Experience-based health state valuation using the EQ VAS: a register-based study of the EQ-5D-3L among nine patient groups in Sweden

In the present study, patient valuations of their health states using the EQ VAS was explored. Valuations of nine selected EQ-5D-3L health states showed general consistency by severity. Similarly, moderate to strong correlations were found between EQ VAS valuations based on modelled EQ VAS data and the EQ-5D-3L index. A change in the EQ VAS scores of the same health states over time and differences across patient groups was also observed. Models of EQ VAS score regressed on EQ-5D-3L dimensions showed mostly consistent decrements by severity level in each dimension both at baseline and at 1-year follow-up.

Patients’ values, across nine selected EQ-5D-3L health states, had good face validity: poor health states had lower values than mild health problems, suggesting that these methods provide a reasonable means of broadly reflecting how good or bad patients consider health states to be. The consistency of EQ VAS scores of health states with severity was also reported by a study of the general population on data from 15 countries [43]. A number of common health states like in the present study were also reported preoperatively and postoperatively in a study of a knee arthroplasty population in the UK [31].

In the EQ VAS scores of the same health states at baseline and at 1-year follow-up, relatively larger increases were observed in valuations of many of the health states from the intervention-based registers than in the diagnosis-based registers. This could be associated with the mainly surgical interventions employed in most of the intervention-based registers relieving pain and mobility problems which are common among such patients [44]. In relation to this, the results also seem to support the face validity of using EQ VAS as patient valuations. The differences in EQ VAS scores of the same health states across patient groups and over time could be associated with the broader nature of EQ VAS compared to the EQ-5D-3L, hence capturing broader aspects of health than the EQ-5D dimensions. This was also reported by a study in the UK patient-reported outcome measures program [13].

The overall consistency between the EQ VAS and the EQ-5D-3L dimensions was also shown in the significant correlations between EQ VAS score and EQ-5D-3L index which increased from baseline to 1-year follow up in all the patient groups. Comparable findings were reported in similar studies among women with cervical pre-cancer (0.51) in the UK [34], and among patients with Parkinson’s disease (0.68) in Spain [45].

The OLS models of EQ VAS regressed on the EQ-5D-3L dimensions both at baseline and 1-year follow-up showed mostly consistent decrements by severity level in each dimension in the different patient groups and the general population. An overall similar finding was reported in a study which compared health state valuation between non-specific low back pain patients and the general population in the Netherlands [32]. Similarly, overall consistent models were reported in previous studies based on preoperative and postoperative data of patients undertaking hip replacement in Sweden [28] and knee replacement in the UK [31]. In addition, studies in the general population conducted through experience-based [8, 9, 18, 19, 46, 47] and hypothetical [48,49,50] perspectives also showed consistent valuations of EQ-5D health states (3L as well as 5L) through VAS.

In the OLS models, inconsistencies were noted mainly in the self-care dimensions in all the patient groups at baseline and in several patient groups at 1-year follow-up. Inconsistencies in the mobility dimension were also shown among several patient groups at baseline and at 1-year follow-up. Similarly, a study in Sweden exploring valuations among patients undergoing hip replacement the self-care dimension showed inconsistency both in preoperative and postoperative valuations [28]. In another study in the UK among patients in four clinical groups (stroke, low back pain, colposcopic investigation and cytological surveillance), the self-care dimension was found to not be statistically significant in either of the severity levels 2 and 3 [33]. Inconsistencies in the self-care and mobility dimensions were also noted among low back pain patients in the study from the Netherlands [32]. In the present study, the inconsistencies were also noted among in the general population.

One of the possible reasons for the inconsistencies noted in the present study in the self-care and to a certain extent in the mobility dimensions could be the relative importance of the different dimensions depending on whether one is valuing their own health state or a hypothetical health state. This has been discussed in a previous study, based on EQ-5D-3L data from the US, which compared experienced and hypothetical health states where the self-care dimension followed by pain/discomfort were the most important dimensions in the valuation of hypothetical health states. In contrast, usual activities and anxiety/depression dimensions were the most important in experience-based valuations [51]. In relation to this, the study also showed that in the experience-based valuations severity levels 2 and 3 of the mobility and self-care dimensions were closest to each other compared to other dimensions and to hypothetical valuations [51]. This was in line with the findings across the patient groups as well as in the general population in the present study. The findings here show that that the aspects of health important to patients are different from those of the general public who are asked to imagine health problems. This, in turn, could yield different results when measuring effectiveness of interventions depending on whether patients’ own perspectives or imagined health states by the general public are used.

The second related possible explanation for the inconsistencies could be the relatively small number of individuals reporting severity level 3 problems in the self-care and mobility dimensions. In the mobility dimension, the fact that level 3 is presented as ‘confined to bed’ could have contributed to fewer individuals reporting that level. In relation to that, individuals with more severe problems choosing mobility severity level 2 could possibly contribute to lower EQ VAS scores. In a number of studies where inconsistencies in one or both of the dimensions were shown, the relative number of responses falling in severity level 3 were small accounting for 1% or less of the all the problem levels in studies from Sweden and the UK [28, 32,33,34]. Notably, in the study from the UK among women with low-grade cytological abnormalities (pre-cancer), severity levels 2 and 3 in mobility, self-care, and usual activities dimensions were combined due to very few number of individuals reporting problems in these dimensions [34]. Comparatively, EQ-5D-5L has been shown to provide better discriminatory power between severity levels than EQ-5D-3L [52] and lower ceiling effects the potential implications of which, on valuation, have been discussed [52, 53]. In the context of the present study the categorization of responses in the ‘no problem’ or ‘moderate’ levels, which would otherwise be in between in EQ-5D-5L, could lead to under/over estimation of valuations.

In the present study, anxiety/depression showed the highest decrements in most patient groups at baseline and at 1-year follow-up indicating it to be the most important dimension to patients. A similar finding was also shown in the general population data. In a study based on data from different groups—people with varicose veins, chest pain, chronic obstructive pulmonary disease, irritable bowel syndrome, osteoarthritis, low back pain, elderly women and patients in intensive care unit – anxiety/depression was the dimension with the highest decrement, similar to the present study [30]. Similar findings were also shown in several other studies in Sweden [28, 47], the Netherlands [32] and the UK [16, 31, 33, 34], employing patient valuations of their own health.

The dimensions with the highest decrements remained the same from baseline to 1-year follow up in most of the patient groups. However, in patients from the ankle and heart failure registers, a change in the dimension with the highest decrement with time was shown. In the ankle register, pain/discomfort had the highest decrement at baseline and anxiety/depression at 1-year follow-up. Heart failure patients on the other hand, had the highest decrement in the usual activities dimension at baseline and in the anxiety/depression dimension at 1-year follow-up. This could relate to the change in the relative importance of the different dimensions depending on the disease/condition patients have and how they experienced them before and after intervention/treatment. It is also notable that the dimension with the highest decrement at 1-year follow-up in the two patient groups had become similar to that of the general population in the study.

In the anxiety/depression dimension, the highest decrement was recorded among patients with bipolar disorder both at baseline and at 1-year follow-up. This seems to show the ability of the EQ-5D instrument to indicate the importance of specific dimensions to patients in line with their diagnosis/conditions. Comparably large decrements in the anxiety/depression dimension was noted in a study from the UK [33]. The decrements in severity level three of the anxiety/depression dimension were comparable to the general population. This could possibly show the emphasis given to experiencing mental health problems in the general population as well, as considerable level of mental health problems are reported in general population samples in Sweden [54, 55].

Following anxiety/depression dimension, while comparable to usual activities dimensions at baseline, pain/discomfort had larger decrements at severity level three mainly among patients from musculoskeletal registers. On the other hand, patients with heart failure, respiratory failure and bipolar disorder assigned large decrements to severity level three of the usual activities dimension. The importance of pain and usual activities dimensions for the respective patient groups seems to go in line with the overall symptoms and the associated implications of the disease/conditions in terms of pain or limitation of day-to-day activities.

At baseline, in most of the patient groups, 20–30% of the variances were explained by EQ-5D-3L dimensions while in the data of respiratory failure (about 14%) and bipolar patients (about 41%) the lowest and the highest proportions were recorded. All were lower than r squared in the model of the general population data (48%). The explained variance increased in all the patient groups at 1-year follow up ranging between 35 and 60% for most patient groups. It showed about 30% explained variance in the model for the patients from the respiratory failure register. The explained variance still remained lower than in the general population for most patient groups with higher proportions in the patients from spine and hip registers and comparable proportions noted among patients from bipolar register. Although not directly comparable, a number of r squared statistics have been reported in regression models of different patient groups including 32% with cervical pre-cancer [34], 39% in those undergoing knee replacement [31] and 47.1% in the eight patients groups cited above [30].

One of the strengths of the present study is the large sample size of patients which allowed investigation of experience-based valuation of health by patients through the EQ VAS and comparison with a large general population sample. The comparison across many patient groups is also an important strength enabling assessment of how specific diseases and associated experience relate to valuation of health states. Furthermore, the study investigated how patients’ valuation of their health changes from baseline to 1-year follow-up. In addition, the study compared OLS models with multilevel models and covariate adjusted models through sensitivity analyses.

On the other hand, an important limitation to take into consideration is possible differences in the way EQ-5D data were collected across the different registers to which some of the difference in valuation could be attributed. In addition, as the state dead was not anchored in the present study, its immediate use in economic evaluations could be limited. However, studies among a sample of patients to get their valuations of the state dead could remedy this in addition to the current discussion on whether anchoring dead is necessary and other alternatives [56]. VAS/ EQ VAS not providing obvious choice or trade-off in the valuation process and the end aversion bias may have had an implication in the EQ VAS valuation in the context of using it in economic valuation [20, 21]. In relation to this, the level of correlation between EQ VAS score and EQ-5D-3L in the different patient groups, even though moderate to high, some level of discrepancy remains between the two measures.

The present study has important implications including showing the feasibility and importance of timing of patient valuations as dimensions important to patients could depend on the type of disease/condition and its stage (e.g., pre- vs. post-operative). This, together with other clinical measures, could facilitate identification of certain aspects of health that may be available for intervention. The broader coverage of EQ VAS than the EQ-5D-3L dimensions was also demonstrated which could emphasize the importance of EQ VAS as a relatively simple but important measure of patients’ overall health. The present study also showed that patient valuations based on EQ VAS scores, elicited through experience-based perspectives, have a potential to be used in the calculation of quality-adjusted life years (QALYs) in comparing different interventions in decision contexts that take patient perspectives into consideration. Furthermore, the study adds information for a discussion on the reconsideration of the need for severity level three in the self-care dimension and to some extent the mobility dimension considering the inconsistencies found in many patient groups. In addition, the findings showed patient valuations could arguably be more appropriate for use in situations where QALYs do not need to be calculated as well; such as summarizing population health survey data and assessing changes in health following surgical and other clinical interventions. The findings also highlighted the importance of mental health among patients with otherwise mainly physical diseases. This provides important information that the mental health aspect is a crucial component in the care of the patients.

The application of clinimetric approaches in future studies of EQ VAS and EQ-5D, besides the current mainly psychometric ones, in assessing patients’ valuations of their health, could provide useful insights in general and in clinical contexts [57, 58].

留言 (0)

沒有登入
gif