Define
Outcome coded 1 for elevated measured blood pressure and 0 otherwise. Predictors were defined before modelling; age was scaled in 10-year units.
DEMONSTRATION CASE USING SYNTHETIC DATA · SYNTHETIC CASE STUDY · R
From a binary outcome to
an honest adjusted model.
A reproducible demonstration of how a public-health question becomes a logistic-regression workflow—with recoding, missingness, diagnostics and evidence limits visible.
After accounting for age and sex, are smoking and low physical activity associated with elevated blood pressure in this synthetic adult sample?
WORKFLOW
Four decisions before interpretation.
The model is only the middle of the analysis. Coding, missingness and diagnostic decisions determine what its coefficients can mean.
Audit
Ranges, duplicates and cross-field logic were checked. Seven rows with missing outcome were excluded; predictor missingness was described, not silently imputed.
Fit
A pre-specified multivariable logistic model used age, sex, current smoking and low physical activity. No stepwise p-value hunting was used.
Check
Linearity in the logit for age, influential observations, variance inflation, calibration and discrimination were reviewed before reporting.
INSPECTABLE R CODE
The method is not hidden behind screenshots.
The downloadable script imports the synthetic CSV, preserves source fields, scales age, fits the declared binomial model and prints adjusted odds ratios with confidence intervals.
Download the complete .R scriptfit <- glm(
elevated_bp ~ age_10_years + sex +
current_smoker + low_physical_activity,
data = data,
family = binomial(link = "logit")
)
exp(cbind(OR = coef(fit), confint.default(fit)))ILLUSTRATIVE RESULTS
Adjusted estimates, not causal claims.
Values below are synthetic and demonstrate reporting structure. They are not findings about a real population.
| Predictor | Adjusted OR | 95% CI | p-value |
|---|---|---|---|
| Age, per 10 years | 1.28 | 1.11–1.48 | 0.001 |
| Current smoker | 1.74 | 1.10–2.77 | 0.019 |
| Low physical activity | 1.46 | 0.94–2.28 | 0.090 |
| Female | 0.92 | 0.61–1.39 | 0.690 |
RESPONSIBLE INTERPRETATION
Holding the listed variables constant, synthetic current smokers had estimated odds of elevated blood pressure 1.74 times those of non-smokers (95% CI 1.10–2.77). Because the design is cross-sectional, this association does not establish temporal order or causality. Residual confounding and measurement error remain possible.
MODEL REVIEW
The result passed a reporting gate—not a truth test.
No concerning collinearity signal among the included predictors.
Moderate discrimination; not a clinical validation claim.
Observed-versus-predicted risk was reviewed across probability groups.
Sensitivity review did not justify undocumented deletion.
YOUR STUDY WILL NEED ITS OWN MODEL
Use the case study to inspect our thinking—then scope your question.
We will not copy this model onto a different design. Start with objectives, variables, sampling and the decision the analysis must support.