03. Compare designs using recruitment and timelines
Lesson
LEARNING OBJECTIVES
Compare designs using their selection mechanisms, align eligibility and follow-up, and identify when existing records cannot answer the intended question.
A DESIGN COMPARISON MATRIX
For each candidate design, specify the population, entry rule, exposure assessment, outcome timing, denominator, intended effect measure and main validity threats. Add resources and ethical constraints. This is more informative than choosing a label from a hierarchy.
A cross-sectional survey can describe a defined period but may not establish whether exposure preceded outcome. A cohort needs a credible population at risk and follow-up. A case–control study needs a source population and controls selected according to a defensible sampling scheme. A randomized study requires a clearly specified intervention and allocation process; qualitative work is appropriate when understanding experiences or mechanisms is central.
WORKED FICTIONAL COMPARISON
Question: Would a reminder programme improve attendance at scheduled visits?
A current register can describe recorded attendance if its appointments and outcomes are complete. Comparing existing reminder users with non-users may be confounded by motivation, phone access and service organization. A pragmatic randomized evaluation could address programme assignment if ethical and feasible; interviews could investigate why reminders are or are not usable.
The available register is not automatically sufficient for every design. Audit whether eligibility, reminder timing, outcomes and relevant covariates are recorded reliably. Missing exposure dates can prevent a credible temporal comparison.
ALIGN THE START OF FOLLOW-UP
For an intervention-effect question, write a hypothetical trial protocol: eligibility, intervention strategies, assignment, outcomes, follow-up and contrast. In observational emulation, align eligibility, classification into strategies and the start of outcome counting. Do not label someone “exposed from baseline” solely because they receive the programme later; surviving until that future exposure can create an artificial event-free period.
This planning framework does not make observational assignment random. Unmeasured confounding, poor measurements and limited overlap can remain. Advanced situations with changing strategies require specialist methods, not merely a new date field.
YOUR TASK
Create a comparison matrix for three plausible approaches to your question. Draw each timeline on paper, marking eligibility, exposure classification, follow-up start, outcome assessment and end of observation. Identify which data exist at each point.
Write a paragraph defending one design and rejecting an alternative for a specific methodological or feasibility reason. State what information would change your choice.
SELF-CHECK
Does the word “retrospective” specify how participants were selected? Does an explicit hypothetical-trial protocol remove all confounding?
FEEDBACK
No. Explain the actual selection and timing. A transparent target trial clarifies the question and can prevent some design mistakes, but observational data still require substantive assumptions and adequate measurement.
PORTFOLIO OUTPUT
A design matrix, aligned timeline and evidence-based justification.
FURTHER READING
BMJ: Target trial emulation: applying principles of randomised trials to observational studies.
https://www.bmj.com/content/378/bmj-2022-071108.long
CDC Field Epidemiology Manual: Designing and Conducting Analytic Studies in the Field.
https://www.cdc.gov/field-epi-manual/php/chapters/design-conduct-analyze-field-studies.html
Compare designs using their selection mechanisms, align eligibility and follow-up, and identify when existing records cannot answer the intended question.
A DESIGN COMPARISON MATRIX
For each candidate design, specify the population, entry rule, exposure assessment, outcome timing, denominator, intended effect measure and main validity threats. Add resources and ethical constraints. This is more informative than choosing a label from a hierarchy.
A cross-sectional survey can describe a defined period but may not establish whether exposure preceded outcome. A cohort needs a credible population at risk and follow-up. A case–control study needs a source population and controls selected according to a defensible sampling scheme. A randomized study requires a clearly specified intervention and allocation process; qualitative work is appropriate when understanding experiences or mechanisms is central.
WORKED FICTIONAL COMPARISON
Question: Would a reminder programme improve attendance at scheduled visits?
A current register can describe recorded attendance if its appointments and outcomes are complete. Comparing existing reminder users with non-users may be confounded by motivation, phone access and service organization. A pragmatic randomized evaluation could address programme assignment if ethical and feasible; interviews could investigate why reminders are or are not usable.
The available register is not automatically sufficient for every design. Audit whether eligibility, reminder timing, outcomes and relevant covariates are recorded reliably. Missing exposure dates can prevent a credible temporal comparison.
ALIGN THE START OF FOLLOW-UP
For an intervention-effect question, write a hypothetical trial protocol: eligibility, intervention strategies, assignment, outcomes, follow-up and contrast. In observational emulation, align eligibility, classification into strategies and the start of outcome counting. Do not label someone “exposed from baseline” solely because they receive the programme later; surviving until that future exposure can create an artificial event-free period.
This planning framework does not make observational assignment random. Unmeasured confounding, poor measurements and limited overlap can remain. Advanced situations with changing strategies require specialist methods, not merely a new date field.
YOUR TASK
Create a comparison matrix for three plausible approaches to your question. Draw each timeline on paper, marking eligibility, exposure classification, follow-up start, outcome assessment and end of observation. Identify which data exist at each point.
Write a paragraph defending one design and rejecting an alternative for a specific methodological or feasibility reason. State what information would change your choice.
SELF-CHECK
Does the word “retrospective” specify how participants were selected? Does an explicit hypothetical-trial protocol remove all confounding?
FEEDBACK
No. Explain the actual selection and timing. A transparent target trial clarifies the question and can prevent some design mistakes, but observational data still require substantive assumptions and adequate measurement.
PORTFOLIO OUTPUT
A design matrix, aligned timeline and evidence-based justification.
FURTHER READING
BMJ: Target trial emulation: applying principles of randomised trials to observational studies.
https://www.bmj.com/content/378/bmj-2022-071108.long
CDC Field Epidemiology Manual: Designing and Conducting Analytic Studies in the Field.
https://www.cdc.gov/field-epi-manual/php/chapters/design-conduct-analyze-field-studies.html
My study notes
Stored only in this browser. Do not enter patient information.Saved in this browser. No account needed.