As of date, there are no studies that has surveyed the health state utility values (HSUV) of transfusion-dependent thalassemia (TDT) patients in Malaysia and this study aims to fill that gap. The availability of these values would aid policy makers in the decision-making process related to disease management for these group of patients, especially since TDT is a chronic and expensive disease.
A ceiling effect was seen in the sample with 67.35% of patients reporting perfect health and this may be attributed to the EQ-5D-3L instrument used. Although the EQ-5D-5L instrument has been shown to be superior compared to the EQ-5D-3L23, the EQ-5D-3L instrument was chosen because of studies which support the validity of its use in children24, 25, 33 in the absence of the EQ-5D-Y value sets, since both the pediatric and adult population was sampled. In addition to that, a 3-level response would be simpler to administer compared to a 5-level response.
This study found differences between the domain responses of pediatric and adult patients. The lower number of reported problems on the self-care domain of older patients may be explained by the independence and knowledge that the patient gains over the years when coping with the condition13, 36. Pediatric patients also tend to be more dependent on their caregivers for daily tasks, hence explaining the higher number of reported problems on the self-care domain. As patients grew older, the burden of treatment may increase with higher volume of blood required, onset of complications and the need for higher dosages of iron chelating therapy, possibly contributing to the higher number of reported problems on the mobility and anxiety/depression domain, although the difference was only significant in the mobility domain. Respondents who self-reported reported higher number of problems in the mobility and anxiety/depression domain. This is expected as they would be able to perceive better a limitation in their daily routine. Respondents who proxy-reported reported a higher number of problems on the self-care domain, as caregivers are the one who would be helping patients in their self-care routine, and would be less able to perceive the other domains which may not be observed physically such as pain/discomfort and anxiety/depression.
The mean (SD) HSUV of 0.893 (0.167) in this study was slightly higher than the HSUVs surveyed among TDT patients using iron chelating therapy in Iran. The study, which utilized US and Iran’s time trade-off value set, had a HSUV range of 0.81 to 0.8634. A similar trend was also observed on the EQ VAS scale, where our patients had a mean (SD) score of 81.22 (19.92) compared to a score of 72.9 (1.1) in Iran34. The utility value range between 0.8 to 0.9 had no reported frequencies as the plausible range on the Malaysian EQ-5D-3L tariff set had a gap between the health state 11111 (utility value = 1.00) and 11112(utility value = 0.756)19.
The distribution of utility values sampled were skewed, with many patients reporting perfect health and this was also evident in previous studies28, 30. The skewed distribution violates ordinary least squares (OLS) regression assumptions. Hence, a two-part model (TPM) was used to derive the HSUVs, controlling for age, the type of iron overload and type of iron chelator. In the two-part model, the disutility score (i.e. 1-EQ-5D-3L utility value)28, 31 was used to predict the HRQoL scores based on the variables. The two-part model is useful for models with mixed discrete continuous outcomes. In this study, the first part of the two-part model predicts the probability of obtaining a disutility score of 0(perfect health), followed by the second part of the model that predicts the disutility score using a generalized linear model(GLM)32. The GLM model was chosen as it allows the outcome variable to be a link function of the linear index of the covariates instead of the outcome variable simply being a linear function of the covariates. This also avoids the problem of retransformation inherent in models that transformed the outcome variable to meet OLS assumptions35.
Based on the full and restricted sample, the health state utility values using a subcutaneous iron chelator resulted in a utility decrement of 2.7% and 17.9% compared to when an oral iron chelator is used. The study conducted in Iran showed a decrement of 6.9% when using a subcutaneous iron chelator or combination therapy compared to an oral chelator or a monotherapy34. Although these findings of decrement are consistent with previous studies, the utility decrement in this study and the study conducted in Iran is much smaller compared to a time trade-off study conducted in Australia37 and the United Kingdom38 that investigated the utility associated with the use of oral and subcutaneous iron chelating therapy. In both studies, a decrement of 28.2% and 21.4% respectively occurred when using a subcutaneous ICT compared to the oral ICT. However, it should be noted that these two studies utilized a direct elicitation valuation method with respondents from the community, while the current study and the study in Iran utilized a generic preference-based instrument with TDT patient. This highlights the effect of different valuation methods and respondents in eliciting HSUVs.
Amongst the various iron overload complications, cardiac and diabetes complications result in the highest disutility of 8.2% and 8.4% (full sample) or 6.5% and 7.3%(restricted sample) respectively compared to respondents without complications. These values were higher compared to the study conducted in Iran which had a disutility of 3.6% (cardiac) and 6.0% (diabetes) when compared to patients who did not have the complication34. A cost-utility analysis conducted in the United States used the assumption that TDT patients with cardiac disease would have a 15% decrement in utility40 based on TTO values for heart failure as reported in a longitudinal cohort study of health status and HRQoL41. These findings imply that the prevention of iron overload complications is crucial in preserving the HRQoL of patients. Based on the marginal effects, as with other chronic conditions39, this study showed that the presence of coexisting comorbidities in TDT is associated with a lower HRQoL. The decrease in HRQoL becomes statistically significant when the number of complications exceeds two. Compliance to iron chelation therapy is essential in the prevention of the development of iron overload complications. Apart from the increased risk of morbidity and mortality, poor compliance to iron chelating therapy has also shown to increase the cost of treating the disease42. The choice of iron chelators has been highlighted as a determinant of HRQoL in TDT patients43-45 and the findings of this study further emphasizes the preference for oral iron chelation.
This study has a few limitations. First, it was not compared to a healthy general population, hence making it difficult to truly estimate the impact of the disease on the HRQoL. In addition, due to time constraints and the variation in documentation of medical records across centres, it was a challenge to obtain additional clinical parameters which could have been used to assess the severity of the disease such as the serum ferritin levels. As is common for quality of life data, the results were skewed in a way that higher scores were reported more than lower scores. A ceiling effect was also seen in our results (67.35% reported perfect health) indicating that a complete variation in health states were not fully captured. The use of EQ-5D-3L in children also presents methodological concerns such as the relevance of the domains, appropriateness of the instrument based on the cognitive ability and the concordance between self and proxy reports. Agreement between the self-report and proxy-report was also not examined as respondents were only surveyed once. Unfortunately, the absence of other suitable utility tariffs in the country, the evidence that support the validity of EQ-5D-3L use in children24, 25, 33 and the simple administration of the EQ-5D-3L instrument made it attractive for use. Finally, based on a literature search, the minimum clinically importance difference (MCID) which represents the smallest amount of benefit that the patient can recognize and value has not been defined for TDT. This limits our ability to compare the actual impact of using different treatment regimens and route of administrations on the HRQoL and to assess if patients have significantly improved, declined or remained stable46. Further empirical work is required in this area.