- Email: [email protected]
Comparative study of electronic vs. paper VAS ratings: a randomized, crossover trial using healthy volunteers
Pain 99 (2002) 341–347 www.elsevier.com/locate/pain
Comparative study of electronic vs. paper VAS ratings: a randomized, crossover trial using healthy volunteers Robert N. Jamison a,b,*, Richard H. Gracely c, Stephen A. Raymond d, Jonathan G. Levine e, Barbara Marino d, Timothy J. Herrmann a, Margaret Daly a, David Fram e, Nathaniel P. Katz a a
Department of Anesthesiology, Perioperative and Pain Medicine, Brigham and Women’s Hospital, Harvard Medical School, 75 Francis Street, Boston, MA 02115, USA b Department of Psychiatry, Brigham and Women’s Hospital, Harvard Medical School, 75 Francis Street, Boston, MA 02115, USA c National Institute of Dental and Craniofacial Research, National Institutes of Health, Bethesda, MD 20892, USA d PHT Corporation,500 Rutherford Avenue, Boston, MA 02129, USA e Lincoln Technologies, Inc., 315 Common Street, Belmont, MA 02478, USA Received 25 May 2001; received in revised form 26 March 2002; accepted 21 May 2002
Abstract The visual analogue scale (VAS) is an established, validated, self-report measure usually consisting of a 10 cm line on paper with verbal anchors labeling the ends. Palmtop computers (PTCs also known as personal digital appliances) have incorporated VAS entry by use of a touch screen. However, the validity and psychophysical properties of the electronic VAS have never been formally compared with the conventional paper VAS. The aim of this study is to determine the agreement between the electronic (eVAS) and paper (pVAS) modes. Twenty-four healthy volunteers were recruited for this study. Each study participant provided input using both measurement methods by marking the eVAS and pVAS in response to two kinds of stimuli, cognitive and sensory. A verbal rating scale of seven descriptors of intensity represented the cognitive stimuli. Participants were asked to mark the location that best corresponded to the pain intensity described by each word on scales from ‘no pain’ to ‘worst possible pain’. The sensory stimuli used were a set of test weights consisting of plastic containers ranging from 7 to 129 g. The VAS for sensory stimuli ranged from 0 (no weight) to ‘reference weight’ (the heaviest weight outside the range of test weights). There were two types of input stimuli and two modes for recording responses for a total of four experimental conditions. Two evaluators independently measured and recorded all the pVAS forms to the nearest millimeter. A total of 2016 stimuli were rated. The overall correlation for ratings of both sensory and cognitive stimuli on eVAS and pVAS was r ¼ 0:91. For paired verbal stimuli the correlation was r ¼ 0:97. For paired sensory stimuli the correlation was r ¼ 0:86. The correlation between group eVAS and pVAS ratings to common verbal stimuli was r ¼ 0:99. For common sensory stimuli the group correlation was r ¼ 0:99. The median of correlations comparing eVAS and pVAS ratings was 0.99 for verbal stimuli and 0.98 for sensory stimuli. Multivariate analyses showed equivalent stimuli to be rated much the same whether entered on paper VAS or PTC touch screen VAS (P , 0:0001). Support was found for the validity of the computer version of the VAS scale. q 2002 International Association for the Study of Pain. Published by Elsevier Science B.V. All rights reserved. Keywords: Visual analogue scales; Palmtop computers; Electronic data; Verbal rating scale; Pain
1. Introduction The visual analogue scale (VAS) is an established, validated, self-report measure of pain intensity usually consisting of a 10 cm line on paper with verbal anchors labeling the ends (Collins et al., 1997; Ferraz et al., 1990; Harms-Ringdahl et al., 1986; Price et al., 1983). VASs have been used in
* Corresponding author. Pain Management Center, Brigham and Women’s Hospital, 75 Francis Street, Boston, MA 02115, USA. Tel.: 11-617-278-1063; fax: 11-617-731-5453. E-mail address: [email protected] (R.N. Jamison).
many studies to measure a number of constructs including pain (Badley and Papageorgiou, 1989), asthma (Dhand et al., 1988), dyspepsia (Nyren et al., 1987), mood (Monk, 1989), appetite (Stubbs et al., 2000), ambulation (Welsh et al., 1993), and vitality (Wood et al., 1990). The VAS has been shown to have advantages over verbal rating scales and numerical scales in sensitivity to changes in pain intensity (Ohnhaus and Adler, 1975) and in the capacity to provide ratio scale measures of experimental pain (Price et al., 1994). They are particularly useful for populations with language barriers (Soh and Ang, 1992) and can be presented in various forms (Stephenson and Herman, 2000). They
0304-3959/02/$20.00 q 2002 International Association for the Study of Pain. Published by Elsevier Science B.V. All rights reserved. PII: s0 304-3959(02)00 178-1
342
R.N. Jamison et al. / Pain 99 (2002) 341–347
have also been adapted to computers with interactive animation (Swanston et al., 1993). Recent developments in hand-held computer technology have created the potential for collection of patient selfassessment data through electronic methods. Electronic diaries can facilitate more complete and timely collection of patient diary data than paper diaries (Jamison et al., 2001). Palmtop computers (PTC) have incorporated VAS entry by use of a touch screen. The PTC is capable of using an electronic visual analogue scale (eVAS) that is quite similar to the conventional paper-based VAS (pVAS). While the two versions of the VAS are similar, the eVAS differs from the pVAS in its shorter overall length and ease in moving the mark along the line without having to erase a former mark. Unlike the pVAS, the eVAS is marked by touching the line with a stylus. Clinical trials have used eVAS to obtain patient self-report data (Tiplady et al., 1995), but concerns remain that the validity and psychophysical properties of the eVAS have never been formally compared with the present pVAS standard. The aim of this study is to assess the validity of the eVAS and to determine whether patient input via eVAS is equivalent to input via pVAS. The intent was to use standard psychophysical techniques to evaluate the psychometric properties of self-report VASs.
2. Methods This is a single-center, randomized, crossover study using healthy volunteers. Twenty-four subjects were recruited for this study. The participants ranged in age from 19 to 57 (mean ¼ 34.4) and were mostly females (79.2%). Seventy nine percent were Caucasian, 17% African–American, and 4% Hispanic. Each study participant received 5 min of introductory training in the use of both the eVAS and the pVAS. The same test administrator using a written script provided the training in both methods. Experimental data collection for each study participant took place in a single session lasting approximately 1 h. Participants provided input by marking the eVAS and pVAS in response to two kinds of stimuli, cognitive and sensory. The cognitive stimuli were represented by a verbal rating scale of seven descriptors of intensity ranging from ‘faint’ to ‘extremely intense’ (Gracely et al., 1978). Participants were asked to mark the location that best corresponded to the pain inten-
sity described by each word on scales from ‘no pain’ to ‘worst possible pain’. Two evaluators independently measured and recorded all the pVAS forms to the nearest millimeter. The sensory stimuli used were a set of weights consisting of plastic canisters. Two canisters were marked as the reference weights (heavy weight and light weight), and seven groups of test canisters served as the study weights. The canisters were identical in size and weights increased progressively from 6.8 (empty canister), to 16.4, 25.4, 43.8, 61.7, 89.1, and 129.1 g, respectively. The reference weight weighed 165.2 g. A pilot trial demonstrated that these weights could be readily differentiated from each other when presented as pairs under conditions where the subjects were asked to identify the heavier weight of the pair. Replicates of the test canisters were produced so that each of the seven weights could be presented repeatedly three times without having to present the same actual test canister on more than one occasion. The canisters were filled with cotton (so the weights did not rattle) and marked with a random text code so that they could be distinguished by the experimenter without cueing the subjects. When assessing the heaviness of the weights, participants were asked to hold the weight with their thumb and index finger just below the cap (Flanagan et al., 1995). Participants were then asked to mark on a VAS the location that corresponded to the test weight in relation to the endpoints of the scales, which were anchored by ‘0’ and ‘Reference Weight’ (see Fig. 1). Before the start of each series of weights, subjects were requested to lift the reference heavy weight. 2.1. Device properties The eVAS was implemented on a Palm Pilot IIIxe, with a 6 cm square screen having a 160 £ 160 pixel resolution. The 5 cm VAS occupied 142 of the pixels. The bars marking each end were 0.5 cm high (14 pixels). The VAS marker was 4 mm high (10 pixels), and about 1 mm wide (3 pixels) and centered perpendicularly on the VAS line. The center pixel of the marker was set equal to the pixel identified as ‘touched’ by the palm touch screen. The eVAS capture program linearly converted the pixel touched to an integer between 0 and 100; so that each integer represented 1 or 2 pixels (,1 mm). In validation testing, the plastic stylus, which had a 1 mm diameter spherical tip, was used to drag
Fig. 1. Visual analogue scales.
R.N. Jamison et al. / Pain 99 (2002) 341–347
the marker from one end of the scale to the other and back repeatedly. Correspondingly, as the stylus was moved incrementally, all the integers from 0 to 100 could be caused to appear sequentially in an onscreen window that was set up during testing to display the automatic scoring. Repeated touches in the same spot could be done carefully enough to post identical scores 10 times out of 10, but more commonly the targeting skill of hand-manipulated touches to the screen would result in N ^ 1 variations in the number posted. To the eye, the resolution of markings on the paper could be resolved as well but not better than the marker on the eVAS. The pencil marks were approximately 1 mm wide, and manual measurements resolved the marks to the nearest millimeter ^ 1 mm. Both scales thus had similar resolution of one part in 100, which was sufficiently finely graded so that a user had the impression of a continuous scale.
2.2. Study design There were two categories of stimuli (sensory and verbal) and two modes for recording responses (paper and electronic), for a total of four experimental conditions. The two combinations involving a pVAS were designated A and B, corresponding to verbal and heaviness ratings. The two combinations involving an eVAS were designated a and b. Each volunteer completed all four experimental conditions in one of the following eight orders: ABab, ABba, BAba, BAab, abBA, abAB, baBA, or baAB. Within each stimulus and mode, each subject experienced each intensity level three times, resulting in 21 measurements per stimulus/mode combination. The 21 intensity levels within a stimulus/modality combination were presented in three sequences of seven intensity levels. Each sequence consisted of a randomized ordering of all seven intensities. Thus, there were 21 stimuli given in each of four conditions to the 24 subjects. Random assignment counterbalanced for the order of mode and stimulus was used to control for possible order effects. To facilitate administration by minimizing switching between paperbased and electronic modes, each participant switched only once (i.e. each was randomized to start with pVAS and then switched to eVAS, or to start with eVAS and then switched to pVAS). The length of the pVAS scale was 10 cm, and the length of the eVAS scale was 5 cm. The pVAS forms were printed (not xeroxed) to preserve dimensional accuracy. The eVAS forms were programmed on the PTC to resemble the pVAS forms. For the verbal test, the pVAS and eVAS forms displayed the stimulus words. For the heaviness test, the pVAS and eVAS forms displayed a coded number (which was meaningless to the subject) that corresponds to the weight canisters being given to the subject. There was a separate code for each canister to prevent the subjects from adjusting their responses to repeated codes.
343
2.3. Subjects Twenty-four healthy volunteers were recruited by advertisement and signed an informed consent form to participate in the study. Each subject was paid US $50 for his or her participation. The inclusion criteria were (1) age 18 or older at time of participation and (2) willingness and ability to use the PTC to record ratings. Exclusion criteria consisted of (1) an inability to differentiate between the lightest and heaviest weight, as determined during the training session, (2) marked inability to rank order the stimuli, or (3) an inability to comply with the protocol. None of the volunteers were excluded. 2.4. Subject training The purpose of the experiment, to compare two methods of recording measurements on a linear scale, was explained to each participant. Inclusion and exclusion criteria were reviewed, including verifying that the participant could correctly distinguish the heaviest and lightest test weights. Each participant was shown how to hold the weight with his or her thumb and index finger just below the cap. Each participant was assigned a sequential numeric code, which was used to determine the sequence of modes, stimulus categories, and stimulus levels to be used with that participant. Each participant used the pVAS and the eVAS in an order determined by the randomization plan. Participants were trained in each modality immediately before using that modality. In the training, the participant was shown how to mark the scale and was asked to try marking the scale four times corresponding to verbal inputs of ‘faint’ and ‘extremely intense’ and to weight inputs of the lightest and heaviest weights. The test administrator assisted the participant if the participant had difficulty marking the scale or reversed the sense of the scale (i.e. marked ‘faint’ above ‘extremely intense’, or the lighter weight above the heavier). 2.5. Test administration Tests were administered in the order defined by the randomization plan. Before each group of verbal rating tests, the participant was shown a list of the words, ordered from least to greatest pain level. Before each group of heaviness tests, the participant was asked to hold the reference weight using a consistent grip between thumb and index finder. During the actual test, the participant marked his/her assessment on a new pVAS sheet or a new eVAS display. Results of previous markings were not visible. The pVAS/ eVAS included the word to be rated or the coded label of the weight. The pVAS forms were collected by the test administrator and each sheet was marked with the participant’s sequence number. All the pVAS forms were evaluated by two evalua-
344
R.N. Jamison et al. / Pain 99 (2002) 341–347
tors, each of whom measured and recorded the VAS marking independently to the nearest millimeter. Each evaluator filled out a series of paper worksheets containing one row for each measurement. A single 100 mm metal ruler was used for all measurements. Each row contained the participant number, the stimulus type (verbal or weight), the coded or plain text stimulus value appearing on the form, and the result of measuring the VAS response with the ruler. The evaluators were instructed to use their best judgement in the case of ambiguous markings and to record missing markings as ‘not available’. The data from the evaluation sheets were keyed into spreadsheets, resulting in two independently determined results for each VAS marking. The pairs of results were compared, and cases of disagreement were noted. If the two values agreed within 1 mm, the two results were averaged. If, on the other hand, the two values were more than 1 mm apart, the evaluators were asked to redo the measurements in question. The data from the eVAS forms were collected and transmitted to a web server, using telephonic techniques similar to those used in clinical trials (Jamison et al., 2001). The recordings included the participant number, the stimulus identifier (stimulus type and level), and the measurement as captured by the PTC. Data from all participants were recorded and entered. To be included in the statistical evaluation, a participant must have provided at least one full replicate for each of the four modality/stimulus combinations. 2.6. Statistics For both cognitive and sensory stimuli, we sought to answer two questions: (1) are pVAS and eVAS ratings equivalent for each stimulus value? and (2) are pVAS and eVAS equally precise? Equivalence, defined as similarity between the average eVAS and pVAS ratings after each subject repeatedly rated the same stimulus on both eVAS and pVAS, was determined by tests of analyses of variance (ANOVAs) with and without repeated measures between the two modes for each stimulus. Precision pertains to the variability of the repeated ratings for each of the two rating modes. If the variability of the ratings in one rating method is less than the variability in the other rating method, the first rating method is considered more precise. If eVAS and pVAS are equivalent, then the more precise mode is preferable. In all statistical procedures, both the pVAS and eVAS values were scaled from 0 to 100. Based upon previous research (Jamison et al., 2001), the variance of the difference between eVAS and pVAS scores was estimated to be 118 (corresponding to a standard deviation of approximately 10.86) when the scores for both eVAS and pVAS ranged from 1 to 100 (this is equivalent to assuming that both eVAS and pVAS have a standard deviation of 24.29, and the correlation between eVAS and pVAS is 0.90). Therefore, with a sample size of 24, this study had 90% power to detect a difference in
mean eVAS and pVAS values of 2.22 or greater. For a particular stimulus, because there will be half the number of ratings, this study had 90% power to detect a difference in eVAS and pVAS of 5.87 or greater.
3. Results A total of 2016 stimuli were presented and rated (24 subjects £ 4 combinations of response mode and stimulus modality £ 7 stimuli £ 3 presentations of each stimulus). For each combination of response mode and stimulus modality (condition), the three responses to each stimulus were averaged to yield a mean response for each of the seven stimuli. One paper data point for a verbal rating was accidentally skipped and the mean for this subject and stimulus was computed from the remaining two observations. The overall correlation for ratings of both sensory and cognitive stimulus types was significant (r ¼ 0:91). For 503 paired verbal stimuli in the 24 subjects the correlation between paper and PTC ratings were highly significant (r ¼ 0:97; range 0.95–0.98), while for 504 paired sensory stimuli the ratings were lower but also significantly correlated (r ¼ 0:86; range 0.81–0.92). r 2 correlations were then computed between the eVAS and pVAS responses to common stimuli for each stimulus modality. The correlation between group eVAS and pVAS ratings to common verbal stimuli was r 2 ¼ 0:997. For common sensory stimuli the group correlation was r 2 ¼ 0:999. Scatterplots showing these realtionships are presented in Fig. 2a,b. These figures show that, for the group, scoring of equivalent stimuli was found to be highly correlated whether entered on paper 10 cm VAS or PTC touch-screen 5 cm VAS. Individual analyses were run for each subject by first computing the same correlations shown for the group in Fig. 2a,b. Scatter plots of these correlations are shown in Fig. 3. This figure shows general agreement for all but one subject and excellent agreement for all but six subjects. For all subjects, the median and the mean for the verbal stimuli were 0.99, and 0.99 (range 0.92–0.99) and for the sensory stimuli the median and mean were 0.98 and 0.96 (range 0.72–0.99). A one-way repeated measures ANOVA using stimulus intensity as a repeated factor with seven levels was run for both stimuli and modes. The results of multivariate analyses of the combined verbal and weight stimulus ratings with and without repeated measures in the model are presented in Table 1. The mode by stimuli interactions were found to be highly significant, although there was less agreement for the weight than for the verbal stimuli. Mean variances ratio tests were compared for eVAS and pVAS for the cognitive and sensory stimuli. The variance ratios ranged from 0.53 and 2.15 for the cognitive stimulus and from 1.62 and 0.22 for the sensory stimulus. None of the differences between pVAS and eVAS were shown to be significant for the seven cognitive stimulus, however, four
R.N. Jamison et al. / Pain 99 (2002) 341–347
Fig. 2. (a) Scatter plot of correlations between eVAS and pVAS to common verbal stimuli. (b) Scatter plot of correlations between eVAS and pVAS to common sensory stimuli.
345
variate analyses between electronic and paper measurements were very highly significant for both types of stimuli, cognitive (verbal pain intensity ratings) and sensory (weights). Since scales that are valid for measurement of various non-painful sensory modes are generally valid for measuring pain (Price et al., 1983,1994), it is reasonable to conclude that electronic VAS measurements are essentially equivalent to paper VAS measurements for pain as well. The variance ratios between electronic and paper VAS ratings were also similar, although for the weights, the variance of the electronic scores was less than the variance of the paper scores. It is intriguing to speculate why the subjects, who had more difficulty with weights than verbal ratings, were able to record their responses more precisely on the electronic medium. The inherent benefits with using an eVAS include less chance for ambiguous markings, errors in measurement, and skipped data. There also may be other properties of a PDA that may help to improve accuracy of responding that are yet to be determined. This finding is worthy of replication. We found differences between the two experiments, with the cognitive tests (verbal scale) showing higher levels of significance than the sensory tests (weights). On average, the subjects could discriminate among the seven stimuli for the verbal stimuli, but had greater difficulty judging among the weight stimuli. This was not surprising since it would be expected that the order of intensity would be clearer with the verbal ratings than with the subjects’ ratings of heaviness. Even though this protocol followed standard psychophysical testing design, the weights did not increase logarithmically, which would make the judgment among the various weights more difficult, as was especially evident among with lighter weights. Overall, however, subjects reliably rated the intensities of both stimuli in a linear manner.
of the seven of the sensory stimulus were significantly different (P , 0:05). Fig. 4 shows a simple measure of the variability of each method by showing the standard errors of the mean (the higher the F-value the lower the standard error). The eVAS generally shows similar standard errors for the verbal or sensory stimuli, while the standard errors of the pVAS for the sensory stimuli are about double the pVAS standard error to the verbal stimuli. Exit interviews of subjects indicated that the sensory discrimination of adjacent weights was a difficult task, which would be consistent with the greater variances and lower correlations found between paired ratings.
4. Discussion This study followed the design used in standard psychophysical techniques for evaluating the psychometric properties of a patient self-report scale (Gracely, 1988, 1989). The results of this study demonstrate that a VAS on a palmtop computer is remarkably similar to a VAS on paper. Multi-
Fig. 3. Comparison of correlations between eVAS and pVAS to common stimuli plotted for each subject.
346
R.N. Jamison et al. / Pain 99 (2002) 341–347
Table 1 One way analysis of variance with and without repeated measures for the verbal and weight stimuli observations a Stimulus
No. of observations
Degrees of freedom
r2
F
Repeated measures Verbal Weight
1007 1008
259 259
0.98 0.92
131.22 *** 33.34 ***
Without repeated measures Verbal Weight
1007 1008
197 197
0.97 0.90
155.32 *** 38.40 ***
a
***P , 0:0001.
Some individual differences among subjects were found that are worthy of discussion. For most subjects, the relationship between rating and response with the cognitive stimuli was essentially linear. For a few of the subjects, however, the relationship was binary (expressed as either high or low on both versions of the VAS), and for one subject the relationship was non-linear (suggesting that this person had trouble discriminating differences). These findings are in keeping with differences usually found in self-report pain ratings among pain patient populations (Jamison et al., 1998). Differences among the subjects were also noted for the weight stimuli. Again, most of the subjects revealed a linear relationship between rating and response, even though these data were more variable than the verbal data. Two subjects seemed to have particular trouble with the task. Again, this might be expected since some individuals possess better discrimination skills in assessing weights than others. Some subjects were also less consistent than others, which accounted for lower correlations even within the same modes (paper to paper r 2-values). This finding lends further support to the similarity among different assessment techniques. Although electronic versions of VAS scales have been used by pain patients in the past without any identifi-
Fig. 4. Standard errors of the mean for each mode (pVAS and eVAS) and stimuli (verbal and weight).
able difficulties (Jamison et al., 2001), further study with chronic pain patients could be helpful to establish the validity and reliability with these patients. In summary, electronic VAS scores were found to be highly correlated with paper VAS scores for both cognitive (verbal intensity) and sensory (weight) stimuli. No support was found for the hypothesis that mode of data collection (paper vs. electronic VAS) has any biasing effect on the data collected. Thus, self-reported ratings using electronic VAS do not differ from such ratings on paper. VAS data from electronic diaries are valid and hold intrinsic benefits for future clinical trials. Acknowledgements A portion of this study was presented at the 20th Scientific Meeting of the American Pain Society, Phoenix, AZ, April 21, 2000. The authors wish to express appreciation to Jamie Bell and Kelly Zou for their assistance, and Jaylyn Olivo who reviewed an earlier version of this manuscript. The study was funded in part by PHT Corporation. References Badley EM, Papageorgiou AC. Visual analogue scales as a measure of pain in arthritis: a study of overall pain and pain in individual joints at rest and on movement. J Rheumatol 1989;16:102–105. Collins SL, Moore RA, McQuay HJ. The visual analogue pain intensity scale: what is moderate pain in millimetres? Pain 1997;72:95–97. Dhand R, Kalra S, Malik SK. Use of visual analogue scales for assessment of the severity of asthma. Respiration 1988;54:255–262. Ferraz MB, Quaresma MR, Aquino LR, Atra E, Tugwell P, Goldsmith CH. Reliability of pain scales in the assessment of literate and illiterate patients with rheumatoid arthritis. J Rheumatol 1990;17:1022–1024. Flanagan JR, Wing AM, Allison S, Spenceley A. Effects of surface texture on weight perception when lifting objects with a precision grip. Percept Psychophys 1995;57:282–290. Gracely RH. Pain psychophysics. In: Chapman CR, Loeser JD, editors. Advances in pain research and therapy, New York, NY: Raven Press, 1989. Gracely RH, Lota L, Walter DJ, Dubner R. A multiple random staircase method of psychophysical pain assessment. Pain 1988;32:55–63. Gracely RH, McGrath F, Dubner R. Ratio scales of sensory and affective verbal pain descriptors. Pain 1978;5:5–18. Harms-Ringdahl K, Carlsson AM, Ekholm J, Raustorp A, Svensson T, Toresson HG. Pain assessment with different intensity scales in response to loading of joint structures. Pain 1986;27:401–411.
R.N. Jamison et al. / Pain 99 (2002) 341–347 Jamison RN, Raymond SA, Levine JG, Slawsby ES, Nedeljkovic SS, Katz NP. Electronic diaries for monitoring chronic pain: one-year validation study. Pain 2001;91:277–285. Jamison RN, Raymond SA, Slawsby ES, Nedeljkovic SS, Katz NP. Opioid therapy for chronic noncancer back pain. Spine 1998;23:2591–2600. Monk TH. A visual analogue scale technique to measure global vigor and affect. Psychiatry Res 1989;27:89–99. Nyren O, Adami HO, Bates S, Bergstrom R, Gustavsson S, Loof L, Sjoden PO. Self-rating of pain in nonulcer dyspepsia. A methodological study comparing a new fixed-point scale and the visual analogue scale. J Clin Gastroenterol 1987;9:408–414. Ohnhaus EE, Adler R. Methodological problems in the measurement of pain: a comparison between the verbal rating scale and the visual analogue scale. Pain 1975;1:379–384. Price DD, Bush FM, Long S, Harkins SW. A comparison of pain measurement characteristics of mechanical visual analogue and simple numerical rating scales. Pain 1994;56:217–226. Price DD, McGrath PA, Rafii A, Buckingham B. The validation of visual analogue scales as ratio scale measures for chronic and experimental pain. Pain 1983;17:45–56.
347
Soh G, Ang HG. Comparison of two pain rating scales among Chinese cancer patients. Chin Med J (Engl) 1992;105:953–956. Stephenson NL, Herman JA. Pain measurement: a comparison using horizontal and vertical visual analogue scales. Appl Nurs Res 2000;13:157– 158. Stubbs RJ, Hughes DA, Johnstone AM, Rowley E, Reid C, Elia M, Stratton R, Delargy H, King N, Blundell JE. The use of visual analogue scales to assess motivation to eat in human subjects: a review of their reliability and validity with an evaluation of new hand-held computerized systems for temporal tracking of appetite ratings. Br J Nutr 2000;84:405–415. Swanston M, Abraham C, Macrae WA, Walker A, Rushmer R, Elder L, Methven H. Pain assessment with interactive computer animation. Pain 1993;53:347–351. Tiplady B, Crompton GK, Brackenridge D. Electronic diaries for asthma. Br Med J 1995;310:1469. Welsh EM, Gettinby G, Nolan AM. Comparison of a visual analogue scale and a numerical rating scale for assessment of lameness, using sheep as a model. Am J Vet Res 1993;54:976–983. Wood C, Magnello ME, Jewell T. Measuring vitality. J R Soc Med 1990;83:486–489.
Comparative study of electronic vs. paper VAS ratings: a randomized, crossover trial using healthy volunteers Robert N. Jamison a,b,*, Richard H. Gracely c, Stephen A. Raymond d, Jonathan G. Levine e, Barbara Marino d, Timothy J. Herrmann a, Margaret Daly a, David Fram e, Nathaniel P. Katz a a
Department of Anesthesiology, Perioperative and Pain Medicine, Brigham and Women’s Hospital, Harvard Medical School, 75 Francis Street, Boston, MA 02115, USA b Department of Psychiatry, Brigham and Women’s Hospital, Harvard Medical School, 75 Francis Street, Boston, MA 02115, USA c National Institute of Dental and Craniofacial Research, National Institutes of Health, Bethesda, MD 20892, USA d PHT Corporation,500 Rutherford Avenue, Boston, MA 02129, USA e Lincoln Technologies, Inc., 315 Common Street, Belmont, MA 02478, USA Received 25 May 2001; received in revised form 26 March 2002; accepted 21 May 2002
Abstract The visual analogue scale (VAS) is an established, validated, self-report measure usually consisting of a 10 cm line on paper with verbal anchors labeling the ends. Palmtop computers (PTCs also known as personal digital appliances) have incorporated VAS entry by use of a touch screen. However, the validity and psychophysical properties of the electronic VAS have never been formally compared with the conventional paper VAS. The aim of this study is to determine the agreement between the electronic (eVAS) and paper (pVAS) modes. Twenty-four healthy volunteers were recruited for this study. Each study participant provided input using both measurement methods by marking the eVAS and pVAS in response to two kinds of stimuli, cognitive and sensory. A verbal rating scale of seven descriptors of intensity represented the cognitive stimuli. Participants were asked to mark the location that best corresponded to the pain intensity described by each word on scales from ‘no pain’ to ‘worst possible pain’. The sensory stimuli used were a set of test weights consisting of plastic containers ranging from 7 to 129 g. The VAS for sensory stimuli ranged from 0 (no weight) to ‘reference weight’ (the heaviest weight outside the range of test weights). There were two types of input stimuli and two modes for recording responses for a total of four experimental conditions. Two evaluators independently measured and recorded all the pVAS forms to the nearest millimeter. A total of 2016 stimuli were rated. The overall correlation for ratings of both sensory and cognitive stimuli on eVAS and pVAS was r ¼ 0:91. For paired verbal stimuli the correlation was r ¼ 0:97. For paired sensory stimuli the correlation was r ¼ 0:86. The correlation between group eVAS and pVAS ratings to common verbal stimuli was r ¼ 0:99. For common sensory stimuli the group correlation was r ¼ 0:99. The median of correlations comparing eVAS and pVAS ratings was 0.99 for verbal stimuli and 0.98 for sensory stimuli. Multivariate analyses showed equivalent stimuli to be rated much the same whether entered on paper VAS or PTC touch screen VAS (P , 0:0001). Support was found for the validity of the computer version of the VAS scale. q 2002 International Association for the Study of Pain. Published by Elsevier Science B.V. All rights reserved. Keywords: Visual analogue scales; Palmtop computers; Electronic data; Verbal rating scale; Pain
1. Introduction The visual analogue scale (VAS) is an established, validated, self-report measure of pain intensity usually consisting of a 10 cm line on paper with verbal anchors labeling the ends (Collins et al., 1997; Ferraz et al., 1990; Harms-Ringdahl et al., 1986; Price et al., 1983). VASs have been used in
* Corresponding author. Pain Management Center, Brigham and Women’s Hospital, 75 Francis Street, Boston, MA 02115, USA. Tel.: 11-617-278-1063; fax: 11-617-731-5453. E-mail address: [email protected] (R.N. Jamison).
many studies to measure a number of constructs including pain (Badley and Papageorgiou, 1989), asthma (Dhand et al., 1988), dyspepsia (Nyren et al., 1987), mood (Monk, 1989), appetite (Stubbs et al., 2000), ambulation (Welsh et al., 1993), and vitality (Wood et al., 1990). The VAS has been shown to have advantages over verbal rating scales and numerical scales in sensitivity to changes in pain intensity (Ohnhaus and Adler, 1975) and in the capacity to provide ratio scale measures of experimental pain (Price et al., 1994). They are particularly useful for populations with language barriers (Soh and Ang, 1992) and can be presented in various forms (Stephenson and Herman, 2000). They
0304-3959/02/$20.00 q 2002 International Association for the Study of Pain. Published by Elsevier Science B.V. All rights reserved. PII: s0 304-3959(02)00 178-1
342
R.N. Jamison et al. / Pain 99 (2002) 341–347
have also been adapted to computers with interactive animation (Swanston et al., 1993). Recent developments in hand-held computer technology have created the potential for collection of patient selfassessment data through electronic methods. Electronic diaries can facilitate more complete and timely collection of patient diary data than paper diaries (Jamison et al., 2001). Palmtop computers (PTC) have incorporated VAS entry by use of a touch screen. The PTC is capable of using an electronic visual analogue scale (eVAS) that is quite similar to the conventional paper-based VAS (pVAS). While the two versions of the VAS are similar, the eVAS differs from the pVAS in its shorter overall length and ease in moving the mark along the line without having to erase a former mark. Unlike the pVAS, the eVAS is marked by touching the line with a stylus. Clinical trials have used eVAS to obtain patient self-report data (Tiplady et al., 1995), but concerns remain that the validity and psychophysical properties of the eVAS have never been formally compared with the present pVAS standard. The aim of this study is to assess the validity of the eVAS and to determine whether patient input via eVAS is equivalent to input via pVAS. The intent was to use standard psychophysical techniques to evaluate the psychometric properties of self-report VASs.
2. Methods This is a single-center, randomized, crossover study using healthy volunteers. Twenty-four subjects were recruited for this study. The participants ranged in age from 19 to 57 (mean ¼ 34.4) and were mostly females (79.2%). Seventy nine percent were Caucasian, 17% African–American, and 4% Hispanic. Each study participant received 5 min of introductory training in the use of both the eVAS and the pVAS. The same test administrator using a written script provided the training in both methods. Experimental data collection for each study participant took place in a single session lasting approximately 1 h. Participants provided input by marking the eVAS and pVAS in response to two kinds of stimuli, cognitive and sensory. The cognitive stimuli were represented by a verbal rating scale of seven descriptors of intensity ranging from ‘faint’ to ‘extremely intense’ (Gracely et al., 1978). Participants were asked to mark the location that best corresponded to the pain inten-
sity described by each word on scales from ‘no pain’ to ‘worst possible pain’. Two evaluators independently measured and recorded all the pVAS forms to the nearest millimeter. The sensory stimuli used were a set of weights consisting of plastic canisters. Two canisters were marked as the reference weights (heavy weight and light weight), and seven groups of test canisters served as the study weights. The canisters were identical in size and weights increased progressively from 6.8 (empty canister), to 16.4, 25.4, 43.8, 61.7, 89.1, and 129.1 g, respectively. The reference weight weighed 165.2 g. A pilot trial demonstrated that these weights could be readily differentiated from each other when presented as pairs under conditions where the subjects were asked to identify the heavier weight of the pair. Replicates of the test canisters were produced so that each of the seven weights could be presented repeatedly three times without having to present the same actual test canister on more than one occasion. The canisters were filled with cotton (so the weights did not rattle) and marked with a random text code so that they could be distinguished by the experimenter without cueing the subjects. When assessing the heaviness of the weights, participants were asked to hold the weight with their thumb and index finger just below the cap (Flanagan et al., 1995). Participants were then asked to mark on a VAS the location that corresponded to the test weight in relation to the endpoints of the scales, which were anchored by ‘0’ and ‘Reference Weight’ (see Fig. 1). Before the start of each series of weights, subjects were requested to lift the reference heavy weight. 2.1. Device properties The eVAS was implemented on a Palm Pilot IIIxe, with a 6 cm square screen having a 160 £ 160 pixel resolution. The 5 cm VAS occupied 142 of the pixels. The bars marking each end were 0.5 cm high (14 pixels). The VAS marker was 4 mm high (10 pixels), and about 1 mm wide (3 pixels) and centered perpendicularly on the VAS line. The center pixel of the marker was set equal to the pixel identified as ‘touched’ by the palm touch screen. The eVAS capture program linearly converted the pixel touched to an integer between 0 and 100; so that each integer represented 1 or 2 pixels (,1 mm). In validation testing, the plastic stylus, which had a 1 mm diameter spherical tip, was used to drag
Fig. 1. Visual analogue scales.
R.N. Jamison et al. / Pain 99 (2002) 341–347
the marker from one end of the scale to the other and back repeatedly. Correspondingly, as the stylus was moved incrementally, all the integers from 0 to 100 could be caused to appear sequentially in an onscreen window that was set up during testing to display the automatic scoring. Repeated touches in the same spot could be done carefully enough to post identical scores 10 times out of 10, but more commonly the targeting skill of hand-manipulated touches to the screen would result in N ^ 1 variations in the number posted. To the eye, the resolution of markings on the paper could be resolved as well but not better than the marker on the eVAS. The pencil marks were approximately 1 mm wide, and manual measurements resolved the marks to the nearest millimeter ^ 1 mm. Both scales thus had similar resolution of one part in 100, which was sufficiently finely graded so that a user had the impression of a continuous scale.
2.2. Study design There were two categories of stimuli (sensory and verbal) and two modes for recording responses (paper and electronic), for a total of four experimental conditions. The two combinations involving a pVAS were designated A and B, corresponding to verbal and heaviness ratings. The two combinations involving an eVAS were designated a and b. Each volunteer completed all four experimental conditions in one of the following eight orders: ABab, ABba, BAba, BAab, abBA, abAB, baBA, or baAB. Within each stimulus and mode, each subject experienced each intensity level three times, resulting in 21 measurements per stimulus/mode combination. The 21 intensity levels within a stimulus/modality combination were presented in three sequences of seven intensity levels. Each sequence consisted of a randomized ordering of all seven intensities. Thus, there were 21 stimuli given in each of four conditions to the 24 subjects. Random assignment counterbalanced for the order of mode and stimulus was used to control for possible order effects. To facilitate administration by minimizing switching between paperbased and electronic modes, each participant switched only once (i.e. each was randomized to start with pVAS and then switched to eVAS, or to start with eVAS and then switched to pVAS). The length of the pVAS scale was 10 cm, and the length of the eVAS scale was 5 cm. The pVAS forms were printed (not xeroxed) to preserve dimensional accuracy. The eVAS forms were programmed on the PTC to resemble the pVAS forms. For the verbal test, the pVAS and eVAS forms displayed the stimulus words. For the heaviness test, the pVAS and eVAS forms displayed a coded number (which was meaningless to the subject) that corresponds to the weight canisters being given to the subject. There was a separate code for each canister to prevent the subjects from adjusting their responses to repeated codes.
343
2.3. Subjects Twenty-four healthy volunteers were recruited by advertisement and signed an informed consent form to participate in the study. Each subject was paid US $50 for his or her participation. The inclusion criteria were (1) age 18 or older at time of participation and (2) willingness and ability to use the PTC to record ratings. Exclusion criteria consisted of (1) an inability to differentiate between the lightest and heaviest weight, as determined during the training session, (2) marked inability to rank order the stimuli, or (3) an inability to comply with the protocol. None of the volunteers were excluded. 2.4. Subject training The purpose of the experiment, to compare two methods of recording measurements on a linear scale, was explained to each participant. Inclusion and exclusion criteria were reviewed, including verifying that the participant could correctly distinguish the heaviest and lightest test weights. Each participant was shown how to hold the weight with his or her thumb and index finger just below the cap. Each participant was assigned a sequential numeric code, which was used to determine the sequence of modes, stimulus categories, and stimulus levels to be used with that participant. Each participant used the pVAS and the eVAS in an order determined by the randomization plan. Participants were trained in each modality immediately before using that modality. In the training, the participant was shown how to mark the scale and was asked to try marking the scale four times corresponding to verbal inputs of ‘faint’ and ‘extremely intense’ and to weight inputs of the lightest and heaviest weights. The test administrator assisted the participant if the participant had difficulty marking the scale or reversed the sense of the scale (i.e. marked ‘faint’ above ‘extremely intense’, or the lighter weight above the heavier). 2.5. Test administration Tests were administered in the order defined by the randomization plan. Before each group of verbal rating tests, the participant was shown a list of the words, ordered from least to greatest pain level. Before each group of heaviness tests, the participant was asked to hold the reference weight using a consistent grip between thumb and index finder. During the actual test, the participant marked his/her assessment on a new pVAS sheet or a new eVAS display. Results of previous markings were not visible. The pVAS/ eVAS included the word to be rated or the coded label of the weight. The pVAS forms were collected by the test administrator and each sheet was marked with the participant’s sequence number. All the pVAS forms were evaluated by two evalua-
344
R.N. Jamison et al. / Pain 99 (2002) 341–347
tors, each of whom measured and recorded the VAS marking independently to the nearest millimeter. Each evaluator filled out a series of paper worksheets containing one row for each measurement. A single 100 mm metal ruler was used for all measurements. Each row contained the participant number, the stimulus type (verbal or weight), the coded or plain text stimulus value appearing on the form, and the result of measuring the VAS response with the ruler. The evaluators were instructed to use their best judgement in the case of ambiguous markings and to record missing markings as ‘not available’. The data from the evaluation sheets were keyed into spreadsheets, resulting in two independently determined results for each VAS marking. The pairs of results were compared, and cases of disagreement were noted. If the two values agreed within 1 mm, the two results were averaged. If, on the other hand, the two values were more than 1 mm apart, the evaluators were asked to redo the measurements in question. The data from the eVAS forms were collected and transmitted to a web server, using telephonic techniques similar to those used in clinical trials (Jamison et al., 2001). The recordings included the participant number, the stimulus identifier (stimulus type and level), and the measurement as captured by the PTC. Data from all participants were recorded and entered. To be included in the statistical evaluation, a participant must have provided at least one full replicate for each of the four modality/stimulus combinations. 2.6. Statistics For both cognitive and sensory stimuli, we sought to answer two questions: (1) are pVAS and eVAS ratings equivalent for each stimulus value? and (2) are pVAS and eVAS equally precise? Equivalence, defined as similarity between the average eVAS and pVAS ratings after each subject repeatedly rated the same stimulus on both eVAS and pVAS, was determined by tests of analyses of variance (ANOVAs) with and without repeated measures between the two modes for each stimulus. Precision pertains to the variability of the repeated ratings for each of the two rating modes. If the variability of the ratings in one rating method is less than the variability in the other rating method, the first rating method is considered more precise. If eVAS and pVAS are equivalent, then the more precise mode is preferable. In all statistical procedures, both the pVAS and eVAS values were scaled from 0 to 100. Based upon previous research (Jamison et al., 2001), the variance of the difference between eVAS and pVAS scores was estimated to be 118 (corresponding to a standard deviation of approximately 10.86) when the scores for both eVAS and pVAS ranged from 1 to 100 (this is equivalent to assuming that both eVAS and pVAS have a standard deviation of 24.29, and the correlation between eVAS and pVAS is 0.90). Therefore, with a sample size of 24, this study had 90% power to detect a difference in
mean eVAS and pVAS values of 2.22 or greater. For a particular stimulus, because there will be half the number of ratings, this study had 90% power to detect a difference in eVAS and pVAS of 5.87 or greater.
3. Results A total of 2016 stimuli were presented and rated (24 subjects £ 4 combinations of response mode and stimulus modality £ 7 stimuli £ 3 presentations of each stimulus). For each combination of response mode and stimulus modality (condition), the three responses to each stimulus were averaged to yield a mean response for each of the seven stimuli. One paper data point for a verbal rating was accidentally skipped and the mean for this subject and stimulus was computed from the remaining two observations. The overall correlation for ratings of both sensory and cognitive stimulus types was significant (r ¼ 0:91). For 503 paired verbal stimuli in the 24 subjects the correlation between paper and PTC ratings were highly significant (r ¼ 0:97; range 0.95–0.98), while for 504 paired sensory stimuli the ratings were lower but also significantly correlated (r ¼ 0:86; range 0.81–0.92). r 2 correlations were then computed between the eVAS and pVAS responses to common stimuli for each stimulus modality. The correlation between group eVAS and pVAS ratings to common verbal stimuli was r 2 ¼ 0:997. For common sensory stimuli the group correlation was r 2 ¼ 0:999. Scatterplots showing these realtionships are presented in Fig. 2a,b. These figures show that, for the group, scoring of equivalent stimuli was found to be highly correlated whether entered on paper 10 cm VAS or PTC touch-screen 5 cm VAS. Individual analyses were run for each subject by first computing the same correlations shown for the group in Fig. 2a,b. Scatter plots of these correlations are shown in Fig. 3. This figure shows general agreement for all but one subject and excellent agreement for all but six subjects. For all subjects, the median and the mean for the verbal stimuli were 0.99, and 0.99 (range 0.92–0.99) and for the sensory stimuli the median and mean were 0.98 and 0.96 (range 0.72–0.99). A one-way repeated measures ANOVA using stimulus intensity as a repeated factor with seven levels was run for both stimuli and modes. The results of multivariate analyses of the combined verbal and weight stimulus ratings with and without repeated measures in the model are presented in Table 1. The mode by stimuli interactions were found to be highly significant, although there was less agreement for the weight than for the verbal stimuli. Mean variances ratio tests were compared for eVAS and pVAS for the cognitive and sensory stimuli. The variance ratios ranged from 0.53 and 2.15 for the cognitive stimulus and from 1.62 and 0.22 for the sensory stimulus. None of the differences between pVAS and eVAS were shown to be significant for the seven cognitive stimulus, however, four
R.N. Jamison et al. / Pain 99 (2002) 341–347
Fig. 2. (a) Scatter plot of correlations between eVAS and pVAS to common verbal stimuli. (b) Scatter plot of correlations between eVAS and pVAS to common sensory stimuli.
345
variate analyses between electronic and paper measurements were very highly significant for both types of stimuli, cognitive (verbal pain intensity ratings) and sensory (weights). Since scales that are valid for measurement of various non-painful sensory modes are generally valid for measuring pain (Price et al., 1983,1994), it is reasonable to conclude that electronic VAS measurements are essentially equivalent to paper VAS measurements for pain as well. The variance ratios between electronic and paper VAS ratings were also similar, although for the weights, the variance of the electronic scores was less than the variance of the paper scores. It is intriguing to speculate why the subjects, who had more difficulty with weights than verbal ratings, were able to record their responses more precisely on the electronic medium. The inherent benefits with using an eVAS include less chance for ambiguous markings, errors in measurement, and skipped data. There also may be other properties of a PDA that may help to improve accuracy of responding that are yet to be determined. This finding is worthy of replication. We found differences between the two experiments, with the cognitive tests (verbal scale) showing higher levels of significance than the sensory tests (weights). On average, the subjects could discriminate among the seven stimuli for the verbal stimuli, but had greater difficulty judging among the weight stimuli. This was not surprising since it would be expected that the order of intensity would be clearer with the verbal ratings than with the subjects’ ratings of heaviness. Even though this protocol followed standard psychophysical testing design, the weights did not increase logarithmically, which would make the judgment among the various weights more difficult, as was especially evident among with lighter weights. Overall, however, subjects reliably rated the intensities of both stimuli in a linear manner.
of the seven of the sensory stimulus were significantly different (P , 0:05). Fig. 4 shows a simple measure of the variability of each method by showing the standard errors of the mean (the higher the F-value the lower the standard error). The eVAS generally shows similar standard errors for the verbal or sensory stimuli, while the standard errors of the pVAS for the sensory stimuli are about double the pVAS standard error to the verbal stimuli. Exit interviews of subjects indicated that the sensory discrimination of adjacent weights was a difficult task, which would be consistent with the greater variances and lower correlations found between paired ratings.
4. Discussion This study followed the design used in standard psychophysical techniques for evaluating the psychometric properties of a patient self-report scale (Gracely, 1988, 1989). The results of this study demonstrate that a VAS on a palmtop computer is remarkably similar to a VAS on paper. Multi-
Fig. 3. Comparison of correlations between eVAS and pVAS to common stimuli plotted for each subject.
346
R.N. Jamison et al. / Pain 99 (2002) 341–347
Table 1 One way analysis of variance with and without repeated measures for the verbal and weight stimuli observations a Stimulus
No. of observations
Degrees of freedom
r2
F
Repeated measures Verbal Weight
1007 1008
259 259
0.98 0.92
131.22 *** 33.34 ***
Without repeated measures Verbal Weight
1007 1008
197 197
0.97 0.90
155.32 *** 38.40 ***
a
***P , 0:0001.
Some individual differences among subjects were found that are worthy of discussion. For most subjects, the relationship between rating and response with the cognitive stimuli was essentially linear. For a few of the subjects, however, the relationship was binary (expressed as either high or low on both versions of the VAS), and for one subject the relationship was non-linear (suggesting that this person had trouble discriminating differences). These findings are in keeping with differences usually found in self-report pain ratings among pain patient populations (Jamison et al., 1998). Differences among the subjects were also noted for the weight stimuli. Again, most of the subjects revealed a linear relationship between rating and response, even though these data were more variable than the verbal data. Two subjects seemed to have particular trouble with the task. Again, this might be expected since some individuals possess better discrimination skills in assessing weights than others. Some subjects were also less consistent than others, which accounted for lower correlations even within the same modes (paper to paper r 2-values). This finding lends further support to the similarity among different assessment techniques. Although electronic versions of VAS scales have been used by pain patients in the past without any identifi-
Fig. 4. Standard errors of the mean for each mode (pVAS and eVAS) and stimuli (verbal and weight).
able difficulties (Jamison et al., 2001), further study with chronic pain patients could be helpful to establish the validity and reliability with these patients. In summary, electronic VAS scores were found to be highly correlated with paper VAS scores for both cognitive (verbal intensity) and sensory (weight) stimuli. No support was found for the hypothesis that mode of data collection (paper vs. electronic VAS) has any biasing effect on the data collected. Thus, self-reported ratings using electronic VAS do not differ from such ratings on paper. VAS data from electronic diaries are valid and hold intrinsic benefits for future clinical trials. Acknowledgements A portion of this study was presented at the 20th Scientific Meeting of the American Pain Society, Phoenix, AZ, April 21, 2000. The authors wish to express appreciation to Jamie Bell and Kelly Zou for their assistance, and Jaylyn Olivo who reviewed an earlier version of this manuscript. The study was funded in part by PHT Corporation. References Badley EM, Papageorgiou AC. Visual analogue scales as a measure of pain in arthritis: a study of overall pain and pain in individual joints at rest and on movement. J Rheumatol 1989;16:102–105. Collins SL, Moore RA, McQuay HJ. The visual analogue pain intensity scale: what is moderate pain in millimetres? Pain 1997;72:95–97. Dhand R, Kalra S, Malik SK. Use of visual analogue scales for assessment of the severity of asthma. Respiration 1988;54:255–262. Ferraz MB, Quaresma MR, Aquino LR, Atra E, Tugwell P, Goldsmith CH. Reliability of pain scales in the assessment of literate and illiterate patients with rheumatoid arthritis. J Rheumatol 1990;17:1022–1024. Flanagan JR, Wing AM, Allison S, Spenceley A. Effects of surface texture on weight perception when lifting objects with a precision grip. Percept Psychophys 1995;57:282–290. Gracely RH. Pain psychophysics. In: Chapman CR, Loeser JD, editors. Advances in pain research and therapy, New York, NY: Raven Press, 1989. Gracely RH, Lota L, Walter DJ, Dubner R. A multiple random staircase method of psychophysical pain assessment. Pain 1988;32:55–63. Gracely RH, McGrath F, Dubner R. Ratio scales of sensory and affective verbal pain descriptors. Pain 1978;5:5–18. Harms-Ringdahl K, Carlsson AM, Ekholm J, Raustorp A, Svensson T, Toresson HG. Pain assessment with different intensity scales in response to loading of joint structures. Pain 1986;27:401–411.
R.N. Jamison et al. / Pain 99 (2002) 341–347 Jamison RN, Raymond SA, Levine JG, Slawsby ES, Nedeljkovic SS, Katz NP. Electronic diaries for monitoring chronic pain: one-year validation study. Pain 2001;91:277–285. Jamison RN, Raymond SA, Slawsby ES, Nedeljkovic SS, Katz NP. Opioid therapy for chronic noncancer back pain. Spine 1998;23:2591–2600. Monk TH. A visual analogue scale technique to measure global vigor and affect. Psychiatry Res 1989;27:89–99. Nyren O, Adami HO, Bates S, Bergstrom R, Gustavsson S, Loof L, Sjoden PO. Self-rating of pain in nonulcer dyspepsia. A methodological study comparing a new fixed-point scale and the visual analogue scale. J Clin Gastroenterol 1987;9:408–414. Ohnhaus EE, Adler R. Methodological problems in the measurement of pain: a comparison between the verbal rating scale and the visual analogue scale. Pain 1975;1:379–384. Price DD, Bush FM, Long S, Harkins SW. A comparison of pain measurement characteristics of mechanical visual analogue and simple numerical rating scales. Pain 1994;56:217–226. Price DD, McGrath PA, Rafii A, Buckingham B. The validation of visual analogue scales as ratio scale measures for chronic and experimental pain. Pain 1983;17:45–56.
347
Soh G, Ang HG. Comparison of two pain rating scales among Chinese cancer patients. Chin Med J (Engl) 1992;105:953–956. Stephenson NL, Herman JA. Pain measurement: a comparison using horizontal and vertical visual analogue scales. Appl Nurs Res 2000;13:157– 158. Stubbs RJ, Hughes DA, Johnstone AM, Rowley E, Reid C, Elia M, Stratton R, Delargy H, King N, Blundell JE. The use of visual analogue scales to assess motivation to eat in human subjects: a review of their reliability and validity with an evaluation of new hand-held computerized systems for temporal tracking of appetite ratings. Br J Nutr 2000;84:405–415. Swanston M, Abraham C, Macrae WA, Walker A, Rushmer R, Elder L, Methven H. Pain assessment with interactive computer animation. Pain 1993;53:347–351. Tiplady B, Crompton GK, Brackenridge D. Electronic diaries for asthma. Br Med J 1995;310:1469. Welsh EM, Gettinby G, Nolan AM. Comparison of a visual analogue scale and a numerical rating scale for assessment of lameness, using sheep as a model. Am J Vet Res 1993;54:976–983. Wood C, Magnello ME, Jewell T. Measuring vitality. J R Soc Med 1990;83:486–489.