Clinical Research · Consumer Intelligence
Lifestyle Risk Factors
& NAFLD Severity
in an Ultrasound Cohort
A statistical analysis of 341 patients examining whether diet, exercise, ultra-processed food consumption, sugar intake, and comorbidities predict the severity of non-alcoholic fatty liver disease — conducted by True Stats Private Limited.
Study Overview
341 Ultrasound Patients
Patients who underwent abdominal ultrasound were graded from Grade 1 (mild steatosis) to Grade 4 (severe steatosis). Demographics included age and gender alongside lifestyle variables: diet, exercise, UPF consumption, sugar intake, and comorbidities.
Five Analytical Approaches
Descriptive statistics, Chi-square independence tests, Spearman rank correlation, OLS linear regression, binary logistic regression, and ROC curve analysis were applied to evaluate the association between lifestyle factors and NAFLD severity.
Lifestyle Alone Doesn't Predict Severity
No lifestyle factor showed a statistically significant independent association with NAFLD severity. The logistic regression model achieved an AUC of only 0.58, indicating limited discriminative ability — suggesting metabolic and genetic factors play a larger role.
Risk Factor Association Summary
Results across all tested demographic and lifestyle variables
| Factor | Test Used | Result | Association | p-value |
|---|---|---|---|---|
Age Group |
Chi-square (χ² = 19.65) | Trend — older patients show slightly higher grades | Weak Trend | p = 0.074 |
Gender |
Chi-square | No difference between male (49.3%) and female (50.4%) | None | p = 0.506 |
Dietary Preference |
Chi-square | Vegetarian majority across all grades; no trend | None | p = 0.894 |
Exercise Intensity |
Spearman ρ = 0.071 | Extremely weak positive; not significant | None | ρ = 0.071 |
Ultra-Processed Food |
Spearman ρ = 0.043 | No meaningful correlation with NAFLD grade | None | ρ = 0.043 |
Sugar Intake (moderate) |
OLS Regression β = −0.044 | Statistically significant but R² = 0.018 (negligible explanatory power) | Borderline | p = 0.046* |
Comorbidities (DM, HTN, CKD…) |
Chi-square (all) | No comorbidity reached significance across 7 conditions tested | None | all p > 0.05 |
* Sugar moderate consumption showed p = 0.046 in OLS regression but with R² = 0.018; the association should be interpreted cautiously as it may reflect random variation rather than clinical significance.
Key Findings
Age Shows a Non-Significant Trend
Patients over 55 showed the greatest variability in NAFLD grade and a higher frequency of severe cases. Chi-square analysis approached but did not reach significance (χ² = 19.65, p = 0.074). Younger patients predominantly had Grade 1 disease.
ROC AUC of 0.58 — Limited Prediction
The logistic regression model using all lifestyle variables achieved an AUC of 0.58 — only marginally above random classification (0.50). This confirms that lifestyle factors alone are insufficient predictors of clinically significant NAFLD in this cohort.
Sugar Significant — But Barely
Moderate sugar consumption produced p = 0.046 in the regression model (β = −0.044), but overall model R² was just 0.018. Dietary preference showed no significant association across vegetarian, omnivore, and other diet groups.
Early-Stage Disease Dominated the Cohort
77.7% of patients had Grade 1 fatty liver, indicating this emergency ultrasound population skewed toward mild disease. This concentration limits the power to detect associations with higher severity grades and should inform future study design.
Conclusion
In this emergency ultrasound cohort of 341 patients, no lifestyle variable — diet, exercise, ultra-processed food consumption, sugar intake, or comorbidities — demonstrated a statistically significant independent association with NAFLD severity. While a trend toward higher severity with advancing age was observed, it did not reach conventional significance. The logistic regression model's limited AUC of 0.58 reinforces that the investigated lifestyle factors are insufficient to predict clinically significant NAFLD. Metabolic, genetic, and environmental determinants likely play a more dominant role in disease progression.
Study Limitations
- Cross-sectional design limits causal inference between lifestyle and NAFLD severity
- Dietary intake and exercise behavior were self-reported, introducing potential recall bias
- Emergency ultrasound cohort may not represent the general NAFLD population
- Grade distribution heavily skewed toward mild disease (77.7% Grade 1), reducing statistical power for severe-grade analysis
- Future studies should incorporate metabolic markers, genetic factors, and longitudinal follow-up
Interested in a similar analysis for your clinical dataset? True Stats delivers rigorous statistical reporting for healthcare and life sciences.
Commission a Report →