HTLH 207 Week 7 - Cross-Sectional Designs

Cross-sectional design

snapshot of the exposure and outcome statuses of a population at a single point in time

Outcome of frequency and measure of association used in cross-sectional designs

• Outcome frequency: prevalence

• Measures of association: prevalence ratios and differences

4 basic steps of designing and conducting a cross-sectional study

1. State the study hypothesis(es) based on the research question

2. Design the cross-sectional study

3. Conduct teh cross-sectional study

4. Analyze and report the data

2 aspects of stating the study hypothesis(es) based on the research question

1. Define the outcome(s) with case definitions

2. Define the exposure(s) and their reference group

2 aspects of designing the cross-sectional study

1. Identify the source and sample populations

2. Determine whether the select individuals based on their group membership or based on their exposure status

2 aspects of conducting the cross-sectional studu

1. Select the sample population, obtain informed consent, and measure the exposure and outcome statuses of the participants

2. Collect data on key covariates

3 aspects of analyzing and reporting data

1. Calculate measures of frequency and association

2. Report findings using the STROBE reporting guidance

3. Consider the strengths and limitations of the study

2 goals of cross-sectional studies

1. Estimate the prevalence of health-related exposures and/or outcomes in a particular population (descriptive epidemiology)

2. Test one of more hypothesis about the relationship between exposures and health-related outcomes in a particular population (analytic epidemiology)

How can cross-sectional studies be used?

can be used to obtain estimates of the prevalence of exposures and outcomes and therefore are useful for understanding and describing the current health status of a population

5 questions that cross-sectional study hypotheses should address

1. What is the population of interest?

2. What is the health outcome of interest and how is it defined?

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?

How are cross-sectional studies often used to explore associations?

Cross-sectional designs are often used to explore relationships between many exposures and outcomes

Source population

group of people from which the study participants will be selected

Sample population

the people who are actually selected from the source population to participate in the study

2 ways that the sample population can be selected

1. Group membership

2. Exposure status

Odds ratio calculated from a cross-sectional study

Prevalence odds ratio

Prevalence odds ratio (POR) calculation

Prevalence odds ratio (POR) = (A*D)/(B*C)

What analysis method can researchers use when the outcome is dichotomous?

, investigators can use logistic regression - a more advanced analytic technique, which estimates odds ratios

What defines a “rare” outcome of interest in a cross-sectional study?

Typically <=10%

POR significance for rare outcomes of interest

When the outcome of interest is rare (typically <= 10%) in both the exposed and unexposed groups, the prevalence odds ratio will be reasonably close to the prevalence ratio

POR significance for common outcomes of interest

When the outcome is not rare, the prevalence odds ratio will provide an overestimate of the corresponding prevalence ratio (i.e. the association will appear stronger than it really is)

Cross sectional surveys

can be used to conduct public health surveillance, whereby a sample of individuals from a source population is surveyed periodically (i.e. annually) to obtain prevalence estimates of health behaviors and outcomes over time

Serial cross-sectional surveys

Repeated surveys that are particularly useful for tracking changes in trends over time

Example of serial cross-sectional survey

CDC’s Behaviora Risk Factor Surveillance System (BRFSS)

How is it ensured that serial cross-sectional surveys produce results that are representative of the source population?

Although each sample contains different individuals, the sampling method used ensures that those included in the sample population are representative of the source population

When may before and after results of serial cross-sectional surveys not be able to be compared?

Responses are not comparable if they use different methodology and/or have different study questions

What error may occur in cross-sectional studies that investigate exposures or outcomes with a short duration? When may this NOT be problematic>?

• May underestimate prevalence

• However, this may not be problematic if the study goal is to estimate prevalence at a particular moment in time

How may study questions be altered to obtain a more accurate estimate of prevalence during a certain time frame?

Specifying the specific range of time or duration for participants

Survival bias

occurs when those who survive and are available to be included in the study are meaningfully different from those who die and therefore are not available to be included in the study

4 strengths of cross-sectional designs

1. Allows for the study of multiple exposures and outcomes

2. Less expensive and require less time than other epidemiological study designs

3. No risk of loss to follow-up

4. Ideal for evaluating the burden of exposures and outcomes in a particular population; this information can be used to inform public health action or for future study planning

3 limitations of cross-sectional designs

1. Temporality may be difficult to establish because the exposure and outcome are measured at the same time

2. Prevalence of outcomes or exposures with short duration may be underestimated because investigators only capture a single point in time

3. May not be well-suited to study exposures that increase or decrease the likelihood of survival because differential mortality among groups may lead to a biased estimate of prevalence at the time the study is conducted (survival bias)