User Guide

Everything you need to know about NHANES analysis

Getting Started

NHANES to Lancet automates the process of analyzing NHANES data for epidemiological research.

Step 1: Choose Your Research Topic

Select a phenotype (Obesity, Diabetes, CVD, etc.) to auto-populate recommended variables, or manually enter NHANES variable codes.

Step 2: Configure Analysis

Select survey cycles (2017-2018 recommended), analysis type, and optionally upload a Word research proposal.

Step 3: Review Results

Review tables, figures, and generated manuscript. Download the complete results as a ZIP file.

Survey Weights Explained

NHANES uses a complex, multistage probability sampling design. Valid analyses MUST use:

  • Survey weights (WTMEC2YR for exam data) — account for selection probabilities and non-response
  • PSU (SDMVPSU) — primary sampling units
  • Strata (SDMVSTRA) — sampling strata
Multi-cycle: When combining cycles, divide the weight by the number of cycles (e.g., WTMEC2YR / 2 for two cycles).

Analysis Methods

Survey-Weighted Logistic Regression

For binary outcomes (e.g., hypertension yes/no). Produces adjusted odds ratios (OR) with 95% CI.

Kaplan-Meier Survival Analysis

Estimates survival functions over time. Generates survival curves with log-rank test.

Cox Proportional Hazards

Evaluates multiple variables on time-to-event outcomes. Produces hazard ratios (HR).

Fine-Gray Competing Risk

When competing events exist (e.g., CVD death vs non-CVD death). Produces sub-hazard ratios.

Interpreting Results

MeasureInterpretation
OR > 1Higher odds of outcome per unit increase in exposure
OR < 1Lower odds of outcome
HR > 1Higher hazard (risk) of event
P < 0.05Statistically significant at alpha = 0.05
95% CI excludes 1Significant at alpha = 0.05

STROBE Checklist

All generated papers follow STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) guidelines.

  • Title and abstract indicate study design
  • Structured abstract with Background, Methods, Findings, Interpretation
  • Scientific background and rationale
  • Eligibility criteria and data sources
  • Statistical methods with survey weight handling
  • Baseline characteristics table
  • Main results with effect sizes and confidence intervals
  • Limitations and generalizability discussion

FAQ

The platform provides a solid starting point. Always verify results against your own analysis and add domain expertise before submission.

Use the most recent cycle(s) with your variables of interest. 2017-2018 is the most recent complete cycle.

NHANES data is publicly available and de-identified. Most institutions consider NHANES analyses exempt from IRB review.