Seroprevalence and determinants of SARS-CoV-2 anti-nucleocapsid IgG in a vaccinated Iranian adult population: A cross-sectional study, Summer 2023

ABSTRACT

Anti-nucleocapsid (anti-N) IgG serology is increasingly used to distinguish natural SARS-CoV-2 infection from vaccine-induced immunity. However, post-vaccination seroepidemiological data from highly immunized Middle Eastern populations remain limited. We assessed anti-N IgG seroprevalence and its predictors in a cohort of Iranian adults during the summer of 2023. A total of 251 individuals aged 18–65 years were enrolled through community outreach at Gholhak Laboratory, a high-throughput diagnostic center in Tehran. Participants completed structured questionnaires that captured demographic characteristics, COVID-19 vaccination history (vaccine platform, dose count, and time since last dose), smoking status, occupation, pregnancy status, and prior COVID–19–related hospitalization. Anti-N IgG levels were measured using a commercial ELISA (Pishtaz Teb, Tehran, Iran); seropositivity was defined by a signal-to-cutoff ratio ≥ 1.1. Prevalence ratios (PRs) were estimated via Poisson regression with robust variance, adjusting for covariates including age, gender, BMI, smoking, and vaccination characteristics. The overall anti-N seroprevalence was 55.8% (95% CI: 49.7–61.7). Seropositivity was significantly associated with female gender (PR for males 0.70, p = .002), older age (3% higher per decade, p = .017), and nonsmoking status (PR for smokers 0.70, p = .008). No associations were observed with vaccine platform, dose count, or recency. Sensitivity analyses using alternative ELISA classifications and penalized regression yielded similar results. Despite near-universal vaccination, more than half of the participants exhibited serological evidence of past infection. These findings suggest that additional in this urban Iranian setting, additional booster doses may offer limited additional protection against infection and underscore the influence of gender, age, and smoking on natural infection risk.

Introduction

The coronavirus disease 2019 (COVID-19) pandemic has caused more than 700 million confirmed cases and over 6 million deaths worldwide, reshaping public-health priorities on an unprecedented scale.¹ Rapid deployment of highly efficacious vaccines markedly reduced hospitalizations and fatalities, yet none of the currently authorized platforms provide durable sterilizing immunity; breakthrough infections remain common, particularly beyond six to twelve months after the last dose.^2,3 Consequently, population-based serosurveillance is essential for quantifying cumulative infection incidence and refining vaccination policies in the post-licensure era.

Routine serological surveys typically measure antibodies to the spike (S) protein, as anti-S titers correlate with neutralizing activity and vaccine response.⁴ However, widespread use of S-based vaccines complicates the distinction between vaccine-induced and infection-induced immunity. The nucleocapsid (N) protein is absent from licensed vaccines; thus, anti-N immunoglobulin G (IgG) serves as a specific marker of natural infection.⁵ Large-scale studies from North America and Europe have used this distinction to show that, despite high vaccine coverage, an estimated 70–90% of adults had been infected with SARS-CoV-2 by mid-2023.^6,7 Comparable data from the Middle East remain limited, and most Iranian reports pre-date mass vaccination or rely on anti-S assays that cannot disentangle vaccine response from prior viral exposure.^8,9

Iran’s national vaccination program, launched in February 2021, initially relied on the inactivated Sinopharm vaccine, later incorporating adenoviral-vectored (AstraZeneca, Sputnik V), protein-subunit (SpikoGen, PastoCovac), and domestically-produced whole-virus platforms (Barkat).¹⁰ By December 2022, over 90% of adults had completed a two-dose schedule, and more than half had received at least one booster.¹⁰ In this context of diverse vaccine platforms and near-universal coverage, it is unclear whether additional boosters substantially reduce infection risk or whether demographic and behavioral factors now dominate transmission dynamics.

This study evaluates anti-N IgG seropositivity as a marker of prior SARS-CoV-2 infection in a highly vaccinated adult cohort in Iran. Our primary objective was to estimate the prevalence of anti-N IgG and to identify demographic, clinical, and vaccination-related correlates of seropositivity during the Omicron period. Because most vaccine platforms used globally do not induce anti-N, anti-N positivity in this setting largely reflects prior infection, with potential contributions from inactivated whole-virion vaccines. We hypothesized that demographic and behavioral factors (e.g., sex, age, and city of residence) would be more strongly associated with anti-N seropositivity than vaccine-related metrics (platform, dose number, timing), consistent with widespread community transmission in this cohort.

To address this hypothesis, we recruited 251 adults from a high-throughput diagnostic center in Tehran during the summer wave of 2023. Using a validated indirect ELISA specific for anti-N IgG, we (i) quantified the seroprevalence of natural infection more than two years into Iran’s vaccination campaign, and (ii) examined the independent contributions of demographic, clinical, and vaccination variables to infection risk. By integrating detailed vaccination histories with anti-N serology, we aimed to clarify whether residual susceptibility in a highly vaccinated Middle Eastern population is driven primarily by vaccine-related factors or by demographic and behavioral characteristics.

Materials and methods

Study design and setting

This cross-sectional study was conducted between 15 June and 30 August 2023. Participants were recruited from a single high-throughput diagnostic facility, Gholhak Laboratory in northern Tehran, during routine visits for a blood grouping test. Although the center attracts individuals from diverse demographic backgrounds and surrounding satellite cities, recruitment from a single urban healthcare facility may introduce selection bias. Individuals attending medical diagnostic centers may differ systematically from the general population in health awareness, healthcare access, baseline health status, or exposure risk, potentially limiting generalizability. To contextualize the representativeness of our single-center urban sample, we benchmarked key sociodemographic and vaccination indicators against national data. Sex distribution and urbanization were drawn from the 2024 National Population and Housing Census (Statistical Center of Iran)¹¹ and UN/World Bank urbanization series;¹² age structure (median age) from UN DESA World Population Prospects 2024 (2023 estimates);¹³ and national COVID-19 vaccination coverage from peer-reviewed syntheses and curated dashboards (Our World in Data/WHO).¹⁴ Because many dashboards report coverage as a share of the total population, we also cite adult-specific vaccination rates where available to align with our adult cohort.

The study followed the STROBE reporting guideline for observational studies and was approved by the Institutional Review Board of Shahid Beheshti University of Medical Sciences (IR.SBMU.MSP.REC.1403.353). All participants provided written informed consent.

Participants and recruitment

Adults aged 18–65 years attending the laboratory for routine blood tests or SARS-CoV-2 screening were approached consecutively by trained research nurses. Exclusion criteria were: (i) acute respiratory symptoms within the preceding seven days, (ii) current immunosuppressive therapy, or (iii) refusal to give venous blood. The target sample size (~250 participants) was calculated to estimate anti-N IgG seroprevalence with a ± 6.1% half-width at 95% confidence, assuming p = .50 (conservative). This precision was adequate for prevalence estimation but not specifically powered to detect small effect sizes in multivariable models. A total of 251 participants were enrolled, meeting the precision target.

Data collection

Demographic and behavioral data were collected via a structured interviewer-administered questionnaire and verified against electronic health records when available. Variables included: age, gender, height, weight, smoking status (current vs never/former), pregnancy status, occupation (medical-related vs other), residential location (Tehran vs other cities), prior COVID-19–related hospitalization, and complete vaccination history (vaccine platform, number of doses, and calendar month of most recent dose). Body mass index (BMI) was calculated as weight in (kg) divided by height (m²). We did not systematically capture self-reported prior PCR/antigen test results or detailed symptom histories; anti-N IgG served as the primary marker of previous infection.

Laboratory procedures

Five milliliters of venous blood were collected from each participant and centrifuged within two hours. Serum aliquots were stored at −80°C and analyzed in weekly batches. Anti-nucleocapsid IgG concentrations were measured with a commercial indirect ELISA kit specific for antibodies against the SARS-CoV-2 nucleocapsid protein (Pishtaz Teb, Tehran, Iran; catalog PT-CoV-N-IgG). All assays were performed in duplicate according to the manufacturer’s instructions, with positive, negative, and blank controls on every microplate. The threshold for seropositivity was set at 1.1 times the mean optical density (OD) of three negative-control wells containing pooled pre-pandemic sera. Samples above this threshold were classified as positive; those below were classified as negative. Results within ±10% of the cutoff were considered indeterminate. Inter-assay and intra-assay coefficients of variation were < 6% and < 4%, respectively.

Variable definitions

Vaccination history was recorded primarily at the platform level. Platforms were categorized as: (i) whole virion (e.g., BBIBP-CorV/Sinopharm, COVIran Barekat), (ii) protein-subunit (e.g., SpikoGen, and (iii) adenovirus-vectored (e.g., ChAdOx1/AstraZeneca, Sputnik V). Brand names were not systematically recorded for all participants. The database field labeled “non-vector” pooled recipients of inactivated and subunit vaccines; these platforms differ in their ability to elicit anti-N antibodies, and this heterogeneity was considered a limitation in analysis. Dose count was coded from 0 to 5 (0 = unvaccinated). The time since the last dose was calculated from the sample collection date to the most recent vaccination and grouped into 1–6, 7–12, 13–18, 19–24, and > 24 months. Indeterminate ELISA results (n = 16, 6.4%) – defined as OD values within ± 10% of the assay cutoff were retested in duplicate. Classification was based on the mean OD of the repeat tests compared with the predefined cutoff. For primary analyses, indeterminate results were excluded; in sensitivity analyses, they were recoded as positive or negative to assess robustness. The primary outcome is anti-N IgG positivity versus negativity.

Statistical analysis

Analyses were performed using Stata 17 (StataCorp, College Station, TX, USA). Continuous variables are reported as means ± standard deviation or medians with interquartile range, and categorical variables as counts and percentages. Exact binomial 95% confidence intervals (CIs) were calculated for overall seroprevalence. Associations between participant characteristics and anti-N IgG seropositivity were assessed using Poisson regression with robust (sandwich) variance to estimate prevalence ratios PRs). The initial (saturated) model included gender, age, BMI, smoking, status, city of residence, occupation, pregnancy status, vaccine platform, dose number, and months since last dose. Age and BMI were also modeled as restricted cubic splines with four knots; interaction terms (Smoking × Vaccination, Gender × Vaccination) were added sequentially. Models were compared using Akaike’s Information Criterion (AIC), with the lowest-AIC specification retained as primary. Goodness-of-fit was assessed with deviance residuals. Model discrimination was evaluated using the area under the receiver-operating characteristic curve (AUC). Ninety-five percent CIs for AUCs were obtained via 1,000-fold bootstrap resampling stratified by outcome. AUCs were computed for a multivariable logistic regression model and a class-weighted Random Forest using the same predictor set and participant sample. Covariates were coded as: sex (male versus female); age (continuous, per 1-year increase); BMI (continuous, per 1 kg/m²); city of residence (Tehran versus other cities); occupation (healthcare worker versus non-healthcare); vaccine platform (inactivated whole-virion, adenovirus-vectored, protein-subunit); and dose number (0–5). These codings were applied in both the Poisson regression (PRs) and logistic regression (ORs with 95% CIs for Figure 3). Statistical significance was defined as a two-sided α = 0.05.

Figure 3.

Multivariable logistic regression for antiN IgG seropositivity. Points show adjusted odds ratios; bars show 95% CIs. Reference categories: female (gender), other cities (city of residence), Nonhealthcare (occupation), inactivated (vaccine platform), and 2 doses (dose number). Age is modeled per 1 year increase; BMI per 1 kg/m² increase. Exact ORs and CIs are provided to two decimal places in the panel labels. Indeterminate ELISA results were excluded from the primary model (see methods); sensitivity analyses yielded consistent estimates.

Results

Cohort description and vaccination history

Two hundred fifty-one adults were enrolled (response rate = 90.3%); their median age was 36 years (inter-quartile range [IQR] 30–45), and 143 (57%) were female. Mean body-mass index (BMI) was 25.2 ± 4.3 kg m²; BMI categories comprised 3% under-weight, 44% normal-weight, 36% overweight, and 17% obese. Current smoking was reported by 38 participants (15%). Thirty-four women were of reproductive age (15–50 years), three of whom were pregnant at sampling. Nearly all participants (240/251, 95.6%) had received at least one COVID-19 vaccine dose; the distribution of cumulative doses was: 0 doses 11 (4.4%), 1 dose 11 (4.4%), 2 doses 104 (41.4%), 3 doses 101 (40.2%), and ≥ 4 doses 24 (9.6%). Due to Iran’s extensive national COVID-19 vaccination program, the vast majority of adults had received at least one vaccine dose, resulting in very few unvaccinated individuals in the sample (n = 11). This sparse representation substantially limits the statistical power to estimate the true effect of vaccination metrics and accurately interpret associations involving vaccination status.

Whole-virus or subunit products (Sinopharm, SpikoGen, PastoCovac, Barkat) accounted for 188 vaccinations (75%), whereas adenovirus-vectored vaccines (AstraZeneca, Johnson or Sputnik V) accounted for 52 (21%). Median time since the most recent dose was 11 months (IQR 9–16). Internal data checks revealed no logical discordance between the recorded platform and dose count (Table 1).

Table 1.

Participant characteristics.

Variable	Value
Age, median (IQR), years	36 (30–45)
Sex	Female: 143 (57.0%)
	Male: 108 (43.0%)
BMI, mean ± SD (kg/m2)	25.2 ± 4.3
BMI Categories	Underweight: 3.2%
	Normal: 44.2%
	Overweight: 35.9%
	Obese: 16.7%
Current smoker	38 (15.1%)
Pregnant (among women 15–50y)	3/34 (8.8%)
Occupation	Medical-related: 74 (29.5%)
	Other: 177 (70.5%)
Residence	Tehran: 187 (74.5%)
	Other cities: 64 (25.5%)
Vaccination status	≥1 dose: 240 (95.6%)
	Unvaccinated: 11 (4.4%)
Dose count	0: 11 (4.4%)
	1: 11 (4.4%)
	2: 104 (41.4%)
	3: 101 (40.2%)
	≥4: 24 (9.6%)
Time since last dose (median, IQR)	11 months (9–16)
Vaccine platform	Inactivated/subunit: 188 (74.9%)
	Adenovirus-vectored: 52 (20.7%)
	Unvaccinated: 11 (4.4%)

Demographic, clinical, and vaccination characteristics of the study cohort (N = 251). Age and time since last dose are presented as medians with interquartile ranges (IQR). Body mass index (BMI) is expressed as mean ± standard deviation (SD). Vaccination status includes all major platforms administered in Iran. BMI categories were defined using WHO adult cutoffs. Vaccination platform 1 includes Sinopharm, SpikoGen, PastoCovac, Barkat, Pfizer, and Johnson; platform 2 includes AstraZeneca and Sputnik V.

Relative to national benchmarks, our cohort (all urban attendees of a high-throughput diagnostic center) is likely more urban, somewhat older than the national median age (Iran, 2023 median age ≈33 years),¹³ and slightly more male than the near-balanced national sex distribution (2024 census ≈ 51% male, 49% female).¹¹ Nationally, by December 2022, an estimated 77.5% of the total population had received ≥ 1 dose and 69.7% had completed two doses; among adults ≥18 years, coverage was higher (first dose 91.7%, second dose 84.2%).¹⁴ Published summaries for mid2023 to 2024 place ≥ 1dose coverage at ~ 77.6–79% and fullseries coverage at ~ 71% in the total population.¹⁵ Our cohort’s very low unvaccinated share (4.4%), therefore, indicates coverage above national averages, consistent with selective attendance at an urban diagnostic center.

Anti-nucleocapsid IgG seroprevalence: Univariate patterns

Overall, 140 participants were anti-N IgG positive, yielding a crude seroprevalence of 55.8% (95% confidence interval [CI] 49.7–61.7). Stratification by gender showed positivity in 63.6% of females versus 46.3% of males (χ² = 11.9, p = .0006). Prevalence rose modestly with age, from 49% in the youngest decile (18–24 years) to 66% in the eldest (56–65 years), although the overall χ² trend across deciles was not significant (p = .15). Smokers exhibited lower seroprevalence than nonsmokers (44.7% vs 57.7%, P = .017). No univariate differences were observed across BMI categories (p = .11), residential location (Tehran vs other, p = .69), occupation (medical vs other, p = .73), pregnancy status (p = .83), vaccine platform (p = .86), cumulative dose count (p = .52), or time since last dose (Cochran – Armitage trend p = .74) (Table 2; Figures 1 and 2).

Table 2.

Anti-N IgG seroprevalence by subgroups.

Subgroup	n	Seropositive, n	Seroprevalence (95% CI)
Overall	251	140	55.8% (49.7–61.7)
Sex
Female	143	91	63.6% (55.4–71.2)
Male	108	50	46.3% (36.8–55.9)
Age Group
18–24 years	30	15	50.0% (31.9–68.1)
25–34 years	60	31	51.7% (38.2–64.9)
35–44 years	65	36	55.4% (42.9–67.3)
45–54 years	56	37	66.1% (52.7–77.7)
55–65 years	40	26	65.0% (48.3–78.8)
Smoking Status
Current smoker	38	17	44.7% (29.5–60.8)
Nonsmoker	213	123	57.7% (50.9–64.2)
BMI Category
Underweight	8	4	50.0% (15.0–85.0)
Normal	111	68	61.3% (51.9–70.0)
Overweight	90	48	53.3% (42.7–63.6)
Obese	42	20	47.6% (32.0–63.6)
Vaccine Dose Count
0 doses	11	7	63.6% (31.6–87.6)
1 dose	11	5	45.5% (16.7–76.6)
2 doses	104	59	56.7% (46.9–66.0)
3 doses	101	56	55.4% (45.5–64.9)
≥4 doses	24	13	54.2% (33.2–74.2)
Vaccine Platform
Inactivated/subunit	188	105	55.9% (48.7–62.8)
Adenovirus-vectored	52	30	57.7% (43.2–71.2)
Non-vaccinated	11	5	45.5% (16.7–76.6)
Time Since Last Dose
1–6 months	25	13	52.0% (31.3–72.2)
7–12 months	84	47	56.0% (45.3–66.1)
13–18 months	72	39	54.2% (42.3–65.5)
19–24 months	42	29	69.0% (52.1–82.4)
>24 months	17	12	70.6% (44.0–89.7)

Crude anti-nucleocapsid IgG seropositivity rates across demographic, behavioral, and vaccination-related subgroups. Exact binomial 95% confidence intervals (CI) are shown. Seroprevalence was highest in older age groups and females, and lowest among current smokers. Subgroups with n < 10 are included for completeness but should be interpreted cautiously. Seropositivity was defined as S/CO ≥ 1.1 per ELISA manufacturer guidelines.

Figure 1.

Age- and sex-specific anti-N IgG seroprevalence. Seroprevalence of SARS-CoV-2 anti-nucleocapsid IgG antibodies stratified by sex and age group (N = 251). Seroprevalence was consistently higher among women across all age groups, with the largest sex gap observed in the youngest strata. Gray = female; green = male. Values above bars indicate % seropositive.

Figure 2.

Seroprevalence by vaccination dose count and platform. Comparison of anti-N IgG seropositivity across vaccination dose counts and platforms. Gray bars represent recipients of inactivated/subunit vaccines (e.g., Sinopharm, SpikoGen), and green bars represent adenovirus-vectored vaccines (e.g., AstraZeneca, Sputnik V). Seropositivity did not differ significantly by dose number or platform.

Independent predictors of seropositivity

In the saturated Poisson model with robust variance, including age, BMI, gender, smoking, location, occupation, pregnancy, vaccine platform, dose count, and months since last dose, three variables retained statistical significance. Male gender remained inversely associated with seropositivity (adjusted prevalence ratio [aPR] 0.70, 95% CI 0.55–0.88, p = .002). Current smoking was likewise protective (aPR 0.70, 95% CI 0.54–0.91, p = .008). Age exerted a small but significant positive effect: each additional year corresponded to a 1% higher probability of anti-N positivity (aPR per year 1.01, 95% CI 1.00–1.02, p = .017). Neither BMI nor any vaccination metric – receipt of any vaccine (aPR 0.86, 95% CI 0.53–1.40, p = .55), cumulative dose, platform type, nor months since last dose – was independently associated with seropositivity. Introducing restricted cubic splines for age and BMI, and interaction terms for Smoking × Vaccination and Gender × Vaccination, did not materially alter model fit; the spline + interaction specification yielded the lowest Akaike Information Criterion yet confirmed the same trio of significant predictors (Table 3; Figure 3).

Table 3.

Multivariable poisson model of independent predictors of seropositivity.

Variable	Adjusted Prevalence Ratio (aPR)	95% CI	p-value
Sex: Male (vs Female)	0.70	0.55–0.88	.002
Age (per year)	1.01	1.00–1.02	.017
Smoking: Yes (vs No)	0.70	0.54–0.91	.008
BMI	NS	–	>.20
Vaccine platform	NS	–	>.50
Dose count	NS	–	>.50
Last dose recency	NS	–	>.40
Pregnancy	NS	–	>.30
Location	NS	–	>.30
Occupation	NS	–	>.30

Adjusted prevalence ratios (aPR), 95% confidence intervals, and p-values from a robust Poisson regression model evaluating associations between participant characteristics and anti-N IgG seropositivity. Sex, age, and smoking status remained significant after adjustment. NS = Not statistically significant. Variables not reaching significance were retained in the model to adjust for confounding.

Sensitivity and robustness analyses

Re-classifying ELISA-indeterminate results as positive or as negative left all vaccination-related aPRs qualitatively unchanged (range 0.88–0.96, p > .55) and preserved the significance of gender, age, and smoking. These results, along with predictive model performance metrics for alternative classification algorithms, are summarized in Table 4. Firth-penalized logistic regression to address sparse unvaccinated strata, as well as models restricted to vaccinated participants only, produced concordant estimates. Platform and dose distributions for the full cohort are provided in Supplementary Table S1, and anti-N IgG serostatus by platform in Supplementary Table S2.

Table 4.

Sensitivity analyses and predictive model performance.

Analysis Type	Key Results
Indeterminate reclassified as Positive	aPR range: 0.88–0.96; p > .55
Indeterminate reclassified as Negative	aPR range: 0.88–0.96; p > .55
Firth-penalized logistic regression	Effect estimates consistent with the primary model
Model restricted to vaccinated participants	Same trio of predictors significant; vaccine terms NS
Logistic regression (Predictive model)	AUC = 0.64
Random forest (Predictive model)	AUC = 0.64
Gradient boosting (Predictive model)	AUC = 0.56

Results of sensitivity analyses evaluating robustness of findings to alternative assumptions and modeling strategies. Predictive model performance (AUC) is shown for three classification algorithms. AUC = Area under the receiver-operating characteristic curve. Logistic regression used Firth penalization as noted. All models confirmed the limited predictive power of vaccination metrics alone in determining seropositivity.

Severe outcomes and vaccine breakthrough events

Four participants (1.6%) reported COVID–19–related hospitalization; three had received ≥ 2 vaccine doses. Penalized logistic regression adjusting for age suggested lower – but imprecisely estimated – odds of hospitalization among the vaccinated (odds ratio 0.68, 95% CI 0.07–6.3). No association emerged between time since last dose and hospital admission within the vaccinated subset (p = .58). Owing to the extremely small number of events, these findings are considered descriptive.

Model performance and predictive value

Penalized logistic regression, random forest, and gradient boosting classifiers incorporating all demographic and vaccination variables achieved mean five-fold cross-validated areas under the receiver-operating characteristic curve (AUC) of 0.64, 0.64, and 0.56, respectively, consistent with the limited feature set available (demographics, geography, vaccine platform/dose, BMI). Accuracy and F1scores were concordant with these AUCs (Figure 4).

Figure 4.

ROC curves models predicting antiN IgG positivity. Curves and 95% bootstrap CIs are shown for multivariable logistic regression and classweighted random forest fit on identical predictors (gender, age, BMI, city, occupation, vaccine platform, dose number). AUCs range from 0.56 to 0.64, consistent with modest discrimination using demographics and coarse vaccine variables alone. Diagonal line = chance performance.

Estimated impact of hypothetical booster scenarios

Setting the time since last vaccination to ≤6 months for all vaccinated individuals reduced the model-predicted number of infections from 133 to 129 (2.9% preventable). Simulating universal uptake of ≥ 4 doses increased the modeled infection count by 3.5%, reflecting the absence of a protective dose–response relationship in the empirical data (Figure 5).

Figure 5.

Counterfactual impact of alternative booster strategies. Predicted infection counts under three simulated vaccination scenarios. Relative to the observed data (gray), universal recent boosting (green) is estimated to reduce infection burden by 2.9%, while universal uptake of ≥ 4 doses (blue) show no protective effect and marginally increases modeled infections. Labels indicate % change from observed scenario.

Discussion

This cross-sectional investigation provides a detailed snapshot of natural SARS-CoV-2 infection in a highly vaccinated urban Iranian population two-and-a-half years into the national immunization program. Using an anti-nucleocapsid ELISA that distinguishes natural infection from vaccine-induced immunity, we found that more than half of adults tested at a large diagnostic center carried anti-N IgG. Seroprevalence was independently associated with female gender, older age, and absence of current smoking, yet was not measurably influenced by any vaccination metric, receipt of vaccine, platform type, cumulative dose count, or recency of the last dose. These findings add to growing evidence that, while vaccination remains highly effective at preventing severe disease, its protection against infection wanes, particularly with immune-evasive Omicron sub-lineages.

Few Middle-Eastern datasets combine detailed vaccination records with anti-N serology. In a nationwide Iranian study from late 2020, before large-scale vaccination, anti-N seroprevalence was 17%.⁹ The six-fold rise to 56% in our survey highlights intense community transmission during the Omicron era despite near-universal vaccination. Our null association between dose count or time since last dose and infection aligns with findings from England’s VIVALDI cohort, where anti-N prevalence during the BA.4/BA.5 wave was largely independent of booster status.¹⁶ The marked female excess in infection risk mirrors earlier Iranian work¹⁷ but contrasts with several Western cohorts where gender differences dissipated after adjustment for occupation and household size.¹⁸ The higher anti-N IgG seroprevalence among women in our study may largely reflect increased exposure through gendered roles, rather than biological susceptibility. As we did not collect occupational or household exposure data, these remain plausible confounding factors, and we caution against causal interpretation of sex differences in seroprevalence without further targeted behavioral data.

The observed inverse association between smoking and seropositivity (smokers: 44.7% versus nonsmokers: 57.7%, p = .017) is consistent with the “smoker paradox” reported in previous research.¹⁹ This relationship may be influenced by confounding factors, including behavioral and occupational variables. For example, smokers may have different patterns of occupational or social exposure, or distinct healthcare-seeking behaviors that affect testing and diagnosis rates. Mechanistic hypotheses include reduced ACE-2 expression in smokers’ lungs, though evidence is mixed. While intriguing, this association warrants cautious interpretation, and future studies should incorporate detailed occupational, behavioral, and immunologic data to clarify underlying mechanisms.

Our models exhibited limited discrimination (AUC 0.56–0.64), as expected for frameworks lacking direct exposure metrics, timing of infection/vaccination, immunologic correlates (e.g., neutralization titers, calibrated anti-S IgG), and host genetic factors. Prior work shows that household and other exposure variables are among the strongest predictors of infection, and that immune markers and genetics explain additional variance in susceptibility. Future studies should incorporate structured exposure histories, standardized immunologic measurements, and, where feasible, genetic data to improve predictive performance.

Strengths of the study include the use of a nucleocapsid-specific assay, detailed vaccination histories across multiple platforms, and robust analytical methods with extensive sensitivity analyses. Nevertheless, several limitations temper inference.

Participants were recruited from a single high-throughput diagnostic center, potentially selecting for individuals with greater health awareness or healthcare access, which may bias estimates and limit generalizability. Additionally, the cross-sectional design also precludes causal inference, and our estimates of natural infection may be conservative due to antibody waning and potential seroreversion. Additionally, prior PCR/antigen results and symptom histories were not collected, preventing classification by documented infection status or exposure intensity. Given anti-N waning and potential seroreversion, this may underestimate cumulative infection and attenuate associations toward null.

The small number of unvaccinated participants (n = 11) limits statistical power for vaccine-related analyses and increases the risk of type II error. The total sample size (n = 251) may also have reduced power for detecting subtle associations in smaller subgroups. The high number of predictors relative to sample size raises the risk of overfitting, potentially reducing generalizability, and the very limited number of COVID–19–related hospitalizations (n = 4) precluded. Although we sought to include all relevant covariates to adjust for confounding, future studies with larger sample sizes or the application of regularization techniques (e.g., Lasso or Ridge regression) may better mitigate this risk and enhance model robustness.

Predictive modeling based on demographic and vaccination variables yielded only modest discriminative performance (AUC ~0.6), suggesting substantial unmeasured heterogeneity in infection risk. To improve predictive capacity and mechanistic insight, future models should incorporate detailed exposure histories, host genetic factors, mucosal immune profiles, and social contact data.

The high burden of natural infection despite widespread vaccination suggests that additional booster campaigns aimed solely at preventing transmission may offer diminishing returns unless more effective sterilizing vaccines become available. Our counterfactual modeling estimated that universal recent boosting would avert, at most, three infections per 100 vaccinated adults under current antigenic conditions. By contrast, the pronounced gender- and smoking-related differentials suggest that behavioral and biological factors may now exert a greater influence on exposure heterogeneity than immunization status alone. Tailored public health messaging for high-exposure female occupational groups, along with mechanistic research on the smoker effect, could yield greater dividends. Importantly, the absence of a vaccine effect on infection should not be conflated with the well-documented protection against severe disease, which our study was underpowered to assess.

Recruitment from a single, urban, high-throughput diagnostic center likely selected individuals who differ from the general Iranian population in age, sex balance, healthcare access, and vaccination coverage. National data show a near-balanced sex distribution (2024 census ≈ 51% male, 49% female), 2023 median age ≈33 years, and ~ 75% urban residence – contrasting with our 100% urban sample and a male-leaning, somewhat older profile.¹³ Moreover, while national coverage around the study period was high (≥1 dose ~ 77.6–79%; full series ~ 71% in the total population; adults often higher¹⁴), our cohort’s unvaccinated fraction was only 4.4%, implying higher vaccine uptake than the national average. These differences may inflate or attenuate seroprevalence and association estimates in directions that limit generalizability to younger, rural, or less vaccinated groups; accordingly, we interpret external validity with caution and report these benchmarks in line with STROBE guidance on generalizability.

Future research should ensure brand-specific vaccine ascertainment, incorporate paired anti-N/anti-S testing (and neutralization assays) to distinguish vaccine from infection-induced immunity, and include longitudinal follow-up to track antibody kinetics. Prospective cohort designs with serial serologic measurements, viral sequencing, and cellular immunity profiling will help clarify how hybrid immunity evolves in Middle Eastern populations. Larger multicenter samples will allow finer stratification by vaccine schedule and comorbidities, and linkage to hospital records will enable robust estimation of vaccine effectiveness against severe outcomes. Mechanistic studies exploring sex-hormone modulation of mucosal immunity and the impact of nicotine or combustion by-products on airway ACE-2 expression could further elucidate the epidemiological patterns observed here.

Conclusions

Anti-nucleocapsid IgG surveillance in a heavily vaccinated Iranian cohort reveals widespread natural infection and indicates that demographic and behavioral factors, rather than vaccination metrics, now dominate infection risk. Continued emphasis on booster uptake to prevent severe disease remains essential, but curbing transmission will likely require next-generation vaccines or non-pharmaceutical interventions directed at the specific exposure patterns highlighted here.

Biography

Farshid Yeganeh is an Associate Professor of Immunology at the Department of Immunology, School of Medicine, Shahid Beheshti University of Medical Sciences, Tehran, Iran. His research centers on vaccine immunology and translational immunotherapy. He leads studies on antigen discovery, adjuvant systems, delivery and release kinetics, and correlates of durable protection for vaccines against infectious diseases, including Leishmania. In parallel, he investigates regenerative immunology using mesenchymal stem cells (MSCs) and their derivatives (extracellular vesicles/secretome) to recalibrate dysregulated immunity in chronic autoinflammatory and autoimmune diseases, including via biomaterial-assisted delivery. His broader interests include immunotherapeutics and the cellular and molecular bases of host–pathogen and host–tissue interactions.

Funding Statement

This research was supported by the Deputy of Research, School of Medicine, Shahid Beheshti University of Medical Sciences [Grant No. 43009932].

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The data supporting the findings of this study are available from the corresponding author upon reasonable request.

Ethical statement

The study was approved by the Institutional Review Board of Shahid Beheshti University of Medical Sciences (IR.SBMU.MSP.REC.1403.353). All participants provided written informed consent.

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/21645515.2025.2560063