Guide overview
The essential idea
Pooling adjacent NHANES cycles can improve sample size and precision, but more rows do not automatically create a valid combined estimate. NCHS advises analysts to examine sample-design changes, item comparability, proper multi-cycle weights, and the assumption of no meaningful trend across the pooled period. Compatibility must be demonstrated variable by variable and decision by decision.
Practical framework
What to apply
Content compatibility
Does each variable represent the same construct in every cycle?
- Compare exact wording, response options, units, labels, special codes, derivations, and eligible ages.
- Read each component's analytic notes, method documents, codebook, and revision history.
- Confirm that the variable was collected in all planned cycles and under compatible inclusion and exclusion rules.
- Create an explicit crosswalk when names or categories change but a defensible common representation remains possible.
Method and instrument compatibility
Could collection or laboratory changes alter the measurement?
- Compare instruments, specimen type, collection protocol, calibration, laboratory method, detection limits, and quality-control notes.
- Distinguish a documented scientific method change from a wording or documentation revision.
- Review conversion guidance, bridge studies, crossover data, or NCHS recommendations when available.
- Plan sensitivity analyses or cycle-specific estimates when residual discontinuity remains plausible.
Design and weight compatibility
Will the combined estimate represent the intended population correctly?
- Identify oversampling and sample-design changes that affect subgroup composition or precision.
- Select the most restrictive applicable full-sample or subsample weight across all analysis variables.
- Construct the multi-cycle weight according to NCHS guidance for the exact periods combined.
- Retain the masked variance strata and primary sampling units and use survey-capable analysis procedures.
Trend and interpretation compatibility
Is pooling scientifically preferable to showing change over time?
- Examine cycle-specific estimates and uncertainty before imposing one pooled result.
- Assess whether policy, population, collection, or secular changes make a no-trend assumption unreasonable.
- Use cycle terms, interactions, stratified estimates, or trend models when the question concerns change.
- Define the midpoint and interpretation of the pooled period clearly.
Before you finish
A short quality review
- Each variable has a cycle-by-cycle definition, eligibility, coding, unit, and method comparison.
- Sample-design and oversampling changes have been reviewed.
- The correct combined full-sample or subsample weight has been constructed.
- Cycle-specific estimates have been examined before pooling.
- Every discontinuity has an evidence grade and analysis action.
- The manuscript states the pooled period, weighting method, compatibility checks, and sensitivity analyses.
Pool only after the compatibility gate. A larger sample can produce a more precise answer to the wrong combined question. Harmonization protects construct meaning before it protects sample size.