolute measures Risk Ratio / Incidence rate ratio; risk difference.

Acknowledgements. Andrew Boulle Nisha Jacob Jennifer Moodley Taryn Young.