Utility Values in Colorectal Cancer Patients Treated with Chemotherapy

Abstract Calculating utility values in colorectal cancer (CRC) patients under chemotherapy treatment is important for studies of economic evaluations. The EQ-5D-5L and composite time trade-off (cTTO) were used to calculate utility values in 105 patients with CRC in Iran. The mean EQ-5D-5L index and cTTO values were 0.45 ± 0.03 and 0.51 ± 0.02, respectively. Anxiety and pain were the most common problems reported by the patients. The BetaMix showed that lower mean utility values were significantly associated with females, aging, a low level of income, a greater number of comorbidities, and an advanced stage of cancer.


Introduction
Colorectal cancer (CRC) is one of the most commonly diagnosed cancers in both sexes (1). According to the latest report by International Agency for Research on Cancer in Iran, the most common cancer types diagnosed in males were the stomach, prostate, colorectal and in females were breast, colorectal, and stomach. In 2020, CRC with a prevalence rate of 9.1% in Iran and 10.0% in the world is the third populated cancer for both sexes (2).
The major part of the high prevalence of this cancer is attributed to the treatments, such as surgery, chemotherapy, radiotherapy, and hormonal therapy (3). Moreover, treatment costs imposed an economic burden on patients and countries' health systems. For example, a study in Iran showed that direct medical costs accounted for 76.5% of direct costs in 2012 (4), and another study in South Korea found that over 80% of direct costs were related to direct medical costs in 2010 (5). High rates of colorectal cancer and its costly interventions have necessitated using costutility analysis (CUA) to assess the interventions. To analyze the cost-utility of interventions, it is necessary to calculate utility values for patients with CRC.
EuroQoL five-dimension (EQ-5D), which is the most popular indirect method in measuring utility, is widely used to estimate utility values. This instrument is usually employed in two versions: The EQ-5D 3levels (EQ-5D-3L) (6) and the EQ-5D 5level (EQ-5D-5L) (7). All studies which estimated the utility values for CRC patients had used the EQ-5D-3L version (8)(9)(10), except two studies that used the EQ-5D-5L (11,12). The empirical evidence showed that EQ-5D-5L is better than EQ-5D-3L in terms of ceiling effects, sensitivity to changes, and reliability and construct validity (13). The ceiling effect is calculated as the proportion of subjects who report no problems on the EQ-5D-5L index (i.e., state 11111). Nevertheless, the weights for both versions of EQ-5D were derived from a representative sample from the general public who have not actually experienced the health states to reflect patients' own preferences. To use patients' preferences directly in measuring utility, the direct methods including the rating scale, the standard gamble, and the time trade-off (TTO) can be used. The use of patients' preferences directly in measuring utilities makes more realistic economic evaluations of health interventions. The TTO is the most commonly used direct method for estimating utility values in health economic studies. Nevertheless, TTO is criticized for some limitations, such as different valuation procedures for states worse and better than dead. To deal with such problems, the composite TTO (cTTO), which is an improved version of the TTO, is developed (14).
Adjuvant chemotherapy is one of the most common treatments for CRC that has high costs and significant side effects on patients (15). Thus, the assessment of the treatment outcomes is needed to provide further evidence about it. CUA is one of the most commonly used forms of economic evaluation for assessing health interventions. To perform CUA of the treatment, it is necessary to estimate utility values for CRC patients under chemotherapy treatment. Aim of the current study, therefore, was to estimate utility values from patients' perspective (use cTTO) and the general public' perspective (use EQ-5D) in this group of patients and identify factors associated with them.

Study design and data collection
A number of 108 inpatients and outpatients with CRC were selected using a consecutive sampling method from the chemotherapy ward of the Oncology Center of Omid, Isfahan Province, Iran over a 4-month period in 2020. Isfahan with a population of more than 5 million is one of the populated provinces in Iran. The patients were referred to the center from all over the country.
The data were collected using the EQ-5D-5L instrument and cTTO questions through the face-to-face interview in the patients' rooms during a single visit, and clinical data were extracted from the medical records of the patients. The ability to speak, the pathological confirmation of the diagnosis of cancer, and the completion of written informed consent were the inclusion criteria for the study. This study was approved by the SSU Ethics Committee, with the following identification: IR.SSU.SPH.REC.1399.016.

EQ-5D-5L instrument
The EQ-5D-5L consists of a classification system of five dimensions (mobility, self-care, usual activities, pain/discomfort, and anxiety/depression) a visual analog scale (VAS). Each EQ-5D-5L dimension has five levels of response options: no problems, slight problems, moderate problems, severe problems, and extreme problems. VAS is a line that respondents could determine their overall health status from 0 (the worst imaginable health) to 100 (the best imaginable health) (7). To use the EQ-5D-5L, we employed the Iranian interim EQ-5D-5L value set because there is currently no Iranian EQ-5D-5L value set (16). Persian version of the EQ-5D-5L was confirmed by the EuroQol group (17).

Composite TTO
We used, in addition to the EQ-5D5L, the accepted composite TTO (cTTO) approach to the valuation of utility. The approach includes the TTO for valuing the health states better than dead (BTD) and lead-time TTO for valuing the health states worse than dead (WTD). To do so, first, each patient was asked to choose between 10 years of full health followed by death (Life A) and 10 years of patient current health status(x) followed by death (Life B). Then, 10 years in full health (t) varied until the patient is indifferent between choices A and B. The value of the status is given by x/t (Figure 1(a)). If the patient chose zero years in full health in comparison with 10 years in her/his current health status, we asked the patient to choose between 10 years of leadtime in full health (L) followed by 10 years of full health (t) (Life A) and 10 years of lead-time in full health followed by 10 years in her/his current health status (x) (Life B). Next, option "Life A" varied until the patient is indifferent between choices A and B. The value of the status is given by (x À L)/t (x t þ L) (Figure 1(b)).

Statistical analysis
Since skewness and kurtosis, multimodality, truncations, ceiling effects, and top and bottom bounded are features of the distribution of EQ-5D utility scores, the use of basic models, such as OLS is not recommended. Therefore, we used the beta mixture (BetaMix) model which was recently developed to deal with the distributional characteristics of the EQ-5D instrument. The data distribution was not normal (Kolmogorov-Smirnov test, p > 0.05). According to our sample size was large enough (n ¼ 105) we used the t-test and ANOVA in assessing differences. STATA version 15.0 for Windows was used for all analyses. Significant level in all tests were considered EQUAL TO5 %.

Results
Of 108 patients who completed the EQ-5D-5L questionnaire, three patients (0.03%) did not complete the interview process of the C-TTO. The socio-demographic characteristics of 105 patients have been presented in detail in Table 1. The mean age of the patients was 59, and the majority of them were in the age group Table 1. Mean values of EQ-5D-5L index and cTTO according to socio-demographic characteristics and univariate analyses.  45-59 years. The majority of the patients were male and secondary educated, and employed/selfemployed. All of the patients had basic insurance, while only 50.48% had complimentary insurance. Patients with income between 13,000,000 and 29,999,999 RLL at the time of the interview were the largest in the sample.
The clinical characteristics of patients are shown in Table 2. More than half of patients diagnosed with regional stage. The average duration of cancer was 2.4 years, and the majority of them have a cancer duration of <6 months. More than half of the patients reported having no comorbidity (62.86%) and only 25.71% of patients received both surgery and chemotherapy.
Distribution of EQ-5D-5L dimensions Figure 2 shows patients-reported problems on each of the five EQ-5D-5L dimensions. The response "unable or extreme" was reported by the patients on all dimensions. This response on dimensions of mobility, self-care, usual activities, pain/discomfort, and anxiety/depression was 1.9, 3.81, 8.57, 7.62, and 9.52%, respectively.

Univariate analyses
The mean EQ-5D-5L index and cTTO values were 0.45 ± 0.03 and 0.51 ± 0.02, respectively. The univariate analyses showed that the difference between the mean EQ-5D-5L index values were significant for age, education status, income level, stage of cancer, and comorbidity number (Tables 1 and 2). Also, the results showed that the difference between the mean cTTO values was significant for age, income level, employment status, stage of cancer, and comorbidity number (Tables 1 and 2).

Regression analyses
Results of the BetaMix model revealed that gender, income level, comorbidity number, and stage of cancer were predictive of EQ-5D-5L index values; and the variables of gender, age, income level, and stage of cancer were predictive of cTTO values (p-value < 0.05) ( Table 3). Among the variables, the highest income level had the most marginal effect on EQ-5D-5L (0.919) and cTTO (1.01).

Discussion
The present study presents utility values using the EQ-5D-5L index and cTTO values in patients with CRC under chemotherapy treatment and impacts of socio-demographic and clinical factors on utility values. However, the high incidence of CRC is a significant health burden and public health concern in developing countries, such as Iran. The utility is a multidimensional concept that can convey important information including physical, emotional, social, and cognitive domains for assessing the overall burden of cancer and the effectiveness of interventions. In addition, the findings of utility values can be used in calculating quality-adjusted life years which is a key determinant of cost-effectiveness analyses.
The mean values of the EQ-5D-5L index were lower than those of two studies that have ever assessed the utility values using EQ-5D-5L in patients with CRC in Iran and China (11,12). Our study found that patients had significantly lower EQ-5D-5L values than those of the study by Yousefi et al, on patients with CRC in Iran (0.45 vs. 0.71). The lower values of EQ-5D-5L obtained in this study compared to the results of the study by Yousefi et al. can be explained by differences in the socio-demographic profile of patients. Compared to the study by Yousefi et al., this study assessed utility in patients who were more female and older. Studies showed that utility scores in patients with CRC were lower in women than in men (18)(19)(20). The other reason for the lower mean utility values in this study in comparison with the study by Yousefi was the higher mean age of patients (59 vs. 55). Aging often has a negative relationship with utility scores in patients with CRC in different contexts from countries (11,21,22). The reason for this relationship can be due to a higher prevalence of multiple comorbidities and physical functioning limitations in older patients (23). Our results showed that 67% of older patients had !2 comorbidities (data not shown). The mean EQ-5D-5L index values of 0.45 were also lower when compared to those of 0.61 in the study carried out in China (24). A significant part of this difference can be explained by using the different tariffs of EQ-5D-5L in the two studies. The Chinese EQ-5D-5L tariff was derived from preferences of the Chines general public that is different from the Iranian general public in terms of culture, demographic backgrounds, etc. This difference is also revealed when comparing the results with those of other developed countries including Japan (0.842) (8) and the UK (0.79) (9).
The findings of this study support that the mean utility scores were lower than those in the developed countries. It may be partly explained by the fact that in the countries with better economic countries, patients are more supported, especially economically and socially. The majority of the patients reported having financial problems (0.90) and limitations of social support (69%). Previous studies evaluating utility values of patients with colorectal cancer found a direct association between income level and health utility scores (11,25,26). Another reason for the differences in mean utility scores between countries may be related to social supports. Sapp et al, demonstrated that social support leads to better scores of utility in patients with colon, especially in terms of the mental health domain (27). In addition, Sultan et al. found that patients with CRC who were socially supported having $6-7 points higher in the domain of mental health than those who have no social support available (19). The use of the different versions of the EQ-5D can explain a significant part of the difference between the mean utility values among the countries. We used the EQ-5D-5L, while the other two studies used the EQ-5D-3L for calculating utility. Empirical evidence shows that the EQ-5D-3L generated higher values than EQ-5D-5L because ceiling effects in EQ-5D-3L are more common than those in EQ-5D-5L.
The mean cTTO values of 0.51 were higher than those of 0.41 in the only study that assessed utility using cTTO in patients with CRC (15). The main part of this difference can be attributed to the different years were considered for the current health state in cTTO task. The number of years of the current health state was considered 10 years in cTTO task for all stages of cancer in the present study, while that was 4, 3, 2, and 1 year for stage 1, 2, 3, and 4, respectively, in the other study. Less number of years of current health state might lead to more patients trading time for an increase in quality of life because patients believe that this short time will not increase their quality of life.
The problems reported by patients on each of the EQ-5D dimensions showed that the percentage of very severe problems reported (i.e., report "Level 5") on the dimension of anxiety (9.52%) was more than other dimensions. This frequency was consistent with the results obtained from the emotional functioning scale of the EROTC QLQ-C30 in patients with CRC. Patients' performance on the emotional functioning scale, which is theoretically very similar to the anxiety/depression of the EQ-5D, was worse among the scales of the EROTC QLQ-C30 in Germany (22) and Brazil (28). Hence, paying attention to anxiety/depression is an important issue in the management of patients with cancer. In addition, this finding was consistent with that of the EQ-5D-5L in China (11). The distribution of problems also showed that the majority of patients (76%) experienced pain, similar to that obtained from EQ-5D for patients with CRC in China (11), the UK, and the Netherlands (29).
Univariate analyses showed that the difference between the mean utility scores obtained from EQ-5D-5L was significant for education level, income level, stage of cancer, and comorbidity number; and those obtained by cTTO were significant for age, income level, employment status, stage of cancer, and several comorbidities. The older patients had lower utility scores, similar to those reported in patients with CRC in other countries (11,21,22). Older patients often have more comorbidities and advanced stages of cancer and having physical functions limitations, which impact on quality of life. Our results showed that low income was associated with low mean utility scores. Financial constraints usually limit individuals from seeking private services and make less opportunity for better healthcare (18). Most patients reported that the reason for choosing the public hospital was financial limitations. In this regard, financial supports can help patients to have a free choice of the service provider. The mean utility scores showed that a higher level of education was associated with higher utility scores.
Higher education level increases patients' perceptions of their disease and treatment. This result was in line with findings reported in other studies (11,18). The low level of education can often contribute to delay in the recognition of symptoms and search for care in health services. The difference between mean utility scores in the variable of employment status was significant, and mean utility was higher in employed patients. The employed may lead to individuals earn more income, and subsequently, they had more opportunities for better healthcare. We also found that mean utility values in patients with advanced stage of cancer were lower than that of the early stage of cancer. This finding with consistent with those reported in Australia (30), China (11), and some European countries for patients with CRC (25,31). Mean utility scores by the number of comorbidities showed that patients with a greater number of comorbidities had significantly lower utility scores. There is strong evidence of an indirect relationship between the number of comorbidities and QoL in CRC (32).
The regression analysis showed that gender, age, stage, and the number of comorbidities were significant independent predictors for utility scores in this study. Aging and having a larger number of comorbidities and advanced stage of cancer were significantly associated with lower EQ-5D scores. These relationships were observed in other studies conducted on colorectal cancer (11,25,(30)(31)(32).
The present study has some limitations that should be noted. The present study was carried out a cross-sectional survey of patients with CRC. The results from a cross-sectional survey do not allow us to establish a causal relationship between outcome variable (utility values) input variables (socio-demographic and clinical characteristics). A longitudinal study requires to be conducted for understanding how the effects of patients' characteristics on utility values. The participants have recruited the Oncology Center in one province, who were not be perfectly representative of patients; hence, generalization of the results needs to be cautious. Another limitation is the use of a crosswalk value set in a calculating value set for the EQ-5D-5L, which is based on the UK preference weights. However, the crosswalk value set is the only technique to generate EQ-5D-5L value set in countries that do not currently have the EQ-5D-5L value set based on their own context.

Conclusion
The findings of this study provide baseline health utility values for patients with CRC, which can be used by researchers in calculating qualityadjusted life years, an indicator essential for health economic evaluations, including cost-utility analyses. The study also revealed that patients with CRC have poor utility values in both EQ-5D-5L and cTTO. Depression/anxiety and pain were the most common problems reported by patients on the EQ-5D-5L. The low utility values of patients with CRC are associated with aging, low income, more advanced stages of CRC, the presence of comorbidities more than 2. According to patients' reports, financial problems were the underlying cause of anxiety. It seems that is necessary to focus on financial and social supports for patients with CRC.

Declaration of interest
The authors declare that they have no conflict of interest.