Beyond numeric scales: validity of the EQ-5D-5L pain/discomfort dimension in postoperative pain assessment after minimally invasive colorectal cancer surgery Authors: Sooah Cho, Soo-Hyuk Yoon, Ho-Jin Lee, Jae Hyun Park, Min Jung Kim, Ji Won Park, Seung-Bum Ryoo, Seung-Yong Jeong Published: 2026 ABSTRACT BackgroundThis study examined the validity of the EuroQol-5 Dimension 5-Level (EQ-5D-5L) pain/discomfort dimension for acute postoperative pain assessment.MethodsThis retrospective study analyzed prospectively collected data from the Korean Enhanced Recovery After Surgery (K-ERAS) registry, including adults undergoing elective minimally invasive colorectal cancer surgery at a single tertiary center between October 2023 and December 2025. Convergent validity was assessed using Spearman correlations between the EQ-5D-5L pain/discomfort dimension and numeric rating scale (NRS) scores at rest and during coughing on postoperative days (PODs) 1–3. Correlations with other EQ-5D-5L domains were compared using Steiger’s Z-test. Known-group validity was evaluated by area under the curve analysis for discrimination between adjacent levels and by comparisons between ERAS and conventional groups using cumulative link mixed models. Responsiveness was assessed using standardized response means.ResultsA total of 333 patients were included. The pain/discomfort dimension demonstrated moderate correlations with NRS scores at rest (ρ = 0.44–0.53) and during coughing (ρ = 0.50–0.57) across PODs 1–3 (all P < 0.001), and showed stronger associations with other EQ-5D-5L dimensions than the NRS. It exhibited acceptable discriminative ability without meaningful floor or ceiling effects and responsiveness comparable to the NRS. The dimension also successfully differentiated clinical groups, with lower pain levels observed in the ERAS group.ConclusionsThe EQ-5D-5L pain/discomfort dimension has potential as a valid and responsive measure for acute postoperative pain, complementing numeric scales by capturing the functional impact of pain. Further validation in diverse surgical populations is warranted. INTRODUCTION Effective postoperative pain management is a cornerstone of perioperative care and essential for patient comfort and successful outcomes. Therefore, accurate pain assessment is critical to achieving optimal pain control [1]. In clinical practice, pain is commonly evaluated using unidimensional tools such as the numeric rating scale (NRS) or visual analog scale (VAS). Although widely adopted because of their simplicity, these instruments assess only pain intensity and fail to reflect the multidimensional nature of postoperative pain [1,2]. Moreover, the practical use of numeric scales presents several limitations. Unidimensional numeric scales require patients to map their internal sensory experience onto a linear numerical continuum. In the absence of clear qualitative benchmarks, patients may have difficulty distinguishing between specific points on the scale, leading to inconsistent reporting [3]. Consistent with this limitation, the NRS has been reported to exhibit information overlap between adjacent scores in elderly patients, resulting in inferior discriminative power compared with the 4-level verbal descriptor scale [4]. The EuroQol-5 Dimension 5-Level (EQ-5D-5L) questionnaire provides a structured alternative for assessing health status across five domains: mobility, self-care, usual activities, pain/discomfort, and anxiety/depression [5]. Each domain is rated on five severity levels, ranging from “no problems” to “extreme problems.” The “pain/discomfort” dimension directly reflects pain control, while the remaining domains are closely related to pain-related functional and psychological recovery, enabling assessment of pain’s multidimensional impact. Compared with the earlier EQ-5D-3L, the EQ-5D-5L offers superior descriptive sensitivity and reduced ceiling effects through its granular response options, allowing more precise detection of subtle changes during recovery [6]. As a word-based descriptive scale, it also provides intuitive descriptive anchors that reduce cognitive burden and promote more consistent self-assessments than abstract numeric mapping. Although conventionally used to assess long-term postoperative quality of life, recent studies have applied this instrument in the early postoperative period to monitor immediate recovery [7–9]. Building on prior institutional evaluation of the Enhanced Recovery After Surgery (ERAS) program in colorectal cancer (CRC) surgery using the EQ-5D-5L questionnaire (unpublished), this study explored the potential of the “pain/discomfort” dimension as a measure of postoperative pain. Convergent validity, known-group validity, and responsiveness of the EQ-5D-5L pain/discomfort dimension were retrospectively assessed in comparison with the standard NRS in patients undergoing minimally invasive CRC surgery. MATERIALS AND METHODS 1. Study design and population This study was a retrospective analysis of prospectively collected data from the Korean ERAS (K-ERAS) Registry [10]. Patients treated at a single tertiary academic medical center between October 2023 and December 2025 were included. The study protocol was approved by the Institutional Review Board of Seoul National University Hospital (IRB no. H-2601-061-1707). Informed consent was obtained for enrollment in the K-ERAS registry, and the requirement for additional consent for this retrospective analysis was waived because only de-identified data were used. The K-ERAS registry enrolled adults aged ≥ 19 years who underwent elective minimally invasive CRC surgery. Exclusion criteria for registry enrollment were metastatic disease, emergency surgery, transanal minimally invasive surgery, bowel perforation or obstruction, hereditary CRC, prior CRC treatment, treatment for other malignancies within the previous 5 years, acute active infection, or mental impairment or illiteracy precluding understanding of the ERAS protocol. For the present study, additional exclusion criteria were applied: conversion to open surgery, missing EQ-5D-5L data from the preoperative period through postoperative day (POD) 3, or missing NRS pain scores during the same period. 2. Data collection The K-ERAS is a web-based, cloud-hosted registry developed to facilitate ERAS protocol implementation, monitor compliance, and audit perioperative outcomes in the Korean clinical setting. It was established by a multidisciplinary team of surgeons and anesthesiologists and prospectively collects comprehensive perioperative data with real-time monitoring to ensure data integrity. Data are recorded across multiple domains, including patient demographics, anesthesia details, surgical variables, and postoperative outcomes. All study variables were extracted directly from the registry. Collected data included demographic and clinical characteristics (age, sex, body mass index, smoking status, American Society of Anesthesiologists physical status, and Eastern Cooperative Oncology Group performance status), primary diagnosis (colon vs. rectal cancer), surgical details (approach, procedure type, operative time, and diversion status), and ERAS implementation. Patient-reported outcomes, including NRS pain scores and EQ-5D-5L responses, were retrieved from the registry. These outcomes were collected using self-administered questionnaires that were distributed and explained preoperatively by a trained ERAS coordinator. Patients independently completed the questionnaires daily from POD 1 until discharge, and the forms were collected by the coordinator before discharge. Assessments were scheduled at 24, 48, and 72 hours postoperatively; however, on discharge day, questionnaires were completed in the morning regardless of the scheduled time. Because most patients in the ERAS group were discharged on POD 3, analyses were restricted to data collected up to this time point. Postoperative pain intensity was assessed using an 11-point NRS, ranging from 0 (“no pain”) to 10 (“worst pain imaginable”). Pain scores were recorded separately at rest and during coughing. Postoperative quality of recovery was measured using the EQ-5D-5L. 3. Perioperative pain management Before ERAS implementation, postoperative analgesia for minimally invasive CRC surgery at the authors’ institution primarily relied on fentanyl-based intravenous patient-controlled analgesia (IV-PCA). The PCA solution contained fentanyl 20 μg/mL and was programmed to deliver 20 μg demand boluses with a 10-minute lockout interval and no background infusion; a 50 μg loading dose was administered at the end of surgery. IV acetaminophen (1 g) was administered approximately 30 minutes before the end of surgery and every 8 hours until the morning of POD 1. From POD 1 until discharge, the patients received oral extended-release acetaminophen (650 mg) and zaltoprofen (80 mg) every 8 hours. Following ERAS implementation, a multimodal opioid-free analgesic strategy was adopted. Preemptive analgesia was administered with oral acetaminophen (650 mg), zaltoprofen (80 mg), and gabapentin (300 mg) or pregabalin (75 mg). After the induction of anesthesia, an ultrasound-guided bilateral subcostal transversus abdominis plane block was performed using 20 mL of 0.375% ropivacaine on each side. Approximately 30 minutes before the end of surgery, the patients received a fixed-dose combination of IV acetaminophen (1 g) and ibuprofen (300 mg), which was continued every 6 hours until the morning of POD 1. From POD 1 until discharge, oral extended-release acetaminophen (650 mg) and zaltoprofen (80 mg) were administered every 8 hours. 4. Study outcomes The primary outcome was convergent validity between the EQ-5D-5L pain/discomfort dimension and NRS pain scores at rest and during coughing on POD 1, 2, and 3. Secondary outcomes included: (1) correlations between the three pain measures (EQ-5D-5L pain/discomfort, NRS pain score at rest, and NRS pain score during coughing) and the other four EQ-5D-5L dimensions (mobility, self-care, usual activities, and anxiety/depression); (2) known-groups validity, assessed by the discriminative ability of the EQ-5D-5L pain/discomfort dimension across adjacent levels and by comparisons of pain/discomfort trajectories between the ERAS and conventional groups from POD 1–3; (3) responsiveness of the three pain measures over PODs 1–3; and (4) floor and ceiling effects of the three pain measures. 5. Statistical analysis Continuous variables are presented as mean ± standard deviation (SD) or median with interquartile range, as appropriate, and categorical variables as frequencies and percentages. Construct validity was assessed through convergent validity, correlations with other health domains, and known-group validity. Convergent validity was assessed using Spearman’s rank correlation coefficients (ρ) between the EQ-5D-5L pain/discomfort dimensions and NRS pain scores at rest and during coughing. As the data involved repeated measurements in the same patient, correlations were assessed separately for each POD and interpreted as day-specific cross-sectional associations rather than pooled patient-independent estimates. Associations between pain measures (both EQ-5D-5L and NRS pain scores) and remaining EQ-5D-5L dimensions (mobility, self-care, usual activities, and anxiety/depression) were examined using Spearman correlations. It was hypothesized that pain severity is closely associated with physical and mental function recovery. Steiger’s Z-test was used to compare correlation coefficients between pain measures to determine which measure demonstrated significantly stronger associations with other domains. Known-group validity was assessed by comparing NRS pain scores across adjacent regrouped EQ-5D-5L levels using the Mann-Whitney U-test, which was performed separately for each POD. Discriminative ability was quantified using the area under the curve (AUC, C-statistic), representing the probability that patients with higher EQ-5D levels reported higher NRS scores. Owing to the small number of responses in the “no” (n = 4, 4, 10 for POD 1, 2, 3, respectively) and “extreme” categories (n = 2 at each time point), these levels were combined with the “slight” or “severe” category to ensure robust statistical analysis. Known-group validity was further evaluated by comparing EQ-5D-5L pain/discomfort dimensions between the ERAS and conventional groups using generalized estimating equations (GEE) with a cumulative logit link, under the hypothesis that the ERAS group would be associated with lower pain/discomfort levels. This approach was chosen to accommodate the ordinal nature of the EQ-5D-5L pain/discomfort dimension and repeated-measures within individuals. Owing to the sparsity of data in the extreme response categories (no or extreme problems), the five levels of the pain/discomfort dimension were regrouped into three categories: no or slight, moderate, and severe or extreme problems to ensure model stability and convergence. The model included “Group” (ERAS vs. Conventional), “Time” (POD 1, 2, 3), and their interaction as independent variables. In order to account for within-subject correlations, a “time-exchangeable” local odds ratio structure was employed. When the interaction was not statistically significant, a main-effects model was used to estimate the overall group effects. Conversely, if the interaction was significant, post hoc comparisons were performed at each time point. Results are presented as cumulative odds ratios (CORs) with 95% confidence intervals (CI), representing the odds of reporting lower pain/discomfort levels; CORs > 1 for the ERAS group indicate a higher likelihood of reporting lower pain/discomfort compared with the conventional group. Responsiveness was evaluated using the standardized response mean (SRM), calculated as the mean change in scores from POD 1 to 3 divided by the SD of the change. Differences in SRMs between the EQ-5D-5L pain/discomfort dimension and NRS pain scores were evaluated using a bootstrap method with 2,000 replications; differences were considered significant when the 95% CI did not include zero. In addition, Spearman’s rank correlation coefficients (ρ) were calculated between the change scores of the EQ-5D-5L pain/discomfort dimension and NRS scores. These correlations were evaluated for the overall period (POD 1–3) and daily intervals (POD 1–2 and POD 2–3) to determine the correspondence between changes in pain intensity. Floor and ceiling effects were assessed to evaluate discriminative performance at the extremes of the scale and were defined as the percentage of patients reporting the lowest (NRS 0; EQ-5D level 1, “no problems”) or highest (NRS 10; EQ-5D level 5, “extreme problems”) possible scores. Effects exceeding 15% were considered significant, indicating limited discriminative ability at the extremes of the scale [11]. All statistical analyses were performed using R software (version 4.4.1; R Foundation for Statistical Computing). A P value < 0.05 was considered statistically significant. RESULTS Of 362 initially identified patients, 29 were excluded due to conversion to open surgery (n = 5) or missing EQ-5D-5L data (n = 24). Consequently, 333 patients were included in the final analysis. Baseline demographic and clinical characteristics are summarized in Table 1. Fig. 1 illustrates the convergent validity of the EQ-5D-5L pain/discomfort dimension through its correlation with NRS pain scores at rest and during coughing on POD 1–3. The EQ-5D-5L pain/discomfort dimension showed significant positive correlations with NRS pain scores at rest (ρ = 0.44–0.53, all P < 0.001) and during coughing (ρ = 0.50–0.57, all P < 0.001) across the three POD. Fig. 2 presents the correlations between the three pain measures (EQ-5D-5L pain/discomfort, NRS at rest, and NRS during coughing) and the other four EQ-5D-5L dimensions (mobility, self-care, usual activities, and anxiety/depression). All three pain measures were significantly correlated with the other dimensions from POD 1 to 3 (P < 0.05), except for NRS at rest and during coughing with anxiety/depression on POD 1 and NRS during coughing with the anxiety/depression dimension on POD 2. The EQ-5D-5L pain/discomfort dimension consistently showed stronger correlations with mobility, self-care, and usual activities compared with both NRS measures. Table 2 summarizes the known-group validity of the EQ-5D-5L pain/discomfort dimension. The discriminative probability between adjacent levels, measured using AUC, ranged from 0.651 to 0.806. The highest discriminative ability was observed for pain intensity during coughing on POD 3 between the “moderate” and “severe or extreme” categories (AUC = 0.806), whereas the lowest was observed for the same comparison on POD 2 (AUC = 0.651). Across all of the time points, the EQ-5D-5L demonstrated an acceptable ability to differentiate between “no or slight problems” and “moderate problems,” for pain during coughing (AUC range: 0.756–0.794). Fig. 3 illustrates the distribution of the EQ-5D-5L pain/discomfort dimension responses from the preoperative baseline to POD 3 in the ERAS and conventional groups. In the GEE model, no significant group-by-time interactions were observed (P = 0.371 for POD 2, and P = 0.365 for POD 3). Therefore, the overall treatment effect was estimated using a main effects model. The ERAS group demonstrated significantly higher CORs of reporting lower pain/discomfort levels when compared to the conventional group throughout the first three PODs (Overall COR = 1.83; 95% CI: 1.32–2.55; P < 0.001). These findings suggest that the pain/discomfort dimension of the EQ-5D-5L effectively reflects the clinical benefits of the ERAS protocol for postoperative pain management, thus supporting its known-group validity in this clinical setting. Fig. 4 presents the responsiveness of the three pain measures during POD 1–3. The EQ-5D-5L pain/discomfort dimension showed an SRM of 0.38, indicating small-to-moderate responsiveness, comparable to the responsiveness of NRS pain scores at rest (SRM = 0.34) and during coughing (SRM = 0.31). Pairwise comparisons using the bootstrap method revealed no statistically significant differences in responsiveness between the three measures. Regarding the correspondence between change scores, those of the EQ-5D-5L pain/discomfort dimension significantly correlated with those of NRS at rest and during coughing for the overall period (POD 1–3: ρ = 0.310 and 0.438, respectively; both P < 0.001). Similarly, daily change scores for the POD 1–2 (ρ = 0.328 and 0.380, respectively) and POD 2–3 (ρ = 0.353 and 0.369, respectively) were all found to be significantly correlated (all P < 0.001). Table 3 summarizes the floor and ceiling effects of the three pain measures. None exceeded the 15% threshold at any time point from POD 1 to 3, indicating that all instruments maintained adequate discriminative ability at the extremes of the pain scale. DISCUSSION The findings of this study provide preliminary evidence for the potential validity and responsiveness of the EQ-5D-5L pain/discomfort dimension in assessing postoperative pain following minimally invasive CRC surgery. While the pain/discomfort dimension exhibited adequate discriminative ability without significant floor or ceiling effects and had responsiveness comparable to NRS scores, it showed moderate correlations with NRS pain scores at rest and during coughing. However, the pain/discomfort dimension showed stronger associations with other EQ-5D-5L dimensions than with NRS scores. Additionally, the differences observed between the ERAS and conventional groups indicate that this dimension is sensitive to variations in analgesic management. Collectively, these findings suggest that the EQ-5D-5L pain/discomfort dimension may serve as a valid indicator of postoperative pain control in the context of general recovery assessment. The validity of the EQ-5D-5L pain/discomfort dimension as a pain measure has been demonstrated in various clinical contexts. In survivors of critical illness, this single dimension showed strong correlations with the Brief Pain Inventory, indicating robust convergent validity for pain severity (r = 0.73–0.80) and interference (r = 0.71–0.83) [12]. Similarly, in patients with hip or knee osteoarthritis, it correlated strongly with the Western Ontario and McMaster Universities Osteoarthritis Index pain scales [13]. In heterogeneous populations, the pain/discomfort dimension correlated more strongly with 5-level verbal pain severity scales (r = 0.83) than with discomfort severity scales (r = 0.58), supporting the interpretation that this dimension primarily reflects physical pain rather than nonspecific discomfort [14]. However, in the present study, correlations between the EQ-5D-5L pain/discomfort dimension and NRS scores were only moderate across the first three PODs, lower than those reported previously. This likely reflects differences in temporal perspective: acute post-operative pain is dynamic, and while the NRS captures a momentary snapshot of pain intensity, the EQ-5D-5L pain/discomfort dimension reflects the patient’s overall daily experience. Additionally, the concept of “discomfort” encompasses a wide range of physical and emotional experiences [15], which may be particularly salient in the acute postoperative setting. Surgical patients often experience non-nociceptive symptoms, including fatigue, malaise, and nausea, which may influence responses to the EQ-5D-5L pain/discomfort dimension but are not captured by a unidimensional scale such as the NRS. Early postoperative fatigue and cognitive load may also compromise the reliability of NRS responses, as converting a subjective symptom into a numeric value requires abstract cognitive processing that can be difficult when pain is intense and patients are physiologically stressed (Fig. 1). Although NRS values corresponding to each EQ-5D-5L pain/discomfort level were more distinct on POD 3, separation between categories was less clear on POD 1 and 2. This reduced discrimination in the early recovery phase may suggest that numeric pain ratings may be limited in reflecting the complexity or burden of postoperative symptoms during this period. These findings are consistent with the authors’ previous observations in the post-anesthesia care unit, where a 4-level verbal rating scale showed a significantly higher response rate than the NRS, likely due to the lower cognitive demand of categorical scales in the immediate postoperative setting [16]. Further studies are warranted to determine assessment tools that provide the most reliable and clinically informative data during this phase. Another notable finding was that the EQ-5D-5L pain/discomfort dimension showed stronger correlations with other EQ-5D-5L dimensions than with NRS scores at rest or during coughing. This suggests that the pain/discomfort dimension may reflect broader aspects of functional recovery rather than pain intensity alone. Several factors may explain this observation. First, the EQ-5D-5L uses descriptive response categories that capture a range of discomfort-related experiences, many of which can affect postoperative functional recovery. Second, the unidimensional NRS may not completely reflect the overall pain burden. Prior studies in patients with chronic pain have demonstrated substantial discordance between NRS pain intensity and perceived pain tolerability, highlighting the limitations of relying solely on numeric ratings to represent a complex subjective experience [17]. Similarly, studies in patients on long-term opioid therapy have reported weak associations between NRS scores and pain-related disability, mental health, and overall quality of life [18]. Support for descriptive approaches to pain assessment comes from the authors’ previous evaluation of the Korean version of the Defense and Veterans Pain Rating Scale (DVPRS), a composite pain assessment tool incorporating numeric, descriptive, and facial expression components [19]. DVPRS demonstrated stronger correlations with the EQ-5D-5L index and EQ-VAS scores at 48 hours postoperatively than the NRS alone [19]. Although the EQ-5D-5L is not a composite pain scale, these findings suggest that pain assessments incorporating descriptive or experiential dimensions may better reflect patient-reported health status than numeric intensity ratings alone. Moreover, the EQ-5D-5L pain/discomfort dimension differed between the ERAS and conventional groups, which may reflect variations in the perioperative care and recovery pathways. This suggests that this dimension may capture clinically relevant differences in postoperative recovery, including the functional consequences of pain control strategies. Overall, these findings suggest that the EQ-5D-5L pain/discomfort dimension is a useful outcome measure for postoperative pain management, particularly when assessing functional recovery in addition to pain intensity. This perspective aligns with the 2017 US Joint Commission report, which emphasized that postoperative pain management should prioritize functional restoration, such as the ability to sit or walk without limitation, rather than solely reducing pain scores [20]. In addition, while the 11-point NRS provides a higher resolution for pain intensity, the lower-resolution 5-level EQ-5D-5L pain/discomfort dimension may offer distinct practical advantages in the immediate postoperative period. This simplified categorization could reduce the cognitive burden on fatigued surgical patients and help clinicians filter out “clinical noise,” such as minor fluctuations in the pain scores that are clinically insignificant. Consequently, this allows for a greater focus on meaningful shifts in a patient’s functional state. Furthermore, along with the other EQ-5D-5L dimensions, it provides a multidimensional perspective on how postoperative pain affects physical and psychological functions. Further studies are warranted to clarify its clinical utility relative to conventional measures, such as the NRS and VAS, in the acute postoperative setting. This study has several limitations. First, its retrospective, single-center design limits generalizability of the findings. The analysis was also restricted to patients undergoing minimally invasive CRC surgery, which may not represent other surgical populations with different pain characteristics and recovery patterns. Second, the two instruments differ in their temporal assessment frames, which may have contributed to variability and attenuated correlations. Third, this study focused on psychometric comparison between the EQ-5D-5L pain/discomfort dimension and the NRS, without examining direct associations with clinical outcomes, such as analgesic effectiveness or treatment adjustments. Thus, while the findings support the validity of the EQ-5D-5L pain/discomfort dimension in the postoperative setting, they do not establish its clinical superiority over the NRS. Fourth, the repeated-measures nature of the data means that these values reflect day-specific cross-sectional associations and do not capture longitudinal within-patient correlation dynamics. Finally, although early postoperative fatigue may affect patients’ ability to reliably report numeric pain scores, formal cognitive assessment was not performed. Consequently, the hypothesized influence of cognitive factors on NRS reliability during the early postoperative period was not directly evaluated. The EQ-5D-5L pain/discomfort dimension shows potential as a valid and responsive tool for postoperative pain assessment. Although its correlations with the NRS were moderate, it demonstrated stronger associations with functional recovery and effectively captured significant differences in outcomes between distinct clinical groups. Thus, this dimension may serve as a useful indicator of perioperative pain control. Further studies are needed in order to validate its clinical utility compared to standard pain measures across broader surgical populations.