3.1. Patient baseline characteristics
Table 1 presents the baseline characteristics of the study population. The cohort comprised 10,041 individuals, with 8531 (85.0%) residing in urban areas and 1510 (15.0%) in rural areas. Significant differences were observed across several variables prior to PSM. Compared to urban patients, rural patients were older (proportion of patients aged 65–85: 48.9% vs 39.7%, P < .001), predominantly White (95.7% vs 89.9%, P < .001), and more likely to be married (71.1% vs 67.7%, P = .009). Additionally, rural patients had significantly lower median household incomes. A median household income below 75,000 USD was recorded for 46.3% of urban patients, compared with 95.1% of rural patients (P < .001). After PSM, 3020 urban glioblastoma patients and 1510 rural glioblastoma patients were included. The baseline characteristics of the 2 groups were balanced after matching (all P > .05). The West China Hospital cohort comprised 265 patients, including 50 rural patients and 215 urban patients.
Table 1
Baseline characteristics of included patients before and after propensity matching.
CharacteristicBefore propensity matchingP valueAfter propensity matchingP valueOverallUrbanRuralUrbanRuralYear of diagnosis.097.863 2000–20095903 (58.8%)4986 (58.4%)917 (60.7%)1826 (60.5%)917 (60.7%) 2010–20194138 (41.2%)3545 (41.6%)593 (39.3%)1194 (39.5%)593 (39.3%)Age (yr)<.001.670 20–491774 (17.7%)1572 (18.4%)202 (13.4%)415 (13.7%)202 (13.4%) 50–644144 (41.3%)3574 (41.9%)570 (37.7%)1186 (39.3%)570 (37.7%) 65–742612 (26.0%)2149 (25.2%)463 (30.7%)900 (29.8%)463 (30.7%) 75–851511 (15.0%)1236 (14.5%)275 (18.2%)519 (17.2%)275 (18.2%)Sex.475.865 Male5942 (59.2%)5061 (59.3%)881 (58.3%)1770 (58.6%)881 (58.3%) Female4099 (40.8%)3470 (40.7%)629 (41.7%)1250 (41.4%)629 (41.7%)Race<.001.950 White9114 (90.8%)7669 (89.9%)1445 (95.7%)2894 (95.8%)1445 (95.7%) Black502 (5.0%)455 (5.3%)47 (3.1%)89 (2.9%)47 (3.1%) Others425 (4.2%)407 (4.8%)18 (1.2%)37 (1.2%)18 (1.2%)Marital status.009.448 Married6851 (68.2%)5777 (67.7%)1074 (71.1%)2115 (70.0%)1074 (71.1%) Unmarried3190 (31.8%)2754 (32.3%)436 (28.9%)905 (30.0%)436 (28.9%)Median household income*<.001.999 ˂75,000 USD5384 (53.6%)3948 (46.3%)1436 (95.1%)2872 (95.1%)1436 (95.1%) ˃75,000 USD4657 (46.4%)4583 (53.7%)74 (4.9%)148 (4.9%)74 (4.9%)Local surgery.021.141 Yes8033 (80.0%)6858 (80.4%)1175 (77.8%)2407 (79.7%)1175 (77.8%) None/unknown2008 (20.0%)1673 (19.6%)335 (22.2%)613 (20.3%)335 (22.2%)Radiotherapy.024.624 Yes8105 (80.7%)6918 (81.1%)1187 (78.6%)2393 (79.2%)1187 (78.6%) None/unknown1936 (19.3%)1613 (18.9%)323 (21.4%)627 (20.8%)323 (21.4%)Chemotherapy.082.896 Yes6528 (65.0%)5576 (65.4%)952 (63.0%)1910 (63.2%)952 (63.0%) None/unknown3513 (35.0%)2955 (34.6%)558 (37.0%)1110 (36.8%)558 (37.0%)
USD = United States dollars.
*
Adjusted to 2022 US median household income.
3.2. Urban–rural disparities in treatment utilization
As shown in Table 1, rural patients had lower crude rates of local surgery than urban patients (77.8% vs 80.4%, P = .021) and lower rates of radiotherapy (78.6% vs 81.1%, P = .024). Table 2 presents the associations between residential status and the receipt of specific treatments, including local surgery, radiotherapy, and chemotherapy. Before adjustment for covariates, rural patients were less likely than urban patients to undergo local surgery (odds ratio [OR] = 0.86, 95% confidence interval [CI] = 0.75–0.98, P = .021) and radiotherapy (OR = 0.86, 95% CI = 0.75–0.98, P = .024). However, there was no statistically significant difference in the receipt of chemotherapy between rural and urban patients (OR = 0.91, 95% CI = 0.81–1.01, P = .082). After adjustment for covariates, rural patients remained less likely than urban patients to undergo local surgery (OR = 0.84, 95% CI = 0.72–0.96, P = .009), radiotherapy (OR = 0.84, 95% CI = 0.72–0.97, P = .012), and chemotherapy (OR = 0.89, 95% CI = 0.80–0.99, P = .046).
Table 2
Odds ratios for receiving specific treatments in rural versus urban glioblastoma patients.
TreatmentUnivariable*Multivariable†OR (95% CI)P valueOR (95% CI)P valueLocal surgery0.86 (0.75–0.98).0210.84 (0.72–0.96).009Radiotherapy0.86 (0.75–0.98).0240.84 (0.72–0.97).012Chemotherapy0.91 (0.81–1.01).0820.89 (0.80–0.99).046
CI = confidence interval, OR = odds ratio.
*
Odds ratios for receiving specific treatments (local surgery, radiotherapy, chemotherapy) in patients with glioblastoma living in rural areas compared with patients with glioblastoma living in urban areas.
†
Adjusted covariates included year of diagnosis, age, sex, race, marital status, and median household income.
3.3. Urban–rural disparities in cancer-specific survival
As shown in Figure 1, rural patients exhibited poorer CSS compared to urban patients both before PSM (hazard ratio [HR] = 1.22, 95% CI = 1.15–1.29, P < .001; Fig. 1A) and after PSM (HR = 1.12, 95% CI = 1.05–1.20, P < .001; Fig. 1B). Table 3 displays the 3-, 6-, 12-, and 24-month CSS rates for patients with glioblastoma. Before PSM, the 3-, 6-, 12-, and 24-month CSS rates for urban patients were 79.7%, 64.8%, 42.6%, and 18.7%, respectively, while for rural patients, these rates were 74.2%, 57.9%, 34.4%, and 13.8%, respectively. After PSM, the 3-, 6-, 12-, and 24-month CSS rates for urban patients were 76.8%, 60.9%, 39.0%, and 16.7%, respectively. External validation results from the West China Hospital cohort are shown in Figure S1, Supplemental Digital Content 1. Rural patients showed a trend toward worse CSS than urban patients, but the difference was not statistically significant (HR = 1.33, 95% CI = 0.96–1.84, P = .085; Fig. S1, Supplemental Digital Content 1).
Table 3
Cancer-specific survival rates at 3, 6, 12, and 24 months in urban and rural glioblastoma patients before and after propensity score matching.
ResidenceCancer-specific survival rate (%, 95% CI)P value*3 months6 months12 months24 monthsBefore PSM Overall78.9 (78.1–79.7)63.8 (62.8–64.7)41.4 (40.4–42.3)18.0 (17.2–18.7)<.001 Urban79.7 (78.9–80.6)64.8 (63.8–65.8)42.6 (41.5–43.7)18.7 (17.9–19.6) Rural74.2 (72.0–76.5)57.9 (55.5–60.5)34.4 (32.1–36.9)13.8 (12.2–15.7)After PSM Overall76.0 (74.7–77.2)59.9 (58.5–61.4)37.5 (36.1–39.0)15.8 (14.7–16.9)<.001 Urban76.8 (75.3–78.3)60.9 (59.2–62.7)39.0 (37.3–40.8)16.7 (15.4–18.2) Rural74.2 (72.0–76.5)57.9 (55.5–60.5)34.4 (32.1–36.9)13.8 (12.2–15.7)
CI = confidence interval, PSM = propensity score matching.
Open multimedia modal
Figure 1.
Open multimedia modal
Cancer-specific survival according to urban–rural residence before and after propensity score matching. (A) Kaplan–Meier curves comparing cancer-specific survival between urban and rural patients before propensity score matching. (B) Kaplan–Meier curves comparing cancer-specific survival between urban and rural patients after propensity score matching. CI = confidence interval, HR = hazard ratio.
Figure 2 illustrates the subgroup analysis based on the year of diagnosis. Significant urban–rural differences in CSS were observed among patients diagnosed with glioblastoma during 2000 to 2009 (HR = 1.15, 95% CI = 1.06–1.25, P < .001; Fig. 2A). Although the urban–rural disparities in CSS narrowed among patients diagnosed during 2010 to 2019, they remained statistically significant (HR = 1.10, 95% CI = 1.01–1.20, P = .046; Fig. 2B). Figure 3 presents the subgroup analysis based on median household income. Significant urban–rural disparities in CSS were observed in patients with a median household income of <75,000 USD (HR = 1.11, 95% CI = 1.04–1.19, P = .001; Fig. 3A) as well as those with a median household income of >75,000 USD (HR = 1.08, 95% CI = 1.02–1.20, P = .036; Fig. 3B).
Open multimedia modal
Figure 2.
Open multimedia modal
Cancer-specific survival according to urban–rural residence stratified by year of diagnosis. (A) Kaplan–Meier curves comparing cancer-specific survival between urban and rural patients diagnosed during 2000 to 2009. (B) Kaplan–Meier curves comparing cancer-specific survival between urban and rural patients diagnosed during 2010 to 2019. CI = confidence interval, HR = hazard ratio.
Open multimedia modal
Figure 3.
Open multimedia modal
Cancer-specific survival according to urban–rural residence stratified by median household income. (A) Kaplan–Meier curves comparing cancer-specific survival between urban and rural patients with a median household income of <75,000 USD. (B) Kaplan–Meier curves comparing cancer-specific survival between urban and rural patients with a median household income of >75,000 USD. CI = confidence interval, HR = hazard ratio, USD = United States dollars.
Table 4 presents the results of the multivariable Cox regression analysis. After adjusting for covariates, residential location remained an independent predictor of CSS in patients with glioblastoma. Compared to urban patients, rural patients demonstrated significantly worse CSS (HR = 1.15, 95% CI = 1.08–1.22, P < .001). Additionally, older age (20–49 as reference, 50–64, HR = 1.57, 95% CI = 1.42–1.73, P < .001; 65–74, HR = 2.32, 95% CI = 2.09–2.57, P < .001; 75–85, HR = 3.19, 95% CI = 2.84–3.58, P < .001), unmarried status (HR = 1.09, 95% CI = 1.02–1.17, P = .011), and the absence of local surgery (HR = 1.78, 95% CI = 1.65–1.92, P < .001), radiotherapy (HR = 1.56, 95% CI = 1.42–1.71, P < .001), and chemotherapy (HR = 1.61, 95% CI = 1.48–1.75, P < .001) were significant adverse factors for CSS. Conversely, female patients had better CSS outcomes (HR = 0.88, 95% CI = 0.83–0.94, P < .001).
Table 4
Univariable and multivariable Cox regression analyses of cancer-specific survival in patients with glioblastoma.
CharacteristicUnivariable*Multivariable†HR (95% CI)P valueHR (95% CI)P valueYear of diagnosis 2000–2009RefRef 2010–20190.86 (0.81–0.92)<.0010.94 (0.88–1.00).065Age 20–49RefRef 50–641.54 (1.40–1.69)<.0011.57 (1.42–1.73)<.001 65–742.38 (2.15–2.63)<.0012.32 (2.09–2.57)<.001 75–853.74 (3.34–4.18)<.0013.19 (2.84–3.58)<.001Sex MaleRefRef Female0.95 (0.90–1.01).1320.88 (0.83–0.94)<.001Race WhiteRefRef Black0.81 (0.67–0.97).0200.90 (0.75–1.08).260 Others0.85 (0.64–1.13).2610.88 (0.66–1.17).381Marital status MarriedRefRef Unmarried1.12 (1.05–1.20)<.0011.09 (1.02–1.17).011Median household income‡ ˂75,000 USDRefRef ˃75,000 USD0.95 (0.83–1.09).4941.01 (0.88–1.16).854Residence UrbanRefRef Rural1.12 (1.05–1.20)<.0011.15 (1.08–1.22)<.001Local surgery YesRefRef None/unknown2.12 (1.97–2.28)<.0011.78 (1.65–1.92)<.001Radiotherapy YesRefRef None/unknown2.63 (2.44–2.84)<.0011.56 (1.42–1.71)<.001Chemotherapy YesRefRef None/unknown2.26 (2.12–2.41)<.0011.61 (1.48–1.75)<.001
CI = confidence interval, HR = hazard ratio, USD = United States dollars.
*
Univariable Cox regression analyses evaluated the association of each characteristic with cancer-specific survival separately.
†
The multivariable Cox regression model simultaneously included year of diagnosis, age, sex, race, marital status, median household income, residence, local surgery, radiotherapy, and chemotherapy.
‡
Median household income was adjusted to 2022 USD.
Figure 4 illustrates the subgroup analysis based on predefined variables after adjusting for covariates. Rural glioblastoma patients exhibited worse CSS compared to urban patients across all subgroups, except for those aged 20 to 49 (HR = 1.09, 95% CI = 0.91–1.30, P = .339) and 65 to 74 (HR = 1.10, 95% CI = 0.98–1.23, P = .114), as well as patients identifying as Black (HR = 1.10, 95% CI = 0.73–1.67, P = .642) or other races (HR = 0.91, 95% CI = 0.47–1.75, P = .783).
Open multimedia modal
Figure 4.
Open multimedia modal
Association between rural residence and cancer-specific survival across predefined subgroups. Adjusted hazard ratios for cancer-specific survival among rural patients compared with urban patients are shown according to year of diagnosis, age, sex, race, marital status, and median household income. Analyses were adjusted for year of diagnosis, age, sex, race, marital status, and median household income, except for the variable used for stratification. CI = confidence interval, HR = hazard ratio, USD = United States dollars.
