Background
Chronic noncancer pain is one of the most common harmful experiences worldwide. It is not only responsible for impaired quality of life (QoL) and active daily life but is also associated with high costs because of long medical treatment and loss of work productivity [
1,
2]. Among the Japanese population, 22.9 % suffer from chronic pain reported by numeric rating scale (NRS) to be a severity
≥ 5 and longer than 3 months duration. Of these patients, 55.9 % reported hospital visits due to pain, and 45.2 % of patients who had a hospital visit reported dissatisfaction with the care they received [
3].
Health-related quality of life (HRQOL) instruments, particularly the EuroQol-5D (EQ-5D) [
4], are widely used in health surveys and medical research. Lamers et al. showed the association between lower level EQ-5D utility scores and efficiency loss/absenteeism in individuals with low back pain [
5]. Kovacs et al. pointed out a notable association between pain, disability, and QoL in patients with low back pain; however, they also postulated that clinically relevant improvements in pain severity exerted much less influence than did improvements in disability and QoL. This suggested that disability and/or QoL should be assessed when evaluating the effect of any form of treatment for low back pain [
6]. In addition, these results suggested the importance of identifying target improvements in patients’ HRQOL and pain intensity and their association with success of chronic noncancer pain treatment. Several reports have suggested a minimal clinically important change (MCIC) of EQ-5D utility score in the treatment of low back pain [
7‐
9]; moreover, several studies on musculoskeletal pain have reported MCIC of a pain intensity score [
10‐
13]. However, no study has reported the MCIC of the Japanese version EQ-5D with an emphasis on chronic noncancer pain.
Although QoL assessment has become common in medical research in Japan, its interpretation still remains unclear. Suka et al. reported that the mean difference of EQ-5D utility scores between patients with low back pain and healthy volunteers was 0.09 in Japanese population [
14], but MCIC was not evaluated in this study. A solid, methodologically sound, Japanese-specific MCIC of the EQ-5D utility score is essential for QoL assessment using the EQ-5D questionnaire in real-world management of chronic noncancer pain.
In the present study, we conducted a retrospective post hoc analysis to identify the optimal cutoff point (OCP) MCIC of the Japanese version EQ-5D utility score as well as the NRS score and to evaluate the association between the two scores using data from a multicenter post-marketing surveillance (PMS) of a tramadol/acetaminophen combination tablet in patients with chronic noncancer pain.
Methods
Data source
In this study, a post hoc data analysis was conducted using PMS data (data cutoff: September 2014, Identifier: UMIN000015901) collected from 1,316 patients with moderate chronic noncancer pain who were treated with a tramadol/acetaminophen combination tablet (tramadol 37.5 mg and acetaminophen 325 mg per tablet) in routine clinical practice [
15]. PMS is a drug-specific registry that evaluates risks and benefits of a specific product. The PMS protocol was reviewed by internal review board members, including the ethical point of view, and was approved by the Japan Pharmaceuticals and Medical Devices Agency, Tokyo (
https://www.pmda.go.jp/english/index.html). The PMS was conducted under the Japanese regulation (Ministry of Health, Labour and Welfare Ministerial Ordinance No. 171) of good post-marketing study practice (GPSP). All patients registered under the PMS received tramadol/acetaminophen combination tablets for up to 52 weeks based on regulatory approved label information in Japan. In Japan, tramadol/acetaminophen combination is indicated for patients with chronic noncancer pain that is unrelieved by non-opioid analgesic medication. The daily dose was determined at the discretion of each responsible practicing investigator. No restrictions were placed on concomitant medications or physical and psychological therapies.
Measures and dataset
The sociodemographic data and name of the causal disease of pain were retrieved from PMS data. We also retrieved the NRS (0 to 10 points, 0: have no pain and 10: worst imaginable pain) for severity of the pain experienced by patients in the previous 24 h, the Japanese version of EQ-5D 3L [
16] as patient-reported outcome measures, and the dichotomous physician’s global impression of treatment effectiveness (PGI; discrete choice of “effective” or “not effective”). The EQ-5D scale contains five dimensions assessing “mobility,” “self-care,” “usual activities,” “pain/discomfort,” and “anxiety/depression.” Each dimension has three levels: “no problem,” “some problem,” and “extreme problem.” The collected EQ-5D scale was translated into a global utility score by using a Japanese version of time trade-off value set. Higher EQ-5D global scores represent better QoL. While the original study collected data up to 52 weeks, relatively a small number of participant responded at 52 weeks. Since we intended to estimate the MCIC without any imputation of missing data, this might cause biased results. Therefore, because of the applicability of our results to the clinical practice and the sample size, the MCICs were estimated based on using data at week 4. Patients with complete EQ-5D, NRS, and PGI at baseline and Week 4 were included in the analyses.
Statistical methods
Patient demographics were evaluated using descriptive statistics; categorical data were summarized by frequency and proportion, continuous data were summarized as mean with standard deviation (SD), median, maximum, and minimum.
A linear regression model was applied to evaluate correlation (r) and slope (β) between absolute changes from baseline of EQ-5D (dEQ-5D) and NRS (dNRS). The MCIC of EQ-5D and NRS at Week 4 were evaluated using receiver operating characteristic (ROC) curve anchored by PGI at Week 4. An area under the ROC curve (AUC) > 0.70 was considered as a sufficient level of prediction to identify MCIC [
17]. MCIC, the OCP of ROC, was determined using the maximal point of the sum of sensitivity and specificity [
18]. Subsequently, the ratio of patients who achieved MCIC of EQ-5D and NRS was calculated. In addition, logistic regression analysis was performed to investigate the influence of dNRS and dEQ-5D on PGI judgment. Standardized (Z score) dNRS and dEQ-5D were utilized for this logistic regression analysis in terms of adjustment to make the mean equal to zero and the SD equal to 1.0. The formulae for calculating standardized dNRS and dEQ-5D in each patient are
A two-sided p value < 0.05 was considered statistically significant. All statistical analyses were performed using R version 3.2.1 (Foundation for Statistical Computing, Vienna, Austria) and JMP 9.0 (SAS Institute Japan Ltd. Tokyo, Japan).
Discussion
In our study population, the most common cause of pain was musculoskeletal disorders. In a majority of patients, baseline NRS scores corresponding to pain intensity could be qualified as moderate-to-severe pain. While Nawata et al. had reported an average EQ-5D utility score of 0.83 (95 % CI, 0.82–0.84;
n = 2937) in Japanese elderly (mean age, 72.3 [SD, 6.1] years) volunteers [
19], the mean EQ-5D utility score at baseline in our study population (mean age, 66.7 [SD, 14.4] years) was remarkably lower (0.51 [SD, 0.20])
In general, pain intensity scores are associated with QOL scale in pain disorders [
6,
20,
21]. In our results, a significant correlation (
r = 0.53,
p < 0.001) was observed between dNRS and dEQ-5D scores, and both dNRS and dEQ-5D supported the treatment effect of tramadol/acetaminophen therapy. The simple linear regression model indicated that a 1-point reduction in dNRS corresponded to a 0.045-point improvement in dEQ-5D. This conversion ratio may be applicable to clinical practice for chronic pain management when practitioners evaluate clinical outcomes using the NRS or EQ-5D scales. However, there is a possibility that other factors strongly associate with the change of NRS and EQ-5D. For example, an individual patient who had severe side effects would result in a deterioration of QoL. Therefore we need to evaluate QoL and functionality along with pain severity in the treatment of chronic pain patients.
The primary aim of this study was to estimate the OCP for MCIC of the NRS and the Japanese version of EQ-5D, with an emphasis on noncancer chronic pain using simple and conventional statistical analysis methods. The OCP of MCIC was estimated to be a 0.10-point increase in EQ-5D and a 2.0-point reduction in NRS.
Two-thirds of physicians have agreed on an OCP of 2- to 4-point reduction in NRS scores as the MCIC [
22]. Salaffi et al. reported a 2-point reduction in NRS as the minimal treatment target for chronic pain [
10]. The mean change in the pain intensity score from baseline by using a 10-mm visual analog scale at 4 weeks after tramadol/acetaminophen therapy was −22.35 mm (SD, 21.19) in the clinical registration trial in Japanese patients with chronic noncancer pain [
23]. These results support that the MCIC of NRS we propose is clinically relevant in the Japanese population. In addition, several reports have suggested a minimal clinically important difference in EQ-5D utility score as approximately 0.08 [
7] to 0.17 [
8] in the treatment of low back pain. The results of our study that used Japanese real-world data are consistent with these previous reports. We identified an MCIC that is clinically comparable to that derived from medical research conducted in multi-countries and that may provide the reasonable minimum treatment target for Japanese patients with chronic pain in real-world practice. The distribution-based approach using one- half of standard deviation at baseline value was frequently used to determine the OCP of MCIC [
24]. In this study, one- half of standard deviation at baseline of EQ-5D was 0.1 (Table
2) that was same as OCP of MCIC produced by the anchor- based approach using PGI. We did not see the discrepancy between the conclusions of both approaches. As a limitation of our result, the estimated MCICs were based on the analysis of complete data at baseline and week 4. We assumed missing at random for missing data mechanism and no imputation for missing data has been performed.
The PGI (effectiveness) and the rate of MCIC achievement of EQ-5D (<0.1) and NRS (< −2.0) were 86.5 % (614/710), 57.0 % (405/710), and 68.6 % (487/710), respectively. The rate of MCIC achievement in both patient-reported outcomes was lower and seemed more conservative than physician’s judgment. A larger discrepancy was observed between physician’s judgment and the ratio of MCIC achievement in EQ-5D than in NRS. In addition, logistic regression analysis using standardized dNRS and dEQ-5D suggested that improvements in both the NRS score (p < 0.001) and the EQ-5D utility score (p < 0.001) were independently associated with a more favorable PGI judgment. The absolute value of the standardized coefficient was compatible between dNRS and dEQ-5D, but the coefficient of dNRS was higher (−1.490) than the coefficient of dEQ-5D (0.795), suggesting that physician’s judgment is influenced more by changes in NRS scores than by changes in EQ-5D utility scores. This result was consistent with the result of ROC analysis that the sensitivity of OPC of MCIC on dNRS was higher than dEQ-5D. This may suggest that when judging improvement, physicians should consider improvements in pain severity than improvements in the patient’s QoL.
Conclusion
The post treatment changes from baseline in NRS scores and Japanese version of EQ-5D utility scores were well correlated in Japanese patients with chronic noncancer pain. Simple ROC using the PGI-anchored approach estimated a 0.10-point increase as the OCP of MCIC in EQ-5D utility scores and a 2.0-point reduction in NRS scores. Our results demonstrated novel cutoff criteria for the Japanese version of EQ-5D, focusing on patients with chronic noncancer pain. The obtained criteria were fairly consistent and could be utilized as an evaluation tool in medical research on noncancer chronic pain in Japan, providing additional functionality and usability for QoL assessment in pain management practice. The EQ-5D scale is a useful evaluation tool for QoL assessment in real-world treatment of Japanese patients with chronic noncancer pain.
Ethics approval and consent to participate
Not applicable
Acknowledgements
The authors are indebted to the physicians who collected this study data in 194 hospitals; Koji Shimamoto, MS, MediStatLab. Co., Ltd. Tokyo, Japan for validation of statistical analysis programing; and Ryoko Nakagawa, Janssen Pharmaceutical K.K. for technical proofreading and verification of numeric results.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (
http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (
http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
Competing interests
The authors are full-time employees of Janssen Pharmaceutical K.K., which manufactures the tramadol/acetaminophen combination tablet in Japan.
Authors’ contributions
KY and KK designed the post hoc study and executed data collection. YO executed data management. TY and KY performed statistical analysis. KY, HK, and MF prepared the manuscript. All authors read and approved the final manuscript.