Marital status and competing risks of mortality in non-small cell lung cancer: a SEER-based nomogram analysis

Introduction

Non-small cell lung cancer (NSCLC) accounts for approximately 85% of all lung cancer cases, with its associated mortality remaining among the highest worldwide (1). Although significant advances have been achieved in surgery, radiotherapy, and systemic therapy in recent years, the overall survival (OS) of NSCLC patients continues to be unsatisfactory (2). Given the increasing emphasis on supportive care in oncology, understanding how marital status and related social support influence not only treatment receipt but also patients’ psychological resilience and quality of life is of great clinical relevance.

Beyond traditional clinical prognostic factors, psychosocial variables such as marital status have been increasingly recognized as potentially important influences on cancer patient outcomes (3). Studies have suggested that married patients typically receive stronger social support, maintain better treatment adherence, and experience superior psychological well-being, ultimately translating into improved survival outcomes (4). Such findings have been validated across multiple cancer types, including prostate, breast, and colorectal cancers (5,6).

In patients with NSCLC, several large-scale population-based studies have similarly indicated that married individuals are more likely to be diagnosed at an earlier stage, to receive definitive therapy, and to achieve superior survival compared to their unmarried counterparts (7,8). However, these studies often could not fully control for confounding factors such as socioeconomic status, insurance coverage, and comorbidities, leaving the mechanisms by which marital status influences cancer-specific mortality unclear (9,10).

Moreover, previous investigations have largely focused only on OS or cancer-specific survival (CSS), without accounting for competing risks of other causes of death, which is especially important in elderly patients and those with multiple comorbidities (11,12). Competing risk models may provide more realistic absolute risk estimates in such settings, thereby offering a more comprehensive evaluation of the impact of marital status on outcomes among NSCLC patients (13).

Therefore, based on a large-scale Surveillance, Epidemiology, and End Results (SEER) population-based cohort, this study aimed to systematically explore the impact of marital status on cancer-specific death (CSD) and other-cause death (OCD) among NSCLC patients using propensity score matching (PSM) and the Fine-Gray competing risk model. Furthermore, we constructed a nomogram that incorporates marital status to enable individualized risk prediction, thereby providing an evidence-based foundation for supportive interventions, survivorship care, and quality-of-life management strategies in patients with NSCLC (14,15). We present this article in accordance with the TRIPOD reporting checklist (available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-1691/rc).

Methods

Study population

This study was designed as a retrospective cohort analysis based on data extracted from the SEER database, aiming to investigate the impact of marital status on survival outcomes in patients with NSCLC. The SEER program covers approximately 28% of the U.S. population and provides comprehensive demographic, tumor, and treatment information. Data were retrieved using SEER*Stat software (version 8.3.9) for the period 2010 to 2015. Since the SEER database does not contain personally identifiable information, this study met all ethical standards, was exempt from informed consent requirements, and was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The analytic window was restricted to 2010–2015 to ensure harmonized American Joint Committee on Cancer (AJCC) 7th edition staging, stable coding of key variables, and sufficient follow-up for 5-year endpoints.

Inclusion and exclusion criteria

Inclusion criteria:

• Age ≥18 years at diagnosis;
• Pathologically confirmed NSCLC, including adenocarcinoma [ADC; International Classification of Diseases for Oncology, 3rd Edition (ICD-O-3) codes 8140–8147, 8255, 8260, 8310, 8323, 8480, 8481, 8490, 8550, 8572] and squamous cell carcinoma (SCC; 8050–8052, 8070–8078);
• Unilateral tumor (left or right side);
• Survival time longer than one month;
• Complete and classifiable marital status information.

Exclusion criteria:

• Patients with other primary malignant tumors;
• Missing or inaccurate demographic, clinicopathological, treatment, or follow-up information;
• Bilateral tumors or pathological types other than adenocarcinoma and squamous cell carcinoma.

To minimize model instability from rare/heterogeneous strata, we excluded cases with bilateral laterality (approximately 0.4% of the cohort) and those coded as “Other” histology (approximately 3.5–3.6%); laterality was analyzed as left vs. right, and histology as adenocarcinoma vs. squamous cell carcinoma.

The detailed patient selection flowchart is presented in Figure 1.

Figure 1 Flow diagram of patient selection and cohort construction. ADC, adenocarcinoma; NSCLC, non-small cell lung cancer; PSM, propensity score matching; SCC, squamous cell carcinoma; SEER, Surveillance, Epidemiology, and End Results.

Study variables

Demographic variables included age, sex (male/female), race (White/non-White), and marital status (married/unmarried). In SEER, marital status is recorded at diagnosis and is not updated thereafter; accordingly, it was analyzed as a baseline, fixed covariate in this study. In the main analysis, age was modeled as a categorical variable (≤65 vs. >65 years); a sensitivity analysis treating age as a continuous variable is provided in the Table S1. Tumor characteristics included grade (I–IV), histologic subtype (ADC vs. SCC), T category (T1–T4), N category (N0–N3), M category (M0/M1), and laterality (left/right). Treatment variables included surgery (yes/no), radiotherapy (yes/no), and chemotherapy (yes/no), based on SEER-registered records. In SEER, treatments are recorded as first-course therapy initiated at or shortly after diagnosis; accordingly, surgery, radiotherapy, and chemotherapy were modeled as baseline binary covariates. An exploratory subclassification of marital status into never-married, divorced, widowed, and separated is presented in the Figure S1, Tables S2,S3.

Study endpoints

The endpoints included:

• OS: defined as the time from diagnosis to death from any cause;
• CSS: defined as the time from diagnosis to death due to NSCLC;
• CSD: defined as death from NSCLC, evaluated under the competing risk framework;
• OCD: defined as death from non-NSCLC causes, also evaluated under the competing risk framework.

Primary endpoints were CSD and OCD, analyzed within a competing-risk framework. Cumulative incidence functions (CIFs) were compared using Gray’s test, and subdistribution hazard ratios (sHRs) were estimated with the Fine-Gray model. OS and CSS from Kaplan-Meier analyses with log-rank tests were prespecified as supportive endpoints.

Statistical analysis

Baseline characteristics across different marital status groups were compared using t-tests or Pearson’s Chi-squared tests. The Kaplan-Meier method was used to plot OS and CSS curves, with log-rank tests for comparison. For competing risk analysis, patients’ statuses were categorized as alive, CSD, or OCD, and the Fine-Gray subdistribution hazard model was applied to assess the independent effect of marital status on CSD risk, implemented through the “cmprsk” package. To mitigate immortal-time and indication biases inherent to baseline coding of first-course treatments, patients with overall survival ≤1 month were excluded and treatment indicators were kept as baseline covariates in all models.

To reduce confounding bias, PSM was performed to balance baseline variables between married and unmarried groups. A 1:1 nearest-neighbor matching algorithm was applied with a caliper width of 0.2 times the standard deviation of the logit of the propensity score, as recommended in methodological literature. Standardized mean differences (SMDs) ≤0.1 were considered indicative of adequate balance. Effect sizes were reported as sHRs with 95% confidence intervals (CI). CIFs were compared using Gray’s test (χ²).

Nomogram construction and validation

For clinical application, the entire cohort was randomly divided into training and validation sets in a 1:1 ratio. A nomogram was then developed in the training cohort based on the Fine-Gray multivariable model to predict 1-, 3-, and 5-year CSD probabilities. The scores for each variable were weighted according to their sHRs, and the total points corresponded to estimated probabilities. The performance of the nomogram was assessed by calculating time-dependent receiver operating characteristic (ROC) curves and the area under the curve (AUC), while calibration was evaluated using calibration curves generated through 1000 bootstrap resamples to assess agreement between predicted and observed probabilities. In addition to discrimination and calibration, overall accuracy was quantified using time-dependent Brier scores calculated with standard censoring-adjusted methods; lower values indicate better accuracy, and ~0.25 approximates a non-informative benchmark. Clinical utility was assessed by decision curve analysis (DCA), which estimates net benefit across clinically relevant threshold probabilities and compares the nomogram with treat-all and treat-none strategies.

Software and statistical standards

All statistical analyses and graphical visualizations were performed using R software (version 4.0.3, http://www.r-project.org), with primary packages including cmprsk (for competing risk analysis), matching (for PSM), rms (for nomogram construction), and survival (for Kaplan-Meier survival analysis). All tests were two-sided, and P values <0.05 were considered statistically significant.

Results

Descriptive analysis

A total of 47,170 patients with NSCLC were included, comprising 25,864 married (54.8%) and 21,306 unmarried (45.2%) individuals (Table 1). Before PSM, the two groups differed across several clinicopathological characteristics (Table 1). Specifically, sex and race: married patients were more often male (60.5% vs. 40.2%, P<0.001) and White (81.0% vs. 80.1%, P=0.01).

Grade and histology: lower grades (I–II) were more frequent in married patients (54.1% vs. 51.4%), whereas higher grades (III–IV) were more frequent in unmarried patients (48.6% vs. 45.9%). Regarding histology, ADC was more common in the married group (68.7% vs. 65.2%), while SCC was more common in the unmarried group (34.8% vs. 31.3%) (all P<0.001).

TNM categories: T1 was slightly more frequent among married patients (30.4% vs. 29.9%), whereas T4 was slightly more frequent among unmarried patients (17.9% vs. 16.4%); M1 was also higher in the unmarried group (30.3% vs. 29.2%, P=0.01). N category differed as well (P=0.04).

Treatment utilization: married patients were more likely to undergo surgery and receive chemotherapy (50.2% vs. 43.7% and 41.6% vs. 34.0%, respectively; both P<0.001), while radiotherapy was comparable between groups (P=0.52).

After 1:1 PSM, 18,708 patients remained in each group. Post-matching balance was evaluated using SMDs, with all SMDs ≤0.10, indicating adequate balance; the corresponding Love plot is shown in Figure S2. For completeness, P values are retained in Table 1 as descriptive information.

In addition, 9,754 patients did not enter the matched cohort. Compared with matched patients, this excluded subgroup comprised predominantly older individuals (>65 years, 82.9%), men (71.3%), and White patients (83.0%); T1–T2 accounted for 66.3% and M1 for 26.5%. Regarding treatment utilization, no surgery was recorded in 52.3%, and no radiotherapy or no chemotherapy in 83.6% and 58.8%, respectively. These distributions indicate limited covariate overlap, explaining their exclusion from the matched analysis.

Survival analysis

As prespecified, OS/CSS Kaplan-Meier curves are presented as supportive analyses given the presence of competing risks. In the overall cohort before PSM, Kaplan-Meier survival analysis demonstrated significantly superior OS for married patients compared with unmarried patients (P<0.001, Figure 2A). This difference remained significant after matching (P<0.001, Figure 2B), suggesting a potential protective effect of marital status on overall survival.

Figure 2 Supportive Kaplan-Meier survival curves comparing OS and CSS between married and unmarried NSCLC patients. (A) OS before PSM; (B) OS after PSM; (C) CSS before PSM; (D) CSS after PSM. OS/CSS are presented as supportive analyses only. Primary endpoints (CSD and OCD) were analyzed under a competing-risk framework using CIFs and Gray’s tests, with effects estimated by Fine-Gray models. CSD, cancer-specific death; CSS, cancer-specific survival; NSCLC, non-small cell lung cancer; OCD, other causes death; OS, overall survival; PSM, propensity score matching.

Similarly, for CSS, married patients had significantly higher survival probabilities than unmarried patients before PSM (P<0.001, Figure 2C), and this survival advantage persisted after matching (P<0.001, Figure 2D). Overall, married patients consistently showed higher survival probabilities over the observation period, with a slower decline in their survival curves compared to unmarried patients.

These results indicate that marital status is an important survival-related factor for NSCLC patients, even after controlling for age, sex, race, stage, histology, and treatment factors, with married status potentially conferring a prognostic advantage.

Competing risk analysis

Under the competing-risk framework, cumulative CIFs and Gray’s tests were used to compare outcomes between marital groups (Table 2).

Before PSM, the CSD CIFs at 1, 3, and 5 years were 29.33%, 46.47%, and 53.02% in married patients, lower than the 33.67%, 50.63%, and 56.78% observed in unmarried patients (Gray’s χ²=95.56, P<0.001).

For OCD, married patients also showed lower CIFs at 1, 3, and 5 years (4.53%, 8.86%, and 12.13% vs. 5.41%, 10.14%, and 13.59%, respectively; Gray’s χ²=27.16, P<0.001).

After PSM, the differences remained significant. The 1-, 3-, and 5-year CSD CIFs were 30.37%, 47.00%, and 53.46% in married patients versus 32.87%, 49.92%, and 56.11% in unmarried patients (Gray’s χ²=30.26, P<0.001).

For OCD, the corresponding CIFs were 4.66%, 8.62%, and 11.73% in married patients versus 5.17%, 9.67%, and 13.00% in unmarried patients (Gray’s χ²=24.53, P<0.001).

Taken together, married status consistently conferred a lower cumulative incidence of both CSD and OCD, and these associations remained robust following covariate adjustment through PSM.

For improved interpretability, the observed CSD/OCD event counts, numbers at risk, and censored cases at 1, 3, and 5 years were additionally provided for both the pre- and post-PSM populations in Table S4. These data confirm that the reported cumulative incidence estimates are supported by adequate event accrual and follow-up information.

In addition, unmarried subcategories (single, separated, divorced, widowed) were further evaluated in a supplementary sensitivity analysis. Despite measurable heterogeneity across subtypes, all unmarried groups consistently showed less favorable cumulative incidence patterns compared with married individuals, and the marital survival advantage remained unchanged. Detailed results are presented in Figure S1, Tables S2,S3.

Subgroup analysis

To further explore the prognostic impact of marital status across different clinical subgroups, cumulative incidence curves were plotted for the entire cohort (Figure 3A) as well as for prespecified subgroups (Figure 3B-3I). Nine representative subgroup plots were selected for presentation.

Figure 3 CIF curves for CSD and OCD in married versus unmarried NSCLC patients. (A) Overall cohort; (B,C) stratified by sex (male, female); (D-G) stratified by T stage (T1–T4); (H,I) stratified by histologic subtype (adenocarcinoma, squamous cell carcinoma). CIF, cumulative incidence function; CSD, cancer-specific death; NSCLC, non-small cell lung cancer; OCD, other causes death; T, tumor.

In the overall population (Figure 3A), married patients had a significantly lower cumulative incidence of CSD than unmarried patients (P1<0.001), and a significantly lower risk of OCD as well (P2<0.001).

In the sex subgroups (Figure 3B,3C), both men and women showed lower CIFs of CSD and OCD in the married group (men: P1<0.001, P2=0.002; women: P1<0.001, P2<0.001), indicating that the association was present in both sexes.

In the T-category subgroups (Figure 3D-3G), both outcomes were lower in the married group for T1 and T2 disease (T1: P1<0.001, P2<0.001; T2: P1<0.001, P2=0.002). For T3 disease, the difference was mainly observed in OCD (P1=0.723, P2<0.001), whereas for T4 disease, the difference was primarily observed in CSD (P1=0.002, P2=0.359).

In the histologic subgroups (Figure 3H,3I), both CSD and OCD were lower in married patients with adenocarcinoma (both P<0.001). Among those with squamous cell carcinoma, CSD did not differ significantly (P1=0.298), while OCD was lower in married patients (P2=0.019).

Overall, marital status was associated with a lower cumulative incidence of CSD and/or OCD in most subgroups, with a minority showing outcome-specific patterns.

Multivariable analysis

A multivariable Fine-Gray competing-risk model was applied to identify independent predictors of CSD (Table 3), with reference categories set as follows: ≤65 years, female, White race, married, grade I–II, T1 stage, N0 stage, M0 stage, left laterality, adenocarcinoma, and no radiotherapy/chemotherapy/surgery.

Patients older than 65 years had an elevated subdistribution hazard of CSD compared with those ≤65 years (sHR 1.13, 95% CI: 1.10–1.17, P<0.001). Male sex was also associated with a higher risk (sHR 1.18, 95% CI: 1.15–1.21, P<0.001), whereas non-White race showed a slightly but significant lower risk compared with White patients (sHR 0.94, 95% CI: 0.91–0.98, P<0.001). Unmarried status emerged as an independent adverse factor (sHR 1.11, 95% CI: 1.08–1.14, P<0.001).

Higher tumor burden was consistently associated with increased CSD risk. Compared with grade I–II disease, grade III–IV demonstrated a higher subdistribution hazard (sHR 1.22, 95% CI: 1.19–1.26, P<0.001). A progressive increase was observed across T categories (T2: sHR 1.58, 95% CI: 1.52–1.64, P<0.001; T3: sHR 1.86, 95% CI: 1.79–1.94, P<0.001; T4: sHR 1.96, 95% CI: 1.87–2.04, P<0.001), N categories (N1: sHR 1.66, 95% CI: 1.59–1.74, P<0.001; N2: sHR 1.74, 95% CI: 1.68–1.81, P<0.001; N3: sHR 1.74, 95% CI: 1.65–1.83, P<0.001), and in M1 disease (sHR 1.91, 95% CI: 1.85–1.98, P<0.001). Laterality showed no statistical significance (right vs. left: sHR 0.99, 95% CI: 0.97–1.02, P=0.67).

Regarding pathology, squamous cell carcinoma had a slightly but significantly higher risk than adenocarcinoma (sHR 1.05, 95% CI: 1.02–1.08, P<0.001). Receipt of radiotherapy (sHR 0.95, 95% CI: 0.92–0.98, P<0.001), chemotherapy (sHR 0.73, 95% CI: 0.70–0.75, P<0.001), and surgical resection (sHR 0.42, 95% CI: 0.40–0.43, P<0.001) was associated with significantly reduced subdistribution hazards.

Collectively, tumor burden (grade and T/N/M classification) and active treatment modalities showed consistent directional effects on CSD, and unmarried status remained independently associated with higher subdistribution hazards after full adjustment, aligning with subgroup findings.

Sensitivity analysis: modeling age as a continuous predictor in the Fine-Gray model did not change the main findings. For each 1-year increase in age, the subdistribution hazard of CSD increased by 1% (sHR 1.01, 95% CI: 1.01–1.01, P<0.001), and unmarried status remained an independent adverse factor (sHR 1.11, 95% CI: 1.05–1.16, P<0.001). Estimates for the other covariates were directionally and quantitatively consistent with the primary model, with overlapping CIs (Table S1).

Nomogram construction and validation

A prognostic nomogram was developed based on the multivariable Fine-Gray competing-risk model by incorporating marital status, age, sex, histologic subtype, tumor grade, T stage, N stage, M stage, and treatment variables including radiotherapy, chemotherapy, and surgery, to estimate the 1-, 3-, and 5-year risk of CSD in patients with NSCLC (Figure 4). The nomogram is used by assigning points to each covariate according to its contribution to risk, summing the total points, and projecting the corresponding CSD probability at each specific timepoint. For example, a total score of 637 points corresponded to approximately 5.5%, 13.8%, and 18.9% probabilities of 1-, 3-, and 5-year CSD, respectively (indicated by the red reference line in Figure 4).

Figure 4 Nomogram for predicting 1-, 3-, and 5-year CSD in NSCLC patients based on the Fine-Gray model. ADC, adenocarcinoma; CSD, cancer-specific death; NSCLC, non-small cell lung cancer; SCC, squamous cell carcinoma; T/N/M, tumor/node/metastasis stage.

Regarding discrimination, the AUCs for predicting CSD at 1, 3, and 5 years were 0.840, 0.834, and 0.828 in the training cohort, and 0.841, 0.817, and 0.806 in the validation cohort, respectively (Figure 5A,5B), indicating consistently robust predictive performance. Calibration plots demonstrated excellent agreement between predicted and observed probabilities in both cohorts (Figure 5C,5D).

Figure 5 Performance of the competing-risk nomogram. Time-dependent ROC curves in the training (A) and validation (B) cohorts; calibration curves for 1-, 3-, and 5-year CSD in the training (C) and validation (D) cohorts; DCA for 1-, 3-, and 5-year CSD in the training (E) and validation (F) cohorts. AUC, area under the curve; CSD, cancer-specific death; DCA, decision curve analysis; ROC, receiver operating characteristic.

For overall accuracy, Brier scores were 0.1434, 0.1669, and 0.1703 in the training cohort and 0.1518, 0.1742, and 0.1763 in the validation cohort for 1-, 3-, and 5-year predictions, all well below the theoretical non-informative benchmark of 0.25, supporting strong prediction accuracy across all time horizons.

Decision curve analysis (Figure 5E,5F) revealed higher net clinical benefit across a wide range of risk thresholds compared with treat-all or treat-none strategies, supporting its applicability in guiding risk-adapted surveillance and treatment planning.

Taken together, this nomogram demonstrated excellent discrimination, calibration, overall accuracy, and net clinical benefit, providing a practical and clinically informative tool for individualized CSD risk stratification in patients with NSCLC.

Discussion

This study, based on 47,170 NSCLC patients from the SEER database, applied PSM and competing risk models to systematically evaluate the impact of marital status on CSD and OCD, and further developed a nomogram prediction model to support clinical decision-making. The results demonstrated that marital status exerted a significant and independent protective effect on the prognosis of NSCLC patients, with both CSD and OCD cumulative incidence rates consistently lower among married patients before and after matching. These findings align with prior studies and emphasize the importance of social support and marital relationships in cancer care (1,8). Such supportive relationships improve adherence and access to care, while also enhancing psychological well-being—crucial domains of supportive oncology.

Compared to previous studies that only analyzed OS or CSS, our work introduced the Fine-Gray competing risk model to separate CSD from OCD, thereby reflecting the multidimensional nature of mortality outcomes more realistically (14). In our analysis, after matching, the five-year CSD in unmarried patients reached 56.11%, significantly higher than 53.46% in married patients, and the five-year OCD cumulative incidence was also elevated (13.00% vs. 11.73%), suggesting that unmarried patients may face a higher risk of non-cancer death due to poorer systemic status, comorbidities, and psychological stress (10,11).

Interestingly, our multivariable Fine-Gray analysis further found that male sex, age ≥65 years, higher T/N/M categories, and not receiving surgery, chemotherapy, or radiotherapy were independently associated with a higher subdistribution hazard of CSD, whereas surgery, chemotherapy, and radiotherapy were each associated with a lower risk. This result is consistent with the findings of Wang et al., who reported the worst prognosis among older, unmarried male lung cancer patients in a SEER-based study (13,16,17).

In the subgroup analyses, the survival advantage associated with marriage remained statistically significant in both males and females, suggesting that the benefit of social support is not primarily contingent upon sex-related differences in care-seeking or coping behaviors. Notably, although male sex was identified as an independent adverse prognostic factor in the multivariable model, marriage still conferred pronounced survival benefits among men. This observation is socio-behaviorally and psycho-oncologically plausible: prior studies indicate that men tend to have higher baseline risks and relatively weaker symptom recognition, healthcare-seeking, and treatment adherence, whereby the decision-making assistance and caregiving support provided by spouses may exert a “risk-buffering” effect, yielding a more measurable net benefit (7,18).

Stratified by T category, the “nonlinear” pattern—significance in T1, T2, and T4, but not in T3—likely reflects heterogeneity in resectability, perioperative multimodality management, and multidisciplinary treatment pathways for T3 disease; we also cannot exclude the possible influence of sample composition or treatment-selection differences, warranting further mechanistic investigation (19). By histology, the protective association persisted in adenocarcinoma but attenuated to non-significance in SCC, which is epidemiologically reasonable given SCC’s stronger association with smoking-related lung function impairment and higher comorbidity burden, potentially attenuating the measurable gains of social support in terms of treatment tolerance and follow-up adherence (20,21). Collectively, these stratified findings align with the growing body of evidence demonstrating the multifactorial and context-dependent role of social determinants in lung cancer outcomes, underscoring the need for prospective and interventional studies to clarify actionable translational pathways (10).

Synthesizing our results with previous evidence, it is reasonable to hypothesize that the survival benefits of marriage for NSCLC patients may derive from multiple pathways. Specifically, marriage helps reduce loneliness and psychological distress through stable social networks (15), provides tangible support that improves treatment adherence and monitoring (4), and is often associated with greater financial security and access to healthcare resources (9). These findings support proactive, risk-adapted supportive care—such as early social-work referral, patient navigation, and psycho-oncology—for patients with limited social support, particularly unmarried individuals, and for others identified as high-risk by the proposed nomogram.

Nevertheless, marriage is not an absolute protective factor and is also moderated by patients’ cultural background, individual psychological status, and access to social resources. For instance, some studies have shown that the prognostic association of marriage is less pronounced among female NSCLC patients, possibly related to differences in social gender roles and caregiving burdens (2). Similarly, elderly NSCLC patients with poor marital quality or lacking emotional support may obtain limited benefits even if formally “married” (13).

Regarding predictive performance, the proposed nomogram demonstrated stable and favorable discrimination in both the training and validation cohorts, with AUC values consistently greater than 0.80 across all time points. It also exhibited close calibration, low Brier scores, and higher net clinical benefit on decision curve analysis, supporting its practical applicability for individualized prognostic assessment. Compared to traditional TNM or AJCC staging alone, a multifactorial prediction model incorporating marital status and social determinants may, to some extent, overcome the limitations of conventional oncology models that neglect social dimensions (1).

In addition, our findings echo similar observations in other tumor types. For example, studies on colorectal, prostate, and breast cancers have all confirmed that marital status consistently provides survival benefits across different cancer types (5,22,23). At the same time, scholars have also noted that in the field of radiotherapy and other local treatments, the role of marital status deserves further exploration (24). Future fine-grained assessments of marital quality, psychological interventions, and social support networks may help further elucidate the mechanisms behind the “marriage protection effect” (14).

This study benefits from its large sample size, rigorous methodology, and integration of competing risk modeling with nomogram construction; however, several limitations remain. First, SEER first-course treatments were coded as baseline binary indicators, and patients with ≤1-month survival were excluded to mitigate immortal-time and confounding-by-indication biases; nevertheless, inconsistent treatment start dates precluded formal time-varying or landmark analyses, so residual time-related confounding may remain. Second, the cohort window [2010–2015] ensured uniform AJCC 7th staging and coding stability but predates the widespread uptake of immunotherapy and targeted agents; absolute risks may not directly generalize to current practice, although relative associations are likely to remain informative. Third, SEER lacks data on marital quality and the intensity of social support, and marital status was captured only at diagnosis rather than longitudinally, which may introduce non-differential misclassification and bias the association toward the null (11). Additionally, the SEER database does not include information on smoking history, which precluded direct adjustment for this behavioral confounder. Prior population-based studies have shown that marital status is strongly associated with smoking behavior—married individuals are generally less likely to smoke and more likely to quit compared with unmarried counterparts, partly due to spousal influence and social support mechanisms (25,26). Although this limitation may introduce unmeasured confounding, the large sample size and robust multivariable adjustment in our study likely mitigate its impact on the observed association between marriage and cancer-specific mortality. Future prospective cohorts or qualitative research incorporating the continuity, intimacy, and emotional support dimensions of marriage are needed to validate these conclusions more comprehensively.

Future efforts should integrate sociological, psychological, behavioral, and supportive care perspectives to explore how the positive effects of marriage or social support can be translated into feasible clinical interventions, thereby supporting longer and higher-quality survival for NSCLC patients (1,27).

Conclusions

This large-scale SEER-based study systematically evaluated the impact of marital status on cancer-specific and other-cause mortality in NSCLC, and developed an individualized risk prediction tool using competing risk models and a nomogram. We found that marital status is an important social determinant of long-term survival, particularly among patients with advanced-stage disease or those not receiving systemic therapy. Importantly, this study integrates a competing risk framework with a nomogram in this context, providing evidence to guide survivorship care planning and follow-up management in NSCLC patients.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the TRIPOD reporting checklist. Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-1691/rc

Peer Review File: Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-1691/prf

Funding: This work was supported by the Hangzhou Health Science and Technology Project (grant No. B20253339) and the Zhejiang Provincial Administration of Traditional Chinese Medicine (grant No. 2021ZA106).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-1691/coif). All authors report that this study was supported by the Hangzhou Health Science and Technology Project (grant No. B20253339) and the Zhejiang Provincial Administration of Traditional Chinese Medicine (grant No. 2021ZA106). The authors have no other conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Zhao D, Zhang R, Yang L, et al. The independent prognostic effect of marital status on non-small cell lung cancer patients: a population-based study. Front Med (Lausanne) 2023;10:1136877. [Crossref] [PubMed]
2. Jatoi A, Novotny P, Cassivi S, et al. Does marital status impact survival and quality of life in patients with non-small cell lung cancer? Observations from the mayo clinic lung cancer cohort. Oncologist 2007;12:1456-63. [Crossref] [PubMed]
3. Krajc K, Miroševič Š, Sajovic J, et al. Marital status and survival in cancer patients: A systematic review and meta-analysis. Cancer Med 2023;12:1685-708. [Crossref] [PubMed]
4. Siddiqui F, Bae K, Langer CJ, et al. The influence of gender, race, and marital status on survival in lung cancer patients: analysis of Radiation Therapy Oncology Group trials. J Thorac Oncol 2010;5:631-9. [Crossref] [PubMed]
5. Li Z, Wang K, Zhang X, et al. Marital status and survival in patients with rectal cancer: A population-based STROBE cohort study. Medicine (Baltimore) 2018;97:e0637. [Crossref] [PubMed]
6. Xu Y, Shu X, Xu W, et al. Marital status as an independent prognostic factor for survival in women with vaginal cancer: evidence from the SEER database analysis. Eur J Cancer Prev 2025;34:483-92. [Crossref] [PubMed]
7. Qiu LH, Song JQ, Jiang F, et al. Marital status impacts survival of stage I non-small-cell lung cancer: a propensity-score matching analysis. Future Sci OA 2024;10:FSO926. [Crossref] [PubMed]
8. Saito-Nakaya K, Nakaya N, Fujimori M, et al. Marital status, social support and survival after curative resection in non-small-cell lung cancer. Cancer Sci 2006;97:206-13. [Crossref] [PubMed]
9. Yang KB, Zhang YZ, Chen ZH, et al. Contribution of insurance status to the association between marital status and cancer-specific survival: a mediation analysis. BMJ Open 2022;12:e060149. [Crossref] [PubMed]
10. Lai Y, Guan T, Zhang H, et al. Association of marital status with cardiovascular death risk in patients with lung cancer: A population-based study. Prev Med Rep 2024;45:102846. [Crossref] [PubMed]
11. Chen ZH, Yang KB, Zhang YZ, et al. Assessment of Modifiable Factors for the Association of Marital Status With Cancer-Specific Survival. JAMA Netw Open 2021;4:e2111813. [Crossref] [PubMed]
12. Wang Y, Jiao Y, Nie J, et al. Sex differences in the association between marital status and the risk of cardiovascular, cancer, and all-cause mortality: a systematic review and meta-analysis of 7,881,040 individuals. Glob Health Res Policy 2020;5:4. [Crossref] [PubMed]
13. Chambers A, Damone E, Chen YT, et al. Social support and outcomes in older adults with lung cancer. J Geriatr Oncol 2022;13:214-9. [Crossref] [PubMed]
14. Aizer AA, Chen MH, McCarthy EP, et al. Marital status and survival in patients with cancer. J Clin Oncol 2013;31:3869-76. [Crossref] [PubMed]
15. Wang H, Zhao Y, Feng S, et al. Impact of marriage on diagnosis and survival in patients with urological cancers. Sci Rep 2025;15:15443. [Crossref] [PubMed]
16. Wang Y, Lyu D, Cheng C, et al. Preoperative nomogram for predicting spread through air spaces in clinical-stage IA non-small cell lung cancer using (18)F-fluorodeoxyglucose positron emission tomography/computed tomography. J Cancer Res Clin Oncol 2024;150:185. [Crossref] [PubMed]
17. Xu J, Zeng H, Zhang G, et al. Development and validation of a nomogram based on preoperative variables for predicting recurrence-free survival in stage IA lung adenocarcinoma. Thorac Cancer 2023;14:3108-18. [Crossref] [PubMed]
18. Frost MG, Jensen KJ, Jimenez-Solem E, et al. Sex disparities in advanced non-small cell lung cancer survival: A Danish nationwide study. Lung Cancer 2025;202:108485. [Crossref] [PubMed]
19. Kim SS, Cooke DT, Kidane B, et al. The Society of Thoracic Surgeons Expert Consensus on the Multidisciplinary Management and Resectability of Locally Advanced Non-small Cell Lung Cancer. Ann Thorac Surg 2025;119:16-33. [Crossref] [PubMed]
20. Jiang H, Huang G, Feng D, et al. Comorbidity of lung cancer and chronic obstructive pulmonary disease: correlation and optimization of treatment strategies. Transl Lung Cancer Res 2025;14:2296-308. [Crossref] [PubMed]
21. Liao S, Wang Y, Zhou J, et al. Associations between chronic obstructive pulmonary disease and ten common cancers: novel insights from Mendelian randomization analyses. BMC Cancer 2024;24:601. [Crossref] [PubMed]
22. Khan S, Nepple KG, Kibel AS, et al. The association of marital status and mortality among men with early-stage prostate cancer treated with radical prostatectomy: insight into post-prostatectomy survival strategies. Cancer Causes Control 2019;30:871-6. [Crossref] [PubMed]
23. Liu L, Chi YY, Wang AA, et al. Marital Status and Survival of Patients with Hormone Receptor-Positive Male Breast Cancer: A Surveillance, Epidemiology, and End Results (SEER) Population-Based Study. Med Sci Monit 2018;24:3425-41. [Crossref] [PubMed]
24. Kishi N, Matsuo Y, Hanazawa H, et al. Lack of an association between marital status and survival in patients receiving stereotactic body radiotherapy for early-stage non-small-cell lung cancer. PLoS One 2022;17:e0269463. [Crossref] [PubMed]
25. Ramsey MW Jr, Chen-Sankey JC, Reese-Smith J, et al. Association between marital status and cigarette smoking: Variation by race and ethnicity. Prev Med 2019;119:48-51. [Crossref] [PubMed]
26. Parekh TM, Wu C, McClure LA, et al. Determinants of cigarette smoking status in a national cohort of black and white adult ever smokers in the USA: a cross-sectional analysis of the REGARDS study. BMJ Open 2019;9:e027175. [Crossref] [PubMed]
27. Richardson CL, Saeed M, Sharp L, et al. The association between marital status and treatment initiation in lung cancer: A systematic review and meta-analysis of observational studies. Cancer Epidemiol 2023;87:102494. [Crossref] [PubMed]