The abbreviated burn specific pain anxiety scale: a multicenter study

The abbreviated burn specific pain anxiety scale: a multicenter study

Burns 25 (1999) 493±497 www.elsevier.com/locate/burns The abbreviated burn speci®c pain anxiety scale: a multicenter study p L.A. Taal a,*, A.W. Fab...

122KB Sizes 0 Downloads 17 Views

Burns 25 (1999) 493±497

www.elsevier.com/locate/burns

The abbreviated burn speci®c pain anxiety scale: a multicenter study p L.A. Taal a,*, A.W. Faber b, N.E.E. van Loey c, C.L.L. Reynders d, H.W.C. Ho¯and e a

Dutch Burns Foundation, Red Cross Hospital, Beverwijk, Netherlands b Burn Centre Martiniziekenhuis, Groningen, Netherlands c Burn Centre Stuivenberg Ziekenhuis, Antwerp, Belgium d Burn Centre Gasthuisberg Ziekenhuis, Leuven, Belgium e Burn Centre Zuiderziekenhuis, Rotterdam, Netherlands Accepted 17 February 1999

Abstract The authors examined ratings on a scale of pain-related anxiety in 173 burn patients in three groups: patients with small burns, patients with moderate burns and patients with extensive burns. The data suggest a greater degree of anxiety during procedures and before procedures in the burn patients with extensive burns than in burn patients with small and moderate burns. This study introduces a novel measure of pain-related anxiety in clinical burn patients, the abbreviated Burn Speci®c Pain Anxiety Scale (BSPAS), which showed a high degree of reliability. The alpha coecients were high for the BSPAS subscales. # 1999 Elsevier Science Ltd and ISBI. All rights reserved. Keywords: Pain; Anxiety; Self-report scale; Con®rmatory factor analyses

1. Introduction Burn injuries are among the most painful types of trauma. Perry et al. [1] report that 84% of 52 burn patients in the New York Hospital described the pain during treatment of wounds as unbearable, despite having been given on average 8.9 mg intramuscular morphine. Most of these patients only felt slight pain while resting. Because burn patients must undergo daily, or even several times daily, painful therapeutic procedures like dressing changes, wound cleansing and physiotherapy, physicians, nurses and physiotherapists need to be in touch with the feeling of anticipatory anxiety in their patients. Since up until now there has been no simple, objective rating of the level of anxiety exhibited by burn patients, doctors and nurses used to

The study was all made possible by the involvement of the Burn Centre Universiteitshospital, Gent, Belgium. * Corresponding author. Tel.: +31-251-278-444; fax: +31-251-278455. p

rely on their own personal intuitive appraisal and act accordingly. However, the existing literature [2±6] suggests that the patients' self-reports tend to correlate poorly with their nurses' observation assessments. In a previous study [7] we described preliminary results with the Burn Speci®c Pain Anxiety Scale (BSPAS), a newly developed instrument to assess anticipatory anxiety in burn patients. In that report we concluded that: (1) while other existing measures of state-anxiety might be of great utility, there remains a need for an instrument that has the capacity to measure state-anxiety speci®cally related to the situation with which the burn patient is confronted and (2) that such an instrument should be very short to be suitable to ®eld situations. A general drawback of many state-anxiety scales is that the questionnaires tend to be rather long for use as routine screening instruments of anticipatory anxiety in large numbers of hospitalised burn patients. Our previous, relatively small (N = 35), study investigated the criterion-related validity of the BSPAS by correlating it with patients' pain perception and nurses' observation ratings of ten-

0305-4179/99/$20.00+0.00 # 1999 Elsevier Science Ltd and ISBI. All rights reserved. PII: S 0 3 0 5 - 4 1 7 9 ( 9 9 ) 0 0 0 3 4 - 0

494

L.A. Taal et al. / Burns 25 (1999) 493±497

sion immediately before and during dressing changes. The scale had, as hypothesised, a high correlation with procedural pain and a low correlation with non-procedural pain. The present study concentrates on the structure, reliability and validity of a revised version of the BSPAS. The original questionnaire contains two items on fear for wound-healing and recovery, two items on pain severity, one item on a cognition about pain, two items on behavioural reactions during dressing changes and two items on a€ective reactions. As a result, the original BSPAS has a rather unbalanced structure. This leaves the validity of our claim that the BSPAS is useful in the assessment of anticipatory anxiety more or less in the dark. In our view this shortcoming was reparable. It was decided to abbreviate the BSPAS and to develop a scale with a more balanced structure. We eliminated ®ve items on the original BSPAS on the basis of an analysis of the concept of anticipatory anxiety. This report introduces the abbreviated version of the Burn Speci®c Pain Anxiety Scale as a measure that correspond closely to both the common sense and the DSM-IV de®nition of anxiety [8] and can be completed in about 3 min.

BSPAS were selected after an iterative careful inspection of the item-total correlations. Each item in the BSPAS is scored on a 100 mm visual analogue line with two reference points given values of 0 and 100. The reference points are also identi®ed by the expressions `not al all' and `the worst imaginable way'. Thus the BSPAS covers the whole range of dressing change anxiety, from no anxiety at all to extreme anxiety. The BSPAS is scored as the mean of the item responses across all items. The shortened version of the BSPAS consists of 5 items re¯ecting only the level of pain related anxiety during and before dressing changes. Items re¯ecting pain intensity (for example, the items ``The pain is sometimes so severe that I must cease all activities'' and ``The pain can be so severe that I become frightened of losing control'') and worry about wound healing (for example, the item ``I can suddenly ®nd myself feeling unsure about my recovery when I see my wounds'') were eliminated. 2.3. Procedure

2. Materials and methods

On average patients were asked to complete the original BSPAS in the presence of the researchers 7 to 14 days after admission to the burn centre. For patients who were unable to write, the researchers marked the chosen position on the visual analogue scales of the BSPAS-items.

2.1. Subjects

2.4. Statistical analyses

The patients sample was composed of 173 primary hospital admissions to 5 burn centres during the period 1 January 1997 to 31 June 1998. Each centre was located in The Netherlands or in Belgium. To qualify for inclusion, the patients had to be (a) Dutch speaking; (b) older than 16 years of age and (c) not having a pre-morbid psychiatric diagnosis. The sample included 129 males and 44 females, with a mean age of 37.4 years (S.D.=13.4). Mean percentage total burned surface area was 12.9 (S.D.=11.2). The average total surface burned (TBSA) of the sample was 12.9% (S.D.=11.2). The mean percentage full thickness burns of the sample was 3.2% (S.D.=5.6).

Con®rmatory factor analysis o€ers a useful tool for the investigation of the dimensionality of a psychometric assessment instrument. It allows the investigator to impose substantively motivated constrains determining which items in a psychological measure are a€ected by which common factor. Statistical tests can be performed to determine whether data con®rm the substantively generated model. We compared two con®rmatory factor models: model 1 was an unrestricted factor model with one factor (`Anx_iety'). Model 2 was a restricted factor model with two factors, in which three BSPAS-items depended upon only one of the factors (anxiety and tension during dressing changes: `Proc_anx') while the remaining two BSPAS items depended only on the other factor (anxiety before dressing changes: `Ant_anx'). ADF-estimators were used and the bootstrap approach was employed for model comparison. Goodness of ®t was assessed by Chi-square and RMR (root mean square residual). The psychometric quality of the abbreviated BSPAS was investigated by calculating Cronbach alphas and the Pearson correlation between the original and the abbreviated BSPAS. Finally, as validity assessment a one-way analysis of variance was used to test a linear

2.2. Materials 2.2.1. Burn speci®c pain anxiety scale Initially, a pool of 27 dressing change anxiety items was developed on the basis of open interviews with clinical burn patients. Most of the items were written during the work of the ®rst author as a therapist in the burn centre of the Zuiderziekenhuis in Rotterdam, The Netherlands. The nine items of the original

L.A. Taal et al. / Burns 25 (1999) 493±497

Fig. 1. Scatterplot with ®t line of correlation between abbreviated and original BSPAS.

increase in average BSPAS scores across a set of ordered means based on total burned surface area.

3. Results 3.1. Preliminary analyses First, we expected that the correlation between the revised and the original BSPAS would be high. As shown in Fig. 1, the correlation was 0.96. Second, multivariate analyses were used to assess the reliability of the original and abbreviated BSPAS. Coecient alpha, a measure of internal consistency, was 0.91 for the original 9-item BSPAS and 0.90 for the abbreviated 5-item BSPAS. A second measure of internal consistency, the average of each item's correlation with the uncorrected total score, was 0.69, with a range of 0.62 to 0.81, for the original BSPAS, and 0.75, with a range of 0.67 to 0.81, for the abbreviated BSPAS (Table 1).

495

Fig. 2. Path diagram and standardised estimates of restricted two factor analysis abbreviated BSPAS.

This set of ®ndings on reliability, internal consistency and the strength of the correlation coecient, con®rmed that the abbreviated BSPAS has good psychometric qualities. Thus, to the extent that the original BSPAS assess state-anxiety in burn patients, so does the abbreviated BSPAS.

3.2. Internal structure of the abbreviated BSPAS The original sample of 173 burn patients served as the population for purposes of bootstrap sampling. First, 1000 bootstrap samples were generated by sampling with replacement from the original sample for each of the two models. Second, model 1 (unrestricted factor model with one factor) and model 2 (restricted two factor model) were ®tted to every bootstrap sample. Third, after each analysis, we calculated the discrepancy between the

Table 1 Means, standard deviations and item-total correlations BSPAS items Item description

Mean

Standard deviation

Item-total correlation

(1) I ®nd it impossible to relax when my burns are being treated (2) I feel my muscles getting tense when the treatment actually begins (3) I am frightened of the pain during and/or after the treatment (4) The pain makes me nervous and restless (5) I ®nd myself worrying about the possible pain I might have to endure for every medical operation

27.29

32.19

0.75

29.53 26.02 25.29 21.99

32.62 29.47 29.87 29.60

0.78 0.81 0.67 0.76

496

L.A. Taal et al. / Burns 25 (1999) 493±497

Table 2 Fit measures for two competing models (standard errors in parentheses) Model

Mean discrepancy

Root mean square residual

Chi-square

P

(1) Unrestricted one factor model (2) Restricted two factor model

24.24 (0.17) 20.67 (0.18)

80.50 43.20

12.61 7.60

0.027 0.107

implied moments from the b-th bootstrap sample and the sample moments from the original sample. Finally, we calculated the average across 1000 bootstrap samples of the discrepancies from the previous step. The average discrepancies for the two competing models are shown in Table 2 along with values of the Chi-square and the RMR-criteria. As shown in Table 2, there is considerable empirical evidence against model 1, the unrestricted factor model with only one factor. The di€erence among the mean discrepancies of model 1 and model 2 are large compared to their standard errors. The lowest mean discrepancy (20.67) occurs for model 2, con®rming the model choice based on RMR and chi-square criteria. Fig. 2 shows the path diagram of model 2 along with standardised regression weights and squared multiple correlations. The two ellipses in the Fig. 2 are labelled `proc_anx', procedural anxiety and `ant_anx', anticipatory anxiety. They represent the unobserved variables that are directly measured by the ®ve BSPAS items. Standardised regression weights are displayed near single-headed arrows and squared multiple correlations near the rectangles. The estimate displayed near the double headed arrow indicates a correlation. In Fig. 2, the squared multiple correlations can be interpreted as follows. For example, 58% of the variance in BSPAS item 4 is accounted for by the variance in `ant_anx', anticipatory anxiety. The remaining 42% of the variance in BSPAS item 4 cannot be explained by the model and is thus attributed to the unique latent variable e4. The unique factor e4, however, does not only represent measurement error, but may comprise systematic unique variance components. Thus, the ®gure 0.58 has to be regarded as a lower-bound estimate of the reliability of BSPAS item 4. As shown in Fig. 2, the subscales of the abbreviated BSPAS were highly correlated. The level of internal consistency as assessed by Cronbach's alpha was 0.89 for the anxiety during procedures subscale and 0.79 for the anticipatory anxiety sub scale. 3.3. Validity assessment of abbreviated BSPAS We expected that patients with an extensive total surface burned would di€er signi®cantly on BSPAS scores from patients with small burns. We tested a linear increase in average BSPAS scores for three groups

of patients: patients with small burns (N = 58, mean TBSA=3.34, min/max TBSA=0 to 6.5), those with moderate burns (N = 56, mean TBSA=9.94, min/max TBSA=7 to 14) and those with extensive burns (N = 57, mean TBSA=25.64, min/max TBSA=15 to 60). The null hypothesis was that the linear contrast between the three groups on the subscales scores of the abbreviated BSPAS was 0-that is, there was no linear e€ect. Analysis of variance (ANOVA) showed a signi®cant linear increase in average BSPAS subscale scores as we moved across from small burns to extensive burns, the F statistic was 20.8 (df=1, P = 0.000) for the anxiety during procedures subscale, and 13.85 (df=1, P = 0.000) for the anxiety before procedures subscale. The means, standard deviations and ranges are presented in Table 3. 3.4. Norms and distributions Initial normative data consisting of the 20th, 35th, 50th, 65th, 80th and 95th percentiles for males and females, are shown in Table 4. This table of the range of variation in BSPAS subscale scores and total score for males and females may serve as baseline data to others who may use the BSPAS.

4. Discussion This study examined the internal structure and psychometric properties of the abbreviated BSPAS in a sample of 173 burn patients. Results of a con®rmatory factor analysis provided support for the multidimenTable 3 Means, standard deviations and ranges for the BSPAS subscales by groups Group Proc_anx: 1 2 3 Ant_anx 1 2 3

N

Mean

Standard deviation

Range

58 56 57

18.09 23.78 40.81

23.23 22.47 33.08

0±88 0±85 0±100

58 56 57

16.83 20.27 34.55

22.12 22.95 32.22

0±85 0±100 0±100

L.A. Taal et al. / Burns 25 (1999) 493±497

497

Table 4 The 20th, 35th, 50th, 65th, 80th, 95th percentiles, means, and standard deviations for subscales and total score of the BSPAS by sex Sex Male Very high (>95) High (80±95) Above average (65±80) Average (50±65) Below average (35±35) Low (20±35) Very low (<20) Mean (standard deviation) Female Very high (>95) High (80±95) Above average (65±80) Average (50±65) Below average (35±50) Low (20±35) Very low (<20) Mean (standard deviation)

Anxiety before procedures

Anxiety during procedures

Total BSPAS score

>83 40±82 20±39 10±20 0 0 0 20.27 (26.29)

>84 51±83 26±50 14±25 3±13 0±2 0 24.02 (28.09)

>78 46±77 24±45 12±23 6±11 0±6 0 22.52 (25.14)

>80 63±79 40±62 26±39 20±25 5±19 0±5 33.52 (27.09)

>86 67±85 48±66 33±47 20±32 13±19 0±13 38.14 (26.76)

>84 63±83 55±82 32±54 20±31 10±19 0±10 36.29 (25.99)

sionality of pain-related anxiety as assessed by the abbreviated BSPAS. Although all BSPAS items had adequate item-total correlations, inspection of the squared multiple correlations suggested that one item (``The pain makes me nervous and restless'') may be a poor measure of the corresponding latent factor. Future research might further evaluate the need for re®ning this item. A technical issue that must be considered is one that surrounds the methodology used in the validity assessment of the abbreviated BSPAS. We reasoned that scores on the BSPAS were, in part, a function of the amount of TBSA. The use of TBSA as an external criterion for pain-related anxiety may seem a weakness of this study; however, in the absence of a valid and independent measure of anxiety in burn patients, the division of the inpatient sample on the basis of TBSA was the only way to establish a criterion. Since the BSPAS is, to the best of our knowledge, the only burn speci®c anxiety instrument available, its validity could not be determined by the correlation with a `gold standard'. Furthermore, future research should examine (a) the scale's sensitivity to therapeutic interventions, (b) the stability of the items with repeated administrations over brief periods and (c) the magnitude of the correlations between the abbreviated BSPAS and measures of control beliefs and pain intensity. In summary, the results of the e€ort to develop and test a valid and reliable, brief, 5-item self report stateanxiety measure suggests that the abbreviated BSPAS meets initial criteria of reliability, validity and utility. It appears to be responsive to the severity of the burn wounds and may be useful for clinical practice or research purposes.

Acknowledgements Financial support for this study was given by the Dutch Burns Foundation. This study was made possible by the wonderful collaboration of the burn centres of the Zuiderziekenhuis Rotterdam, the Martiniziekenhuis Groningen, The Netherlands, and the burn centres of the Stuivenbergziekenhuis Antwerp, Gasthuisbergziekenhuis Leuven and Universiteitshospital, Gent, Belgium. We are very much indebted to Mrs. E. Sondervan for participating in the preparation of the manuscript.

References [1] Perry S, Heidrich G, Ramos E. Assessment of pain by burns patients. J Burn Care Rehab 1981;2:322±6. [2] Van der Does AJW. Patients' and nurses' ratings of pain and anxiety during burn wound care. Pain 1989;39:95±101. [3] Geisser ME, Bingham HG, Robinson ME. Pain and anxiety during burn dressing changes: concordance between patients' and nurses' ratings and relation to medication administration and patient variables. J Burn Care Rehab 1995;16:165±70. [4] Iafrati N. Pain on the burn unit: patient vs nurse perceptions. J Burn Care Rehab 1986;7:413±6. [5] Choiniere M, Melzack R, Girard N, Rondeau J, Paquin MJ. Comparisons between patients' and nurses' assessments of pain and medication ecacy in severe burn injuries. Pain 1990;40:143±52. [6] Walkenstein MD. Comparison of burned patients' perception of pain with nurses' perception of patients' pain. J Burn Care Rehab 1982;3:233±6. [7] Taal LA, Faber AW. The burn speci®c pain anxiety scale: introduction of a reliable and valid measure. Burns 1997;2:147±50. [8] American Psychiatric Association. Diagnostic and statistical manual of mental disorders, 4th ed., revised. Washington DC: American Psychiatric Association, 1994 p. 764.