HLTH 207 Week 5 Lecture and Asynchronous Work - Cohort Designs (ONE)

Basic process of a cohort design

1. Begins with a group of individuals who are at-risk of the outcome of interest

2. Investigators observe (or record) the exposure status of study participants

3. Group is followed forward in time to see who develops the outcome of interest (i.e. to identify incident cases)

Outcome frequency measure(s) used for cohort designs

risk or rate

Measures of association used for cohort deisgns

 risk/rate ratios and differences

Function of cohort design

Investigators designing and conducting cohort studies may be interested in learning about a broad research area (i.e. cancer, CVD, HIV) or understanding the health effects of a specific exposure

Study hypotheses for cohort designs should address what 5 questions?

1. What is the population of interest?

2. What is the health outcome of interest?

3. What is the exposure of interest and how is it defined?

4. What measures of frequency are being compared?

5. What is the direction of the hypothesized association?

Cohort

group of individuals who are followed or traced over a period of time and for whom membership is defined by some common criteria

Birth cohort

is a group of people defined by being born during a particular time

2 types of cohorts

1. General population cohort

2. Special exposure cohort

General population cohort

uses broad admissibility criteria to define membership, such as living in ascertain place or having a certain occupation to evaluate the effect of common exposures

Special exposure cohort

targeted enrollment of individuals with a particular exposure of interest

When is a special exposure cohort useful?

Useful when the exposure is uncommon in the general population (i.e. occupational radiation, chemical spill, etc.) or if it only affects a certain group of individuals

Key element of the cohort design

All study participants are at-risk for the outcome at the beginning of the follow-up period

In what 2 ways can researchers ensure that all study participants are at-risk for the outcome in cohort studies?

1. Only enroll individuals who are at-risk for the outcome

2. Include individuals regardless of their outcome history and then make appropriate exclusions during the analysis stage of the study

When is excluding individuals at the analysis stage of a cohort design particularly useful?

Excluding individuals at the analysis stage is particularly useful if interested in multiple outcomes and to get the most out of resources expended on the study

Unexposed reference group

should be comparable to the exposed group for all factors that may be associated with the outcome of interest

Goal of having an unexposed reference group

approximate the counterfactual, which is unobservable

Internal reference group

comprised of unexposed members of a general population cohort

External reference group

required for special exposure cohorts because selection is based on an individual's exposure status, and thus everyone initially sample is exposed; a separate group of unexposed people is then identified and enrolled

Closed cohort

• defined by a common start time

• No one is added to the cohort during the follow-up period, and individuals only exit the cohort when the outcome occurs, the study ends, or they die

Conducting a closed cohort design is when it is truly appropriate to report ____ based measures

risk

Benefit of closed cohort design

defined by a common start time

3 cases when maintaining a closed cohort design is difficult

1. Follow-up period

2. Population is highly mobile

3. Mortality rates aer high among that population

Open/dynamic cohort

• Members can enroll at leave at different times

• A person could enter, leave, and then later re-enter the cohort

Benefit of open/dynamic cohort

Allows for study participants to contribute vastly different amounts of person-time during the follow-up periods

In open/dynamic cohorts, you must calculate ______ and ______-based measures of association

rates, rate

How does the underlying structure of an open/dynamic cohort study change when it is conducted in real time vs. if it leverages historical records?

Underlying structure is identical - regardless of whether the study happens in real-time or if it leverages historical/existing records

How are studies conducted in real time distinguished from those conducted using historical records? Which is often considered “better”?

• Conducted in real time: prospective (“better”)

• Conducted using historical records: retrospective

Correct this myth: prospective studies are always better than retrospective studies

While investigators have less control over the data collection process when leveraging existing data sources, these studies are not inherently more subject to error than those conducted in real-time

4 measures of association used in cohort designs

1. Risk ratio

2. Risk difference

3. Rate ratio

4. Rate difference

Crude mortality rate

represents the total number of deaths without incorporating any other variable

Hazard ratio

similar to a rate ratio in that it uses person-time

Induction period

the time it takes for an outcome to occur due to an exposure of interest

Induction period varies substantially across ___________________

exposure-outcome relationships

How may induction period be estimated?

May be estimated based on what is known about disease etiology (i.e. the relationship between nutrition and cancer has a long induction period)

Latency period

the time between when an outcome first occurs and when that outcome can be detected

Latency period may vary depending on _____________________

 the methods for detection (i.e. nucleic acid versus antibody test to detect HIV)

Incubation period

unique to infectious diseases and refers to the time between exposure to an infectious agent and the onset of symptoms

What can the incubation period help estimate?

when people may become symptomatic and/or identify the timing or source of an exposure

Incubation period for COVID-19 and its median

2-14 days, with a median of 5 days

Does the COVID-19 incubation period stay the same over time? Why or why not?

the time changes as the virus mutates and variants emerge

3 strengths of cohort designs

1. Temporality between exposure and outcome is established by design because the exposure is observed and/or recorded before the outcome occurs

2. Cohort studies are well-suited for studying rare exposure because individuals may be selected based on their exposure status (i.e. special exposure cohort

3. Examination of associations between multiple exposure and outcomes

4 limitations of cohort designs

1. Loss to follow up could lead to selection bias

2. Cohort studies are not well-suited to study rare outcomes because there may not be enough events during the follow-up period to detect differences in outcomes across exposure groups

3. Exposure-outcome relationships with long induction periods or outcomes with long latency periods require a long period of time to accrue enough outcome events

4. Expensive, time-consuming, and resource intensive