Association between financial toxicity and frailty among older patients with cancer: a cross-sectional study using COST and G8

Abstract

Purpose

Financial toxicity (FT) is increasingly recognized as a critical challenge in cancer care. Older patients can be particularly vulnerable due to frailty-related impairments and accumulated treatment burden. This study aimed to evaluate FT and its association with geriatric vulnerability among older patients with cancer.

Methods

We conducted a single-center cross-sectional study of patients aged ≥ 65 years with cancer treated at a regional hospital in Japan. FT was assessed using the Comprehensive Score for Financial Toxicity (COST) instrument. Frailty-related vulnerability was evaluated using questionnaire-based assessments, including the G8 screening tool. Associations between the COST score and clinical variables were examined using univariate and multivariable linear regression analyses. Supplementary analyses were performed using the conventional G8 cutoff value (≤ 14).

Results

Ninety-one patients aged ≥ 65 years were included. The median COST score was 25 (range, 8–44). Lower G8 scores were significantly associated with greater FT (Pearson’s r = 0.345, p = 0.001). In the multivariable analysis, lower G8 scores (p = 0.048), poorer health-related quality of life (p = 0.036), and the receipt of multimodal therapy (p = 0.013) were independently associated with greater FT. Similar findings were observed when the G8 score was analyzed using the conventional cutoff value.

Conclusion

Among older patients with cancer, FT was independently associated with geriatric vulnerability assessed by the G8 screening tool, in addition to health-related quality of life and treatment burden. The assessment of geriatric vulnerability may complement conventional socioeconomic indicators in identifying patients at an increased risk of FT.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1007/s00520-026-11036-8.

Introduction

Cancer diagnosis and treatment impose substantial psychological, physical, and financial burdens not only on patients but also on their caregivers. Financial toxicity (FT) refers to the distress and economic hardship experienced by patients with cancer resulting from out-of-pocket expenses, lost income, and other indirect costs incurred since diagnosis. Although the term “financial toxicity” was first introduced by Connor in 2009, it has gained widespread recognition over the past decade as an integral issue within oncology [1]. Older adults with cancer are particularly vulnerable to FT due to age-related physiological decline, multimorbidity, and reduced socioeconomic resilience. Frailty-related vulnerability may further compromise patients’ ability to maintain independence, manage treatment-related costs, and cope with cumulative financial stress, thereby increasing the risk of financial distress. FT, in turn, has been associated with impaired health-related quality of life, treatment non-adherence, and altered treatment decision-making, potentially compromising the quality of cancer care [2–4]. Therefore, identifying older patients who are vulnerable to both clinical and socioeconomic consequences has become increasingly important in geriatric oncology.

While FT is a global issue, the nature and extent of this burden vary across regions due to differences in healthcare systems, economic conditions [5], and social factors such as healthcare accessibility, religion, and culture. In Japan, where cancer is highly prevalent and the population is rapidly aging, understanding the impact of economic burden has become increasingly important [6]. Japan operates under a universal health insurance system. Under this system, patients are generally responsible for a co-payment up to 30% of medical expenses, depending on their age and income. Furthermore, the High-Cost Medical Expense (HCME) Benefit system limits the out-of-pocket expenditures by reimbursing costs that exceed a predefined ceiling [7]. Despite these institutional protections, financial distress related to cancer treatment remains a significant challenge for many patients and their caregivers [8].

The aim of this single-center study was to evaluate FT and its association with frailty-related vulnerability among older patients with cancer managed in a regional area of Japan. We specifically examined whether questionnaire-based geriatric vulnerability assessments, including the G8 screening tool, were associated with FT.

Materials and methods

Study design

This observational, cross-sectional study was conducted between January 2024 and August 2025 at the International University of Health and Welfare Narita Hospital, a tertiary referral hospital located in a regional area of Japan.

We surveyed patients with malignancies who visited our hospital for treatment purposes. Patients with cancer attending the participating departments (medical oncology and urology) during the study period were screened for eligibility and invited to participate. For the present study, only patients aged ≥ 65 years were included in the analyses, consistent with the validated target population of the G8 screening tool. Both inpatients and outpatients were eligible to participate, provided they did not decline.

Each participant was asked to complete a series of questionnaire-based assessments. These included evaluations of frailty-related vulnerability, including G8 screening tool (G8), Instrumental Activities of Daily Living (IADL), Mini-Cog, and Charlson Comorbidity Index. Health-related quality of life (HRQoL) was assessed using the European Organisation Research and Treatment of Cancer Quality of Life Questionnaire Core 30 (EORTC QLQ-C30; Global Health Status), and FT was measured via Comprehensive Score for Financial Toxicity (COST). Additionally, a structured interview sheet was utilized to collect socio-demographic and clinical data, including the age, marital status, number of persons living together, whether having a private insurance coverage, annual household income, employment status, educational background, duration since cancer diagnosis, means of transportation when visiting the hospital, travel time to the hospital, presence of a companion when visiting the hospital, restriction of leisure due to the malignancy, and the awareness about the HCME system. Validated Japanese versions were used for all standardized questionnaires, and all participants were native Japanese speakers. Consequently, the study cohort was heterogeneous with respect to cancer type and disease stage. Participants did not receive any financial assistance for their participation in this study. No funding support was received for this study. The study was approved by the Institutional Ethics Committee, and written informed consent was obtained from all participants. We followed the guidelines outlined in the Declaration of Helsinki.

Statistical analysis

Patient characteristics were summarized using descriptive statistics. Continuous variables were expressed as medians with ranges or as means with standard deviations (SD), as appropriate, and categorical variables as frequencies and percentages. Prior to statistical analyses, the distribution of the COST score was examined to assess its normality. Visual inspection of histograms and Q-Q plots was performed, and normality was formally tested using the Shapiro–Wilk test.

The linear association between the G8 score and COST was evaluated using Pearson’s correlation coefficient (r). Correlation analyses were performed overall and in age-stratified subgroups. Sensitivity analyses using Spearman’s rank correlation coefficient were also performed to assess the robustness of the association between G8 and COST. For the age-stratified analyses, an age threshold of 75 years was applied based on its established clinical relevance in geriatric oncology. Differences in correlation coefficients between age groups were assessed using Fisher’s Z transformation.

Univariate analyses were performed using appropriate statistical tests depending on the covariates: simple linear regression for continuous variables, Student’s or Welch’s t-test, as appropriate, for comparisons between two groups, and one-way ANOVA for multi-group comparisons.

To prevent overfitting, the number of variables included in the multivariable model was limited according to the available sample size, maintaining an adequate observations-per-variable ratio. Age and sex were forced into the model regardless of univariate significance. The remaining candidate variables were selected based on a univariate significance level of p < 0.10 while avoiding multicollinearity among closely related variables. Individual treatment modalities were consolidated into a single binary variable representing the receipt of multimodal therapy (defined as ≥ 2 treatment modalities).

In supplementary analyses, G8 scores were dichotomized using the conventional cutoff value of ≤ 14 to define an abnormal screening status, in accordance with previous validation studies [9]. COST scores were compared between the dichotomized groups using Student’s t-test. Additional multivariable linear regression analyses were also performed using dichotomized G8 status as an independent variable. A two-sided p-value < 0.05 was considered statistically significant.

Statistical analyses were performed using the Stata software package (BE-Basic Edition, version 17.0; Stata Corp LLC, College Station, TX, USA).

Results

Between January 2024 and August 2025, 123 patients participated in the study, of whom 91 (74.0%) were aged ≥ 65 years. The median age was 74 years (range from 65 to 88). Of these, 81.3% were male. Forty-five participants (49.5%) had gastrointestinal cancer, 35 participants (38.5%) had urological cancer, 5 participants (5.5%) had hepato-biliary-pancreatic cancer, and 6 participants (6.6%) had cancers from other organs, including cancers of unknown primary origin (Fig. 1, Table 1). Sixty-three patients (69.2%) had advanced-stage disease, whereas 26 (28.6%) had early-stage disease. Most patients with non-advanced disease had localized urological cancers, including resectable prostate or bladder cancer. The majority of the participants belonged to the household annual income range of 2 to 4 million yen. All patients were covered by public health insurance, and 39 (42.9%) of them had additional private insurance coverage (Table 2). Additional socioeconomic factors are shown in Supplementary Table 1.

Fig. 1.

Distribution of primary cancer sites among study participants. Distribution of primary cancer sites in the study cohort

Table 1.

Baseline characteristics of patients aged ≥ 65 years (N = 91)

Variable	N	%	Median or mean ± SD	Range
Age			74	(65–88)
Sex
Male	74	81.3
Female	17	18.7
Origin of cancer
Gastroenterology	45	49.5
Urological	35	38.5
Hepato-biliary-pancreatic	5	5.5
Others	6	6.6
Disease duration (months)			18.7 ± 26.4	(0.5–240)
Disease status
Non-advanced	26	28.6
Advanced	63	69.2
Unknown	2	2.2
Recurrence
Yes	20	22.0
No	67	73.6
Unknown	4	4.4
Marriage
Married (including divorced or widowed)	78	85.7
Single	6	6.6
Unknown	7	7.7

Continuous variables are presented as median (range) or mean ± standard deviation (SD), as appropriate. Categorical variables are presented as n (%)

Table 2.

Financial toxicity, frailty-related vulnerability, and socioeconomic characteristics of the study participants

Variable	N	%	Mean ± SD	Range
COST score			24.6 ± 7.11	(8–44)
Annual household income (yen)
0–2Million	20	22.0
2–4Million	38	41.8
4–6Million	11	12.1
6–8Million	4	4.4
> 8Million	4	4.4
Unknown	14	15.4
G8			11.6 ± 2.6	(4.5–17)
IADL
8	29	31.9
< 8	55	60.4
Missing	7	7.7
Charlson Comorbidity Index
0	74	81.3
≧1	14	15.4
Missing	3	3.3
Mini-Cog
Positive	3	3.3
Negative	84	92.3
Missing	4	4.1
EORTC QLQ-C30			56.4 ± 25.4	(0–100)
Use of public transportation
Yes	3	3.3
No	82	90.1
Unknown	6	6.6
Private insurance
Yes	39	42.9
No	21	23.1
Unknown	31	34.1

Continuous variables are presented as mean ± standard deviation (SD) or median (range), as appropriate. Categorical variables are presented as n (%)

A total of 89 of the 91 participants (97.8%) completed the COST questionnaire. Visual inspection of a histogram and Q-Q plots suggested that the distribution of the COST score was approximately normal. The Shapiro–Wilk test also indicated no significant deviation from normality (p = 0.45). The median COST score was 25 (range, 8–44). The mean COST score was 24.6 ± 7.11 (Table 2).

We also evaluated the mean COST score between the two age subgroups among older patients (65–74 years vs. ≥ 75 years). The mean COST score was 24.0 ± 7.94 for patients aged 65–74 years (N = 46) and 25.2 ± 6.13 for those aged ≥ 75 years (N = 43). There was no statistically significant difference in the COST scores between these two age groups (Welch’s t-test, p = 0.448).

The strength of the association between the G8 and COST scores was assessed using Pearson’s correlation analysis. A weak but positive correlation was observed between the two variables (r = 0.345, p = 0.001; Fig. 2a), indicating that higher G8 scores were associated with lower FT. Similar results were observed in Spearman’s rank correlation analyses (ρ = 0.309, p = 0.003).

Fig. 2.

Relationship between the G8 score and the COST score. a scatterplot showing the correlation between continuous G8 and COST scores. the solid line represents the linear regression trend line. pearson’s correlation coefficient (r = 0.345, p = 0.001) indicates a weak but statistically significant positive correlation. b boxplot comparing COST scores according to G8 screening status using the cutoff value of ≤ 14

To evaluate variables associated with FT, we performed a univariate linear regression analysis (Table 3). Lower COST scores, indicating greater FT, were significantly associated with the following variables. Among patient-related variables, the G8 and HRQoL were significantly associated with FT. Disease- and treatment-related factors were also associated with FT. Patients with progressive disease (p = 0.021), those who received chemotherapy (p = 0.020), and those who underwent multiple treatment modalities (p = 0.004) showed greater FT.

Table 3.

Results of univariate and multivariable analyses of factors associated with the COST score. Continuous variables in univariate analyses are presented as regression coefficients (β) with 95% confidence intervals (CI), whereas binary variables are presented as mean differences with 95% CI. For binary variables, mean differences were calculated as mean(0) − mean(1). only p-values are shown for variables analyzed using one-way ANOVA. p < 0.05. Statistically significant $p$-values ($p<0.05$) are denoted with an asterisk (*)

Variable	Univariate estimate (95% CI)	p-value*	Multivariate β (95% CI)	p-value*
Age	0.13 (−0.14 to 0.40)	0.332	0.24 (−0.04 to 0.49)	0.054
Sex	1.77 (−2.14 to 5.66)	0.371	0.64 (−2.94 to 4.21)	0.723
Marriage	−3.73 (−9.76 to 2.30)	0.222
Living with someone	−1.90 (−5.60 to 1.78)	0.306
Employment status	−1.51 (−5.02 to 1.99)	0.393
Education status	–	0.731
Income	–	0.065
Limitations of leisure	1.17 (−2.04 to 4.39)	0.468
G8	0.93 (0.39–1.47)	0.001*	0.59 (0.05–1.17)	0.048*
IADL < 8	3.09 (−0.07 to 6.24)	0.055	−2.53 (−5.39 to −0.316)	0.081
CCI	0.59 (−0.21 to 1.40)	0.149
Mini-Cog	−3.21 (−7.55 to 1.12)	0.144
QOL	0.10 (0.05–0.16)	< 0.001*	0.06 (0.004–0.12)	0.036*
Disease duration	0.01 (−0.04 to 0.07)	0.621
Recurrence	1.90 (−1.73 to 5.54)	0.301
Progression	3.82 (0.60–7.04)	0.021*	−1.38 (−4.48 to 1.71)	0.376
Surgery	1.82 (−1.18 to 4.82)	0.232
Chemotherapy	4.11 (0.67–7.54)	0.020*
ICI	2.25 (−1.27 to 5.76)	0.207
Radiation	−1.21 (−4.60 to 2.17)	0.478
Multiple treatment modalities	−4.32 (−7.21 to −1.43)	0.004*	−3.71 (−6.60 to −0.81)	0.013*
Time to the hospital	−1.36 (−4.40 to 1.69)	0.379
Use of public transportation	7.14 (−0.84 to 15.11)	0.079

In the multivariable linear regression analysis, the receipt of multiple treatment modalities (β = −3.71; 95% confidence interval (CI), −6.60 to −0.81; p = 0.013), QOL (β = 0.07; 95% CI, 0.004 to 0.12; p = 0.036), and G8 (β = 0.61; 95% CI, 0.05 to 1.17; p = 0.048) emerged as significant and independent predictors of the COST score (Table 3). Other covariates in the final model, including age, sex, disease progression, and IADL, did not show statistically significant associations. The mean variance inflation factor was 1.17, indicating no evidence of multicollinearity.

Although the correlation between the G8 score and COST appeared stronger in patients aged 65 to 74 years (r = 0.485) than in those ≥ 75 years (r = 0.153), this difference did not reach statistical significance (p = 0.089, Fisher’s Z transformation).

In the supplementary sub-analysis using the conventional G8 cutoff value (≤ 14), patients with an abnormal G8 screening demonstrated significantly lower COST scores than those with G8 scores > 14 (24.0 ± 6.6 vs 29.8 ± 7.2, p = 0.006; Fig. 2b). When this dichotomized G8 status was entered into the multivariable model, it showed a marginal trend toward an independent association with lower COST scores (β = −3.61; 95% CI, −7.83 to 0.60; p = 0.092). Notably, the direction of the association between G8 and COST was consistent across both the continuous and dichotomized analyses.

Discussion

FT is an increasingly recognized issue in clinical oncology worldwide, with a high prevalence reported among patients with cancer across both high-income and low-/middle-income countries [10, 11]. Despite the relatively generous financial protection provided by Japan’s universal health insurance system [7], FT still poses a considerable burden on patients, with more than half reporting the use of alternative coping strategies [12].

In the present study, most participants had a household annual income of 2 to 4 million yen, which is lower than the mean household income in Japan in 2022 (5.24 million yen) [13]. The median COST score of 25 (range 8–44) and mean COST score of 24.6 ± 7.11 were slightly higher than, but comparable to, those in previous studies [12, 14].

Several parameters are known to influence FT. Previous studies have reported lower levels of economic distress among older patients, identifying younger age as a risk factor [3, 5, 15]. However, older adults with cancer still face substantial economic jeopardy; older patients with cancer incur significantly higher out-of-pocket expenses than those without cancer [4]. Therefore, the present study focused on factors associated with FT especially among older patients with cancer. Globally and in Japan, cancer incidence increases with advancing age, and the majority of patients with cancer are older adults [16, 17]. In parallel, advances in cancer diagnosis and treatment have substantially improved patient survival, resulting in extended treatment durations and prolonged survivorship [18]. Although treatment duration was not significantly associated with FT in our cohort, prolonged therapy may still contribute to cumulative financial burden in some patients, as suggested previously [19]. Older adults with cancer undergoing long-term treatment may experience not only disease-related symptoms and treatment-related adverse effects, but also cumulative financial distress over time.

The association between HRQoL and FT observed in our study is consistent with recent global evidence. A recent cross-sectional study reported that greater financial distress was independently associated with poorer HRQoL among patients with urologic cancer, supporting the close relationship between economic burden and patient well-being [20]. Our findings extend these observations to a heterogeneous cohort of older patients with cancer.

Recent evidence suggests that intensive multimodal cancer treatment may contribute to financial distress through cumulative treatment burden and productivity loss [21]. Given that treatment intensity is closely related to financial distress across various cancer types, a major methodological strength of our study is that we strictly adjusted for both baseline HRQoL and the experience of multimodal therapy, allowing us to evaluate independent associations within a mixed-primary cohort.

Our findings suggest that FT in older patients with cancer is multifactorial, being associated with patient-reported HRQoL, treatment burden, and frailty-related vulnerability. Among these factors, the continuous G8 score remained independently associated with FT even after adjustment for demographic characteristics, HRQoL, and treatment-related variables, highlighting the potential role of geriatric vulnerability beyond conventional clinical and socioeconomic factors. Patients receiving multiple treatment modalities may experience a greater cumulative treatment burden, including repeated hospital visits, indirect costs, and work disruption, which may collectively contribute to FT.

While high-quality literature often focuses on physical vulnerability in older patients primarily in the context of acute treatment-related toxicities across various oncological therapies [22, 23] our findings suggest an additional perspective for older cancer survivors. Our findings indicate that a lower G8 score is independently associated with higher FT. This association may reflect reduced socioeconomic resilience and increased care-related burden among vulnerable patients. Similar findings were observed when the G8 score was analyzed using the conventional cutoff value of ≤ 14, supporting the consistency of the observed association. To the best of our knowledge, no prior study has directly evaluated the relationship between the G8 screening tool and the COST measure of FT in patients with cancer.

The G8 is one of the most widely used geriatric screening tools in oncology, developed and validated for older adults with cancer. It is widely used in geriatric oncology clinical trials [24] and practice. A major strength of this study is the integration of questionnaire-based assessments related to frailty and vulnerability with a validated measure of FT. Among the assessed variables, the G8 showed similar associations when analyzed as both a continuous variable and using the conventional cutoff value. FT not only burdens patients emotionally and economically but is also associated with worse health outcomes [4, 11]. Although the G8 was not originally developed to assess FT, impairments captured by this tool, including nutritional, functional, and vulnerability-related domains, may overlap with factors contributing to reduced socioeconomic resilience. Therefore, abnormal G8 findings may help clinicians recognize patients who could also be at an increased risk of financial distress, potentially facilitating earlier supportive interventions without imposing substantial additional assessment burden on patients and caregivers. Furthermore, financial distress has also been associated with clinical outcomes; severe financial distress compromises care by driving treatment non-adherence and prescription abandonment [4]. Intercepting FT early may contribute to improved treatment adherence and clinical outcomes. These findings suggest that assessment of geriatric vulnerability may complement conventional socioeconomic indicators when identifying older patients at risk of FT.

Limitation

Several limitations of this study should be acknowledged. First, this was conducted at a single center with a relatively small sample size (n = 91), which may limit statistical power and generalizability of the findings. Second, the heterogeneity in primary tumor sites and treatment modalities, combined with the cross-sectional design, resulted in variability in treatment duration, cancer stage, and timing of survey administration. Third, because this study was conducted within a specific social and healthcare context, particularly Japan’s unique health insurance system and regional characteristics, the findings may not be fully generalizable to other populations or healthcare settings. Accordingly, these findings should be interpreted with caution and warrant confirmation in larger multicenter studies. Nevertheless, this study provides valuable, hypothesis-generating insights as the first to demonstrate a direct association between G8 screening scores and FT.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Momoko Kobayashi (MK) conceived and designed the study, performed the statistical analyses, and drafted the manuscript. Hajime Higuchi (HH) supervised the study, provided critical revisions, and approved the final manuscript. Yutaka Sugawara (YS) and Eri Mannouji (EM) contributed to patient data collection and interpretation. Jun Miyazaki (JM), Shinich Hisasue (SH), Akinobu Katami (AK), and Hiroshi Yuuki (HY) contributed to patient enrollment and data acquisition. All authors have read and approved the final manuscript and meet the authorship criteria.

Funding

Open Access funding provided by International University of Health and Welfare. The publication of this study was supported by research funding from the International University of Health and Welfare.

Data availability

The datasets generated and/or analyzed during the current study are available from the corresponding author on reasonable request, in accordance with institutional regulations and ethical guidelines.

Declarations

Ethics approval and consent to participate

This study was approved by the Institutional Review Board of IUHW Narita Hospital (Approval No.23-Im-067). All participants provided written informed consent for participation in the study and for publication of anonymized data.

Consent for publication

All participants provided consent for publication of anonymized data; no individual participant is identifiable in this manuscript.

Competing interests

The authors declare no competing interests.