Description
NHANES Study
A publication-ready cross-sectional or trend study built on the National Health and Nutrition Examination Survey, analysed the way NCHS specifies rather than the way most submitted NHANES papers do. The distinction matters more every year: NHANES is the most heavily mined public dataset in clinical research, editors know it, and the single most common reason an NHANES manuscript is desk-rejected is that the complex survey design was ignored. This service exists to produce the version that survives that screen.
When This Is the Right Study
NHANES answers questions about the non-institutionalised US population where the exposure and outcome are both measured in the survey, dietary intake, laboratory biomarkers, anthropometry, physical activity, environmental exposures, self-reported conditions, and mortality through the linked National Death Index. It is the right instrument when you need nationally representative prevalence, an exposure-outcome association across a general population, or a trend across cycles.
It is the wrong instrument for anything requiring longitudinal follow-up within individuals, rare disease, hospital-based outcomes, or treatment effects. NHANES is cross-sectional by design; each participant is measured once. If your question needs incidence rather than prevalence, we will say so before you buy rather than deliver a study that a reviewer will reject on that basis.
How the Analysis Is Actually Done
The design is not optional. NHANES uses a stratified, multistage probability sample with unequal selection probabilities, and every published estimate must account for it.
- Weight selection, not weight guessing – The correct weight depends on which variables your analysis uses. Interview-only variables take WTINT2YR; anything drawn from the mobile examination centre takes WTMEC2YR; fasting subsample biomarkers and dietary day-two recalls take their own weights. Using the wrong one silently biases every estimate in the paper.
- Correct multi-cycle reweighting – Combining cycles is not concatenation. Weights are divided by the number of cycles pooled, and we verify that every variable was collected with comparable methodology across the whole window before pooling. Where a laboratory method changed mid-window, NCHS regression-correction equations are applied or the affected cycles are excluded, with the decision documented.
- Subpopulation analysis, not row deletion – Restricting to adults, or to a disease group, is done with a subpopulation option so that variance estimation retains the full sample structure. Dropping the rows instead produces standard errors that are wrong in a direction you cannot predict.
- Taylor series linearisation – Applied with SDMVSTRA and SDMVPSU, with degrees of freedom calculated as PSUs minus strata. Small subgroups that fall below stable-estimate thresholds are reported as such rather than presented with false precision.
- DAG-based confounder selection – The adjustment set is chosen from an explicit causal diagram. We do not use stepwise selection on the outcome, and we do not adjust for mediators or colliders, both of which are routine errors in the published NHANES literature.
- Non-linearity tested, not assumed – Restricted cubic splines with knots at conventional percentiles, plus threshold and segmented analysis where the relationship is genuinely non-monotonic.
- Multiplicity handled – Where several exposures or outcomes are tested, false-discovery-rate correction is applied and reported.
- Missing data addressed explicitly – Pattern examined, mechanism reasoned about, and multiple imputation applied where appropriate, with a complete-case sensitivity analysis alongside.
What You Receive
- Complete manuscript – Introduction, methods, results, discussion and limitations, written to the reporting conventions of your target journal
- Survey-weighted tables – Including the weighted descriptive table reviewers expect first
- Publication-quality figures – Vector output, sized and lettered for the journal
- Full analysis code – R (survey package) or Stata (svy prefix), commented and runnable, so the analysis is reproducible by you or by a reviewer
- Completed STROBE checklist – See our STROBE tutorial for what reviewers actually check
- Analysis log – Every cycle included, every variable derived, every exclusion and why
- One round of author-requested revisions
What Is Not Included
We do not ghost-write authorship. You direct the research question, review the interpretation, and take responsibility for the scientific claims, which is what makes the resulting paper legitimately yours. We do not submit on your behalf, we do not guarantee acceptance at any journal, and we do not respond to peer review as part of this package (that is a separate service). Article processing charges are yours. If the analysis produces a null result, we write up the null result; we do not go looking for a significant one.
When We Will Tell You Not to Buy This
If your question has already been answered several times in NHANES with the same exposure and outcome, a further paper will struggle regardless of how well it is executed. If the variables you need were collected in only one cycle, the sample will not support subgroup analysis. If your outcome has too few events in the weighted sample for a stable estimate, no amount of modelling repairs that. We check all three before starting, and if the answer is that the study should not be done, we say so and refund. The Feasibility & Protocol Assessment exists precisely so that this check can happen before a full commitment.
Data Access
NHANES is fully public. We obtain the data ourselves, there is no application, no data use agreement, and nothing for you to arrange. Mortality-linked analyses use the publicly released NDI linkage files.
Timeline
Standard delivery is 15 business days from an agreed analysis plan. Multi-cycle trend studies and imputation-heavy designs may run longer; if yours will, we tell you before you pay, not afterwards.
Common Questions
Will the analysis be reproducible? Yes. You receive the code and the analysis log, and the data are public, so anyone, including a sceptical reviewer, can rerun it.
Do I need to register the study? Not required, but increasingly expected by some journals even for retrospective work. We cover the reasoning in this article on registering retrospective studies.
Can you use my target journal’s format? Yes. Tell us the journal at the start and the manuscript is written to its structure and word limit.
What if I already have a draft that was rejected? Send it. Rework of a desk-rejected NHANES manuscript is a common request and often costs less than a study built from scratch.
Extended Scope
Multi-cycle trend decomposition, propensity-matched designs, mediation analysis, and linkage of NHANES to other data sources are quoted individually, typically from $3,500. If your question spans more than one database, see the Extended & Multi-Database Registry Study.
Have a Question First?
If you are not sure this is the right fit for your research question, ask before you buy. Send it below and a senior analyst replies within one business day. There is no obligation, and if a different product suits you better, or if the honest answer is that we cannot help, we will say so.
