Patient characteristics
In total, 3947 patients with G-NEN who satisfied the inclusion criteria, comprising 2377 (60.2%) married patients and 1570 (39.8%) unmarried subjects, were identified in the SEER database. Of the unmarried subjects, 408 (10.3%) were divorced or separated, 646 (16.4%) were single, and 516 (13.1%) were widowed. Demographic and clinicopathological characteristics of these patients were described in Table 1, stratified by marital statuses. Chi-square tests showed significant differences in most variables, including age at diagnosis (P < 0.001), sex (P < 0.001), year at diagnosis (P < 0.001), ethnicity (P < 0.001), tumour size (P = 0.02), and surgery performed (P < 0.001).
The effects of marital status on overall and cause-specific survival
We applied Kaplan-Meier curves to evaluate the OS rates of G-NEN patients. As shown in Figure 1A, unmarried status was associated with worse prognosis compared to married status according to the Cox regression model (HR 1.47, 95% CI 1.33-1.64, P < 0.001). After adjusting baseline parameters, including age, sex, year at diagnosis, race, tumour grade, tumour size, and surgery performed, unmarried patients still had poorer prognosis than married counterparts (HR 1.49, 95% CI 1.33-1.67, P < 0.001). The CSS rates of G-NEN patients were also displayed by plotting Kaplan-Meier curves. As shown in Figure 1B, unmarried status contributed to unfavourable prognosis (HR 1.29, 95% CI 1.10-1.51, P = 0.002) according to the Cox model and even after adjusting confounding factors (HR 1.29, 95% CI 1.09-1.54, P = 0.003).
To explore whether different unmarried status led to worse prognosis than married status, we divided unmarried subjects into three subgroups: the divorced/separated, single and widowed. On univariable analysis, windowed patients had a statistically significant higher risk of all-cause mortality (HR 3.35, 95% CI 2.05–2.68, P < 0.001). As shown in Figure 1C, compared with married patients, windows had significantly lower OS rate. On multivariable analysis, unmarried status (including single marital status) remained an independent prognostic factor for increased risk of all-cause mortality, while single status did not indicate higher risk of cancer-specific death compared to married G-NEN patients.
In addition, age, sex, tumour grade, tumour stage, and surgery performed were validated as independent prognosis factors for OS and CSS in the multivariate Cox analyses. The detailed description of each prognostic factor is displayed in Table 2.
Subgroup analysis of the effect of marital status stratified by gender
Since widowed patients had the poorest OS, we analysed whether unmarried status, especially widowed status contributed to the poor survival rates in the subgroups of G-NEN patients stratified by gender. As shown in Table 3, marital status was found to be an independent prognostic factor of OS in both male and female G-NEN patients according to the log-rank tests and Cox regression analysis (Figure 1D, 1E). Particularly, widowhood affected the prognosis more in women than in men.
Clinical outcomes after propensity score matching
To further confirm the findings that married G-NEN patients survived longer and to minimize bias in the previous analysis, we conducted a PS-matching analysis. Using a 1:1 PS-matching method, we matched 506 unmarried patients with 506 married patients. As shown in Table 4, all the baseline variables were clearly well matched (all P > 0.05).
Although the HR was not higher after matching the data than before, unmarried patients still shown poorer OS (HR 1.51, 95% CI 1.19-1.90, P = 0.001) and CSS (HR 1.50, 95% CI 1.10-2.05, P = 0.01) in univariate Cox model. In multivariate analysis (Figure 2), unmarried status was still linked with significantly worse OS (HR 1.39, 95% CI 1.09-1.78, P = 0.008). As shown in Figure 3A and 3B, survival curves for OS and CSS indicated that married patients showed significantly better survival than their unmarried counterparts. Compared with married patients, widowed patients had a significant reduction in both OS and CSS rate (Figure 3C, 3D).