If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Studies Comparing Numerical Rating Scales, Verbal Rating Scales, and Visual Analogue Scales for Assessment of Pain Intensity in Adults: A Systematic Literature Review
Address correspondence to: Marianne Jensen Hjermstad, PhD, Department of Oncology, Regional Center for Excellence in Palliative Care, Oslo University Hospital, Ullevål, P.O. Box 4956, Nydalen, Oslo 0424, Norway.
Regional Center for Excellence in Palliative Care, Department of Oncology, Oslo University Hospital-Ullevål, Oslo, NorwayEuropean Palliative Care Research Center, Department of Cancer Research and Molecular Medicine, Faculty of Medicine, Norwegian University of Science and Technology and Trondheim University Hospital, Trondheim, Norway
European Palliative Care Research Center, Department of Cancer Research and Molecular Medicine, Faculty of Medicine, Norwegian University of Science and Technology and Trondheim University Hospital, Trondheim, NorwayDepartment of Public Health, University of Aberdeen, Scotland, United Kingdom
European Palliative Care Research Center, Department of Cancer Research and Molecular Medicine, Faculty of Medicine, Norwegian University of Science and Technology and Trondheim University Hospital, Trondheim, NorwayRegional Center of Excellence for Palliative Care, Western Norway, Haukeland University Hospital, Bergen, Norway
European Palliative Care Research Center, Department of Cancer Research and Molecular Medicine, Faculty of Medicine, Norwegian University of Science and Technology and Trondheim University Hospital, Trondheim, NorwayNational Resource Centre for Late Effects after Cancer Treatment, Oslo University Hospital, Oslo, Norway
Regional Center for Excellence in Palliative Care, Department of Oncology, Oslo University Hospital-Ullevål, Oslo, NorwayFaculty of Medicine, University of Oslo, Oslo, Norway
European Palliative Care Research Center, Department of Cancer Research and Molecular Medicine, Faculty of Medicine, Norwegian University of Science and Technology and Trondheim University Hospital, Trondheim, NorwayPalliative Medicine Unit, Department of Oncology, St. Olavs University Hospital, Trondheim, Norway
The use of unidimensional pain scales such as the Numerical Rating Scale (NRS), Verbal Rating Scale (VRS), or Visual Analogue Scale (VAS) is recommended for assessment of pain intensity (PI). A literature review of studies specifically comparing the NRS, VRS, and/or VAS for unidimensional self-report of PI was performed as part of the work of the European Palliative Care Research Collaborative on pain assessment.
Objectives
To investigate the use and performance of unidimensional pain scales, with specific emphasis on the NRSs.
Methods
A systematic search was performed, including citations through April 2010. All abstracts were evaluated by two persons according to specified criteria.
Results
Fifty-four of 239 papers were included. Postoperative PI was most frequently studied; six studies were in cancer. Eight versions of the NRS (NRS-6 to NRS-101) were used in 37 studies; a total of 41 NRSs were tested. Twenty-four different descriptors (15 for the NRSs) were used to anchor the extremes. When compared with the VAS and VRS, NRSs had better compliance in 15 of 19 studies reporting this, and were the recommended tool in 11 studies on the basis of higher compliance rates, better responsiveness and ease of use, and good applicability relative to VAS/VRS. Twenty-nine studies gave no preference. Many studies showed wide distributions of NRS scores within each category of the VRSs. Overall, NRS and VAS scores corresponded, with a few exceptions of systematically higher VAS scores.
Conclusion
NRSs are applicable for unidimensional assessment of PI in most settings. Whether the variability in anchors and response options directly influences the numerical scores needs to be empirically tested. This will aid in the work toward a consensus-based, standardized measure.
There is an extensive literature regarding the use of Numerical Rating Scales (NRSs), Verbal Rating Scales (VRSs), and Visual Analogue Scales (VASs) dating from the 1950s. Nearly all of this literature is from the social sciences, notably census and surveys, public opinion polls, and marketing research. Two particular themes emerge in this literature. The first is a focus on determining the optimal number of response options when using NRSs or VRSs; the second relates to the comparative value of VASs and NRSs.
In the area of cancer pain assessment, the main emphasis of most authors has been on comparing VAS scores, the most common measure for pain intensity (PI) in cancer research,
to the scores obtained on 10-step or 11-step NRSs (NRS-10 and NRS-11, respectively). Fewer papers seem to focus on comparisons involving VRSs. Despite the vast body of papers, few articles recommend the use of one scale over the other. Furthermore, the use of terms is often ambiguous. For the purpose of the present paper, we consistently use the abbreviations and terms outlined in the Appendix.
of cancer pain assessment agreed about the top three dimensions to include in a multidimensional assessment of cancer pain: intensity, temporal pattern, and treatment-related factors (exacerbation/pain relief). This is in line with other reports.
Pain measurement tools and methods in clinical research in palliative care: recommendations of an Expert Working Group of the European Association of Palliative Care.
Pain measurement tools and methods in clinical research in palliative care: recommendations of an Expert Working Group of the European Association of Palliative Care.
Kaasa S, Apolone G, Klepstad P, Loge JH, Hjermstad MJ. Consensus expert conference on cancer pain and classification, the need for international consensus: working proposals on international standards. J Clin Oncol, in press.
are recommended for more comprehensive pain assessment. At present, there is no consensus concerning the terminology for temporal factors/breakthrough pain.
Pain measurement tools and methods in clinical research in palliative care: recommendations of an Expert Working Group of the European Association of Palliative Care.
Pain measurement tools and methods in clinical research in palliative care: recommendations of an Expert Working Group of the European Association of Palliative Care.
The first version, primarily focusing on pain, was used in the EPCRC-Computerized Symptom Assessment of pain, depression, and cachexia, an international data collection study including more than 1000 advanced cancer patients (www.epcrc.org).
toward the development of the computerized tool. We have reviewed studies with a specifically stated objective of comparing the use of the NRS, VRS, and/or VAS for unidimensional self-report of PI, in cognitively intact adults. Because of the widespread use of the NRS for the assessment of PI in many disease groups and the fact that it constitutes a major part of more comprehensive assessment tools, specific focus was put on this scale. The major study aim, therefore, was to examine the results from comparative studies on unidimensional assessment of PI using the NRS, VRS, or VAS. The following points were investigated:
•
What was the objective of comparing scales, and which scales were most frequently compared?
•
Did compliance and usability differ between scales?
•
Were different modes of scale administration compared, that is, plastic rulers, computers?
•
Did the number of response options, verbal anchor descriptors, and time frames vary?
•
What kind of statistics was used to report the results?
•
Were patients’ preferences for scales examined?
•
Did the results from cancer patients differ from results in other patient groups?
•
Were any of the scales recommended over the other(s) for research purposes and/or clinical use, and if so, why?
Methods
The literature search was performed in the following databases; MEDLINE (1950–2010, May week 2), PsycINFO (1806–2010, May week 3), and EMBASE (1980–2010 week 20) through OvidSP, and the Social Science Citation Index in Web of Science (1956–2010 update May 22) through ISI Web of Knowledge. Search-term groups representing 1) the NRS/VRS/VAS, 2) evaluation (including assessment and measurement), validation, comparison, clinimetry (including the clinimetric filter for PubMed/MEDLINE of Terwee et al.
), and 3) pain, were applied in all the combinations and adaptations according to the specific database and search engine requirements. Two limitations—“adults” and “English”—were used. The detailed search profiles can be obtained from the corresponding author on request.
All abstracts were read, and papers were selected for further reading if the abstract contained any information related to explicit comparisons of all or any two of the NRSs, VRSs, and VASs for assessment of PI. For inclusion in the present report, the publication had to meet the following criterion: A study with a specifically stated primary or secondary objective of comparing NRS/VRS/VAS for self-report of PI in adults.
Thus, case reports, editorials, letters, commentaries, reviews, and overviews were excluded, as were conference abstracts, and clinical studies simply using different scales for PI assessment, without aiming to compare the use and properties of scales. Pure validation studies of new tools or tools translated from the original language into another also were not included. Specific versions of the scales, that is, Faces Pain Scales (close to the VRSs) and the box variant with horizontal or vertical boxes for each value of the NRS, were not included, nor were studies comparing two types of the same scale, for example, the pen and paper version vs. the plastic version of the VAS, unless also comparing them to another scale (NRS/VRS).
In line with the study objectives and because of the plethora of pain assessment tools available, only the NRSs, VRSs, and VASs used for unidimensional assessment of PI in the included studies are described in detail; other pain tools are listed only in the tables.
The review process was conducted in two steps. First, two independent raters (M. J. H. and I. B.) examined all abstracts according to the eligibility criteria, consulting the full-text papers if in doubt about inclusion. In cases of uncertainty, a third independent classification was performed by a third reviewer (D. F. H.) and subsequently discussed. Second, all full-text papers of the selected abstracts were read to finally decide about inclusion. The Related Articles function in PubMed and the reference lists of the included papers were examined for additional relevant publications meeting the inclusion criteria.
Results
The searches produced 359 hits (MEDLINE 208, Embase 89, PsycINFO 30, SSCI 32) of which 120 were duplicates. After screening the 239 abstracts, 59 were retained for further reading. The main reason for noninclusion at this stage (69%, 125/180) was that comparing scales for unidimensional assessment of PI was not a specifically stated study objective (Fig. 1). Reading of the 59 full-text articles resulted in another 13 papers being excluded, whereas eight additional papers were identified from the reference lists and/or the Related Articles function. Fifty-four papers were finally retained (Fig. 1).
Fig. 1Flowchart of the literature review and selection of papers.
Country of origin showed a wide spread: 13 papers (24%) were from the United States, six from the United Kingdom, three from Australia/New Zealand and Canada, respectively, two from Africa, one from Mexico, and one from China. The 28 remaining papers were from European countries other than the United Kingdom, including 12 from the Nordic countries. The majority, 35 (65%), were published in 2000 or later.
Objectives of Comparing Scales and Study Samples
Most of the 54 studies compared different pain rating scales to find the most applicable scale for clinical use in a given population, as reflected in the samples studied (Table 1). Thirteen studies evaluated postoperative PI,
How patients and physicians rate patients’ pain in a French emergency department using a verbally administered numerical rating scale and a visual analog scale.
Evaluation of the Iowa pain thermometer and other selected pain intensity scales in younger and older adult cohorts using controlled clinical pain: a preliminary study.
Pain intensity assessment in older adults: use of experimental pain to compare psychometric properties and usability of selected pain scales with younger adults.
Compare relationship between scales, examine characteristics of noncompliant patients
Orthopedic surgery, second postoperative night
417
VAS-100 mm
VRS-5
Spearman rank
VAS and VRS scores highly correlated, but a wide range of VAS scores corresponding to each VRS category. Lower VAS completion rate with various impairments.
Compare NRS and VRS for breakthrough pain exacerbations
Advanced cancer patients
240
NRS-11
VRS-6
Percentage consistent ratings, weighted kappa
NRS higher discriminatory capability between background and peak PI, lower proportion of inconsistent ratings, higher reproducibility in PI exacerbations.
NRS preferred, due to higher discriminatory capability and reproducibility.
Compare pain and mathematical equivalence, examine nurses’ responses to ratings
Cancer inpatients
50
NRS-6
VAS-100 mm
Lower VAS ratings than NRS. >3/4 ratings not mathematically equivalent. Nurses provided with fictitious patient scenarios did not provide the same pain medication for equivalent ratings.
No preference. Research into interpretation of scales necessary.
Explore patient preferences for scales, validation
Rheumatoid arthritis outpatients
113
VAS-100 mm, ruler
VRS-5
Pearson’s, Spearman rank, χ2, ICC
High correlation of scales (>0.79). 53% preferred the VRS, 28% the VAS, 19% had no preference. VRS viewed as easier to understand. Patients with lower education (<6 years) preferred the VRS.
No preference. Both scales valid, choice to be based on setting, clinical goal, level of education.
Recommend the best method for assessing PI in the ED
ED convenience sample
1176
VNRS-11
VAS-100 mm
VAS-100 mm, ruler (VASp)
Bland-Altman, ICC
High correlation VNRS/VAS (0.88) and the VASp/VAS (0.92). VASp is probably valid to estimate acute pain, VNRS seems less valid due to wide limits of agreement and variable bias (mainly lower scores).
Compare pain scales before and after afferent stimulation/placebo
Acute orofacial pain patients, EU
80
NRS-101
VAS-100 mm
VRS-6
+2 Modified NRS/VAS+pain relief scales
Pearson’s, logistic regression, ANOVA
Except for VRS, significant correlation between scales at both assessments, good reliability. VRS changes did not correspond to equally large changes on the other scales.
NRS was the preferred scale by patients, also showed low error rates, higher face, convergent, divergent, and criterion validity regardless of age. VAS difficult in the elderly.
NRS preferred, as properties were not age related.
Patients’ ability to use different tools varied. Assessments correlated with each other and with nurses’ estimations. Nurses both underestimated and overestimated patients’ pain. Patients’ verbal pain assessments varied widely.
No preference, not totally clear whether pain tools are usable in the recovery room; further research necessary.
Phase 1: Higher correlation between tools when using same verbal anchors; Phase 2: VDS preferred overall, but had higher failure rates. VAS vertical preferred to VAS horizontal. All tools appropriate.
VAS may be preferred in research due to better sensitivity. Patients' preferences important in the clinic.
Pain intensity assessment in older adults: use of experimental pain to compare psychometric properties and usability of selected pain scales with younger adults.
All scales psychometrically sound, effective in discriminating different levels of pain. VDS was most sensitive and reliable in older. Low failure rates, except for the VAS. NRS preferred by patients.
VDS preferred, due to psychometric properties and patients’ preference.
Evaluation of the Iowa pain thermometer and other selected pain intensity scales in younger and older adult cohorts using controlled clinical pain: a preliminary study.
RR of failure to respond, χ2, Poisson regression, GLM method for scale sensitivity
The IPT lowest failure rate, highest for the VNS and the VAS. Cognitive impairment significantly related to failure on VAS/NRS. All scales sensitive for PI changes. IPT, followed by the FPS most preferred by patients.
The VAS and VNRS highly correlated, but cannot be used interchangeably. Large differences between VNRS/VAS in paired observations, significantly higher scores on the VNRS.
Unidimensional PI scores mainly reflect sensory pain dimensions, supporting the discriminant validity of the NRS/VAS. Separate scales should be used to rate PI and emotions.
Compare PI measures on selected criteria; correct response, relationship between scores
Chronic pain
75
NRS-101
VAS-100 mm
VRS-4
VRS-5
+Behavioral Rating Scale (BRS-6)+Box-11
χ2, correlation coefficients, principal factor analyses
High correlation across scales, similar rate of correct responses and utility, similar predictive validity. NRS easier to use and offers more response options.
NRS-101 may be preferred based on ease of use, sensitivity, and applicability across age group. All scales useful.
Compare the relative sensitivity of three outcome measures and one composite measure for pain relief in two RCTs
Postoperative pain
247
VAS-100 mm
VRS-4
+VRS relief
ANOVA, F scores
Variability in the sensitivity of the pain ratings, VAS better than VRS. Pain relief was related yet distinct from changes in PI. The composite score did not increase the sensitivity of the pain assessment.
No preference, choice to be based on the specific dimension that relates to treatment.
Evaluate fluctuation of symptoms, compare symptom scales
Cancer, hospice inpatients
125
VAS-100 mm
VRS-4
χ2, ANOVA, regression analyses
Significant differences between VRS categories and corresponding VAS scores, but overlap too wide to accurately assign cut-off points. VAS less reliable.
VRS may be better due to large variability of VAS.
Evaluate the quality of the intraindividual pain assessment and the equivalency of scale cut-offs
Musculoskeletal pain, outpatients
80
VAS-100 mm
VRS-5
Pair-wise comparisons, coefficient of monotonic agreement
VAS/VRS not to be used interchangeably, low intrascale agreement, the meaning of the rated PI dependent on pain etiology. Probable underestimation of PI when the VAS was categorized. Overlapping VAS scores within each VRS category.
VRS may be preferred, but pain etiology should be considered.
How patients and physicians rate patients’ pain in a French emergency department using a verbally administered numerical rating scale and a visual analog scale.
All scales valid, but more mistakes with increasing age, most on the VAS. Box scales most preferred, the VDS in the older. In mixed population, box scale is the method of choice.
PI, scale concordance, statistical modeling for research
Long-term undefined pain, prior to body awareness course
43
NRS-7
VAS-100 mm
Statistical modeling of distributions of paired assessments (details in paper)
A certain point on the VAS did not relate to a numerically labeled PI on the NRS. Continuous VAS/NRS offer a false impression of reliable measures expressed in millimeters or numerals.
VRS with clearly described response categories preferred for research.
Sample size varied from 12 to 1387. In 32 studies, repeated pain assessments were performed, in addition to one study with repeated assessment in a subsample only for test-retest purposes.
The different scales were presented to the patients in random order in 25 studies (one of which also used fixed order in a subsample for test-retest purposes) and in fixed order in 16; the order of presentation was not specified in 13 studies.
Overall, the VAS was by far the most frequently used scale. A total of 59 VASs were administered in 52 of the 54 studies, relative to 39 VRSs in 37 studies. In seven studies,
Pain intensity assessment in older adults: use of experimental pain to compare psychometric properties and usability of selected pain scales with younger adults.
a vertical version of the VAS was used, together with the traditional horizontal VAS in four of these. Traditional NRSs were included in 32 studies (33 NRSs), whereas the verbal version, the VNRS (see Appendix), was used in another five (eight VNRSs), plus in two of those that also used the NRS,
Evaluation of the Iowa pain thermometer and other selected pain intensity scales in younger and older adult cohorts using controlled clinical pain: a preliminary study.
Pain intensity assessment in older adults: use of experimental pain to compare psychometric properties and usability of selected pain scales with younger adults.
The NRSs/VNRSs were compared with the VAS in 16 studies, with the VRS in two, and with both the VAS and VRS in 18 studies; the VAS and VRS were compared in the remaining 18 studies. As indicated in Table 1, several other assessment tools for pain or other symptoms also were included in some studies.
The nomenclature used by authors was consistent for the NRSs/VNRSs, with one exception: the acronym VNS used for a VNRS in one paper. Full consistency was found for the VAS scales, although specific acronyms were used for the plastic or mechanical devices substituting for the traditional paper VAS in some studies. For the VRS scales however, four different abbreviations were used; Verbal Pain Scale, Verbal Descriptor Scale, Simple Descriptor Scale, and Graphic Rating Scale. Some of these variants had a number connected to each verbal descriptor. For consistency, Table 1 uses only the standard terms NRS, VRS, and VAS.
Compliance and Usability
When reported, better compliance was reported for the NRSs/VRSs relative to the other scales in 15 studies, whereas 16 studies did not provide any such information. Lower compliance on the VAS was found in nine studies, associated with higher age, degree of trauma, or other impairments. Compliance results were based on the number of patients who were able to perform the ratings, the number of correct answers, and error rates percentages. In some studies, test/retest scores and discriminant validity between patient groups also were used to indicate compliance.
Different Modes of Administration
Six studies used a plastic or mechanical VAS version with a moveable cursor along a line, with anchors at the extremes only, as a substitute for the traditional paper VAS.
concluded that the two versions correlated highly, and that the mechanical version seemed easier and more practical in the emergency room when compared with the VNRS.
used a mechanical VAS that also provided an option for judgment of ratios of perceived PI, which was regarded as a feasible method for research and clinical work.
One study compared the intraindividual variation in repeated scores on traditional tools (NRS/VAS) with electric skin stimulation as the matching stimulus,
and concluded that none of the methods demonstrated systematic disagreement. None of the identified studies aimed to compare electronic or web-based appliances (handheld devices, laptop computers, cell phones, etc.) with traditional paper and pencil versions for PI assessment.
Response Options, Anchor Descriptors, and Time Frames
The NRS-11/VNRS-11 was most frequently used (n=26), but six other versions also were used: NRS-6,
Evaluation of the Iowa pain thermometer and other selected pain intensity scales in younger and older adult cohorts using controlled clinical pain: a preliminary study.
Pain intensity assessment in older adults: use of experimental pain to compare psychometric properties and usability of selected pain scales with younger adults.
allowed the patients to give their score as half integers on the VNRS-11, which may then be regarded as a 21-point scale. One study used two NRSs (0–10 and 0–100) that were erroneously labeled as NRS-10 and NRS-100,
Five different versions of VRS answer categories were used; 12 used a VRS with four response categories (VRS-4), 15 used a VRS-5, seven used a VRS-6, three used a VRS-7, and one used a VRS-11; one study used a 15-category version
where the patients were asked to assign their own verbal descriptors to the numbers between the two anchors “none” and “severe.” All studies using VAS scales used the VAS-100 mm version, also labeled as VAS-10 cm.
As shown in Table 2, the descriptors used for the extremes varied, with 24 different adjectives being used. “No Pain” and “Worst Pain Imaginable” were most frequently used; the terminology was not given in five studies. Twenty-two studies used the same verbal anchors for all scales being compared (three scales or more in nine studies, two scales in 13 studies), and 14 studies used different descriptors for all scales being compared. Among the studies that used the same labels for two of three or more scales, the VRSs were most often labeled differently.
Table 2Overview of Anchor Labels Used with the NRS/VNRS, VAS, and VRS
The exact wording of the probe questions that were used for PI assessment was not reported in all papers, nor were the time spans being covered. However, 36 studies specifically asked for “current pain,” “present pain,” or “pain right now,” supplemented in seven studies with specified ratings of weakest, worst/strongest, recalled, anticipated, or average pain over different periods of time. Another seven papers did not specify the wording, but it was deduced from the objectives and patient samples that current pain was being evaluated. Other formats were PI at rest and when moving/coughing (2); maximal pain last hour (1); worst pain ever experienced (1); and average pain last week (3), last 24 hours (2), last night (1), and last month (1). One study supplemented the 24-hour PI rating with a rating of the most severe PI in the last 24 hours to specifically address pain exacerbations.
Descriptive statistics were used in all the studies (not tabulated). Apart from the statistical modeling papers that used stochastic simulation techniques and other advanced statistics, the majority of studies used various forms of correlation statistics for comparing scale scores, inter-rater reliability, and evaluation of treatment effect, depending on primary study outcomes. However, confidence intervals for the differences between scales were rarely presented and intraclass correlation coefficients were used in five papers only. Most papers reported good correlation between scales (Table 1), particularly so between the NRS and VAS. In cases of discrepancy, the NRS scores were higher than the equivalent VAS scores, particularly so for the verbal NRSs.
Some studies reported a marked variation between numerical and verbal scales, but in different directions. One study reported that VAS scores above 30 mm corresponded to moderate pain or above on the VRS-4, thereby including 85% of those reporting moderate pain,
Pain intensity assessment in older adults: use of experimental pain to compare psychometric properties and usability of selected pain scales with younger adults.
in rheumatoid arthritis (1), geriatric (2), chronic pain (2), and cancer (1) patients, respectively. All studies used a VAS and different VRSs, supplemented by the NRS-11, NRS-21, or NRS-101 in three studies, and by other scales (Table 1). Although patient preferences reflect the tools being used and the population under study, the VRS was preferred by the less educated
Pain intensity assessment in older adults: use of experimental pain to compare psychometric properties and usability of selected pain scales with younger adults.
and the Italian McGill Pain Questionnaire for multidimensional pain assessment, a scale for pain relief, and an integrated PI and duration measure in one.
Although correlations across scales were high in all studies, the recommendations for use in cancer differed. NRS-11 was recommended in three studies based on results and ease of use,
One study in hospice patients concluded that the VAS-100 mm showed no superiority over the VRS in assessing fluctuating symptoms, that there were significant differences between VRS categories and corresponding VAS scores, and the VRS was more appropriate for symptom assessment in those with advanced disease. On the basis of the few studies in cancer, it cannot be concluded that results or recommendations differ from those in other populations.
Study Recommendations
The majority of papers, 29, did not conclude with a preference for one tool over the other(s) (Table 1). Three papers recommended tools other than the NRS/VRS/VAS. The NRS was considered superior in 11 studies,
How patients and physicians rate patients’ pain in a French emergency department using a verbally administered numerical rating scale and a visual analog scale.
primarily because of ease of use and high compliance, although some papers expressed a slight reservation, claiming that the tool may be best suited for a subset of the population only. Seven papers recommended using the VRS
Pain intensity assessment in older adults: use of experimental pain to compare psychometric properties and usability of selected pain scales with younger adults.
for ease of use, low age-dependent failure rates, superior psychometric properties, and better responsiveness to fluctuating symptoms, although depending on the pain etiology.
Few papers specifically stated if the preference was for clinical use, but based on the study objectives, this was likely to be the case in most papers. Two of the statistical modeling papers specifically recommended VRS for research.
The arguments were that although the reliability and power of the VAS and VRS made them equally useful for clinical use, the psychometric properties of the VRS were better for research purposes,
The level of PI at the initial assessment has been shown to be a significant predictor of the complexity of cancer pain management and the time needed to obtain stable pain control.
PI is probably the most clinically relevant dimension of the pain experience regardless of disease. The overarching importance of this domain was accentuated in the present review, in that 89% of the identified studies were performed in populations other than cancer patients. According to expert surveys and consensus conferences,
Pain measurement tools and methods in clinical research in palliative care: recommendations of an Expert Working Group of the European Association of Palliative Care.
Kaasa S, Apolone G, Klepstad P, Loge JH, Hjermstad MJ. Consensus expert conference on cancer pain and classification, the need for international consensus: working proposals on international standards. J Clin Oncol, in press.
PI should be assessed by unidimensional scales based on self-report. The importance of the latter was evidenced by the incongruence in some studies between patient and proxy ratings, with proxies underestimating high pain levels. Despite the apparent consensus on PI assessment, our review showed that PI is monitored by a wide variety of unidimensional scales. The differences were expressed by the number of response options, scales of variable lengths, different verbal descriptors, and the different time spans covered. Reviews also have shown that the development of new tools for various pain domains, including intensity, is a continuously ongoing process
The objectives in most of the reviewed papers were to find the most applicable scale in the population being studied, but only 25 papers concluded with a specific recommendation. This may be because the statistical methods and sample sizes were insufficient to detect significant differences. The use of correlation coefficients alone is misleading to decide whether one scale performs better than another,
and only a few studies provided sophisticated statistical methods. This is presumably because many of the studies were designed to test the applicability of the scale use, not psychometric performance. The latter is supported by the fact that most of the unidimensional scales performed reasonably well in all studies.
Although some studies examined the applicability of mechanical, ruler versions of the VASs, it was a little surprising that none of the identified studies compared computerized and paper versions of the different scales. The rapid development in handheld computer technology provides ample opportunities for self-report of symptoms in most settings and also has been shown as a feasible assessment method in patients with advanced cancer.
Advanced technology may increase the reliability of pain and symptom assessment, facilitate the transfer of information, and yield immediate scores that are readily available for clinical or research purposes. However, these methods do not enhance the validity and clinical utility of the assessments per se, which are dependent on the psychometric properties of the questions that are being presented to the patients.
The exact number of response options used in a scale is important. A scale with only two (e.g., pain/no pain) or three response options offers little opportunity for discrimination. Most of the reviewed papers used scale versions according to current recommendations, primarily NRS-11s, VRSs up to seven categories, and VAS-100 mm.
Pain measurement tools and methods in clinical research in palliative care: recommendations of an Expert Working Group of the European Association of Palliative Care.
An overall conclusion from the general measurement literature is that there is relatively little gain in precision with more than seven options and hardly any above nine.
The most frequently used scale in the reviewed studies was the VAS-100 mm, which is relatively seldom subject to variations in length. This scale potentially offers the greatest opportunities for discrimination, although in practice this is illusory if most respondents are unable to discriminate PI with precision beyond nine or 10 distinct levels. Only one study used a VRS with more than seven points. The NRS-11 has been shown to perform well for PI assessment over the central portion of the continuum (2–8)
and was the most used version of the NRSs. Four studies in this review used NRSs ranging from 0 to 100 (NRS-101). However, on inspection, these were actually presented as verbal scales, having the patient indicate a number between 0 and 100 rather than marking the appropriate number. Thus, there seems to be some ambiguity in recognizing the scales, calling for standardization. It also has been shown that patients tend to treat the NRS-101 scales as NRS-21 or NRS-11 scales.
but it should be noted that some VRS-6 scales may be scored as 0–2–4–6–8–10, thereby complying with the preference for the 0–10 scales.
As far as we know, no specific recommendations exist with respect to anchor labels, as evidenced by the 24 different descriptors used. Although most scales used “no pain” at the lower end, there were more variations at the upper extremes (Table 2), some directly implying a comparison with previous pain experiences (“worst pain experienced”). One study concluded that the anchor labels incorporated a range of personal values, not only a description of the PI domain.
Another study by the same author showed a lack of concordance between patients and of consistency within patients when they were completing VASs and NRSs by using their own descriptions and forced choices.
In our opinion, it seems likely that the labels influence the responses, maybe even more at the upper end of the scale than at the lower end, particularly so in different languages and cultures. However, to what extent and in which direction the actual scores are influenced, remains an empirical question that needs further investigation. Nevertheless, standardization with respect to anchor labels is warranted and the optimal description should be aimed for.
The compliance rates were surprisingly high in all studies reporting this, which may be viewed in context to the different settings. Most papers used the term compliance, regardless of the different statistical or arithmetic methods used to examine this, which actually shows an inconsistent use of concepts, as it covered both compliance and usability in most studies.
It is highly likely that responding to a verbal NRS by saying a number between 0 and 10 or using a plastic ruler held by a nurse in the emergency room yields close to 100% compliance, whereas completion of pen and paper scales in elderly cancer patients is more complicated. It also may be that a selection bias comes into play in these relatively well-controlled studies, with specific emphasis on prompting as many patients as possible to answer. Overall, however, better compliance was reported for the NRS relative to the other scales, whereas the VAS seemed to be more complicated with higher error rates, especially in the elderly or cognitively impaired, as documented previously.
In relation to this, it may be regarded as a study limitation that our results were not differentiated between study populations, for example, the elderly, the cognitively or physically impaired, etc. Pain assessment in cognitively impaired adults, however, implies challenges other than the ones related to the actual pain tools and their content, in relation to mode of administration, visual limitations, layout, print size, actual size, and format of the paper tool, the need to go through the scores with the patient in more detail than with the cognitively intact, and so forth. Thus, it was decided to limit the literature search to those who were cognitively intact, and tabulate specific results from the elderly in some of the studies as appropriate.
The majority of the reviewed papers showed relatively consistent findings with respect to the correlation between scales, and when assessed, most coefficients between changes in scores over time were high, indicating that the scales tended to measure variations in the same direction.
However, several authors pointed to the variation in NRS scores within each bracket of the VRS and reported that ratings were not mathematically equivalent, which was taken as an indication of low interchangeability between scales by some authors.
In addition, the expectation of obtaining direct equivalence between mathematically different scales may have been too optimistic. It is probably not realistic that patients provide equal values on scales with different layouts, response options, and anchor labels. The VRS pain assessment scales that are being considered in this article have response options chosen that are ordinal and generally assumed to have approximately equal intervals—although in the past this equal-interval assumption has rarely been tested and many statisticians would argue that ordinal methods of analysis should be applied.
The 54 papers included in the present review constituted only 23% of the papers originally identified by our search terms. The majority of the papers that were not included did not have a specific aim to compare scales. Thus, it was interesting to notice that several of these used both NRSs and VASs, and simply reported the mean values to conclude on the efficacy of analgesic treatment. It may be that some of these studies would have given additional information about the performance of the scales. Despite our thorough reading of abstracts and several articles from treatment studies, it can never be ruled out that we did not identify all relevant papers. However, the included papers cover a broad spectrum of studies comparing PI assessment tools, so we do not think that we failed to include important information on this subject.
Another limitation of this study is related to the heterogeneity of studies, samples, and variety of scales that may restrict the general relevance of our findings. Additionally, most papers based their conclusions and recommendations on descriptive or correlation statistics and were not designed to investigate the psychometric properties of the tools, which may be fundamentally different in, for example, chronic cancer pain vs. acute postoperative pain. However, the 11 studies specifically recommending the use of NRS were performed in different populations (acute, chronic, or cancer pain) and the recommendations were among others based on feasibility criteria, which are important features both for clinical use and research. The results from the studies in cancer patients did not differ from the other studies in any respect. Thus, the heterogeneity of the included studies describes very well this lack of standardization in PI assessment. PI as a dimension is paramount for all pain management and should follow a standardized assessment methodology, regardless of the patient population or whether it is a part of a multidimensional tool or is used as an unidimensional scale.
This means that the same methodology (same scale, wording, time frame, and format) should be applied when assessing pain over time in the same patient population. Some of the reviewed studies conclude that the choice of scale may depend on factors such as patients’ preferences and/or their level of cognitive functioning.
We only partly agree to this. In addition, it is obvious that certain population characteristics have to be considered, such as age, frailty, literacy level, and cognitive impairment. For example, the higher number of errors on the VAS with increasing age and other impairments makes this scale less applicable in the cognitively impaired, as documented in the literature.
Pain measurement tools and methods in clinical research in palliative care: recommendations of an Expert Working Group of the European Association of Palliative Care.
This is also in line with a recent letter based on a study comparing NRS and VRS emphasizing the need to be selective in the use of scales for clinical use.
However, because the psychometric properties largely depend on certain basic characteristics, the selection of scales is better guided by specific consensus-based recommendations rather than left to the judgment of the individual clinicians. Furthermore, a standardization and consensus-based recommendation on the use of scales will facilitate the interpretation of results from studies and make comparisons across studies possible. It also may be necessary to distinguish between PI assessments for clinical use vs. research. We have reason to believe that the recommendations identified in the present review were mostly intended for clinical purposes, because only three papers specifically presented recommendations for research.
It is important to remember that a complex pain experience requires a multidimensional assessment, in line with the general recommendations in cancer.
Pain measurement tools and methods in clinical research in palliative care: recommendations of an Expert Working Group of the European Association of Palliative Care.
Kaasa S, Apolone G, Klepstad P, Loge JH, Hjermstad MJ. Consensus expert conference on cancer pain and classification, the need for international consensus: working proposals on international standards. J Clin Oncol, in press.
However, directly combining PI scores with other measurements, such as pain interference scores, may be less relevant in clinical settings, as it may obscure the actual scores of each domain.
Fayers PM, Hjermstad MJ, Klepstad P, et al. The dimensionality of pain: palliative and chronic pain patients differ in their reports of pain intensity and interference. Pain 2011 Mar 31. [Epub ahead of print].
Pain measurement tools and methods in clinical research in palliative care: recommendations of an Expert Working Group of the European Association of Palliative Care.
and it has been questioned whether cancer patients with multiple symptoms are able to discriminate between pain and other factors that interfere with their functioning.
Can cancer patients assess the influence of pain on functions? A randomised, controlled study of the pain interference items in the Brief Pain Inventory.
Although cancer pain may differ from acute, postoperative, and chronic pain in many respects, the common feature of any pain, regardless of cause, is its subjective nature, which makes it necessary to assess patients’ pain perception in a standardized manner. In this respect, a promising initiative resulted from a consensus meeting on cancer pain assessment and classification in Milan in September 2009.
Kaasa S, Apolone G, Klepstad P, Loge JH, Hjermstad MJ. Consensus expert conference on cancer pain and classification, the need for international consensus: working proposals on international standards. J Clin Oncol, in press.
In relation to PI, the recommendation was that it should be measured by a 0–10 NRS with the standard endpoints “no pain” and “pain as bad as you can imagine,” with clinically meaningful time frames. This consensus can be supported by the present review. Although the recommendation above was put forward for PI assessment in cancer specifically, it may well be applied to other populations as well. Key factors to remember in this respect in relation to the patient population are level of cognitive function, which may make a verbal NRS the instrument of choice, and the appropriate time frame for monitoring changes in PI over time.
Kaasa S, Apolone G, Klepstad P, Loge JH, Hjermstad MJ. Consensus expert conference on cancer pain and classification, the need for international consensus: working proposals on international standards. J Clin Oncol, in press.
We think it is time to welcome all consensus-based approaches that aim to standardize and facilitate the assessment of the subjective pain experience to improve pain management and promote research.
In conclusion, the results show that NRS-11, VRS-7, or VAS all work quite well. Thus, it is reasonable to say that the most important choice is not the type of scale per se, but the conditions related to its use, which include: a standardized choice of anchor descriptors, methods of administration, time frames, information related to the use of scales, interpretation of cut-offs and clinical significance, and the use of appropriate outcome measures and statistics in clinical trials.
We believe that all these areas can be improved by an international consensus process based on the evidence, which, in our opinion, should include, at least as a first step, perfecting and standardizing the use of NRS-11.
Disclosures and Acknowledgments
The EPCRC is funded by the European Commission’s Sixth Framework Programme (contract no LSHC-CT-2006-037777) with the overall aim to improve treatment of pain, depression, and fatigue through translation research. Core scientific group/work package leaders: Stein Kaasa (project coordinator), Frank Skorpen, Marianne Jensen Hjermstad, and Jon Håvard Loge, Norwegian University of Science and Technology (NTNU); Geoffrey Hanks, University of Bristol; Augusto Caraceni and Franco De Conno, Fondazione IRCCS Istituto Nazionale dei Tumori, Milan; Irene Higginson, King’s College London; Florian Strasser, Cantonal Hospital St. Gallen; Lukas Radbruch, RWTH Aachen University; Kenneth Fearon, University of Edinburgh; Hellmut Samonigg, Medical University of Graz; Ketil Bø, Trollhetta AS, Norway; Irene Rech-Weichselbraun, Bender MedSystems GmbH, Austria; Odd Erik Gundersen, Verdande Technology AS, Norway. Scientific advisory group: Neil Aaronson, The Netherlands Cancer Institute; Vickie Baracos and Robin Fainsinger, University of Alberta; Patrick C. Stone, St. George’s University of London; Mari Lloyd-Williams, University of Liverpool. Project management: Stein Kaasa, Ola Dale, and Dagny F. Haugen, NTNU.
Authors have no conflicts of interest.
Special thanks to librarian Ingrid Riphagen at the Medical Library, NTNU, Trondheim, for the performance of efficient and precise literature searches, and to Irmelin Bergh (I. B.) at the Regional Center for Excellence in Palliative Care, Oslo University Hospital, for examination of abstracts.
Appendix.
Tabled
1Descriptions and Abbreviations for NRS, VRS, and VAS Used in the Article
NRS
Numerical rating scale, commonly from 0 to 10 (NRS-11) or 1 to 10 (NRS-10). Usually, only the two extreme categories are labeled, for example, “No pain at all” and “Worst imaginable pain.” NRS may be called a VNRS/VNS when the scale is explained or shown on paper to the patient, who responds by indicating a number.
VRS
Verbal rating scale. Ordered categorical scale, with each response option consisting of adjectives. For different levels of PI, “no pain,” “mild pain,” “moderate pain,” “severe pain,” “extreme pain,” and the “most intense pain imaginable” form a six-category VRS scale (VRS-6). VRS scales are commonly of lengths four to seven. The adjectives are scored by assigning numbers (0–6) to each response option. The scale also may be called VPS (Verbal Pain Scale), VDS (Verbal Descriptor Scale), or SDS (Simple Descriptor Scale).
VAS
Visual analogue scale, usually 0–100, a straight line with the extreme categories labeled as for NRS. The distance measured from the “No pain” end to the patient’s mark is the VAS score. Usually graduated with labeled marks indicating tens (10, 20, 30, etc.) and sometimes unlabeled marks for the units.
Pain measurement tools and methods in clinical research in palliative care: recommendations of an Expert Working Group of the European Association of Palliative Care.
Kaasa S, Apolone G, Klepstad P, Loge JH, Hjermstad MJ. Consensus expert conference on cancer pain and classification, the need for international consensus: working proposals on international standards. J Clin Oncol, in press.
How patients and physicians rate patients’ pain in a French emergency department using a verbally administered numerical rating scale and a visual analog scale.
00014-5&id=gr1.jpg){kind=link}