Prospective Case-Crossover Design — Epidemiological Method
Prospective Case-Crossover Epidemiological Design · Also known as: prospective case-crossover study, forward-looking case-crossover, prospective self-controlled case-crossover, real-time case-crossover
The prospective case-crossover design is an observational epidemiological study in which each case serves as their own control. Unlike the retrospective variant, exposures are recorded in real time as participants are followed forward, eliminating recall bias. It is particularly suited to investigating transient environmental or behavioral triggers of acute events such as myocardial infarction, asthma attacks, or road-traffic injuries.
Read the full method
Sign in with a free account to read this section.
Method map
The neighbourhood of related methods — select a node to explore.
When to use it
Use this design when you want to estimate the short-term triggering effect of a transient exposure on an acute event and when you can measure exposures prospectively. It is the method of choice when stable individual confounders are a major concern, when real-time exposure data are available (e.g., from sensors or electronic records), and when the outcome is acute, discrete, and accurately timed. Do NOT use it when the exposure of interest is chronic and does not vary meaningfully within individuals over time, when the outcome is gradual in onset or difficult to timestamp precisely, when prospective data collection is infeasible or too costly, when the induction period between exposure and event is long, or when carryover effects between hazard and control windows cannot be ruled out.
Strengths & limitations
- Perfect control of all time-invariant confounders — each subject acts as their own control, eliminating unmeasured stable confounding.
- Prospective exposure measurement removes the recall bias that undermines retrospective case-crossover studies.
- Efficient use of data — only cases are needed, making the design practical when controls are difficult to recruit.
- Well-suited to studying transient environmental, behavioral, or drug-related triggers of acute events.
- Results are not confounded by between-person differences in medical history, genetics, or chronic lifestyle factors.
- Requires precise event timing; acute events without a clear onset time (e.g., gradually worsening conditions) are not suitable.
- Prospective data collection is resource-intensive and requires sustained participant engagement and/or sensor infrastructure.
- Cannot estimate the effect of chronic, stable, or slowly varying exposures, as there is insufficient within-person variability to detect an effect.
- Assumes no carryover effect from hazard to control windows; if the exposure itself alters the risk in subsequent windows, estimates are biased.
- Analysis requires adequate variation in exposure within persons across time; homogeneous exposure profiles yield low statistical power.
Frequently asked
How does the prospective variant differ from the original case-crossover design?
In Maclure's original 1991 formulation, cases were asked to recall their exposure in the period before the event, which introduced recall bias. In the prospective variant, exposures are measured and recorded in real time throughout the follow-up period — using sensors, administrative databases, or electronic monitoring — before any events occur. This eliminates recall bias and often improves exposure measurement precision.
Why use conditional logistic regression rather than ordinary logistic regression?
Because each case's hazard and control windows form a matched stratum (the same person across different time points), standard logistic regression would ignore this matching and produce biased estimates. Conditional logistic regression conditions on the stratum — in effect, using only within-person variation — which is exactly the source of information exploited by the design.
How do I choose between a prospective case-crossover and a prospective cohort design?
The case-crossover estimates the short-term triggering effect of a transient exposure on an acute event, and it is efficient when the outcome is rare and time-varying within-person exposure data exist. A cohort design estimates long-term incidence rates across different exposure levels and can study both acute and chronic outcomes. Choose case-crossover when your research question is specifically about transient triggers and you can timestamp events precisely.
What is the minimum sample size for a prospective case-crossover study?
There is no universal minimum, but power depends on (a) the number of cases with at least some within-person exposure variability and (b) the magnitude of the transient effect. Studies in environmental epidemiology have been conducted with as few as a few hundred cases when exposure variability is high; smaller samples require a stronger expected effect size. A formal power calculation using conditional logistic regression should guide the design.
Can the design be used with continuous outcomes or only binary events?
The classic case-crossover applies to binary acute events (event / no event). Adaptations using conditional Poisson regression can accommodate count outcomes, and some extensions model symptom severity as a continuous outcome within the same within-person framework, but these are non-standard and require careful justification.
Sources
- Maclure, M. (1991). The case-crossover design: a method for studying transient effects on the risk of acute events. American Journal of Epidemiology, 133(2), 144–153. DOI: 10.1093/oxfordjournals.aje.a115853 ↗
- Navidi, W., & Weinhandl, E. (2002). Risk set sampling strategies for case-crossover studies. Epidemiology, 13(1), 100–105. link ↗
How to cite this page
ScholarGate. (2026, June 3). Prospective Case-Crossover Epidemiological Design. ScholarGate. https://scholargate.app/en/epidemiology/prospective-case-crossover-design
Which method?
Set this method beside its closest kin and read them side by side — the library lays the books on the table; the choice is yours.
- Case-control studyEpidemiology↔ compare
- Case-crossover designEpidemiology↔ compare
- Prospective Cohort StudyEpidemiology↔ compare
- Self-Controlled Case SeriesSocial Epidemiology↔ compare