Abstract
Keywords
Quality-adjusted life-years (QALYs) have been recommended as the main outcome measure in economic evaluations of health care interventions.1,2 The amount of QALYs produced by an intervention is commonly calculated by multiplying patients’ remaining life duration by a weight (often referred to as utility) representing the value of their health-related quality of life. One of the most common methods for deriving such values is the time tradeoff method (TTO). 3 In the TTO exercise, the respondents are commonly asked to imagine living in some nonoptimal health state (Q) for a specific number of years (t). They are then presented with an alternative scenario in which they would live in full health but for a shorter period of time (x). By varying x and asking the respondent to choose between the two scenarios, it is possible to determine the value of x for which the respondent is indifferent between the two alternatives. Because in QALY calculations perfect health is commonly normalized to 1, the relative value of Q is found by dividing x by t (abstracting from discounting). The TTO method can be used to value hypothetical health states as well as actual health states; in the latter situation, respondents are asked to imagine that they will live in their own current health condition for t years.
Once values are derived for different health states, these can be used in QALY calculations. These are commonly based on the linear QALY model, multiplying the value of the health state by the duration of that health state. An important assumption underlying the QALY model is that of constant proportional tradeoff (CPTO). This means that if an individual is willing to trade off years of his or her remaining life for an improvement in health status, he or she should be willing to trade off an equal proportion of life-years independent of how long the remaining life is. To give an example, this means that if a respondent is willing to trade off 2 out of 10 years in some imperfect health state (e.g., having diabetes) in order to regain full health, that respondent should also be willing to trade off 4 out of 20 years or 2 out of 10 weeks. Although some studies have found support for the assumption of CPTO,4,5 other studies report violations of this assumption.6–11 Violations of CPTO have been explained in several ways, including failure to account for discounting of future life-years,7–9,12 loss aversion acting on life-years but not health,6,12,13 and maximum endurable time.11,14 If any of these biases are present and not corrected for, the choice of time period in the TTO exercise may affect elicited values as longer time frames will lead to lower TTO values.
A fourth, less frequently studied, potential problem, which can result in violation of the CPTO assumption, is whether the time frame presented in the TTO exercise is perceived as a gain or a loss relative to the respondents’ own expectations regarding remaining life duration. To illustrate, a commonly used time frame in TTO exercises is 10 years, independent of the age of the respondents. People are instructed to choose between living 10 years in the nonoptimal health state, after which they die, and (even) fewer years in perfect health, after which they die. The use of a 10-year time frame decreases the problem of diminishing marginal utility and discounting. However, it may increase the reluctance of trading off life-years, since the majority of respondents will experience a 10-year time frame as short compared with their expected remaining life duration and will feel as though they already have lost some years when they start the exercise. Respondents with different expectations on how long they will live may thus value the 10 years differently. Various studies have used longer time frames as the point of departure, such as the actuarial life expectancy of the respondents. 12 An alternative approach that has been used in a few studies,10,13–17 is to use a time frame based on the respondents’ own subjective life expectancy (SLE). The argument for this approach is that the respondents’ SLE is in line with the respondent’s expected endowment and so when used in a TTO exercise should result in a more meaningful tradeoff. This is because the respondents will not start the exercise feeling as though they have received “bonus years” relative to their own expectations or conversely have been “cheated” out of some years relative to these expectations.
To our knowledge, only two studies have empirically tested whether respondents’ SLE affects TTO valuations. In 2004, van Nooten and Brouwer 18 performed an exploratory study in a general population sample using a TTO exercise taking 80 years as the age of death for all respondents. They found that respondents expecting to live longer than the 80 years indicated in the TTO exercise were less willing to trade off years than respondents expecting to live fewer than 80 years. To investigate whether the influence of SLE was also noticeable in TTO exercises using a more conventional 10-year time frame, an additional study was performed in 2009. 19 This study showed that the SLE of the respondents still influenced TTO responses. In addition, the higher the SLE, the less likely respondents were to trade off years of life.
It is important to note that in both of these studies, the general public was asked to value hypothetical health states. To date, it has not yet been investigated whether SLE also influences the TTO values of patients valuing their own health state. This is of interest, given the debate on whose values are most appropriate to use in the context of economic evaluations: patient values or those of the general public.20–23 The lack of consensus in this matter is reflected in practice. Many national health economic guidelines do not give clear direction on whether to use patient or general public values, while some guidelines (e.g., in the Netherlands 24 and the United Kingdom 1 ) prescribe the use of general public values, and others (e.g., in Sweden 25 ) prescribe the use of patient values. In addition, it could be hypothesized that the effect of SLE will be more pronounced in patient valuations; patients are asked to value their own health and therefore may find it more difficult to detach from their own expectations regarding length of life. A similar effect has been suggested for status quo bias, a bias caused by people valuing goods more highly once they own them. This bias has been seen to be stronger in patient valuations than in valuations of hypothetical health states. 26 The study presented in this article aimed to explore whether and to what extent patients’ SLE affects their TTO valuations. In line with previous studies,18,19 it was expected that patients with longer SLE than t would be less willing to trade off years than patients with shorter SLE than t. In addition, it was expected that the effect on patient valuations would be larger than the effect that previously has been seen in valuations of hypothetical health states among the general public.
Methods
Hypothesis
Based on the previously shown effect of SLE on TTO values elicited from representatives of the general public, our study sought to determine whether patients’ SLE significantly affects patient valuations with the TTO method. It was expected that patients with longer SLE than t would be less willing to trade off years than patients with shorter SLE than t.
Study Population
To test these hypotheses, an empirical study was performed among Swedish patients with diabetes. The study population was a subsample of a study measuring health-related quality of life (HRQoL) among patients with diabetic retinopathy. Most TTO studies among patients with diabetic retinopathy (in contrast with many studies in other disease areas) have used SLE as the time frame of the TTO exercise. The relevant questions for the current study were included as a supplement after the HRQoL study had already been initiated. In total, 87% (n = 145) of the respondents in the HRQoL study were included in this study. Potential study participants were identified during their screening appointment at the eye clinic at Linköping University Hospital in Östergötland County, Sweden, or through the registers of the two low-vision rehabilitation centers in the same region. If patients gave their informed consent, they were contacted to schedule an appointment for a telephone interview. The recruitment process and classification of patients has previously been described in more detail. 27 To be eligible for inclusion in the study, participants had to be diagnosed with diabetes at least 1 year ago and be at least 18 years old. The following patients were excluded: those on dietary treatment only, those with difficulty understanding instructions (e.g., due to Alzheimer’s disease, dementia), those with severe concomitant disease that could affect quality of life (as judged by the recruiting physicians), and those insufficiently fluent in Swedish to complete the interview. The telephone interview included four HRQoL measures (the Health Utilities Index Mark 3 [HUI-3], 28 TTO, EQ-5D, 29 and the National Eye Institute Visual Functioning Questionnaire 25 30 ) and took approximately 45 min. The TTO exercise was introduced after a general introduction of the interview and the HUI-3 questions. Every patient was interviewed by the same trained researcher (E.H.). Information concerning potential clinical confounders was collected through the patients’ clinical records or in the few cases when this information was lacking in the patient records by asking the patients (HbA1c value for one patient and diabetes duration for another).
TTO Question
In the TTO exercise, respondents were first asked to imagine that they were going to live in their current health state for a specific integer number of years, t, followed by death. The time t was based on the statistically expected remaining survival for the sex and age group of the respondent, 31 rounded to the closest 10 years (e.g., if the actuarial expected remaining survival was 38 years, t equaled 40 years). Subsequently, the respondents were asked to imagine that there was a treatment that could make them regain full health but at the same time would reduce their remaining length of life to x (x < t) number of years. The preferences of the respondents were then elicited by asking them if they preferred to live in their own health state for t more years or in full health for x more years. Time x was varied, starting at 75% to 80% or 20% to 25% of t for every second patient and then ping-ponging the number of years for x with minimum steps of half a year until indifference was reached between the two alternatives. After having completed the TTO exercise, the patients’ SLE was elicited by asking them the number of additional years they themselves expected to live.
Data Analysis
The TTO value was calculated conventionally by dividing the time at full health, x, by the time at the patient’s health state, t (x/t). The effect of SLE on the TTO value was tested in simple and multiple regression models, with β representing the unstandardized regression coefficients. In all models, the TTO value served as the dependent variable. In the simple regression model, the difference between t and the patient’s SLE was the only independent variable. In the multiple regression models, SLE and t were included as two separate independent variables, but the models also included potential confounders. SLE was defined as the patient’s remaining SLE.
The base model for the multiple regressions was constructed in two steps. First, the variables that on a priori expectations were most likely to influence TTO values were included. Second, additional variables were included if these improved the model in terms of Akaike’s information criterion 32 (AIC; the model with the smallest AIC value is preferred over those with larger AIC values. For negative AIC values, this means that the model with the most negative value is preferred.) The variables included in the first step were the rounding of the actuarial life expectancy to tens when determining t (a round up was represented by 1 and a round down by 0), whether the respondent was interviewed using a high (represented by 1) or low (represented by 0) starting point for x in the choice procedure, marital status, level of education, and the patient’s self-reported EuroQoL visual analog scale (EQ-VAS) score. Dummies for marital status (with 1 representing married or cohabiting and 0 representing single or widowed) as well as level of education (with 1 representing vocational training after upper secondary school or university studies and 0 representing primary school or upper secondary school) were included, because socioeconomic status has previously been shown to influence TTO values.19,33,34 The variables considered for inclusion in the second step were objective indicators of the patient’s diabetes status (all variables in Table 1). These were considered to test whether these measures would capture additional effects of health status not captured by the subjective EQ-VAS scores. Diabetes duration, HbA1c (Mono S), type of diabetes, severe visual impairment, nephropathy, and neuropathy improved the model in terms of AIC and were included in the model. Moreover, to examine the effect of these clinical variables on SLE, their influence was investigated in a multiple regression model with SLE as the dependent variable.
Background Variables (Total Sample = 145)
Note: EQ-VAS = EuroQol Visual Analogue Scale; SLE = subjective life expectancy; TTO = time tradeoff. Retinopathy was classified according to the worse eye. VI was classified according to the better eye.
To provide more insight into the association between SLE and t, additional analyses were conducted. To examine whether the effect of SLE on the TTO values is similar independent of whether t is shorter or longer than the patients’ SLE, patients were divided into two subsamples: one for patients with a shorter SLE than t and one for patients with a longer SLE than t. The base model was run on both of these subsamples. In addition, an alternative model was constructed replacing t with variables for sex and age. Although age and sex have previously been shown to affect TTO valuations, 32 they were not included in our base model, as their effects were assumed to have already been captured by the presented time frame t, which was based on the actuarial life expectancy of each patient’s sex and age group.
The regression models were tested for nonnormal and heteroscedastic residuals by plotting the residuals toward the fitted values and by applying the Breusch-Pagan 35 and the Ramsey RESET test. 36 There were no signs of nonnormality, but the Breusch-Pagan test and the residual plots indicated that the models did suffer from heteroscedasticity. This was corrected for with White’s robust standard errors (SEr). 37
Results
A total of 145 diabetes patients were included in this study. The mean age of the patients in the sample was 57 years (SD = 14), and on average the patients expected to live another 23 years (SD = 12; Table 1). This average remaining SLE was lower than the average time frame (t) of 27 years (SD 13) presented in the TTO exercises. In other words, on average, the patients expected to live for a shorter time than their rounded Swedish age- and sex-dependent actuarial life expectancy would suggest. More than half of the respondents (57.2%; n = 83) expected to live shorter than t, whereas only 15.9% (n = 23) expected to live longer (the distribution of the patients’ SLE, their actuarial life expectancy, and t is described in Figure 1). For the remaining respondents (26.9%; n = 39), SLE was equal to the t used in the exercise. Patients expecting to live fewer years than t reported lower scores on the EQ-VAS scale than did patients expecting to live equal or more years than t.
Scatter plot showing the relationship between the patients’ subjective remaining life expectancy (SLE) and their actuarial remaining life expectancy. The actuarial life expectancy was rounded to the closest tens when determining the time frame in the TTO exercise (t). Ten years was the lowest t that was used. The areas between the dashed vertical lines indicate values of actuarial life expectancy that were rounded to the same tens when determining t.
The average TTO value was 0.79 (SD = 0.22), and the average EQ-VAS score was 74 (SD = 17) of 100. Almost a quarter of the patients included in our study (23.5%) were not willing to trade off any years in order to regain full health. These patients were significantly older (t test, P = 0.006) but had shorter diabetes duration (P = 0.001) and lower HbA1c value (P = 0.03) than patients who were willing to trade off life-years. In addition, a higher proportion of the nontraders were free of neuropathy (Fisher’s exact test, P = 0.030). Although this may indicate that the nontraders had a better health than the traders did, their average EQ-VAS score was not significantly higher compared with the other patients (t test, P = 0.227). Five of the traders experienced some difficulties with responding to the TTO exercise, and one experienced major difficulties. Despite their difficulties, these six respondents completed the exercise and gave a usable TTO value. As can be seen in Table 2, SLE was significantly positively related to EQ-VAS scores and negatively with diabetes duration.
Regression Analysis Investigating the Effect of Disease Severity on Patients’ SLE (Dependent Variable)
Note: EQ-VAS = EuroQol Visual Analogue Scale; SEr = White’s robust standard errors; SLE = subjective life expectancy.
High educational level is defined as university studies or vocational training.
The difference between t and SLE had a significant effect on the TTO value in our simple and multiple regression models. In the simple model, the difference between the presented time frame t in the TTO exercise and SLE had a significant negative effect on the TTO value (β = −0.016, SEr = 0.003, P < 0.001; Table 3; Figure 2). In other words, if t was longer than the patient’s remaining SLE, they were willing to trade off more years than if t was shorter than SLE. When SLE and t were treated as separate variables (Table 3), remaining SLE had a positive effect (β SLE = 0.017, SEr = 0.003, P < 0.001) on the TTO value, whereas t had a significant negative effect (β t = −0.016, SEr = 0.003, P < 0.001). These effects remained when other variables were included in the model (β SLE = 0.014, SEr = 0.003, P < 0.001; β t = −0.014, SEr = 0.003, P < 0.001; Table 4). As indicated by the almost equal absolute values of the coefficients, SLE and t seem to outweigh each other, leading to a net effect on the TTO value of approximately zero, as long as the remaining SLE and t are equal (which was the case for 26.9% of the respondents). According to the multiple model in Table 3, the effect of t and SLE on the TTO value can be explained by the following equation:
where e represents random variation. This means that when SLE exceeds t, the TTO value increases with every additional succeeding year, and because the β t coefficient is negative, the TTO value decreases with each additional year of difference when t is longer than SLE. Put differently, if the respondents expected to live longer than t, they were inclined to trade off fewer years for every year they expected to live longer than t, and if they expected to live fewer years than t, they had a tendency to trade off more years for every bonus year they were offered. This effect was stable across all the models we considered and did not change when excluding the patients with difficulties with responding the TTO exercise. These findings confirm our hypothesis: Patients’ SLE significantly affects their TTO valuations.
Regression Analysis Investigating the Relation between the TTO Value (Dependent Variable) and the Patients’ SLE
Note: SEr = White’s robust standard errors; SLE = subjective life expectancy; t = time frame in the TTO exercise; TTO = time tradeoff.
Scatter plot showing the relationship between the difference between the time frame used in the time tradeoff (TTO) exercise (t) and the patients’ subjective remaining life expectancy (SLE) and the TTO value. The horizontal line at 0 on the y-axis indicates where t equals SLE.
Multiple Regression Analysis Investigating the Effect of SLE on the TTO Value (Dependent Variable)
Note: AIC = Akaike’s information criterion; BIC = Bayesian information criterion; EQ-VAS = EuroQol Visual Analogue Scale; LE = life expectancy; SEr = White’s robust standard errors; SLE = subjective life expectancy; t = time frame in the TTO exercise; TTO = time tradeoff. The model with the smallest AIC or BIC value is preferred over those with larger AIC values. For negative AIC, values this means that the model with the most negative value is preferred.
High educational level was defined as university studies or vocational training after upper secondary school.
When we reexamined the effects of SLE on TTO values after dividing the patients into two different subsamples based on whether their SLE was shorter or longer than t, the results from both subsamples were similar to those from our base model. In the subsample of patients expecting to live longer than t, each year of difference positively affected the TTO value, whereas in the group expecting to live shorter than t, each year of difference negatively affected the TTO value (Table 5). However, the impact of SLE seemed to be smaller in the subgroup where SLE exceeded t (β SLE = 0.010, SEr = 0.004; β t = −0.012, SEr = 0.005) than in the group where t exceeded SLE (β SLE = 0.024, SEr = 0.007; β t = −0.021, SE = 0.006).
Subgroup Analysis of the Effect of Patients’ SLE on the TTO Value (Dependent Variable) with Respondents Classified according to Whether Their SLE Was Longer or Shorter than the Time Frame of the TTO Exercise (t)
Note: EQ-VAS = EuroQol Visual Analogue Scale; LE = life expectancy; SEr = White’s robust standard errors; SLE = subjective life expectancy; t = time frame in the TTO exercise; TTO = time tradeoff.
None of the patients expecting to live longer than the time frame of the TTO question had severe vision impairment.
High educational level was defined as university studies or vocational training after upper secondary school.
As described, we included several potential confounders in the multiple base model. As it turned out, a high or low starting point in the choice procedure of the TTO exercise did not significantly affect the TTO values. In our base model, there was also no effect seen from the rounding of the actuarial life expectancy to the nearest 10 y when determining the individual time frame to present in the TTO exercise (Table 4). However, the rounding did have a negative impact on the TTO value when t was replaced with the variables for age and sex (Table 4). This means that the TTO value was lower if t had been rounded upward than if it had been rounded downward. A likely explanation is that the rounding of the actuarial life expectancy affects the difference between SLE and t. Thus, when replacing t with age and sex, the rounding captures part of the effect of this difference.
Discussion
Main Findings
This study was initiated to investigate whether patients’ SLE affects their TTO-based valuations of their own health. Our results indicate that SLE does affect patients’ willingness to trade off years within the TTO exercise. Similar results have been seen in previous studies investigating the effects of SLE on TTO values from the general public concerning hypothetical health states.18,19 Our results furthermore indicate that patients who expect to live shorter than the presented time frame in the TTO exercise were proportionally more willing to trade off years as the difference between t and SLE increased. The reverse association was also seen: Patients who expected to live longer than the presented time frame were proportionally less willing to trade off years for every additional year they expected to live compared with t. These findings are in line with the suggestion by Dolan and others 38 that individuals presented with a TTO time frame exceeding their SLE may willingly give up these “extra” years. This idea was also supported by patients thinking aloud during the telephone interview about how the presented time frame related to their own expectations about length of life, expressing thoughts such as, “Hmmm, how old will I be then . . . ? No, I don’t think I will live that long. I could trade a few of those years.” However, to be able to gain insight in the precise nature of the effect of SLE on TTO values, it would be necessary to conduct in-depth interviews, comparable to the research by Van Osch and Stiggelbout 39 and Robinson and others 40 on the cognitive process underlying VAS and TTO valuations.
Our results show a (much) stronger relationship between SLE and TTO valuations (in terms of higher positive or negative regression coefficients) than the two previous studies among the general public.18,19 In our study, every additional year that SLE exceeded t elevated the TTO value with 0.014. With a 5-year longer SLE than t, this could cause an increase in the TTO value of 0.07 (approximating in 2 fewer years less traded off if t = 30: 2/30 = 0.067). In the previous studies where t was fixed at 80 and 10 years, the regression coefficients demonstrating the effects of every additional year longer SLE were 0.005 with the TTO value as the dependent variable 18 and −0.009 with the years traded off as the dependent variable. 19 This difference may be explained by differences in study design. For instance, in the research by Van Nooten and others, an open-ended TTO question was applied directly asking for the respondents’ indifference point, rather than a choice procedure, ping-ponging toward the indifference point, as applied here. Nevertheless, the difference in magnitude of the effect of SLE on TTO values was in line with our expectations. Respondents valuing their own health may indeed have more difficulties in detaching from their SLE than respondents valuing a hypothetical health state, and hence the impact of SLE was expected to be larger than in the general public.
Limitations
Some limitations should be noted regarding this study and its findings. First, the SLE of the patients was correlated with their disease severity (as defined by the EQ-VAS and diabetes duration); that is, patients with worse health expected to live shorter than healthier patients. Using age- and sex-dependent survival derived from the Swedish population instead of disease-specific survival when determining t, patients in worse health are more likely to have a lower SLE than t. Moreover, patients with worse health are expected to trade off more years than more healthy patients. Consequently, the fact that respondents with an SLE < t are more willing to trade off time could be a result of their health status rather than that their SLE is lower than t. Therefore, the effect of SLE on the TTO value is potentially confounded by differences in health status. To correct for this, we included several health status variables in the multiple regression model, such as EQ-VAS scores and diabetes-related clinical variables. When adding these variables, the effect of SLE decreased somewhat but remained significant. Second, a potential explanation for the stronger willingness to trade off years among patients with a relatively low SLE could be that patients with a relatively low SLE may also expect a lower future HRQoL than patients with a relatively high SLE. Van Nooten and others 18 found a significant positive association between expectations of future HRQoL and TTO values. Even though this effect was seen to be less prominent than the effect of SLE, it could have biased our results in the sense that patients with shorter SLE may have been more willing to trade off years (resulting in lower TTO values) partly because of their low expectations regarding future HRQoL, rather than solely due to their SLE. It would be worthwhile to include questions concerning expectations regarding future quality of life in future studies to be able to correct for potential confounding. Third, respondents were orally introduced to the TTO exercise by telephone without any form of visual support. Given the severity of the visual impairment in some of the respondents, using visual aids would still have been difficult even if the questionnaire had been administered face to face. Still, some patients may have found it difficult to fully comprehend the TTO exercise in the telephone interview, and the cognitive challenge could have led them to focus on their own life expectancy as a reference point rather than on the proportion of time to trade off. However, we stress that the respondents did not give the impression of having difficulty understanding the exercise. Moreover, previous research has indicated that TTO questions administered over the phone yield similar results as face-to-face interviews. 41 Fourth, our results are based on a small sample within one specific patient group. Future research is necessary to be able to determine whether the effect of SLE is equally significant in TTO exercises conducted for the self-assessment of health within larger samples using other administration modes among patients suffering from distinct health problems.
Implications
The results of this study indicate that patients’ SLE affects their willingness to trade off life-years in TTO exercises. This may have some important implications. First, TTO valuations elicited with different time frames may not be comparable, which stresses the need for consensus on which TTO time frame is the most appropriate. Second, applying time frames that differ from respondents’ SLE may lead to biased health state valuations, because the respondents’ relative willingness to trade may depend on the deviation between the time frame presented and their real-life expectations, instead of only their assessment of the health state. If our results are confirmed in other studies, the coherent solution would be to apply respondents’ SLE as the TTO time frame. In that case, TTO exercises should logically start with asking the respondents how long he or she expects to live. Since the effect of SLE on TTO values seems to be more pronounced among patients, this may be especially relevant for patient valuations.
This effect of SLE on TTO valuations suggests that SLE may play a role in violations of the CPTO assumption. If patients use their SLE as a reference point for tradeoffs, this could lead to proportionally different tradeoffs in TTO exercises based on different time frames. To give an example, an individual expecting to live 30 additional years may in a 31-y TTO question trade 6 out of 31 years, resulting in a QALY weight of 0.81 (25/31), but in a 29-y TTO question trade 4 out of 29 years, resulting in a QALY weight of 0.86 (25/29). Consequently, respondents’ SLE may need to be adjusted for when investigating violations of the CPTO assumption due to other factors such as decreasing marginal utility, loss aversion, and maximum endurable time.
It should be noted that we have investigated the effect of SLE on TTO values elicited with a TTO time frame based only on the actuarial life expectancy and not on the widely adopted 10-y time frame. A time frame based on actuarial life expectancy may be more likely to trigger the respondents to think about their SLE compared with a 10-y TTO time frame, because in most cases, the actuarial life expectancy is more closely related to respondents’ real-life expectations. Although Van Nooten and others 19 have shown that the SLE of the general public does affect 10-y TTO values, future research is warranted to determine whether this is also the case for a 10-y TTO time frame used for the self-assessment of patients’ health.
Conclusions
Patients’ SLE seems to affect their responses in TTO exercises with time frames based on actuarial life expectancy. Patients presented with a time frame in a TTO exercise exceeding their SLE may trade off more years than patients presented with a TTO time frame shorter than their SLE. The TTO value seems to increase with every additional year SLE exceeds the TTO time frame and to decrease with every additional year SLE is shorter than the TTO time frame. These findings emphasize the importance of choosing an appropriate time frame in TTO exercises. Applying time frames that are systematically deviating from patients’ SLE may bias values resulting from the TTO exercise. Consequently, if these weights are used in economic evaluations, this could bias cost-effectiveness outcomes. The effect of SLE may be stronger within patient populations than within the general public. Further research is needed to learn whether this is also the case for patient valuations with TTOs using a 10-y time frame.
Footnotes
Acknowledgements
The authors are grateful to Arthur Attema, Werner Brouwer, and Job van Exel for their useful comments on the analysis and article.
Financial support for this underlying data collection was provided by Astra Zeneca AB Sweden. The analysis of the data and the preparation of the article were, however, supported by Östergötland County Council. The funding agreement ensured the authors’ independence in designing the study, interpreting the data, writing, and publishing the report.