Original Research Article Pain Quality of Life as Measured by Utilities

Transcription

1 bs_bs_banner Pain Medicine 2014; 15: Wiley Periodicals, Inc. REHABILITATION SECTION Original Research Article Pain Quality of Life as Measured by Utilities Sarah Wetherington, BS,* Laura Delong, MD, MPH, Seema Kini, MD, MSCR, Emir Veledar, PhD, Michael K. Schaufele, MD, Anne M. McKenzie-Brown, MD, and Suephy C. Chen, MD, MS Departments of Dermatology, Orthopaedics, Anesthesia, HSR&D & Division of Dermatology, Atlanta VA Medical Center, Emory University, Atlanta; *Medical College of Georgia, Georgia Regents University, Augusta, Georgia, USA Reprint requests to: Suephy C. Chen, MD, MS, Department of Dermatology, Emory University, 1525 Clifton Road, Atlanta, Georgia 30329, USA. Tel: (404) ; Fax: (404) ; Disclosure: The authors have no conflicts of interest to report. This study was conducted at Emory University. Abstract Objective. Utilities are values of health-related quality of life (HRQoL) based on patient preference for a health status. The purpose of this study was to compare indirect measures to a directly elicited utility. Design. Cross-sectional study. Setting and Patients. Emory Spine Center and the Emory Center for Chronic Pain at Crawford Long Hospital. Patients at least 18 years of age with chronic pain, defined as pain that persists beyond the normal time of healing, usually beyond 6 months. Measures. Chronic pain subjects completed a paper-based, self-administered time trade-off (TTO) survey, EQ-5D survey, and a face-to-face (FTF) TTO interview. Current pain severity was graded using the Numeric Rating Scale-11, subsequently stratified into no (0), mild (1 3), moderate (4 6), and severe (7 10) pain. Results. Paired t test comparing FTF TTO and proxy utility measures stratified by severity revealed FTF TTO utility values significantly higher than EQ-5D utility values for all pain severities (overall mean difference 0.18, standard deviation [SD] 0.30, P < 0.001; Pearson s correlation 0.34, P < ); FTF TTO utility values were lower than paper TTO utility values, reaching statistical significance for mild and moderate pain (overall mean difference 0.09, SD 0.29, P = ; Pearson s correlation 0.38, P < ). Conclusions. This study demonstrates that the EQ-5D overestimates, whereas the paper version of TTO underestimates, the impact of pain on HRQoL compared with the directly elicited FTF TTO utility. Our findings provide preliminary evidence that utilities vary by method, and directly elicited utility values differ from indirectly elicited measures. Key Words. Pain; Quality of Life; Utility; EuroQoL; EQ-5D; Measurement Introduction Studies have widely documented that pain has a significant impact on health-related quality of life (HRQoL) [1 4], perhaps even more so than many other factors such as depression, disability, and fatigue. Improvements for HRQoL are often associated with treatment and improvement of pain [1]. The impact of pain on HRQoL can be quantified using health economic measures called utilities. Utilities are values of HRQoL based on patient preferences for a health status, ranging from 0 (dead) to 1 (perfect health) [5]. In cost-utility analyses, 1 quality-adjusted life year (QALY) is equal to the utility value multiplied by 1 year of life; thus, 1 year of life in less-than-perfect health equates to less than 1 QALY. The utility value associated for a health status multiplied by the total number of years lived 865

2 Wetherington et al. in that state determines the total number of QALYs [6]. The ratio of cost to QALY gained for a particular intervention can also be calculated. Health care resources are often allocated to those interventions with a low cost to QALY gained ratio (i.e., greater QALY gained for lower cost). Given that pain primarily impacts quality of life (rather than survival), and given that new therapies and interventions are being implemented, accurate methods of eliciting utilities are important. Direct elicitation of utilities is performed by asking patients to choose between life with the health state of interest (such as chronic pain) and life without the health state but with some sort of consequence. The consequences may include a certain risk of death (called the standard gamble method) or a reduction of total lifetime (called time tradeoff [TTO]) [7]. Directly elicited utilities are both labor and time intensive. Thus, health service researchers have developed alternative approaches. The EQ-5D is a commonly utilized survey that includes multiple attributes to classify respondents into categories based on assignment of health status; each category has a pre-assigned utility value derived from general population data [8]. Given the increase in new pain therapies and the current health care climate of determining the most cost-effective therapies, the pain field needs to find the most accurate elicitation method. As one of the most common proxy measures is the EQ-5D, the objective of this article is to compare the EQ-5D to a directly elicited utility. The faceto-face (FTF) TTO is a widely accepted and implemented method of direct elicitation [9,10]; so, we used utility values obtained via an FTF TTO interview as our reference values for comparison with proxy measures. We hypothesize that the EQ-5D utility overestimates disease burden when compared with FTF TTO direct utility scores obtained from patients with chronic pain. Because direct elicitation is time consuming, a secondary objective was to determine whether a paper-based TTO survey would be a sufficient reliable proxy measure for utilities. Materials and Methods We obtained approval from the Emory Institutional Review Board to perform this cross-sectional study. Subjects From June to September 2007, patients were recruited from the Emory Spine Center and the Emory Center for Chronic Pain at Crawford Long Hospital. Patients at least 18 years of age with chronic pain, defined as pain that persists beyond the normal time of healing, usually beyond 6 months, were eligible for inclusion. We obtained a convenience sample of subjects by placing flyers in the clinic offices and asking the treating physician to inform patients of the study. We excluded any subjects who were unable to speak or read English or had any other disability that would prevent the subject from completing both the questionnaires and FTF interview. We obtained informed consent from all subjects before including them in the study. Subjects were only included once, regardless of the number of times they attended the pain clinic for the duration of the study. Procedure In order to maximize efficiency, consenting subjects were given the paper-based, self-administered TTO and EQ-5D survey to complete while waiting for their physician. Subjects underwent the FTF TTO interview by a member of the study staff after their physician visit in order to avoid interfering with and slowing down clinic flow. The study staff member was blinded to the subjects responses to the other two paper and pencil measures and pain ratings. A subject information sheet was also administered to all subjects. This included questions regarding patient demographics as well as pain location, severity, timing, duration, quality, and functional limitations. Subjects graded current pain severity using the Numeric Rating Scale-11, which is subsequently stratified into groups of no (0), mild (1 3), moderate (4 6), and severe (7 10) pain. All subjects reported some amount of current pain. FTF TTO and Paper TTO We explained the TTO method to subjects prior to administering the paper TTO survey and FTF TTO interview. For both the FTF TTO and the paper TTO, we presented subjects with two hypothetical states prior to asking about their chronic pain. These states included life with or without paralysis and life with or without health problems in addition to chronic pain. These hypothetical states were used as calibrations; we expected that subjects would rate the calibration states worse than they rated life with chronic pain. If a subject rated his or her current pain worse than either calibration state, the interviewer clarified the TTO technique and confirmed that the subject understood the task and was accurately assessing his or her pain state. The FTF TTO method we used is a standard method used to elicit health utility scores [11]. Our computer program used standard life tables to estimate subjects remaining life years based on gender and race. Respondents were asked to choose between living this estimated time in a poorer health state or trading some of this time in exchange for a better health state. For each state, the options were as follows: 1) live in hypothetical or current state for X year(s) (generated by the utility calculator) or 2) live in hypothetical or current state for X Y year(s) and give up Y year(s) of your life. Subjects could choose either option, or they could state that they believed the two choices to be equal. An iterative ping-pong technique was used to titrate down to the subject s point of indifference. For the paper TTO, the same two hypothetical calibration health states as used in the FTF TTO were printed on the survey in addition to current pain state. Again, respondents were asked to choose between living their 866

3 estimated lifetime in a poorer health state or trading some of this time in exchange for a better health state. The iterative technique was not reproducible for the paper TTO due to the fact that iterations were titrated based on patient answers during the interview. Thus, for each of the three states, three scenarios were presented: 1) 100% relief from hypothetical or current state; 2) 50% relief from hypothetical or current state; and 3) never having experienced hypothetical or current state. For each of these scenarios, subjects chose between four options: 1) give up 5 years of life; 2) give up 10 years; 3) give up more than 10 years, with a blank for the subject to fill in the number of years; and 4) none of the above, with a blank for the subject to complete. The last option allowed for a choice less than 5 years. This questioning method has been used previously as an acceptable form of paper-based TTO, and utility values can be compared with values obtained in the FTF TTO [12,13]. We derived both the FTF TTO and paper TTO utility scores by calculating the ratio of time remaining after the trade to the individual s life expectancy. For example, a patient with a life expectancy of 75 years who was willing to give up 3 years to live without pain would have a utility score of 72/75, or 0.960, for the health state of chronic pain. In other words, this patient would be willing to forfeit 4% of his or her life expectancy to live without pain. The computerized survey used in this study had built-inerror checks to mitigate grossly inconsistent responses; the interviewer employed these same error checks. If a subject s response was inconsistent with earlier responses, we asked for confirmation and pointed out that the current responses conflicted with earlier responses. EQ-5D Survey The EQ-5D survey measures health status based on five dimensions, including mobility, self-care, usual activity (work, study, housework, family, and leisure), pain/ discomfort, and anxiety/depression. Each of the dimensions has three levels: no, some, and extreme problems. We formed a health status profile for each respondent and matched a corresponding health status valuation to this profile. Weighted valuations for each health state have been elicited from respondents in the general population for eight different countries; we utilize the valuations from the United States [14]. Statistical Analyses Descriptive summary statistics are included for the sample demographics and utilities stratified by pain severity in Tables 1 and 2. Continuous variables are reported as means with standard deviations (SDs), and categorical variables are reported as proportions. ANOVA was used to compare differences in utilities between severity groups and utility assessment method. Pearson s correlation coefficient and a paired t test were used to compare the FTF TTO and EQ-5D utilities, as well as the FTF TTO and Table 1 Patient demographics, age, mean (SD), and patient symptom severity Demographic N % Male Caucasian Married Education: any high school Income ($US): 75K Employed: full time Age, mean (SD) 55 (16) Symptom Severity N % Mild Moderate Severe Pain (N = 143). Majority of subjects was married, Caucasian, and had completed some high school education. Average age was 55 (16) and majority of patients reported moderate pain (43%). SD = standard deviation. paper TTO utilities. All analyses were performed using SAS 9.2. (SAS Institute Inc., Cary, NC, USA) A P value of <0.05 was considered statistically significant. Power Analysis We estimated a sample size of 139 in each group, assuming that a paired t test one-sided significance level will have an 80% power to reject the null hypothesis that the EQ-5D and FTF TTO are not equivalent (the difference in means, m T m S, is or farther from zero in the same direction) in favor of the alternative hypothesis that the means of the two groups are equivalent when the expected mean difference is 0.036, assuming the SD of the differences is As the same subject answered Table 2 (SD) Measurement of Pain Utilities Utilities stratified by pain severity, mean EuroQoL FTF T TO Paper T TO N = 138 N = 129 N = 116 Mean (SD) Mean (SD) Mean (SD) Mild 0.74 (0.12) 0.84 (0.26) 0.96 (0.08) Moderate 0.51 (0.20) 0.74 (0.32) 0.86 (0.21) Severe 0.48 (0.17) 0.73 (0.34) 0.74 (0.38) Overall 0.59 (0.21) 0.77 (0.31) 0.89 (0.20) Mean EQ-5D utility values were lower overall and for all pain severities than FTF TTO utility values. Mean paper TTO utility values were higher overall and for all pain severities than FTF TTO utility values. FTF = face-to-face; SD = standard deviation; TTO = time tradeoff. 867

4 Wetherington et al. Table 3 Mean differences and correlations stratified by severity EuroQoL vs FTF TTO Paper TTO vs FTF TTO FTF T TO EuroQoL Correlation FTF T TO Paper T TO Correlation Pain Severity N Mean difference (SD) P value r value P value N Mean Difference (SD) P value r value P value Mild (0.30) (0.21) Moderate (0.24) < (0.33) Severe (0.38) (0.26) Overall (0.30) < < (0.29) < The overall mean difference (SD) for FTF TTO EQ-5D was 0.18 (0.30), P < Overall mean difference (SD) for FTF TTO Paper TTO was 0.09 (0.29), P = FTF = face-to-face; SD = standard deviation; TTO = time trade-off. the EQ-5D and the FTF TTO, we targeted a total of 139. We recruited additional subjects in the event that some subjects had incomplete surveys. Results Demographics and Pain Characteristics (Table 1) Our population is representative of pain populations in the United States [15]. A total of 143 subjects were included, of which the mean (SD) age was 55 (16) years, 38% were male, and 78% were Caucasian. The majority of patients was married and had some high school education or higher. Of all subjects, 23% experience mild pain (N = 33), 43% experience moderate pain (N = 61), and 30% experience severe pain (N = 43). Absolute Utility Means (Table 2) The mean FTF TTO utility score was 0.77 (0.31) compared with 0.89 (0.20) for the paper TTO and 0.59 (0.21) for the EQ-5D. Overall and for all pain severities, paper TTO utility values were highest, and EQ-5D utility values were lowest. For all elicitation methods, subjects who reported greater pain severity had lower utility values. After adjustment for multiple comparisons using Bonferroni correction, utility values obtained via the EQ-5D survey were significantly different between mild and moderate pain severities (P < ) and mild and severe pain severities (P = ); utility values obtained via paper TTO were significantly different between mild and moderate pain severities (P = 0.03) and between mild and severe pain severities (P = 0.03). Comparison of FTF TTO and EQ-5D Utilities Stratified by Severity (Table 3) Paired t test comparing FTF TTO and EQ-5D utility values stratified by severity revealed that for all pain severities, FTF TTO utility values were significantly higher than EQ-5D utility values. Greater differences in valuation were observed for moderate and severe pain levels compared to mild pain. Overall, mean difference (SD) was 0.18(0.30) (P < 0.001) and Pearson s correlation was 0.34 (P < ). Correlation for moderate pain severity was 0.33 (P = 0.003). Correlation between FTF TTO and EQ-5D for mild and severe pain severity was not significant. Comparison of FTF TTO and Paper TTO Utilities Stratified by Severity (Table 3) Paired t test comparing FTF TTO and paper TTO utility values stratified by severity revealed that for all pain severities, FTF TTO utility values were lower than paper TTO utility values, reaching a statistical significance for mild and moderate pain levels. Overall, mean difference (SD) was 0.09 (0.29) (P = ), and Pearson s correlation was 0.38 (P < ). Correlation for moderate pain severity was 0.32 (P = 0.01). Correlation between FTF TTO and paper TTO for mild and severe pain severity was not significant. Discussion Chronic pain carries a significant burden as demonstrated by utilities. In our study, the average subject was willing to give up 23% of his or her life expectancy in order to have life without any pain. Our population s utility was comparable with that published in the chronic pain literature [1]; such a significant condition deserves accurate measures of HRQoL impact so that resource allocation via costeffectiveness analyses and burden-of-disease studies can be done equitably [1]. Quick and easy methods to elicit utilities would be ideal. For all elicitation methods, as pain severity increased, utility scores decreased, therefore indicating a worse HRQoL. However, the low correlation between the EQ-5D and FTF TTO (r = 0.35, P < ) suggests that the EQ-5D is a poor proxy for the FTF TTO. More revealing, our data indicate that the EQ-5D-derived utility estimates greater disease burden in this chronic pain population. Specifically, the EQ-5D indicates a lower utility score and a worse 868

5 HRQoL when compared with the FTF TTO utility overall and for all pain severities. These conclusions are comparable with other populations. In a literature review of studies measuring HRQoL in patients with osteoarthritis, direct methods resulted in higher valuations relative to indirect elicitation methods [7]. Further, in a study assessing patients in different states of breast cancer, FTF TTO values were higher than EQ-5D values [8]. Finally, FTF TTO values were significantly higher than EQ-5D utility values in a study assessing patients with menopause symptoms [16]. The methodology for the FTF TTO differs from that used in the paper TTO and also that to derive values for the EQ-5D scores. The FTF version has been adjusted to account for the estimated remaining number of life years of respondents, whereas the other two versions are based upon 10 years remaining life expectancy [17]. This could partly explain the results. Also, the differences between the EQ-5D and FTF TTO could be a result of different individuals making the valuations (general population vs patients). Existing evidence suggests that the valuation of a health state might be influenced by the duration effect relating to the time spent in that state, and there is evidence to show that patients with first-hand experience place higher values upon dysfunctional health states than members of the general public [17,18]. Interestingly, at greater pain severities (moderate and severe), the difference between EQ-5D and FTF TTO utility values was greater than at mild pain severities. These findings suggest that there is a limitation to the EQ-5D classification system. This approach often only assesses the most basic health status and does not cover diseaseor condition-specific issues [19 23]. The EQ-5D method does not include emotional impact, such as frustration, anger, helplessness, and stigmata, when determining chronic pain utility values. The inclusion of these factors may assist in increasing EQ-5D sensitivity at greater pain severities. Another approach to shortening the utility elicitation is to utilize noniterative approaches such as our paper TTO approach. Replication of the same ping-pong technique used in the FTF TTO is not feasible via a paper survey. Rather, our survey included choices of varying lengths of time for each scenario. This questioning method has been used previously as an acceptable form of paper-based TTO. Utility values can be compared with values obtained in the FTF TTO [12,13]. Unfortunately, our data also suggest that the paper TTO is a poor proxy for the FTF TTO method by the low correlation (r = 0.38, P < ) observed. The paper TTO appears to underestimate disease burden (i.e., indicates a greater utility value and better HRQoL) when compared with the FTF TTO utility as revealed by the paired t test results. This is likely due to the nature of the paper survey and the inability to titrate down to the patient s point of indifference using a preprinted form. Our findings indicate that the paper TTO may not be sensitive enough to capture mild to moderate pain levels. Measurement of Pain Utilities Interestingly, a correlation trending toward significance was seen between the paper TTO and FTF TTO utilities in the severe pain category (r = 0.76, P = 0.08). Further studies with more subjects in the severe pain category are needed to investigate these results. Our study is not without limitations. First, all of our patients were recruited from one institution, thus potentially limiting ability to generalize our results to patients of other demographics and geography. Secondly, because subjects were recruited out of a tertiary care facility and because our sample was self-selecting, we face a potential selection bias in terms of severity of patients. Our sampling procedure could have introduced potential bias to the preference assessment, as paper TTO and EQ-5D surveys were completed prior to the physician visit, whereas the FTF TTO interview was conducted after the visit in order to streamline efficiency during clinic hours. This sampling method introduces the potential for a significant order effect as it has long been documented that the relative position of an item in a series may uniquely influence the way in which a respondent reacts to them [24]. Had the three survey forms been administered in a random order, we may have seen significantly different results. Further, as we aimed to compare indirectly elicited measures with the directly elicited FTF TTO interview, we did not compare EQ-5D and paper TTO utility scores. In addition, our numbers were relatively small; a larger study may have elucidated significant differences in subgroups. Lastly, our data were not normally distributed; thus, nonparametric significance tests could have elicited different results. Despite these limitations, our results demonstrate that as pain severity increases, impact on HRQoL also increases, and the difference between EQ-5D and FTF TTO utility scores widens. Our results show that the average EQ-5Dderived utility value was lower than the FTF TTO utilities overall and for all pain severities, especially at greater pain severities. Thus, the EQ-5D-derived utility appears to overestimate disease burden when compared with the FTF TTO utility in this pain population. Our results also show that the average paper TTO utility value was higher than the FTF TTO utilities overall and for all pain severities, especially at mild pain severities. Thus, the paper TTO appears to underestimate disease burden when compared with the FTF TTO utility in this pain population. Further studies in chronic pain are needed to confirm the results in this sample of chronic pain patients. In addition, future studies could compare various dimensions of the EQ-5D, as all dimensions could be influenced by chronic pain. Our study is potentially significant for future clinical research. Our findings demonstrate that compared with an FTF TTO utility, the EQ-5D underestimates the impact of chronic pain on quality of life in this chronic pain population, whereas the paper version of the TTO overestimates the impact of chronic pain on quality of life in our population. Though these two paper proxy surveys are admittedly less labor intensive, we recommend continued use of FTF TTO interviews in future chronic pain studies. 869

6 Wetherington et al. Conclusion The EQ-5D underestimates and the paper TTO survey overestimates the impact of chronic pain on HRQoL compared with the FTF TTO utility in this pain population. The paper TTO survey, though less labor intensive, did not elicit utility values comparable with those elicited in a FTF interview in this pain population. Thus, researchers should continue to use FTF TTO interviews to elicit utility values in future chronic pain studies. Acknowledgments We would like to thank Drs. Tunisia Finch, Livia Van, Mary Jayne McIlwain, and Shinko Lin for their efforts in data acquisition. References 1 Doth AH, Hansson PT, Jensen MP, Taylor RS. The burden of neuropathic pain: A systematic review and meta-analysis of health utilities. Pain 2010;149(2): Bowling A. What things are important in people s lives? A survey of the public s judgements to inform scales of health related quality of life. Soc Sci Med 1995;41(10): Goossens ME, Vlaeyen JW, Rutten-van Mölken MP, van der Linden SM. Patient utilities in chronic musculoskeletal pain: how useful is the standard gamble method? Pain 1999;80(1 2): Skevington SM. Investigating the relationship between pain and discomfort and quality of life, using the WHOQOL. Pain 1998;76(3): Bakker C, Van der Linden SMJP. Health related utility measurement: An introduction. J Rheumatol 1995; 22(6): Torrance GW, Feeny D. Utilities and quality-adjusted life years. Int J Technol Assess Health Care 1989; 5: Ruchlin HS, Insinga RP. A review of health-utility data for osteoarthritis: Implications for clinical trial-based evaluation. Pharmacoeconomics, ADIS Int Limited 2008;26: Lidgren M, Wilking N, et al. Health related quality of life in different states of breast cancer. Qual Life Res 2007;16: Drummond MF, O Brien B, Stoddart GL, Torrance GW. Methods for the Economic Evaluation of Health Care Programmes, 2nd edition. United States: Oxford University Press, Inc; Gold MR, Siegel JE, Russell LB, Weinstein MC. Cost- Effectiveness in Health and Medicine. New York, NY: Oxford University Press, Inc; van Osch S, Wakker P, van den Hout W, Stiggelbout A. Correcting biases in standard gamble and time tradeoff utilities. Med Decis Making 2004;24(5): Buckingham JK, Birdsall J, Douglas JG. Comparing three versions of the time tradeoff: Time for a change? Med Decis Making 1996;16(4): Schiffner R, Schiffner-Rohe J, Gerstenhauer M, et al. Willingness to pay and time trade-off: sensitive to changes of quality of life in psoriasis patients? Br J Dermatol 2003;148(6): U.S. Valuation of the EuroQol EQ-5D Health States. Rockville, MD: Agency for Healthcare Research and Quality; (accessed August 2013). 15 Pleis JR, Ward BW, Lucas JW. Summary health statistics for U.S. adults: National Health Interview Survey, 2009: National Center for Health Statistics. Vital Health Stat 2010;10(249): Zethraeus N, Johannesson M. A comparison of patient and social tariff values derived from the time trade-off method. Health Econ 1999;8(6): Sackett D, Torrance G. The utility of different health states as perceived by the general public. J Chronic Dis 1981;31: Brazier J, Ratcliff J, Tsuchiya A, et al. Measuring and Valuing Health Benefits for Economic Evaluation. New York: Oxford University Press; Brazier J, Roberts J, Tsuchiya A, et al. A comparison of the EQ-5D and SF-6D across seven patient groups. Health Econ 2004;13: Macran S, Weatherly H, Kind P. Measuring population health: A comparison of three generic health status measures. Med Care 2003;41: Johnson JA, Pickard AS. Comparison of the EQ-5D and SF-12 health surveys in a general population survey in Alberta. Can Med Care 2000;38: Bharmal M, Thomas J. Comparing the EQ-5D and the SF-6D descriptive systems to assess their ceiling effects in the US general population. Value Health 2006;9(4): Herdman M, Gudex C, Lloyd A, et al. Development and preliminary testing of the new five-level version of EQ-5D (EQ-5D-5 L). Qual Life Res 2011;20(10): Perreault WD Jr. Controlling order-effect bias. Public Opin Quart 1975;39(4):