unit 5

Selection of Comparison Group

"Counterfactual Ideal" = Ideal comparison group would be the exact same people who are in the exposed group had they not been exposed.

cohort studies

-A study in which people who are free of disease and that differ according to exposure (e.g., exposed and unexposed) are followed over time for incidence of disease -are the observational equivalent of experimental studies but the researcher does not allocate exposure but rather locates a "natural" experiment to observe relationship between exposure and disease

Case-control designs depend on historical information

-Cases and/or controls may remember or report their past experiences differently -"Recall bias" -"Reporting bias"

Experimental design: the investigator...

-Controls who is exposed to a factor of interest -Assigns subjects randomly to study groups

cases

-Criteria for case definition should lead to accurate classification of disease -Efficient and accurate sources should be used to identify cases: existing registries, hospitals

controls

-Definition: A sample of the source population that gave rise to the cases. -Purpose: To estimate the exposure distribution in the source population that produced the cases.

Observational design: the investigator...

-Does not have control over the exposure factor -Usually is unable to assign subjects randomly to study conditions

Ecologic Studies: Disadvantages

-Ecologic fallacy -Imprecise measurement of exposure

Ethical Responsibilities within Epi Research

-Ethics: norms that distinguish between acceptable and unacceptable behavior. -Ethical guidelines: a set of core values that guide practice.

Selecting Controls

-General population controls -Most often used when cases are selected from a defined geographic population -Sources: random digit dialing, residence lists, drivers' license records Example: Upper Cape Cancer Study

Experimental Studies:

-Investigator intervenes -Feasible and ethical.

Observational Studies (cohort and case-control):

-Investigator watches -Experimental study notfeasible or ethical.

Strengths and Weaknesses

-Investigators have more control in a prospective cohort study. -Retrospective cohort study is quicker and cheaper.

Ecologic Studies: Advantages

-May provide information about the context of health. -Can be performed when individual-level measurements are not available. -Can be conducted rapidly and with minimal resources.

Case-Control Studies

-Offers an alternative to cohort studies that can be much more efficient. -Same goal as cohort studies: to estimate the magnitude of an association between an exposure and outcome. -Uses a different sampling strategy: subjects are defined on the basis of the presence or absence of a disease/outcome of interest. -Cases and controls are compared with regard to past exposures in an attempt to identify differentiating factors. -Data can come from a variety of sources: •personal interviews •medical and hospital records

Randomized Controlled Trials (RCTs)

-The group receiving the new procedure, medicine, or intervention is often referred to as the treatment or study group. -The other group is referred to as the control group (may be receiving no treatment, a placebo, or the previous standard of care such as an existing drug). Note that the controls in an RCT are different from the controls in a case-control study! • In a case-control study, controls are those without the outcome of interest • In an RCT, controls are those who do not receive the novel intervention (exposure)

Cross-Sectional Study objective

-To examine health problem or disease frequency -To examine association between exposure and health problem or disease frequency -Unit of analysis is individual -Exposure and disease status of individual is assessed at the same time

Limitations of Cross-Sectional Studies

-We cannot follow people over time -We don't establish that causal link.

Sources of Exposure Information

-pre-existing records -questionnaires, interviews -direct testing

Attributable risk

-the difference between the incidence rate of a disease in the exposed group and the incidence rate in the non-exposed group. -in a cohort study, refers to the difference between the incidence rate of a disease in the exposed group and the incidence rate in the non-exposed group. Attributable risk = (Incidence rate in the exposed) — (Incidence rate in the non-exposed)

controls provide

-the information for the denominators in a case control study. -a fast and inexpensive means of obtaining the exposure experience in the population that gave rise to the cases.

Case-Control Approach

-we cannot compute the probability of disease in each exposure group (or, incidence), because we no longer have the denominators in the last column. i.e., we don't know the exposure distribution for the entire source population. -However, we can compute the odds of disease in each group.

ecologic studies

...a study in which the units of analysis are populations or groups of people rather than individuals." ex: of groups are nations, states, census tracts, counties. -May be used when individual measurements are not available, but group-level data can be obtained.

Analysis

1. Descriptive Analysis • Table 1 : Are exposed and unexposed groups comparable in terms of all other characteristics? 2. Measures of frequency/association • Set up 2 x 2 table • Cumulative incidence in exposed and unexposed • Risk Ratio comparing exposed and unexposed • Risk Difference comparing exposed and unexposed

Disadvantages of a Randomized Controlled Trial

1. External validity can be limited • Think: the more tightly controlled the study, the less it resembles the "real world." Experiments have a high level of investigator control. • Can be influenced by: location of study, characteristics of participants/volunteer biases; study procedures differ from "real world" 2. Time and cost; can be expensive' 3. Selection bias - concern about differential attrition (i.e., dropping out)across treatment groups 4. Conflict of interest dangers (i.e., Pharmaceutical industry sponsors) 5. Ethics (i.e., is withholding or assigning a specific treatment unethical?)

Types of Cohorts

1. In GENERAL cohort, researcher can study effects of MULTIPLE exposures on one or more outcomes. 2. In SPECIAL cohort (MOST COMMON), researcher can study effects of SINGLE exposure on one or more outcomes.

Cross Sectional Studies are useful for:

1. Measuring the prevalence of health 2. Understand determinants of health, 3. Describing features of a population. 4. Unlike other types of observational studies, cross-sectional studies do not follow individuals up over time. 5. They are usually inexpensive and easy to conduct. 6. They are useful for establishing preliminary evidence in planning a future advanced study.

Types of RCTs

1. Prophylactic or Prevention trial • Designed to test preventive measures 2. Therapeutic trial • Designed to evaluate new treatment methods 3. Clinical trial • Designed to evaluate whether a medical strategy, treatment, or device is safe and effective for humans 4. Cross-over design • Participants are switched between treatment groups partway through the study 5. Cluster-randomized trials • Randomization can occur at the group-level; requires specialized statistics for analysis

Descriptive studies are useful for:

1. Providing clues about disease causation and prevention that are usually investigated further in formal studies. 2. Assessing the health status of a population(e.g. Healthy People 2020) 3. Allocating resources efficiently and targeting populations for education or preventive programs

Measures of Association for Cohort Studies

1. Relative risk (RR) 2. Attributable risk

Epidemiologists select different sets of people who are as similar as possible

1. With respect to other factors that could influence outcome 2. With respect to collection of comparable & accurate information

Cross-Sectional Studies

A type of investigation "...that examines the relationship between diseases (or other health-related characteristics) and other variables of interest as they exist in a defined population at one particular time." A type of prevalence study

When is an Experimental Study Ethical?

An experimental study is only appropriate and ethical when equipoise is present. Equipoise = A state in which investigators truly do not know whether one treatment is better than another* but truly believe that withholding treatment* from study participants will not harm them

Advantages of general population controls

Because of selection process, investigator is usually assured that they come from the same base population as the cases.

Ambidirectional Cohort Study

Contains elements of both prospective and retrospective types of studies

Advantages of a Randomized Controlled Trial

High-quality evidence (gold standard) • Randomization, an inherent part of an RCT, eliminates bias in treatment assignment • Using placebos and/or blinding can help reduce bias elsewhere in the study Ideal study design if assigning exposure/treatment is feasible and ethical.

Minimizing Bias: Randomization

If done correctly, measured and unmeasured factors should be equally distributed across the treatment groups at baseline • Eliminates bias in treatment assignment Randomization provides a valid basis for statistical inference. • Permits use of probability theory to express likelihood that differences in outcome between groups are due to chance. Ultimate Goal = TREATMENT is only the difference between study groups

What illnesses make good hospital controls?

Illnesses that have no relation to the risk factor(s) under study Example: Should respiratory diseases be used as controls for a study of smoking and myocardial infarction? •Do they represent the distribution of smoking in the entire population that gave rise to the cases of MI?

Beneficence

Maximize benefits and minimize harms

What would a quasi-experimental community trial of suicide prevention look like?

Questions to consider: • Who would be your study subjects? • What is the intervention? • What health outcome (or set of outcomes) would you study in relation to your selected intervention? • What is a testable hypothesis to describe what you expect to see?

Belmont Report (1978)

Summarizes ethical principles and codes for research involving human subjects Basic Ethical Principles 1. Respect for Persons 2. Beneficence 3. Justice

Measures of Disease Association

The chance of something happening can be expressed as a risk and/or as an odds:

Disadvantages of general population controls

Time consuming, expensive, hard to contact and get cooperation; may remember exposures differently than cases

How to interpret Attributable Risk (Rate Differences)

Tip #1: If you have cumulative incidence (CI) expressed as percentage, convert % to convenient fractions so that you can express it as the excess risk in a group of people who have the risk factor. Tip #2: Focus your interpretation on the excess risk in the exposed group. -Example:"There were 4 excess wound infections per 100 subjects in the group that had incidental appendectomies, compared to the group without incidental appendectomy."

Hospital controls

Used most often when cases are selected from a hospital population. Example: Study of cigarette smoking and myocardial infarction among women. •Cases identified from admissions to hospital coronary care units. •Controls drawn from surgical, orthopedic, and medical unit of same hospital. •Controls included patients with musculoskeletal and abdominal disease, trauma, and other non-coronary conditions.

Cross-level inference

We cannot assume that associations at the ecological are the same as associations at the individual level

Odds Ratio (OR)

a measure of the association between frequency of exposure and frequency of outcome used in case-control studies. We get an "odds ratio", but calculating the ratio of the "odds" of exposure.

Quasi-experimental Study

a type of research in which the investigator manipulates the study factor but does not assign individual subjects randomly to the exposed and nonexposed groups.

Ecologic correlation

an association between two variables (exposure and outcome) measured at the group level (e.g., population average)

ecological fallacy

assumes that a generalized cultural value applies equally well to all members of the culture

Ecologic comparison study

involves an assessment of the association between exposure rates and disease rates during the same time period.

Special control groups

like friends, spouses, siblings, and deceased individuals. •These special controls are rarely used. •Cases might not be able to nominate controls because they have few appropriate friends, are widowed or are only or adopted children. •Dead controls are tricky to use because they are more likely than living controls to smoke and drink.

In a prospective cohort study...

outcome free individuals are identified and their exposure status is classified at study entry -Subjects are followed into the future to identify the incidence of the outcome in each exposure group

In a retrospective cohort study...

outcome free individuals are identified at a past baseline point and their exposure status is classified at study entry -Subjects are followed into the future to identify the incidence of the outcome in each exposure group

Diagram of a prospective cohort study

study question: how many new cases of illness occur? baseline: healthy people who have had an exposure time passes follow-up: healthy people and sick people

risk

the chances of something happening ---------------------------- the chances of all things happening

odds

the chances of something happening ---------------------------- the chances of it not happening

Relative risk (RR)

the ratio of the incidence rate of a disease or health outcome in an exposed group to the incidence rate of the disease or condition in a non-exposed group Relative risk = Incidence rate in the exposed ------------------------------ Incidence rate in the nonexposed Interpretation: RR ≈ 1 ⇒ association between exposure and disease unlikely to exist. RR > 1 ⇒ increased risk of disease among those that have been exposed. RR < 1 ⇒ decreased risk of disease among those that have been exposed.

Experimental = Intervention Study

• "An investigation involving intentional change in some aspect of the status of the subjects,..." • Randomized control trial (RCT)• Quasi-experimental design • Key Feature: Investigators influence or control the exposure • Exposure can be allocated at either the individual- or group-level: • Community/group trials -> unit of randomization/analysis is the group/community (an experimental ecologic study)

Attributabdle Risk Percent

• Can be derived from attributable risk; the portion of cases in the population attributable to the exposure. • AR% = Incidence (exposed) - Incidence (unexposed) *100 Incidence (exposed)= [(RR-1)/RR]*100

Cohort Studies Strengths

• Clear temporal sequence between E and D, especially if prospective • Efficient for rare exposures • Good information on exposures, confounders, especially if prospective • Can study effects of an exposure on multiple outcomes

Strategies for Follow-Up

• Collect identifiable information -Full name & Address -DOB, SSN -Contact information • Maintain frequent contact with all respondents -Regular mail (questionnaires, newsletters) -Telephone calls -Social media -Personal contact (if possible) • Incentives (gifts, calendars, money)

Community Trials Subtype of quasi-experimental designs

• Community intervention trials determine the potential benefit (or harm) of new policies and programs • Intervention: Any program or other planned effort designed to produce changes in a target population • Community refers to a defined unit, e.g., a county, state, or school district

Cohort Studies Disadvantages

• Expensive and time consuming • Complicated and difficult to carry out • Subjects may be lost to follow-up during the course of the study • Exposures can be misclassified

Minimizing Bias: Compliance

• For example, in the Physicians' Health Study, compliance involved taking a pill every day • Deviations from protocol can occur for many reasons: side effects, illness, level of interest, and length of follow-up • If there is a lack of compliance among participants due to a common reason (such as side effects), this can introduce bias

From the Contingency Table...

• Incidence of disease in the exposed group= A/(A + B) • Incidence of disease in the non-exposed group= C/(C + D) • The relative risk (RR)= [A/(A + B)] ÷ [C/(C + D)]

Matching in Quasi-experimental Designs

• Involves selecting pairs of participants or clusters of individuals who are comparable on important variables (e.g., age, sex, medical history). • Certain statistical methods (e.g., analysis of covariance and propensity score analysis) are sometimes used to minimize confounding after the data are collected for the study.

Cohort Studies Limitations

• May need large numbers of subjects to be followed for long periods of time • Can be expensive and time consuming • Not good for rare diseases or those with long latency, if prospective • Loss to follow-up term follow-up has potential to undermine validity

Respect for Persons/Autonomy

• Participants enter into research voluntarily (i.e., "autonomy") • Additional protection for: - children younger than 18 years of age -those who may be more vulnerable to exploitation in research due to their status, e.g., prisoners -individuals with diminished mental capacity

Justice

• Participants should not be in research studies if they are unlikely to benefit from the results • Fair distribution of the burdens and benefits • Groups expected to benefit should be included in study sample ("bear the burden")

Cohort Studies Advantages

• Permit direct observation of risk • Exposure factor is well defined • Can study exposures that are uncommon in the population • The temporal relationship between factor and outcome is known

Minimizing Bias: Placebos

• Placebos cannot always be used. -Interventions such as surgery, exercise, diet, etc. -Placebos are not always ethical; often use current standard of care instead of placebo • "Placebo Effect" = Participants assigned to placebo group improve because they are told that they will or think they will. This stems from the power of suggestion. • "Nocebo Effect" = Participants develop negative health issues because they are told they may (such as in a list of potential side effects). This also stems from the power of suggestion and a participant's expectations.

An experimental study is only appropriate and ethical when equipoise is present.

• Public health is always best served by proper evaluation, and often RCTs are the best suited to inform policies. • It can be unethical not to perform a trial, because it prevents new knowledge from being obtained and used.

Natural Experiment

• Quasi experiments are sometimes called "natural experiments "because membership in the intervention or "treatment" category is determined by conditions beyond the control of the researcher. • Levels of exposure to a presumed cause differ among a population in a way that is relatively unaffected by outside factors, so the situation resembles a planned experiment • Can be the result of policy changes (e.g., health care reform),major events (e.g., 9/11)

Randomization vs Random Selection

• Randomization (or random assignment) is distinct from random selection or random sampling • Random sampling is a method of choosing participants to be in the overall study. • Random assignment (randomization) is the allocation of a participant to a treatment group; this occurs after participants have been selected to join the overall study. • You may see random sampling in both observational and experimental studies; however, randomization is exclusive to RCTs.

Sources of Comparison Groups

• Rates of morbidity and mortality among a working population are lower than those of the general population • Health requirements for workers (especially physical laborers) tend to be stringent • General population consists of both healthy and ill people • Leads to underestimation of risk when comparing a worker population to the general population

Characteristics of Quasi-experimental Studies

• Relies on participants who -- 1) volunteer to join the study OR 2) are geographically close to the study site OR 3) conveniently turn up (at a clinic, school) while the study is being conducted • The study groups are opportunistically rather than randomly composed, therefore group characteristics (e.g., age, sex, race, SES) may not be balanced before the study begins (at baseline). • Baseline differences between groups may confound the study's results; may use "matching" techniques to limit this issue.

Advantages of hospital controls

• Same selection factors that led cases to hospital led controls to hospital • Easily identifiable and accessible (so less expensive than population-based controls) • Accuracy of exposure recall comparable to that of cases since controls are also sick • More willing to participate than population-based controls

Population Risk Difference

• Similar measure to AR except it is concerned not with the excess rate of disease in the exposed, but the excess rate of disease in the population • PAR is the proportion of the disease incidence in the population (i.e. exposed & non-exposed) that is due to the exposure • Accordingly, it is the disease incidence in the population that would be eliminated if the exposure were eliminated • Calculation:(Incidence in the total population) — (Incidence in the non- exposed segment)

How to Interpret Risk Ratios

• Since the relative risk is a simple ratio, errors tend to occur when the terms "more" or "less" are used. • Because it is a ratio and expresses how many times more probable the outcome is in the exposed group, the simplest solution is to incorporate the words "times the risk" or "times as high as" in your interpretation. "Those who had (name the exposure) had RR 'times the risk' compared to those who (did not have the exposure)."

Minimizing Bias: Blinding (Masking)

• Single-blind design: Subject unaware of group assignment (a placebo may be used if the treatment is a type of medication) • Double-blind design: Neither subject nor experimenter is aware of group assignment • Why? - To maintain the integrity of a study and reduce the potential for bias. Note: There are limitations to blinding

Conducting Community Trials

• Start by determining eligible communities and their willingness to participate • Collect baseline measures of the problem to be addressed in the intervention and control communities • Use a variety of measures, e.g., disease rates, knowledge, attitudes, and practices

follow-up

• The only valid way "out" of a cohort study is to develop disease under study, to die, or for the study to end. • Goal is to obtain complete follow-up information on all subjects regardless of exposure status. -This means following all subjects from exposure until disease, death, or end of follow-up • Very time-consuming process • Losses to follow-up raise doubts about validity of a study. -High levels of LTF can result in a bias (e.g., study results will not be accurate). -Decrease sample size -There is no "magic number" for how much loss to follow-up is OK (>80% is good).

Belmont Principles in Practice

• Universities must be approved to conduct clinical research and clinical trials under Federal Wide Assurance (FWA) license • Clinical protocols and informed consent documents must be approved by the Institutional Research Board (IRB) • Data and Safety Monitoring Boards (DSMB) oversee large federally-funded trials and must review the data on a periodic basis.

Matched Case-Control Study

•A study in which the cases and controls have been matched (i.e., are selected to be similar) according to one or more criteria such as sex, age, race, or other variables. •Reason to match?

Interpretation of an Odds Ratio

•An OR ˃1 suggests a positive association between exposure and disease. •An OR <1 indicates that the exposure might be a protective factor. •An OR = 1.0 indicates no association between exposure and outcome. •Formula for correct interpretation: •The odds of X [your outcome] are X [calculated value for the OR] times higher among those who [Exposure group] compared to those who [Unexposed group].

advantages

•Can be used to study low-prevalence conditions •Having a disease is a criterion for being selected as a case. •Relatively quick and easy to complete •Usually inexpensive •Involve smaller number of subjects

Retrospective Cohort Study

•Cheaper, faster •Efficient with diseases with long latent period •Exposure and confounder data may be inadequate •More vulnerable to bias • Study is conceived after some people have developed outcome of interest, and begin identifying and enrolling subjects after outcomes have occurred. • Investigators make use of historical/pre-existing data to determine exposure level at some baseline in the past. • When studies are based on subject's memories, they may be less reliable.

Strengths of case-control studies

•Efficient for rare diseases and diseases with long induction and latent period. •Can evaluate many risk factors for the same disease so good for diseases about which little is known

Weaknesses of case-control studies

•Inefficient for rare exposures •Vulnerable to bias because of retrospective nature of study •May have poor information on exposure •Difficult to infer temporal relationship between exposure and disease

general cohorts

•Investigate exposures that are common in the general population (e.g., high blood pressure, OC use, smoking, alcohol use) •Examples:- Framingham Heart Study - Black Women's Health Study

special cohorts

•Investigate unusual and rare exposures •Examples: -Vietnam veterans exposed to Agent Orange -Atomic Bomb survivors -Some occupational exposures

Disadvantages

•Measurement of exposure may be inaccurate •Representativeness of cases and controls may be unknown •Provide indirect estimates of risk •The temporal relationship between exposure factor and outcome cannot always be ascertained. •Can only study one outcome

Prospective Cohort Study

•More expensive, time consuming •Not efficient for diseases with long latent periods •Better exposure and confounder data •Less vulnerable to bias

Problem & Alternative Approach

•Problem: we usually wouldn't have resources to get data on all subjects in a population. •Alternative approach—Case control

Odds Ratio as an Estimate of Risk Ratio

•Provided that the disease is uncommon (say <10%), odds ratio is a good estimate of the risk ratio. •However, as the outcomes of interest become more common, the odds ratio gives estimates that are increasing more extreme than the risk ratio would have been.

Important consideration for selecting controls: "the would criterion"

•Review: Controls are a sample of the source population that gave rise to the cases. Purpose is to provide information on the exposure distribution in the source population. •When selecting a control group consider the "WOULD CRITERION": If a member of the control group actually had the disease under study WOULD he/she end up as a case in your study? Answer should be YES.

Disadvantages of hospital controls

•Since hospital based controls are ill, they may not accurately represent the exposure history in the population that produced the cases •Hospital catchment areas may be different for different diseases

OR calculation

•The odds in favor of exposure among the disease group (the cases) = A/C. •The odds in favor of exposure among the no-disease group (the controls) = B/D. •The OR is defined as (A/C) ÷(B/D). •The OR can be expressed as AD/BC.

Measure of Association for Case-Control Studies

•The odds ratio (OR) is a measure of the association between frequency of exposure and frequency of outcome used in case-control studies. •We get an "odds ratio", but calculating the ratio of the "odds" of exposure. OR = Odds of exposure among the diseased Odds of exposure among the non-diseased