Article Text

Download PDFPDF

The end of the line? The Visual Analogue Scale and Verbal Numerical Rating Scale as pain assessment tools in the emergency department
  1. Helen Mohan1,
  2. John Ryan2,
  3. Brendan Whelan3,
  4. Abel Wakai2
  1. 1UCD School of Medicine and Medical Science, Belfield, Dublin, Ireland
  2. 2Emergency Department, St. Vincent's University Hospital, Elm Park, Dublin, Ireland
  3. 3Statistics and Survey Design, TILDA Project, Trinity College, Dublin, Ireland
  1. Correspondence to Mr. John Ryan, Emergency Department, St Vincent's University Hospital, Elm Park, Dublin 4, Ireland; j.ryan{at}


Objectives To compare the Visual Analogue Scale (VAS) and the Verbal Numerical Rating Scale (VNRS), in the assessment of acute pain in the emergency department (ED). Furthermore, to determine the influence of demographics on this agreement and practical limitations of the scales.

Setting St Vincent's University Hospital, Dublin; a 479-bed teaching hospital; annual ED census 36 000 adult patients.

Methods A prospective observational study was conducted on ED patients with acute pain as a component of their presenting complaint. Eligible patients scored their pain on both VAS and VNRS within 1 hour of arrival. They rescored their pain every 30 minutes for 2 hours using both scales. The primary outcome measure was agreement between VAS and VNRS. Secondary outcomes were ease of pain scale use and effect of patient demographics on pain scores. Agreement between scores was evaluated using the Bland-Altman method.

Results 123 patients were included (median age 35; 43.9% male). There was a strong correlation between VAS and VNRS (rs=0.93). However, there was not perfect agreement between the two scales. Patient age (older age, p<0.005, t=−4.448), gender (female sex, p<0.005, t=4.903) and educational level attained (third level education, p<0.005, t=5.575) had a statistically significant influence on the agreement between VAS and VNRS. There was a preference for VNRS in those patients who expressed a preference for one pain scale over the other.

Conclusions VAS and VNRS are not interchangeable in assessing an individual patient's pain over time in the ED setting. VNRS has practical advantages over VAS in this setting.

  • Pain assessment
  • verbal numerical response scale
  • visual analogue scale
  • pain measurement
  • pain measurement

Statistics from


Although over 60% of emergency department (ED) patients present with conditions associated with pain, oligoanalgesia is a problem in emergency medicine.1–7 Use of pain scales has been shown to improve the frequency of analgesic administration in EDs.8 9 The Visual Analogue Pain Scale (VAS) and the Verbal Numeric Rating Scale (VNRS) are commonly used pain assessment scales.

VAS has been well validated and is often used in the research setting.10–14 However, it is not commonly used in day-to-day clinical practice in the ED. It uses a 100-mm line with endpoint descriptors such as ‘no pain’ marked at the left-hand end of the line and ‘worst pain imaginable’ marked at the right-hand end. The patient is asked to mark a point on the line that best represents their pain. VAS score is distance from ‘no pain’ to the patient's mark measured in millimetres (0–100). Previous studies have demonstrated that the minimum clinical significant difference (MCSD) in VAS lies in the range of 0.9–1.3 cm (95% CI 0.6 to 1.7).15–18 An advantage of VAS is that the wording can be written in different languages. Disadvantages of VAS are that it is time-consuming, specific equipment is needed and some patients may have difficulty understanding or performing this score.

VNRS is used frequently in EDs to assess pain and the effect of pain management. The patient rates pain verbally on a numerical scale from 0 to 10. It has the advantage of being quick and simple, requires no equipment and does not depend on intact motor skills. A disadvantage of VNRS is difficulty overcoming language barriers. Despite its regular ED use, the scientific validity of VNRS compared with VAS has undergone limited ED investigation.19–22

There were two specific aims of this study. First, to test agreement between VAS and VNRS in the ED setting. Second, to determine whether the agreement between the two pain scales was altered by patient demographics. In addition, practical limitations of the scales and patient utility data were investigated and the MCSD for VNRS calculated.

Materials and methods

Study design

This was a single-centre prospective observational study of a convenience sample of patients presenting to the ED. Patients triaged with any complaint associated with pain were approached within their first hour of attending the ED. Those who met the inclusion criteria were invited to participate in the study.


The study was conducted from July to August (4 weeks) 2006 in the ED of St Vincent's University Hospital, Dublin, Ireland. This is an urban university hospital with an average annual census of 36 000 adult patients, average admission rate 24.7%. The study was approved by the ethics committee of the hospital. Informed consent was obtained from all participants prior to enrolment in the study.


The study population consisted of a convenience sample of English-speaking patients aged 18 years and over with acute pain as a component of their presenting complaint. Patients were excluded if they had insufficient English, altered mental status or were deemed by the examining clinician to be too acutely unwell to participate. Any patient who gave only one reading on the pain scales and subsequently left the ED was excluded.

Data collection

A single-use paper 100-mm VAS was used with the words ‘no pain’ on the far left and ‘worst pain imaginable’ on the far right. The patient was asked to mark with a pen where their pain was along the line. The VAS was drawn using AutoCAD 2004 and printed, because photocopying has been shown to alter the length of the VAS.23 Patients were asked to score their pain on the VNRS from 0 to 10 with 0 indicating ‘no pain’ and 10 indicating ‘worst pain imaginable’. Patients could give their score as half integers—that is, 0.5. Each patient was reassessed every 30 minutes for up to 2 hours after the initial administration of the study or until their departure from the ED if they left within 2 hours. The order of use of the scales by a participant was maintained for the remainder of that individual's assessments, scoring their pain on both scales every time. The order was alternated between every second participant. This procedure ensured that the samples using the two pain scales were stratified evenly across time periods.

When the patients were asked to re-score their pain they were also asked to use verbal descriptor categories: ‘much better’, ‘a bit better’, ‘about the same’, ‘a bit worse’ or ‘much worse’. Patients were not permitted to view their previous pain scores.

The influence of demographic features such as gender, age group, social class and educational level on agreement between the pain scales was examined. Practical difficulties using the scales were recorded as were patient preferences.

Statistical analysis

Data were entered into Microsoft Excel 2003 and then exported into SPSS 12.0 statistical software for analysis.

To facilitate comparison with the VAS, the VNRS was normalised to a 0–100 scale. This approach is the one adopted in a number of previous studies both in the ED and non-ED settings.19 20 24–26 It should be noted that there is disagreement in the literature regarding the measurement properties of the VNRS as illustrated in a paper by Breivik et al27 in contrast to that by Hartrick et al.26

Agreement between VAS and VNRS was evaluated using the Bland-Altman method.28 This has been used to compare VAS and numerical pain scales, including the VNRS, in numerous previous studies.19 20 24 25 Bland-Altman is the preferred method to determine agreement between two clinical measures. It involves plotting the mean of each pair of measurements against their difference. The 95% limits of agreement are calculated as two standard deviations around the overall mean difference. Two methods can be regarded as being in agreement and used interchangeably if this range is no greater than the MCSD.

Multiple regression was used to examine whether the differences between VAS and VNRS scores were influenced by patient characteristics including patient age, gender, pain type, educational level, nationality, social class, pain location, analgesic administration prior to presentation, triage category, pre-existing chronic pain or which scale was used first.

Outcomes of interest

The primary outcome measure was agreement between VAS and VNRS. Secondary outcomes were: patient utility and effect of patient demographics on pain scores.


Characteristics of study participants

The median age of the study participants was 35 (IQR 51-27). The majority of participants were Irish (80.5%). Twenty-six patients (21.1%) had pre-existing chronic pain (table 1).

Table 1

Demographics of study participants

A total of 531 paired measurements of VAS and VNRS were obtained from the 123 patients enrolled in the study. A paired measurement is both a VAS and VNRS score for a particular patient at a particular point in time. Full datasets were obtained on 84 patients (68.3%). At least three paired measurements were obtained on 110 patients (89.4%). Four patients declined to participate, seven were deemed too unwell and eight had language difficulties that excluded them from the study. One patient withdrew consent to participate further in the study after three readings were obtained. One patient was excluded as he scored his pain as 10/10 on the VNRS and 100 mm on the VAS on the first reading. He then increased his score to 11 and 12 on the VNRS on subsequent readings and extended the VAS line.

Trauma was the source of pain in 44.7% (n=55). Location of pain was limb 50.4%, abdomen 24.4%, chest 8.9%, headache 5.7%, orofacial 4.9%, back 4.1% and testicular 1.6%.

Correlation and agreement between VAS and VNRS

Correlation between VAS and VNRS for all observations was significant (rs=0.93, 95% CI 0.92 to 0.94) (figure 1). The correlation of the first observations for each of the 123 patients was also significant (rs=0.89, 95% CI 0.87 to 0.91).

Figure 1

Scatterplot of Visual Analogue Scale (VAS) versus Verbal Numerical Rating Scale (VNRS).

The Bland-Altman plot was used to demonstrate the variability in differences between VAS and VNRS (figure 2). The difference between VAS and VNRS medians was −5 mm. The difference between VAS and VNRS means was −7.1 mm (95% CI −7.9 to −6.2, p<0.005) As 95% of the differences between VAS and VNRS scores lay between −26 and +12 mm, this range of 38 mm was outside the acceptable MCSD, as recorded in the literature, of 13 mm.15 In all, 24.6% of differences fell outside this maximum MCSD.

Figure 2

Bland-Altman plot showing the mean difference (solid line) and 95% limits of agreement (broken line).

VNRS was equal to VAS in 45 of the paired observations, VNRS was greater than VAS in 398 of the paired observations and VNRS was less than VAS in 88 of the measurements (n=531). The difference between VAS and VNRS did not vary significantly according to which score was measured first (p=0.298).

Effect of demographics on correlation and agreement between VAS and VNRS

The difference between VAS and VNRS varied significantly with age, sex and whether university level education had been attained. Stepwise multiple regression determined that lack of university level education was associated with the largest difference between VAS and VNRS scores (p<0.005, t=5.575). Female sex (p<0.005, t=4.903) and increasing age (p<0.005, t=−4.448) were also associated with a large difference between VAS and VNRS.

Nationality also had a statistically significant effect on the variation in differences between VAS and VNRS, with a greater difference between VAS and VNRS if the patient was non-Irish. Multiple regression determined that pain due to trauma versus non-trauma did not exert a statistically significant difference between VAS and VNRS scores (t=0.102 p=0.919). Similarly, location of pain—for example, abdominal or limb, did not exert a statistically significant difference.

The correlation between scores increased slightly on subsequent measurements for the same patient; rs=0.89 at the first measurement (CI 0.85 to 0.92) and rs=0.95 at the fifth measurement (CI 0.92 to 0.97). This may indicate a small learning effect (table 2).

Table 2

Spearman rank correlation coefficient for VAS and VNRS scores corresponding to time of measurement

Practical limitations of the scales and patient preference


Severe difficulty filling in VAS was a result of (a) fine motor problems or being immobilised (n=6), (b) visual problems (n=4) or (c) comprehension difficulty (n=6), totalling 16 patients (13%).

The number who had mild to moderate mechanical difficulty using VAS was not recorded—for example, dominant hand injury leading to use of non-dominant hand.


No patient in the study had comprehension difficulty with VNRS.

The majority of patients (59.4%) found that using a pain scale helped them describe the severity of their pain. Twenty-two patients (17.89%) found VAS easier, 43 patients (34.96%) found VNRS easier to use. Fifty-eight patients (47.15%) said there was no difference; 47.2% of patients said they would expect pain relief while 52.8% said they did not expect to receive pain relief in the ED. There was no significant correlation between VAS or VNRS scores and patient expectation of pain relief.

Minimum clinically significant difference

For VAS, the MCSD was found to be 11.9 mm (CI 9.7 to 14.9). For VNRS, the calculated MCSD was 1.25 (CI 1.05 to 1.45). Given that the VNRS goes up in a minimum of half integers this suggests that the minimum clinically significant difference for the VNRS is 1. Patient demographics were not found to influence the MCSD.


Appropriate pain assessment, including use of pain scales, has been shown to improve analgesic use in the ED.1 2 VAS or VNRS may be used to assess acute pain in the ED. While this study found an almost linear correlation between the VAS and VNRS, their use is not interchangeable for measuring pain change over time for an individual ED patient as they do not have perfect agreement.

For example, if a patient reported their pain as 7/10 on VNRS and on the next measurement uses VAS, indicating their pain as 84 mm, it is impossible to know if this indicates a true change in pain or if the change is artefactual owing to alternating methods of pain assessment. This was in keeping with the findings of Holdgate et al.19 Therefore it is best to use only one of the scales, rather than changing between them for the same patient.

The study demonstrated that patient age, gender and level of education significantly affected agreement between VAS and VRNS. In contrast, demographic factors were not found to influence individual VAS or VNRS measurements, or the MCSD. This suggests that self-reported pain severity does not differ significantly based on age, gender or cause of pain. Patient expectation of pain relief was not found to correlate with pain scores. The relatively low level of expectation of pain relief merits further study.

Patients tended to score their pain slightly higher on the VNRS than the VAS. This was in keeping with the findings of previous studies. It is an important factor in why there is not perfect agreement between the two pain scales. It may be because the VNRS can only increase in increments no smaller than 0.5/10.

The effect of nationality on the difference in pain scales is interesting. There are cultural components to pain and pain assessment. It is possible that the methodological approach used was more familiar to Irish patients.

VNRS is a practical scale for assessment and re-assessment of pain in the ED. In general, patients find it easier to use. It also requires less explanation to patients. Thirteen per cent of patients in this study were noted to have severe difficulty using VAS. Previous studies have estimated that as many as 15% of ED patients are unable to score their pain using VAS.27

Some patients find it difficult to use any pain scale to describe changes in their pain. Use of clinical judgment in identifying this subset of patients and modifying pain assessment to suit their requirements is essential.


Inter-observer bias was eliminated in this study as only one researcher collected data. However, it necessitated use of a convenience sample of patients, which is a source of potential bias.

Whether the VAS and VNRS can be compared using the Bland-Altman method necessitates two assumptions. One is that the VNRS can be normalised to a numerical magnitude ranging from 0 to 100. This has been widely used in the literature.19 20 24–26 The second assumption is that they incorporate the same level of measurement. There is debate in the literature on this point—that is, whether the VAS and VNRS are ratio, interval or ordinal scales in all circumstances.26 27 29 30 In keeping with a range of other studies, we have adopted the simpler approach of treating both scales as being measured at the same level.19 20 24 25 This may have implications for the interpretation of this study.


VAS or VNRS are useful methods to determine pain changes in the ED. VAS and VNRS cannot be used interchangeably for assessing pain changes over time for an individual ED patient, as the agreement between the scales is not perfect. VNRS is an acceptable alternative to VAS in the ED, with practical advantages and high patient acceptability.



  • Competing interests None.

  • Patient consent Obtained.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.