Construction And Validation Of A Predictive Model For The Risk Of Financial Toxicity In Patients With Head And Neck Cancer

Objective:To localize the Financial Index of Toxicity(FIT)questionnaire into Chinese and analyze its reliability and validity.To describe and analyze the current status of financial toxicity and risk factors among patients with Head and Neck Cancer(HNC).To construct and validate an Nomogram model for HNC patients.Methods:(1)Systematic searches were conducted in both Chinese and English databases.After dual screening,relevant data were extracted.The quality and risk of bias of included studies were evaluated using the National Institutes of Health’s tool for assessing the quality of observational studies.Meta-analysis of incidence rates and risk factors was conducted using Stata 16.0.(2)The FIT questionnaire was translated into Chinese following the Brislin translation model,and a cross-cultural adaptation and pilot study were conducted.Face-to-face questionnaire surveys were administered to HNC patients at a tertiary cancer hospital,and a subset of 50 patients were retested after a 4-week interval.Six experts in relevant fields evaluated the Chinese version of FIT.The reliability and validity of the questionnaire were assessed using SPSS 26.0 software.Receiver Operating Characteristic(ROC)analysis was used to calculate the cutoff value for FIT.(3)A cross-sectional survey was conducted using the translated FIT questionnaire.Sample size calculation was based on the results of the meta-analysis,and a general information questionnaire was developed for patient data collection.Data collection took place from October 2021 to November 2022 at a tertiary cancer hospital in Chengdu.Descriptive analysis,univariate analysis,and binary logistic regression analysis were performed using SPSS 26.0 to determine the incidence rate and risk factors of financial toxicity among HNC patients.(4)Based on the logistic regression results,a pathway model was developed and visualized using the Free Statistical software V1.7.ROC curves,calibration curves,and decision curves were generated using R language 4.2 to evaluate the discrimination,calibration,and clinical utility of the model.The Area Under Curve(AUC),Hosmer-Lemeshow(H-L)test,and decision curve analysis were used for evaluation.Results:(1)After systematic searches and screening,a total of 31 studies were included.The meta-analysis using a random-effects model revealed an overall incidence rate of financial toxicity(FT)in cancer patients of 45%(95%CI:38%-53%,I~2=97.3%,P<0.01).Among them,mild FT accounted for 42%,moderate FT accounted for 17%,and severe FT accounted for 1%.Meta-analysis of 28 studies identified 9 potential risk factors for FT:low income,younger age,lack of commercial insurance,higher out-of-pocket expenses,non-white ethnicity,advanced cancer stage,unemployment,recent diagnosis,and unmarried status.(2)The Chinese version of FIT consisted of three dimensions and nine items.The pre-test results indicated that the questionnaire items were clear and easily understandable for patients.The internal consistency,assessed by Cronbach’s alpha coefficient,was 0.896 for the overall scale and ranged from 0.737 to0.864 for the three dimensions(economic distress,financial hardship,and productivity loss).The test-retest reliability was 0.783(P<0.05).The content validity index for the scale level was 0.83,and for the item level,it ranged from 0.822 to 1.00.Exploratory factor analysis yielded a Kaiser-Meyer-Olkin(KMO)value of 0.908 and a significant Bartlett’s test of sphericity(P<0.01).Three factors were extracted,explaining a cumulative variance contribution of 77.85%.ROC analysis revealed an AUC of 0.946with a standard error of 0.015.The optimal cutoff score for high risk was determined to be 43,with a sensitivity of 0.830 and specificity of 0.915.(3)The modeling group included 391 eligible participants.The overall incidence rate of financial toxicity among HNC patients was found to be 62.2%.Univariate analysis indicated significant differences(p<0.05)in age,long-term residence,education level,employment status,type of medical insurance,family annual income,travel time for medical visits,clinical stage,receipt of immunotherapy,and targeted therapy.After adjusting for confounding factors,binary logistic regression analysis identified education level,employment status,medical insurance,family annual income,travel time to hospital,tumor stage,and targeted therapy as independent factors influencing financial toxicity in HNC patients(P<0.05).(4)The validation group comprised 167 patients.Chi-square tests indicated no significant differences(P>0.05)between the modeling and validation groups.A pathway diagram model was constructed based on the binary logistic regression results,and internal and external validations were performed.The results showed an AUC of 0.846(95%CI:0.808-0.884)and 0.863(95%CI:0.807-0.920)for the internal and external validations,respectively.The H-L test yielded2=5.228(P=0.733)and2=12.45(P=0.132)for the two validations,indicating satisfactory model fit.Decision curves demonstrated the clinical utility of the predictive model.Conclusion:The prevalence of FT was high and influenced by multiple factors.The FIT items are simple and easily understood,and their good reliability and validity have been confirmed,making them suitable for screening financial toxicity in head and neck cancer(HNC)patients in the cultural context of our country.Clinical application results indicate a high incidence of FT among HNC patients,with independent risk factors including education level,employment status,medical insurance,family annual income,travel time to hospital,tumor stage,and targeted therapy.The internal and external validations of the nomogram model demonstrate its applicability to HNC patients,with high predictive discriminative power,accuracy,and clinical utility.This model can aid healthcare professionals in early risk screening for FT in HNC patients.