https://orcid.org/0000-0002-8179-4927
Figures
Abstract
Introduction
The COmprehensive Score for Financial Toxicity (COST) has proven to be a reliable tool for quantifying the impact of financial toxicity (FT) in patients with cancer in clinical and public health settings. However, the COST has not yet been validated in Vietnam. Therefore, we aimed to evaluate its reliability and validity among Vietnamese patients with cancer.
Methods
A cross-sectional study was conducted in a sample of 300 patients with cancer aged 27–95 years (mean: 58.5±11.2) in a tertiary hospital. The COST was translated into Vietnamese and English and adjusted to suit the local culture. Reliability was evaluated using Cronbach’s alpha and McDonald’s omega coefficients. The construct and convergent validities were also assessed.
Results
The COST demonstrated good internal consistency and reliability (Cronbach’s alpha = 0.913; McDonald’s omega = 0.915). The exploratory factor analysis revealed two factors that explained 64.9% of the variance. The adjusted fit indices indicated a good fit of the model (χ2 (39) = 67.78, p = 0.003; standardized root mean squared residual = 0.042; Tucker–Lewis index = 0.971; comparative fit index = 0.979; root mean square error of approximation = 0.061, 90% confidence interval = 0.035–0084). Higher COST scores were significantly correlated with higher health-related quality of life (EQ-5D-5L utility score: r = 0.21, p = 0.002; EQ VAS: r = 0.28, p < 0.001). Multivariate quantile regression analysis revealed that female sex, rural residence, and unstable job/unemployment were associated with lower COST scores. There was no statistically significant difference in other factors, including clinical factors (types of cancer, staging, and treatment modalities).
Citation: Tran BT, Le DD, Nguyen TG, Nguyen MT, Nguyen MH, Dang CK, et al. (2024) Validation of the COmprehensive Score for Financial Toxicity (COST) in Vietnamese patients with cancer. PLoS ONE 19(6): e0306339. https://doi.org/10.1371/journal.pone.0306339
Editor: Le An Pham, Ho Chi Minh City University of Medicine and Pharmacy, VIET NAM
Received: April 8, 2024; Accepted: June 15, 2024; Published: June 28, 2024
Copyright: © 2024 Tran et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All relevant data are within the manuscript and its Supporting Information files.
Funding: This study was supported by Hue University (DHH 2022 – 04–175). The funder played no role in the study design, data collection and analysis, decision to publish, or manuscript preparation.
Competing interests: The authors have declared that no competing interests exist.
Introduction
There is an increase in cancer incidence both globally and in Vietnam [1–3]. Notably, people with cancer experience a high economic burden and financial impact, such as financial distress [4, 5]. Consequently, this has been associated with poor adherence to treatment and reduced health-related quality of life (HRQoL) [6, 7], which has negatively impacted their family (education, work, and poverty) [8, 9]. The quantitative economic burden of cancer is mostly predominantly characterized as out-of-pocket expenses in literature [10]. However, understanding its impact has only become a major focus in recent years [11], and the term financial toxicity (FT) has recently been adopted by De Souza et al. [11, 12].
FT refers to the negative effect of cancer on an individual’s financial situation and its subsequent impact on physical and mental health. Among all instruments for assessing FT, the COmprehensive Score for Financial Toxicity (COST), which was developed and adapted by De Souza et al. (2014), is the most used and is a clinically relevant patient-reported outcome [11, 12]. This self-administered tool comprises 11 items and has demonstrated good psychometric properties in previous studies conducted in high-income countries [13–18]. However, its validity for assessing financial stress in Vietnam remains unexplored. The findings of this study aim to provide robust evidence regarding COST’s ability to measure constructs of interest and its potential applications in Vietnam.
Materials and methods
Participants
We conducted a cross-sectional validation study of the COST in a single adult oncology center in a tertiary hospital, which is the largest referral hospital in the central and highlands of Vietnam. A sample size of 300 inpatients with the most common cancer was recruited [19], following the rule of thumb suggested by Wolf et al. [20]. This rule recommends including 10 cases per variable in the model for validation. COST contains 11 items; therefore, a sample of 300 patients was considered ideal for robust validation [3].
Data collection
After obtaining informed written consent from the inpatients, public health students conducted face-to-face interviews using the constructed questionnaires in a private room. All clinical cancer information was obtained from medical records with the assistance of an oncologist. Data were collected between September 1, 2022, and March 31, 2023.
Instruments and variables
Demographic and clinical characteristics.
A demographic survey assessed age, sex, education level, marital status, household income, health-risk behaviors (smoking and drinking status), and health insurance benefits. Clinical information included the cancer type, clinical stage, and treatment modality.
FT. Translation and adaptation. We obtained the COST instrument through forward-backward translation. This process involved two fluent English oncologists and experts in patient-reported outcomes, who were translating the original English version into Vietnamese. The translated version was then back-translated into English by a different translator who was blinded to the original English version. Discrepancies between the original and back-translated versions were resolved through discussions between the translators and the principal investigator to ensure semantic equivalence.
Instrument description. The Vietnamese version of the COST retains the core structure of the original scale and comprises 11 elements. It utilizes a 5-point Likert scale ranging from 0 to 4 (see Supporting Information [S1 File] for the COST scale in Vietnamese). Higher COST scores (0–44) indicate better financial well-being, with items 2, 3, 4, 5, 8, 9, and 10 being reverse scored [12, 21].
HRQoL assessments. In this study, we used the EQ-5D-5L instruments (comprising the EQ-5D and a visual analog scale (EQ VAS)), a widely used and culturally appropriate scale to measure HRQoL. This scale is particularly valuable because it has been validated with a specific set of Vietnamese values designed to ensure an accurate interpretation of the results within the Vietnamese context [22, 23]. The sets of values for the EQ-5D and EQ VAS were -0.5115 to 1 and 0 to 100, respectively [22, 23].
Statistical analysis
Continuous variables were summarized as means, standard deviations (SD), medians, and interquartile ranges (IQR). Categorical variables are described using frequencies and percentages.
The reliability of the scale was evaluated using Cronbach’s alpha and McDonald’s omega coefficients. A coefficient of ≥0.7 generally indicates good internal consistency [24].
The construct validity involves two steps: exploratory factor analysis (EFA) and confirmatory structure analysis (CFA). EFA identified the underlying factors influencing questionnaire responses. The screen plot and varimax rotation determined the number of factors retained [25]. EFA ensured clear and reliable factors through several strategies: (1) removing items with low loadings (< 0.4) for minimal contribution; (2) requiring factors to have at least three items to adequately capture the underlying construct; and (3) removing items with high cross-loadings (> 0.3) to minimize ambiguity and enhance factor distinction [26]. Bartlett’s sphericity test and the Kaiser-Meyer-Olkin (KMO) confirmed the suitability of the EFA [27, 28]. Building on the initial exploration of the factor structure using EFA (two-factor loading was selected), CFA was used to assess the validity of the identified factor structure [27, 29]. Notably, several fit indices (root mean square error of approximation (RMSEA), goodness-of-fit index (GFI), comparative fit index (CFI), and adjusted goodness-of-fit index (AGFI)) indicated good model fit (p < 0.001).
Convergent validity was evaluated using the Pearson correlation coefficient between the COST score and HRQoL (EQ-5D and EQ-VAS). The Shapiro–Wilk W test (p < 0.001) indicated that the COST scores were not normally distributed due to skewness. Therefore, nonparametric tests were used to analyze the differences in COST scores based on the characteristics. The Mann–Whitney U test was used for comparisons between two groups, and the Kruskal–Wallis test was used for comparisons between three or more groups. Additionally, a multivariate quantile regression analysis (at the 50th percentile) was conducted to investigate the factors associated with the COST score. All independent variables were included for adjustment in the model. Statistical significance was set at p< 0.05. Data analysis was performed using IBM SPSS Statistics (version 29.0) and Amos software (version 22.0).
Results
Sample characteristics
Of the 330 hospitalized patients approached, 300 (with a participation rate of 90.9%) consented to participate and completed the survey. Table 1 summarizes the characteristics and COST scores of these patients. The participants had a mean age of 58.4 years (SD = 11.2; range, 27–95 years). Males comprised more than half of the group (69.3%), and 68.3% had at least a secondary school education. Biliary/pancreatic cancer was the most common diagnosis (31.6%), and chemotherapy was administered to more than half of the participants (58.9%). The median and mean COST scores were 13.00, IQR (11–18), and 15.1 ± 7.3, respectively.
Reliability
Cronbach’s alpha coefficients for each domain of the COST measure were 0.898 and 0.914, respectively (Table 2). Overall, Cronbach’s alpha coefficients and Omega coefficients were 0.913 and 0.915, respectively.
Validity
Construct validity.
Two main components were extracted, representing 64.9% of the cumulative variance. The first principal component consisted of seven items ranging from 0.479 to 0.869, and the second consisted of four items ranging from 0.621 to 0.861. Table 3 shows the principal components and factor loadings of COST.
Fig 1 illustrates the principal components and factor loadings of the COST. The fit of the model was evaluated using several indices. The RMSEA value (0.056) was below the recommended threshold of 0.07, indicating a close fit. Furthermore, the GFI (0.955), CFI (>0.9), and AGFI (0.926) exceeded their respective cutoff values of 0.9, further supporting a well-fitting model (p < 0.001).
Convergent validity.
The convergence validity analysis showed a mildly positive correlation between the COST measure and HRQoL. Specifically, the COST scores were positively correlated with the utility score EQ-5D (r = 0.210, p = 0.002) and EQ VAS (r = 0.273, p < 0.0001) (Fig 2). Bivariate analysis revealed several factors that were significantly associated with the COST score (all p < 0.05). These factors included female sex, residential area, level of education, occupation, economic status of the household, health insurance copayment, cancer type, and treatment modality (Table 1). In the multivariate model, we found that female sex (coefficient: -5.96, 95% confidence interval (CI): -10.39 –-1.54), residence in rural areas (coefficient: -4.57, 95% CI: -7.41 –-1.73) and occupation with unstable income jobs/unemployment (coefficients: -5.39, 95% CI: -8.71 ––2.07; and -4.64, 95% CI: -8.63 –-0.65, respectively) were statistically associated with a lower COST score. There was no statistical significance for other factors, including clinical factors (cancer type, staging, or treatment modalities) (Table 4).
Discussion
In this study, we aimed to validate the use of the COST in a Vietnamese population of hospitalized patients with cancer. Our findings provide evidence for the COST’s reliability, construct validity, and convergent validity, suggesting its potential as a valuable tool for identifying and measuring FT in this population.
Reliability
Cronbach’s alpha coefficients for individual domains (0.8976 and 0.9137) and the general scale (0.913) exceeded the commonly accepted threshold of 0.7, indicating high internal consistency. Furthermore, Omega’s coefficient (ω = 0.915) confirmed high reliability. These findings are consistent with previous studies that have reported a good reliability (Cronbach’s alpha) of COST in different populations such as Brazil, Japan, and India at 0.83, 0.87, and 0.92, respectively [16–18].
Validity
This finding aligns with the theoretical framework of COST and supports its validity. EFA identified two distinct factors that accounted for 64.9% of the cumulative variance. All items were loaded into these factors with loads ranging from 0.479 to 0.869. The first factor included elements FT2, FT3, FT4, FT5, FT8, FT9, and FT10, whereas the second factor included elements FT1, FT6, FT7, and FT11.
Our results are consistent with those reported in China by Yu et al. (2021), who found a cumulative variance of 63.04%. Similar to our study, their analysis revealed two factors, with the first factor containing seven elements and the second containing four. Yu et al. suggested that the COST scale could be conceptualized as measuring two dimensions: positive wealth status and negative psychosocial aspects [15]. However, a study by Shim et al. (2022) in Korea also identified a two-factor structure; however, this study had some differences in item loadings. The first factor included items FT3, FT5, FT8, FT9, and FT10, and the second factor included items FT1, FT4, FT6, and FT7. Items FT4, FT11 (which appeared in both factors in their analysis but with a low load), and FT2 were not included in either factor. The authors attributed this discrepancy to linguistic validation issues; however, they concluded that the general validity of the scale remained good [13].
Our findings differ from those of the original development of the COST scale, which reported a one-factor structure explaining 93% of the variance in an English-speaking sample. Other studies have also reported variations. Dar et al. (2021) identified a one-factor structure that explained 56% of the variance in India [16]. However, Sharif et al. (2021) identified a three-factor structure that explained 65% of the variance in Iran [14]. These discrepancies highlight the importance of COST validation in various contexts. However, in our data analysis, achieving an explained variance of over 80% required five-factor loadings, which is not ideal for the COST scale with only 11 items. Our final two-factor model demonstrated acceptable fit indices consistent with the findings of Yu et al. in China [15]. These comparisons suggest that the COST is likely to measure distinct but related constructs relevant to FT in patients with cancer.
Convergent validity
The significant positive correlation between the COST and HRQoL measured using the EQ-5D or EQ VAS supports the convergent validity of the COST. This finding provides evidence that patients with higher COST scores experience higher HRQoL; consequently, a higher level of FT is associated with worse HRQoL, which is consistent with previous studies [7, 30, 31]. In addition, factors such as low socioeconomic status and cancer treatment bolster evidence for the multidimensionality of the FT [6, 32] and the validity of the COST for application across diverse clinical characteristics and settings [7].
To our knowledge, this is the first study to validate the COST in patients with cancer in Vietnam. Second, the sample size for this study was sufficient, particularly for scale validation, which suggests a range of 30–460 [20].
This study has some limitations. First, the cross-sectional design limited our ability to evaluate how the FT changes over time during treatment. Previous research in China has shown fluctuations in FT throughout treatments [15]. Second, our sample included mainly patients with common cancers, most of whom were in advanced stages. This limits the generalizability of our findings to other types and specific stages of cancer. Third, we did not establish a threshold score for the COST scale because of the lack of a gold standard for measuring FT. Future research should investigate possible threshold scores to identify patients who may benefit from FT-targeted interventions.
Conclusions
Our study provides positive preliminary evidence for the reliability, construct, and convergent validity of the COST in a Vietnamese population of hospitalized patients with cancer. This suggests that the COST could be a valuable tool for healthcare professionals to assess and monitor FT in this population. Future research with larger and more diverse samples is required to confirm these findings and explore the potential applications of COST in clinical and public health settings.
References
- 1. Pham DX, Phung AHT, Nguyen HD, Bui TD, Mai LD, Tran BNH, et al. Trends in colorectal cancer incidence in Ho Chi Minh City, Vietnam (1996–2015): Joinpoint regression and age–period–cohort analyses. Cancer epidemiology. 2022;77:102113. pmid:35078008
- 2. Pham DX, Nguyen HD, Phung AHT, Bui TD, Tran TS, Tran BNH, et al. Trends in incidence and histological pattern of thyroid cancer in Ho Chi Minh City, Vietnam (1996–2015): a population-based study. BMC Cancer. 2021;21(1):296. pmid:33743620
- 3. Nguyen SM, Deppen S, Nguyen GH, Pham DX, Bui TD, Tran TV. Projecting Cancer Incidence for 2025 in the 2 Largest Populated Cities in Vietnam. Cancer Control. 2019;26(1):1073274819865274. pmid:31331188
- 4. Ngan TT, Van Minh H, Donnelly M, O’Neill C. Financial toxicity due to breast cancer treatment in low- and middle-income countries: evidence from Vietnam. Support Care Cancer. 2021;29(11):6325–33. Epub 2021/04/17. pmid:33860362; PubMed Central PMCID: PMC8464564.
- 5. The ACTION Study Group. Catastrophic health expenditure and 12-month mortality associated with cancer in Southeast Asia: results from a longitudinal study in eight countries. BMC Medicine. 2015;13(1):190. pmid:26282128
- 6. Xu B, So WKW, Choi KC, Huang Y, Liu M, Qiu L, et al. Financial Toxicity and Its Risk Factors Among Patients with Cancer in China: A Nationwide Multisite Study. Asia-Pacific Journal of Oncology Nursing. 2024:100443. pmid:38665637
- 7. Pangestu S, Rencz F. Comprehensive Score for Financial Toxicity and Health-Related Quality of Life in Patients With Cancer and Survivors: A Systematic Review and Meta-Analysis. Value Health. 2023;26(2):300–16. Epub 2022/09/06. pmid:36064514.
- 8. Kong YC, Kimman M, Subramaniam S, Yip CH, Jan S, Aung S, et al. Out-of-pocket payments for complementary medicine following cancer and the effect on financial outcomes in middle-income countries in southeast Asia: a prospective cohort study. Lancet Glob Health. 2022;10(3):e416–e28. Epub 2022/02/19. pmid:35180423.
- 9. Hoang VM, Pham CP, Vu QM, Ngo TT, Tran DH, Bui D, et al. Household Financial Burden and Poverty Impacts of Cancer Treatment in Vietnam. BioMed Research International. 2017;2017:9350147. pmid:28904976
- 10. Bestvina CM, Zullig LL, Yousuf Zafar S. The implications of out-of-pocket cost of cancer treatment in the USA: a critical appraisal of the literature. Future oncology. 2014;10(14):2189–99. pmid:25471033
- 11. Witte J, Mehlis K, Surmann B, Lingnau R, Damm O, Greiner W, et al. Methods for measuring financial toxicity after cancer diagnosis and treatment: a systematic review and its implications. Annals of Oncology. 2019;30(7):1061–70. pmid:31046080
- 12. De Souza JA, Yap BJ, Hlubocky FJ, Wroblewski K, Ratain MJ, Cella D, et al. The development of a financial toxicity patient-reported outcome in cancer: The COST measure. Cancer. 2014;120(20):3245–53. Epub 2014/06/24. pmid:24954526.
- 13. Shim S, Kang D, Kim N, Han G, Lim J, Kim H, et al. Validation of Korean Version of the COmprehensive Score for financial Toxicity (COST) Among Breast Cancer Survivors. Cancer Res Treat. 2022;54(3):834–41. Epub 2021/10/15. pmid:34645130; PubMed Central PMCID: PMC9296937.
- 14. Sharif SP, Moshtagh M, Sim OF, Naghavi N, Nia HS. Psychometric evaluation of the comprehensive score for financial toxicity scale among iranian cancer patients. Journal of Cancer Policy. 2021;30:100308. pmid:35559803
- 15. Yu HH, Yu ZF, Li H, Zhao H, Sun JM, Liu YY. The COmprehensive Score for financial Toxicity in China: Validation and Responsiveness. J Pain Symptom Manage. 2021;61(6):1297–304.e1. Epub 2021/01/08. pmid:33412268.
- 16. Dar MA, Chauhan R, Sharma KK, Trivedi V, Dhingra S, Murti K. Assessing the reliability and validity of comprehensive score for financial toxicity (COST) among radiation oncology patients in India: a cross-sectional pilot study. Ecancermedicalscience. 2021;15:1219. Epub 2021/06/24. pmid:34158823; PubMed Central PMCID: PMC8183655.
- 17. Honda K, Gyawali B, Ando M, Sugiyama K, Mitani S, Masuishi T, et al. A prospective survey of comprehensive score for financial toxicity in Japanese cancer patients: report on a pilot study. Ecancermedicalscience. 2018;12:847. Epub 2018/08/07. pmid:30079109; PubMed Central PMCID: PMC6057653.
- 18. De Alcantara Nogueira L, Koller FJ, Marcondes L, de Fátima Mantovani M, Marcon SS, Guimarães PRB, et al. Validation of the comprehensive score for financial toxicity for Brazilian culture. Ecancermedicalscience. 2020;14:1158. Epub 2021/02/13. pmid:33574903; PubMed Central PMCID: PMC7864683.
- 19. Phuong NTT, Thang TB, Anh NDQ, Hanh NM, Chau HM, Thanh NT, et al. The distribution of gastrointestinal cancer in two specialized oncology hospitals in Hue and Da Nang cities, 2022. Vietnam Journal of Preventive Medicine. 2024;33(6 Supp):104–10. (Vietnamese)
- 20. Wolf EJ, Harrington KM, Clark SL, Miller MW. Sample Size Requirements for Structural Equation Models: An Evaluation of Power, Bias, and Solution Propriety. Educational and Psychological Measurement. 2013;73(6):913–34. pmid:25705052
- 21. De Souza JA, Yap BJ, Wroblewski K, Blinder V, Araújo FS, Hlubocky FJ, et al. Measuring financial toxicity as a clinically relevant patient-reported outcome: The validation of the COmprehensive Score for financial Toxicity (COST). Cancer. 2017;123(3):476–84. Epub 2016/10/08. pmid:27716900; PubMed Central PMCID: PMC5298039.
- 22. Mai VQ, Sun S, Minh HV, Luo N, Giang KB, Lindholm L, et al. An EQ-5D-5L Value Set for Vietnam. Quality of Life Research. 2020;29(7):1923–33. pmid:32221805
- 23.
Mai VQ. A feasibility and applicability study of a health-related quality of life measurement in Vietnam. 2023. Doctoral dissertation, Umeå University.
- 24. Tavakol M, Dennick R. Making sense of Cronbach’s alpha. Int J Med Educ. 2011;2:53–5. Epub 2011/06/27. pmid:28029643; PubMed Central PMCID: PMC4205511.
- 25. Costello AB, Osborne J. Best practices in exploratory factor analysis: Four recommendations for getting the most from your analysis. Practical assessment, research, and evaluation. 2019;10(1):7.
- 26. Tavakol M, Wetzel A. Factor Analysis: a means for theory and instrument development in support of construct validity. Int J Med Educ. 2020;11:245–7. Epub 20201106. pmid:33170146; PubMed Central PMCID: PMC7883798.
- 27. Kaiser HF. An index of factorial simplicity. Psychometrika. 1974;39(1):31–6.
- 28. Bartlett MS. The Effect of Standardization on a χ2 Approximation in Factor Analysis. Biometrika. 1951;38(3/4):337–44.
- 29. Hu L-t Bentler PM. Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling. 1999;6(1):1–55.
- 30. Jiang H, Lyu J, Mou W, Jiang Q, Du J. Association between financial toxicity and health-related quality of life in cancer survivors: A systematic review. Asia-Pacific Journal of Clinical Oncology. 2023;19(4):439–57. pmid:36457166
- 31. Ghazal LV, Abrahamse P, Ward KC, Morris AM, Hawley ST, Veenstra CM. Financial Toxicity and Its Association With Health-Related Quality of Life Among Partners of Colorectal Cancer Survivors. JAMA Netw Open. 2023;6(4):e235897. Epub 2023/04/07. pmid:37022684; PubMed Central PMCID: PMC10080378.
- 32. Wang S, Wang J, Kang H, Zeng L, Liu G, Qiu Y, et al. Assessment of the prevalence and related factors of financial toxicity in cancer patients based on the COST scale: A systematic review and meta-analysis. Eur J Oncol Nurs. 2024;68:102489. Epub 2023/12/21. pmid:38118267.