Study population
This study utilized public data from the National Health and Nutrition Examination Survey (NHANES), conducted by the National Center for Health Statistics (NCHS), covering the period from 1999 to March 2020 [9]. NHANES employs a stratified, multistage probability sampling design to ensure national representation of the U.S. population. Each biennial survey cycle includes standardized interviews, physical examinations at mobile examination centers, and laboratory tests. Written informed consent was obtained from all participants.
We included individuals aged 18 years or older with complete data for ALT, AST, platelet count, HBV and HDV test results, required for APRI and FIB-4 calculations. Participants with ALT or AST levels exceeding ten times the upper limit of normal and HCV antibody positive were excluded to eliminate cases of potential acute hepatitis or severe hepatitis, requiring liver protection therapy. After applying inclusion and exclusion criteria, a total of 54,550 participants were analyzed. A flowchart of the participant selection process is provided in Supplementary Fig. 1.
Variable definitions
Key variables in this study were defined based on methodologies established in prior NHANES research [5, 10]. Marital status was categorized as either married/living with a partner or not, while education level was divided into three groups: less than high school, completed high school, and more than high school. Race and ethnicity were classified into five groups: Mexican American, other Hispanic, non-Hispanic White, non-Hispanic Black, and other. Economic status was determined by the Poverty Income Ratio (PIR), with categories for low (< 1), medium [1,2,3], and high (> 3) income levels. Smoking status was identified as never, former, or current smokers, and alcohol and drug use were dichotomized based on survey responses. HBV vaccination history was recorded as having received 3 doses, 1–2 doses, or no doses.
Definitions of hepatitis B and D infections and liver fibrosis
Hepatitis B infection was defined by a positive HBsAg result, and HDV infection was defined by a positive HDV antibody result. Based on infection status, participants were grouped into three categories: NVH (individuals without HBV/HDV/HCV infection), HBV (HBsAg-positive only) and HBV + HDV (HBsAg positive and HDV antibody positive). Given that transient elastography results were only available from the cycle after 2017, liver fibrosis status was assessed with APRI [11] and FIB-4 [12] scores calculated as follows:
$$eqalign{&{rm APRI (11)}cr & =:frac{text{A}text{S}text{T}(text{I}text{U}/text{L})times:100}{text{U}text{p}text{p}text{e}text{r}:text{l}text{i}text{m}text{i}text{t}:text{o}text{f}:text{N}text{o}text{r}text{m}text{a}text{l}:text{f}text{o}text{r}:text{A}text{S}text{T}left(text{U}text{L}text{N}right)left(text{I}text{U}/text{L}right)text{*}text{P}text{l}text{a}text{t}text{e}text{l}text{e}text{t}:text{C}text{o}text{u}text{n}text{t}left({10}^{9}/text{L}right)}}$$
$${rm FIB-4(12)} =:frac{text{A}text{g}text{e}:left(text{y}text{e}text{a}text{r}text{s}right)times:text{A}text{S}text{T}(text{I}text{U}/text{L})}{text{P}text{l}text{a}text{t}text{e}text{l}text{e}text{t}:text{C}text{o}text{u}text{n}text{t}({10}^{9}/text{L})times:sqrt{text{A}text{L}text{T}(text{I}text{U}/text{L})}}text{}$$
Statistical analysis
To account for the extended study period, NHANES weights were recalculated for combined survey cycles, and subsequent analyses used weighted data. The survey package in R was utilized to apply weights appropriately.
Collinearity diagnostics were performed by calculating the Variance Inflation Factor (VIF) for each predictor [13], and no significant collinearity was detected (all VIFs < 5). Linear regression models were used to analyze associations between infection groups (NVH, HBV, HBV + HDV) and fibrosis scores. Regression coefficients (β) indicated the marginal effect of HBV or HBV + HDV infection on APRI and FIB-4 scores, with the NVH group as the reference. A P-trend test assessed the sequential increase in fibrosis scores from NVH to HBV to HBV + HDV groups. Estimated marginal means (emmean) were calculated to assess the significance of differences in APRI and FIB-4 scores between the HBV + HDV and HBV groups.
A multi-model strategy was employed to enhance the robustness of the analysis. Model I was an unadjusted baseline model. Model II included adjustments for demographic factors: age, gender, marital status, education level, birthplace and race/ethnicity. Model III further accounted for lifestyle factors, including smoking, alcohol use, PIR, BMI and drug use. Finally, Model IV incorporated additional adjustments for medical conditions: hypertension, diabetes, HBV vaccine history, kidney disease, heart failure, coronary heart disease, stroke and thyroid disease.
To refine the comparison between HBV and HBV + HDV groups, we performed 1:3 optimal propensity score matching without replacement using the MatchIt package in R [14, 15]. The propensity scores were estimated via logistic regression with the following covariates: Demographics: Age, gender, race, birthplace, marital status; Socioeconomic factors: Education level, PIR. Clinical/lifestyle variables: BMI, hypertension, diabetes, HBV vaccine history, alcohol use, smoking, kidney disease, heart failure, coronary heart disease, stroke and thyroid disease.
Covariate balance was evaluated using standardized mean differences (SMD), with a threshold of |SMD| < 0.1 indicating adequate balance. Balance diagnostics was conducted by visualization in Supplementary Fig. 2. Variables with residual imbalances (SMD ≥ 0.1) were included as covariates in the final post-matching regression model to further mitigate confounding.
Statistical significance was set at P < 0.05, and analyses were performed using R version 4.4.1.