Validity and reliability of the EQ-5D-5 L in family caregivers of leukemia patients

Leukemia is one of the common cancers, ranking in top 11 in cancer incidence and top 10 in causes of cancer death [1]. It can happen in adults as well as in children [2]. In 2012, 352,000 patients were diagnosed with leukemia globally, and 265,000 died from leukemia [3]. China reported 75,300 new cases of leukemia and 53,400 deaths of leukemia patients in 2015 alone [4]. The recent medical advancement has improved the prognosis of leukemia significantly. For example, the five-year survival rate of leukemia patients increased from 19.6% in 2003–05 to 25.4% in 2012–15 in China [5], which has resulted in growing prevalence of patients living with leukemia. The Global Burden of Disease 2015 Study (GBD 2015) estimated that the world now has about 2.3 million people living with leukemia [1, 6, 7].

Leukemia is a devastating event not only to the patients but also to their family. Behind each statistic of a new leukemia case is an individual face, accompanied by the faces of family caregivers (FCs) drawn into this singular event. Caring for a patient with leukemia can impose tremendous toll on the physical and emotional health of the family caregivers (FCs) [8, 9]. Previous studies have applied generic instruments such as the 36-Item Short Form Health Survey (SF-36, [10]) and the World Health Organisation Quality of Life-BREF (WHOQOL-BREF) for assessing health-related quality of life (HRQoL) of the FCs. Although these instruments can offer detailed descriptions about HRQoL across a range of domains [11, 12], they are not able to be converted into a single utility index. A single utility index is needed to reflect the overall preference of the public. This is particularly important when one is rated higher in some domains of HRQoL but lower in other domains than other people. Cost-utility analysis has been widely accepted as a useful tool for resource allocation [13, 14]. In recent years, the EQ-5D developed by the EuroQol group has attracted increasing attention for its simplicity and availability of a utility calculation algorithm based on the preference of general public. More than 140 language versions of EQ-5D have been developed [15].

There are two versions of EQ-5D: one asking respondents to rate their experience across three levels (EQ-5D-3 L) and another one across five levels (EQ-5D-5 L). We attempted to use the EQ-5D-5 L in HRQoL assessment on FCs of leukemia patients because it is less likely than the EQ-5D-3 L to show ceiling effect (reaching the maximum possible score) and may be more responsive to monitor small changes, especially in mild conditions [16‐18]. However, there is a lack of psychometric evidence to support the validity and reliability of the EQ-5D-5 L in Chinese populations. This study aimed to test the validity and reliability of the EQ-5D-5 L in FCs of leukemia patients. Specifically, we tested the convergent validity, divergent validity, known- groups validity, and test–retest reliability of the EQ-5D-5 L instrument.

A cross-sectional questionnaire survey was conducted on FCs of leukemia patients. We selected leukemia for this study for several reasons: (1) Leukemia is one of the top 10 causes of cancer death; (2) Leukemia can happen in adults as well as in children, which presents particular challenges to FCs of patients; (3) High quality of care became critical for improving HRQoL when the prognosis of leukemia is improved significantly.

The patients were recruited from three tertiary hospitals located in the capital city (Harbin) of Heilongjiang province in China. The survey was undertaken between July 2015 and February 2016. The hospitals provided a list of leukemia patients admitted over the period of the survey. Trained interviewers approached the FCs of the patients, seeking informed consent from the FCs for participating in the study. The participating FCs had to be a primary caregiver without receiving any payment, be 18 years or older, and be able to communicate with the interviewers.

The questionnaire was administered through face to face interviews in a private room in the hospitals where the patients were treated. The interviewers were recruited from postgraduate research students in Harbin Medical University. Training was provided to the interviewers about how to approach potential participants, how to explain the purpose and procedure of this study, how to obtain informed consent, how to assess the eligibility of participants, and how to fill the questionnaire. A total of 349 primary FCs of leukemia patients were approached and 298 (85%) returned a questionnaire that was valid for data analyses (Fig. 1).

This study has provided empirical evidence to substantiate the reliability and validity of the EQ-5D-5 L. Such a study is important simply because the EQ-5D-5 L has been widely used in over 100 countries across a range of disease conditions and general populations including in China [31]. Although its psychometric properties have only been reported in a few populations [32‐34], to the best of our knowledge, the current study was the first to evaluate formally validity and reliability of EQ-5D-5 L among FCs of leukemia patients.

The study showed that the EQ-5D-5 L is highly reliable in the FCs of leukemia patients. All of the five dimensions as well as the utility index and VAS scores exhibited very good test-retest reliability, with Kappa and ICC values equaling or exceeding 0.8. A single study on psychometric properties of the EQ-5D-3 L in caregivers of autistic children did not test test–retest reliability of the EQ-5D-3 L instrument because of its design deficiencies [23]. Such a degree of agreements in the current study is higher compared with the findings of studies conducted elsewhere in various patients [32, 35]. This may be a result of several reasons: first, the repeated test was conducted 24 h after the initial survey and the interval length is shorter compared to the other studies; second, FCs are less likely to experience changes in HRQoL than patients who are undergoing treatments; third, we maintained the same interviewers for repeated tests and avoided inter-interviewer variations. Previous studies (test by interviewers and retest by telephone or mail) demonstrated that the presence of interviewers and different administration methods can exert some influence on the results of a questionnaire survey [32, 36].

The study confirmed the convergent validity and the divergent validity of the EQ-5D-5 L through a comparison with the WHOQOL-BREF. Both the WHOQOL-BREF and EQ-5D questionnaires are commonly used HRQoL tools for the FCs of patients. One of the advantages of the EQ-5D-5 L over the WHOQOL-BREF is that it can be used to measure utility values which make cost-utility analysis and economic evaluations of healthcare interventions possible. Two of domains (physical and psychological) of the WHOQOL-BREF can find an equivalent match in the EQ-5D-5 L. Although respondents may include a consideration of the social and environment factors in their VAS ratings, no equivalent constructs can be found in the EQ-5D-5 L for the social and environment domains of the WHOQOL-BREF. These similarities and differences enabled us to test the convergent validity and divergent validity of the EQ-5D-5 L. In addition, one of the strengths of this study rests on the characteristics of the study participants. For FCs of leukemia patients, we were able to directly estimate their burden of care and test its association with HRQoL. As expected, we revealed some moderate correlations between the EQ-5D-5 L scores and the ZBI scores. These findings are consistent with a previous study [23]. All of the evidence points to a perfect fitness with the conceptual hypotheses. However, the correlation between mobility of EQ-5D-5 L and ZBI score was moderate. This finding was not consistent with previous study [23] and our expectation, and should be further studied in the future.

In addition, we were also able to test gradient changes in the EQ-5D-5 L utility index and VAS scores of the FCs along with the performance of the leukemia patients who they cared for and the perceived overall health of themselves: both were confirmed. These findings may indicate that the EQ-5D-5 L has a strong discriminatory power between known groups of FCs. Previous studies have produced different conclusions about the connection between HRQoL of FCs and the functional status of the patients they cared for. A study in the US detected differences of the Quality of Well-Being (QWB) in the FCs caring for children with spina bifida with different levels of functioning [37]. But another study also in the US failed to establish a connection between the Health Utility (HUI-II) of the FCs and the severity of Alzheimer’s disease of their care recipients [38]. The authors argued that the generic instruments may not be sensitive enough to capture the potential differences. We found the EQ-5D-5 L is sensitive in terms of distinguishing preference scores among caregivers by the level of functioning of leukemia patients. This finding is similar to a study on caregiver of autism with EQ-5D-3 L [23]. In addition, this study also found that FCs who were reporting poorer health tend to have lower EQ-5D-5 L and EQ-VAS scores. This also further confirmed the EQ-5D known-groups validity found in previous validation study [30].

There are several limitations in this study that should be acknowledged. This study was not conducted in a representative sample of participants. It is also limited to FCs of leukemia patients. Further studies in a variety of populations are needed in the future. The cross-sectional design prevented us from testing causal relationships. To avoid potential changing conditions, we performed repeated tests in a short period of time. We cannot exclude the possibility of influence of the initial survey on the repeated survey.

The Chinese version of EQ-5D-5 L has satisfactory reliability and validity in FCs of leukemia patients. Given that there is no FCs-specific instrument measuring health utility, the EQ-5D-5 L presents a psychometrically sound instrument for researchers to undertake health economic studies, such as cost-utility analyses.