Ars.els-cdn.com



Supplementary file 2 Psychometric properties of selected instrumentsPsychometricPropertiesMPQMCPGQInternal consistencyYes – Cronbach’s α = 0.768 overall; α = 0.888 for disability subscale; α = 0.836 for pain intensity subscale; α = 0.509 for frequency and duration subscale.Yes – Cronbach’s α = 0.74 (back pain); α = 0.67 (headache); α = 0.71 (TMD).Reliability coefficients of the Guttman scales also acceptable (> 0.70) (Von Korff et al., 1992).Yes – Cronbach’s α = 0.9132. Item-total correlation values: Pearson’s r = 0.6885-0.8285 (Smith et al., 1997).Yes – Cronbach’s α = 0.86 (total questionnaire); α = 0.89 (disability score); α = 0.81 (pain intensity). Item-total correlation values: Pearson’s r = 0.500-0.771 (Salaffi et al., 2006).Test-retest reliabilityNoNoFace ValidityYes – administered to orchestra members.YesContent ValidityYes – as part of the principal component analysis below.UnknownConstruct ValidityYes – Principal component analysis - 3 components with eigenvalues >1. Principal component analysis with varimax rotation to reduce 12 items to 10, grouped into 3 components: frequency and duration; intensity of pain; disability. 10 items explained 76% of the variance.Yes – Using Guttman scaling methods (exploratory Mokken analysis) and goodness of fit (H coefficient) to determine a hierarchical severity scale. 3 components identified as forming a Guttman scale: pain intensity; disability score; disability days. (Von Korff et al., 1992).Yes – Principal component analysis (PCA). One relevant component identified (eigenvalue = 4.80), all others with eigenvalues <0.85. (Smith et al., 1997). Italian version: 2-factor structure identified by PCA, with eigenvalues > 1, explaining 76.4% of the variance: characteristic pain intensity; disability score. Factor loading after varimax rotation confirmed the 2-component structure with 3 items per component (Salaffi et al., 2006).Yes – Convergent validity – Moderate to high Spearman’s r correlation between CPGQ and SF-36 for the bodily pain and physical component dimensions in Smith et al. (1997)(Spearman’s r = 0.7069-0.8389, p < 0.001), and in Salaffi et al. (2006)( Spearman’s r = 0.545-0.623, p < 0.0001).Yes – Discriminant validity (Wilcoxon and Kruskal-Wallis tests) – Women tend to report higher pain intensity scores than men (p = 0.028); higher disability score associated with older age (p < 0.0001); increasing education associated with lower pain and disability scores (p < 0.001); patients with co-morbid conditions had worse pain intensity (p < 0.0001) and disability scores (p < 0.001)(Salaffi et al., 2006). Concurrent ValidityYes – criterion validity with CPGQ.Overall score (r = 0.65, p < 0.01);disability (r = 0.53, p < 0.01); pain intensity (r = 0.99, p < 0.01).Yes – statistically significant (p ≤ 0.001) but using Chi-square tests and General Linear Models to determine association between CPGQ and several behavioural and psychological variables (Von Korff et al., 1992). Predictive ValidityNoYes – but using Chi-square tests of significance. CPG scores at baseline strongly predicted CPG scores one year later (Von Korff et al., 1992).Population Validity(Generalizability) No – tested and designed for a population of professional orchestra musicians, mostly freelancers.Only tested in Quebec / English language.Tested in different patients groups: back pain, headache, temporo-mandibular disorder (TMD) patients, chronic MSK pain. Using a primary care patient population only. Tested in the USA, UK, translated and tested in Italy.SensitivityNoNoSpecificityNoNoResponsivenessNoYes – Concurrent criterion validity with SF-36. Changes in CPG scores were significantly correlated with changes in SF-36 scores. Low to moderate correlations between changes in SF-36 and changes in CPG scores (highest correlation coefficients were Spearman’s r = -0.342, p < 0.01; Spearman’s r = -0.420, p < 0.01)(Elliott et al., 2000). Yes – See below the results from Krebs et al. (2010), detailed in the “Responsiveness” section of this table for the BPI. Floor / Ceiling EffectsUnknownAnomaly with subgroup of back pain patients with high disability but low pain intensity (Von Korff et al., 1992).PsychometricPropertiesNMQNMQ – EInternal consistencyNoNoTest-retestreliabilityPercentage disagreement – rate of non-identical answers between 0 - 23%. Small sample used.Yes – Test-retest: Proportion of observed agreement Po = 0.88-0.98 and K / Kmax = 0.71-0.96 for all 10 dichotomous questions. Point prevalence was the least reliable question.Yes – Reproducibility tests (self-report vs interview): Po = 0.92-0.98 and K / Kmax = 0.76-1.00 for all 10 dichotomous questions.Face ValidityYes – used in many studies.Yes – pilot testing.Content ValidityPartial – Through pilot testing and modification of wording.Partial – Through pilot testing and modification of wording.Construct ValidityNoNoConcurrent ValidityYes – criterion validity by comparison with clinical history, but small sample used. Percentage of disagreement between 0-20%.NoPredictive ValidityNoNoPopulation Validity(Generalizability) NoNoSensitivityYes – 66% to 92%.NoSpecificityYes – 71% to 88%. NoResponsivenessNoNoFloor / Ceiling EffectsUnknownUnknownPsychometricPropertiesLF-MPQ and SF-MPQInternal consistencyYes – Cronbach’s α = 0.70-0.79 (LF-MPQ) and 0.76-0.78 (SF-MPQ) for both sensory and affective dimensions (Menezes Costa et al., 2011).Test-retestreliabilityYes- using consistency of response with small sample of 10 patients. Mean consistency of response = 70.3% (Melzack, 1975).Yes – LF-MPQ: ICC = 0.46 (95% CI: 0.29-0.60) for the PPI score to ICC = 0.80 (95% CI: 0.72-0.86) for the PRI Total score (PRI-T) (Menezes Costa et al., 2011). Yes – SF-MPQ: ICC = 0.55 (95% CI: 0.40-0.67) for the PPI score to ICC = 0.69 (95% CI: 0.57-0.78) for the PRI-T (Menezes Costa et al., 2011).Face ValidityYes – Piloted in many studies in different languages.Content ValidityYes – Classification of 102 words describing pain into 3 major classes and 16 subclasses, using an expert panel. 4 additional subscales were subsequently added. (Melzack, 1975). Construct ValidityYes – Interval scaling procedure used to estimate the pain intensities implied by the words within each subclass: see content validity section (Melzack, 1975).Yes – convergent validity between the four scoring methods of the LF-MPQ. High correlations between the Pain Rating Intensity Score (PRI-S), the PRI Rank placement of each word (PRI-R), and the Number of Words Chosen (NWC): Pearson’s r = 0.89-0.97. Low correlation between the Present Pain Intensity score (PPI) and the other three scores: Pearson’s r = 0.32-0.42 (Melzack, 1975). High correlation between LF-MPQ scores and SF-MPQ scores for the sensory, affective, and total PRI scores (Pearson’s r = 0.65-0.94)(Melzack, 1987).Yes – convergent validity. Moderate to high correlations (statistically significant) between the PRI -T of LF-MPQ, PRI-T of SF-MPQ, and NRS (Pearson’s r = 0.49-0.68) (Menezes Costa et al., 2011). Yes – discriminant validity between eight different pain syndromes (including arthritis and degenerative disc disease). Multiple group discriminant analysis was statistically significant (p < 0.001), showing that each pain type was characterised by a distinctive constellation of verbal descriptors. 77% of patients could be correctly classified into diagnostic groups on the basis of the verbal description of pain (Dubuisson and Melzack, 1976).Yes- Confirmatory factor analysis (CFA) comparing a 3-factor structure of the Pain Rating Index (PRI) subscales (sensory, affective, evaluative) to a 1-factor structure, using goodness-of-fit measures. The 3-factor structure had a significantly better fit than the 1-factor model, using the chi-square difference test. However, there was a large amount of shared variance between the 3 dimensions (Pearson’s r = 0.60-0.80)(Holroyd et al., 1992). Yes – Principal component analysis (PCA). 4-component structure identified with oblique rotation: 2 sensory dimensions, 1 affective dimension, 1 evaluative dimension. A CFA was followed, comparing the 3-factor and 4-factor structures in two different samples of patients. The 4-factor structure had a significantly better fit than the 3-factor model in both samples, using the chi-square difference test. Once again, there was a large amount of shared variance between the 4 dimensions (Pearson’s r = 0.69-0.91), as described for a 3-factor structure (Holroyd et al., 1992). Yes – Hierarchical factor analysis was carried out to determine if second-order factors could explain the high inter-correlations observed between the primary three constructs / dimensions of pain. A single “pain-distress” factor was identified, explaining 62% of the shared variance, suggesting limited discriminant validity and limited clinical utility of the separate PRI subscale scores. CFA was followed, showing that this higher order structure provided a good fit to the data in two different samples of patients (Holroyd et al., 1992). Concurrent ValidityNoPredictive ValidityNoPopulation Validity(Generalizability) The LF-MPQ has been tested in patients groups with different pain problems in several studies: homogeneous and heterogeneous groups of pain patients have been used, bringing controversy to the component / factor structure of the MPQ. Tested in Canada, UK, USA, Brazil. Translated in 17 languages. The SF-MPQ has been translated in 13 languages. Clinimetric properties tested in Portuguese (Menezes Costa et al., 2011).SensitivityNoSpecificityNoResponsivenessInternal responsiveness: moderate effect size (ES). SF-MPQ: ES = 0.47 (84% CI: 0.35-0.59) for PRI Total score. LF-MPQ: ES = 0.55 (84% CI: 0.42-0.69). NRS: ES = 0.76 (84% CI: 0.57-0.94). No differences between LF-MPQ and SF-MPQ (Menezes Costa et al., 2011).External responsiveness: small to moderate correlations with the GPE scale at discharge. SF-MPQ: Pearson’s r = 0.39 for PRI-T. LF-MPQ: Pearson’s r = 0.51for PRI-T. NRS: Pearson’s r = 0.43 (Menezes Costa et al., 2011). External responsiveness using Area Under the ROC Curve (AUC): SF-MPQ: AUC = 0.68 (95% CI: 0.58-0.79) for PRI-T. LF-MPQ: AUC = 0.76 (95% CI: 0.66-0.85) for PRI-T. NRS: AUC = 0.69 (95% CI: 0.59-0.80). The LF-MPQ was the most responsive (Menezes Costa et al., 2011). Floor / Ceiling EffectsNo floor and ceiling effects were detected (Menezes Costa et al., 2011).PsychometricPropertiesBPIInternal consistencyYes - Cronbach’s α = 0.85-0.94 (pain intensity subscale) and 0.90-0.92 (pain interference subscale) in post-surgery patients (Mendoza et al., 2004).Yes - Cronbach’s α = 0.82-0.89 (pain intensity subscale) and 0.93-0.95 (pain interference subscale) in arthritis and low back pain patients (Keller et al., 2004).Yes - Cronbach’s α = 0.86-0.96 for each subscale: pain intensity, mood interference, activity interference. There was no change in reliability if an item was removed. Item-total correlation values were consistently high (Mendoza et al., 2006).Yes – inter-item correlations. Correlations among pain intensity items = 0.57-0.80. Correlations among interference items = 0.44-0.83 (Cleeland et al., 1988).Test-retestreliabilityYes – Correlation coefficients between adjoining assessments = 0.72-0.95 for 10 assessments. One outlier correlation coefficient = 0.58 (Mendoza et al., 2004).Yes – Correlation coefficients between adjoining assessments = 0.67-0.93 for 7 assessments for all 3 subscales (pain, mood, activity)(Mendoza et al., 2006).Face ValidityYes – Used in many studies in different languages.Content ValidityPartial – Through pilot testing and modification of wording (Daut et al., 1983).Construct ValidityYes – Exploratory factor analysis (EFA): 2-factor structure identified (pain intensity, pain interference) with these two factors accounting for 67% of the variance (Cleeland et al., 1988).Yes – Discriminant validity: discriminant function analysis showed that the BPI could distinguish between three groups of cancer patients receiving different analgesia. Pain intensity scores were significantly different between the groups (Kruskal-Wallis H = 18.57, p < 0.05); however, there were no statistical differences for the pain interference scores (Cleeland et al., 1988).Yes – Confirmatory factor analysis (CFA): the 2-factor model was significantly superior to a 1-factor model with regards to all statistical indicators. The 3-factor model was also statistically superior to the 2-factor model: Root-mean-squared error of approximation (RMSEA) = 0.075 (95% CI: 0.058-0.092); comparative fit index (CFI) = 0.953; change in chi-squared given the degrees of freedom values: χ2 (2) = 9.53, p < 0.05. The 3-factor model (pain intensity, activity interference, affective interference) was the best fit for the data. The 2-factor model was treated as a suitable alternative. A multi-group structural analysis showed that the 3-factor model was invariant with regards to disease type, age, and ethnicity groups (Atkinson et al., 2011). Yes – modified BPI with principal axis factoring (PAF): confirmation of the 2-factor structure (pain intensity and pain interference) for each assessment day (Mendoza et al., 2004). Yes – principal component analysis (PCA) followed by principal factor analysis with promax rotation: confirmation of the 2-factor structure (pain intensity and pain interference) in arthritis and low back pain patients, with these two factors accounting for 67% of the variance (Keller et al., 2004). Yes – Convergent validity: moderately strong correlation coefficients. Pearson’s r = 0.58-0.69 between the BPI and the Health Assessment Questionnaire (HAQ) for low back pain patients; Pearson’s r = 0.57-0.81 between the BPI and the Roland Disability Questionnaire (RDQ) for arthritis patients; Pearson’s r = 0.61-0.74 between the BPI and the SF-36 Bodily Pain (SF-36 BP) index for both low back pain (LBP) and arthritis patients (Keller et al., 2004).Yes – Discriminant validity: ANOVA calculations between BPI subscale scores and CPGQ classification showed that the BPI scores could significantly discriminate between groups of arthritis and low back pain patients who were classified according to pain severity grade with the CPGQ (Keller et al., 2004). Yes - modified BPI with principal axis factoring (PAF): a 3-factor structure was identified, with 3 subscales: pain, mood and activity. However, the items “sleep” and “enjoyment of life” did not load on any factor and were subsequently dropped. This 3-factor solution explained 86% of the variance (Mendoza et al., 2006).Yes – Convergent validity: correlation coefficients = 0.55-0.63 between the BPI pain subscale, the pain VAS, and the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC). Correlation coefficients = 0.47-0.62 between each item of the BPI pain subscale, the pain VAS, and the WOMAC pain scale. Correlation coefficients = 0.58-0.65 between the activity BPI subscale and the WOMAC physical function scale (Mendoza et al., 2006).Concurrent ValidityYes – criterion validity by comparing pain intensity subscale scores with a single item “sternotomy pain”. Correlation coefficients = 0.72-0.81 (Mendoza et al., 2004).Yes – criterion validity by comparing the change from baseline to day 7 in BPI subscales when patients were categorised into “improved”, “no change”, and “worse” groups according to the Patient’s Global Assessment of Arthritis (PGAA) questionnaire. Patients who reported improved arthritis with the PGAA questionnaire had significant improvement (0.011 < p < 0.001) in all three BPI subscale scores (pain, mood, activity)(Mendoza et al., 2006).Predictive ValidityNoPopulation Validity (Generalizability) The BPI has been translated and tested in several languages: English, French, Spanish, Portuguese, Italian, Greek, Vietnamese, Taiwanese, Korean, Mandarin Chinese, Arabic, Norwegian, German, Russian, Filipino, Japanese, Hindi. It has been tested for several pain conditions: cancer, HIV/AIDS, acute pain, post-operative pain, MSK pain. SensitivityNoSpecificityNoResponsivenessYes – sensitivity to change over time: ANOVA calculations between the BPI and the RDQ scores showed that the BPI subscales discriminated significantly among groups of arthritis patients according to whether they had improved, stayed the same, or declined over time; similar results were found when comparing the BPI subscale scores with the HAQ scores for low back pain patients (Keller et al., 2004).Yes – Standardised Response Means (SRM): SRM values for MSK pain patients (persistent LBP, hip or knee pain) classified as worse, the same, or better on the Global Rating of Change measure at 12 months showed that the measures used (BPI, CPGQ, RDQ, SF-36 BP) had similar effect sizes within each category of change (Krebs et al., 2010).Yes – Area Under the ROC Curve (AUC): AUC values were similar between all measures (BPI, CPGQ, RDQ, SF-36 BP), with fair to good discriminatory ability (AUC = 0.65-0.85) although the CPGQ and SF-36 BP had the lowest values (Krebs et al., 2010).Yes – Standardised Effect Sizes (SES): SES calculated between the intervention and control groups showed small to moderate intervention group effects (SES = 0.41-0.67) for all patient groups (LBP, hip or knee pain). The CPGQ had low overall values (SES = 0.41-0.43) compared to the BPI (0.56-0.64). When results were stratified by pain location, the responsiveness of the CPGQ (SES=0.32-0.44) and the SF-36 BP (SES=0.39) were poor for hip or knee pain patients compared to the BPI (SES=0.58-0.69) (Krebs et al., 2010). Floor / Ceiling EffectsUnknownReferencesAtkinson TM, Rosenfeld BD, Sit L, Mendoza TR, Fruscione M, Lavene D, Shaw M, Li Y, Hay J, Cleeland CS, Scher HI, Breitbart WS, Basch E. Using confirmatory factor analysis to evaluate construct validity of the Brief Pain Inventory (BPI). Journal of Pain & Symptom Management 2011; 41(3): 558-65.Cleeland CS, Ladinsky JL, Serlin RC, Thuy NC. Multidimensional measurement of cancer pain: comparisons of US and Vietnamese patients. Journal of Pain and Symptom Management 1988; 3(1): 23-7. Daut RL, Cleeland CS, Flanery RC. Development of the Wisconsin Brief Pain Questionnaire to assess pain in cancer and other diseases. Pain 1983; 17: 197-210.Dubuisson D, Melzack R. Classification of clinical pain descriptions by multiple group discriminant analysis. Experimental Neurology 1976; 51(2): 480-7. Elliott AM, Smith BH, Smith WC, Chambers WA. Changes in chronic pain severity over time: the Chronic Pain Grade as a valid measure. Pain 2000; 88(3): 303-8. Holroyd KA, Holm JE, Keefe FJ, Turner JA, Bradley LA, Murphy WD, Johnson P, Anderson K, Hinkle AL, O’Malley WB. A multi-center evaluation of the McGill Pain Questionnaire: results from more than 1700 chronic pain patients. Pain 1992; 48(3): 301-11. Keller S, Bann CM, Dodd SL, Schein J, Mendoza TR, Cleeland CS. Validity of the brief pain inventory for use in documenting the outcomes of patients with non-cancer pain. Clinical Journal of Pain 2004; 20(5): 309-18.Krebs EE, Bair MJ, Damush TM, Tu W, Wu J, Kroenke K. Comparative responsiveness of pain outcome measures among primary care patients with musculoskeletal pain. Medical care 2010; 48(11): 1007-14.Melzack R. The short-form McGill Pain Questionnaire. Pain 1987; 30(2): 191-7. Melzack, R. The McGill Pain Questionnaire: major properties and scoring methods. Pain 1975; 1(3): 277-99. Mendoza T, Mayne T, Rublee D, Cleeland C. Reliability and validity of a modified Brief Pain Inventory short form in patients with osteoarthritis. European Journal of Pain 2006; 10(4): 353-61.Mendoza TR, Chen C, Brugger A, Hubbard R, Snabes M, Palmer SN, Zhang Q, Cleeland CS. The utility and validity of the modified brief pain inventory in a multiple-dose postoperative analgesic trial. Clinical Journal of Pain 2004; 20(5): 357-62.Menezes Costa L da C, Maher CG, McAuley JH, Hancock MJ, de Melo Oliveira W, Azevedo DC, Freitas Pozzi LM, Pereira AR, Costa LO. The Brazilian-Portuguese versions of the McGill Pain Questionnaire were reproducible, valid, and responsive in patients with musculoskeletal pain. Journal of clinical epidemiology 2011; 64(8): 903-12. Salaffi F, Stancati A, Grassi W. Reliability and validity of the Italian version of the Chronic Pain Grade questionnaire in patients with musculoskeletal disorders. Clinical rheumatology 2006; 25(5): 619-31. Smith BH, Penny KI, Purves AM, Munro C, Wilson B, Grimshaw J, Chambers WA, Smith WC. The Chronic Pain Grade questionnaire: validation and reliability in postal research. Pain 1997; 71(2): 141-7.Von Korff M, Ormel J, Keefe FJ, Dworkin SF. Grading the severity of chronic pain. Pain 1992; 50(2): 133-49. ................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download