Within-person incidence analysis
Specify incidence contrasts and adjustments
Conditioning within people removes factors constant during observation, not time-varying confounding. Use a model with interval duration and appropriate age or seasonal structure; pooled event proportions do not reproduce SCCS.
Flow diagram
Steps
Define within-person incidence contrast
Specify risk periods relative to baseline and event definition.
Select conditional count model
Specify a conditional event-count model, conditioning on each person’s total and including interval-duration offsets. Choose the SCCS variant for the exposure and observation process; do not compare pooled counts alone.
Adjust time-varying effects
Prespecify age, season and calendar-time structure.
Handle recurrence and dependence
Justify event-independence assumptions or an appropriate extension.
Specify estimation and checks
Record uncertainty, sparse intervals and model diagnostics.
Within-person incidence analysis
End flow
Tables and supporting information
Within-person model plan
Preserve interval duration and conditional person structure.
Assumptions and rationale
Stable personal characteristics are controlled by design, but event-dependent observation or exposure requires additional assumptions.
Method and protocol evidence
Attach the protocol, methodological sources and versioned software specification supporting these decisions. Record citations and provenance; do not enter participant identifiers.
Conditional rate model notation
Nᵢⱼ is the event count and Tᵢⱼ the duration in person i’s interval j. X indicates a risk period; Z describes time-varying factors. Conditioning on each person’s total removes the person-specific αᵢ; exp(β) is the relative incidence for the specified contrast. Do not estimate the SCCS effect by simply pooling risk-period and baseline counts across people. Standard model assumptions about event-dependent exposure, observation and recurrence require assessment; extensions must match the violation.