Cross-Cultural Adaptation and Validation ofthe Profile of Chronic Pain: Screen for aBrazilian Population

Wolnei Caumo, MD, PhD,*†‡§

Linda S. Ruehlman, PhD,¶ Paul Karoly,¶**Francisléa Sehn, MsH,* Liliane Pinto Vidor, MsH,*Letizzia Dall-Ágnol, MsH,* Mônica Chassot, MsH,*and Iraci L. S. Torres, PhD,*‡§

*Post-Graduate Program in Medical Sciences,School of Medicine, Universidade Federal do RioGrande do Sul (UFRGS), Porto Alegre, Rio Grandedo Sul;

†Pain and Palliative Care Service at Hospital deClínicas de Porto Alegre (HCPA), UniversidadeFederal do Rio Grande do Sul (UFRGS), Porto Alegre,Rio Grande do Sul;

‡Laboratory of Pain & Neuromodulation,HCPA/UFRGS, Porto Alegre, Rio Grande do Sul;

§Pharmacology Department, Instituto de CiênciasBásicas da Saúde, UFRGS, Porto Alegre, Rio Grandedo Sul, Brazil;

¶Goalistics LLC, Department of Psychology, ArizonaState University, Tempe, Arizona;

**Department of Psychology, Arizona State University,Tempe, Arizona, USA

Reprint requests to: Wolnei Caumo, MD, PhD,Laboratory of Pain & Neuromodulation, Hospital deClínicas de Porto Alegre, Rua Ramiro Barcelos, 2350 -CEP 90035-003, Bairro Rio Branco, Porto Alegre,RS 90035-003, Brazil. Tel: (55) 51 33598083;Fax: (55) 51 33085621; E-mail: caumo@cpovo.net.

Declaration of Interests: The authors declare thatthere are no financial or other relationships that mightlead to conflicts of interest.

Competing Interest Statement: The present researchwas supported by the following Brazilian agencies:research grant from CNPq (Dr. I. L. S. Torres andDr. W. Caumo); and assistance administrative supportfrom the Postgraduate Research Group at theHospital de Clínicas de Porto Alegre. The institution(HCPA, UFRGS) received support from the followinggovernmental Brazilian agencies: FAPERGS; CNPqand CAPES.

Name of department where the work was conducted:Pain and Palliative Care Service at HCPA/UFRGS.

Abstract

Objective. To translate the original English versionof the Profile of Chronic Pain: Screen (PCP:S) intoBrazilian Portuguese and examine basic psychomet-ric properties of the translated version. We investi-gated ceiling and floor effects, internal consistency,factor structure, convergent validity, and the abilityof the Brazilian PCP:S (B-PCP:S) to discriminatepersons with pain who were either employed or notworking, or in treatment or not in treatment.

Methods. The Brazilian Portuguese version of theProfile of Chronic Pain: Screen (B-PCP:S) wasadministered to a sample of 414 adults (men = 67).Pain catastrophizing was also assessed. Sub-samples with special conditions (working despitepain [N = 116] vs not working due to pain [N = 122],and not receiving treatment for pain [N = 119] vsreceiving treatment [N = 119]) were identified toinvestigate the discriminative properties of B-PCP:S.

Results. For the B-PCP:S, Cronbach’s a valueswere 0.76 (severity), 0.88 (interference), and 0.87(emotional burden). Confirmatory factor analysissupported the original, English language three-factor structure, with the comparative fit index =0.93, root mean square error of approxima-tion = 0.075, and normed fit index = 0.93. Significantcorrelations were found between pain intensity, paininterference, and emotional burden, and a criterionmeasure of catastrophizing (correlation coefficientsranged from 0.48 to 0.66, P < 0.01). B-PCP:S scores(severity, interference, and emotional burden) werehigher in subjects under a doctor’s care for pain andin those not working due to pain.

Conclusion. This B-PCP:S version was found tobe a reliable instrument, with basic evidence of vali-dity for the evaluation of pain severity, interference,and emotional burden in Brazilian Portuguese adults.The profile of B-PCP:S scores was similar to thatobserved in the original version.

Key Words. Confirmatory Factor Analysis; Cross-Cultural Adaptation; Factorial Analysis; PainAssessment; Profile of Chronic Pain: Screen(PCP:S)

bs_bs_banner

Pain Medicine 2012; *: **–**Wiley Periodicals, Inc.

1

Introduction

Prevalence estimates for chronic noncancer pain in theprimary care settings range from 5% to 33% [1]. A recentmeta-analysis showed that chronic musculoskeletal in aBrazilian sample ranges from 14.1% to 85.5% [2]. In areview of 13 chronic pain studies conducted in variouscountries around the world, the International Associationfor the Study of Pain found prevalence rates ranging from10.1% to 55.2% [3]. The high prevalence of chronic painobserved in Brazil placed it as one of the countries that ismost affected by this problem.

Chronic pain is a debilitating condition whose multifacetednature encompasses physical, psychological, and behav-ioral characteristics [4]. It is associated with significantsuffering, decreased quality of life, higher incidence ofabsenteeism, and more prevalent symptoms of depres-sion, anxiety, and catastrophic thinking, and differs inimportant ways from acute pain [5]. Further, the methodsused for acute pain screening are insufficient to provide acomplete picture of the multidimensional impact ofchronic pain upon the lives of those who endure it.

Chronic pain can be defined as continuous or intermittentpain or discomfort that has persisted for 3 months ormore, and is associated with the frequent seeking of treat-ment or the use of analgesic medications [6,7]. It has beenargued that the assessment of chronic pain should gaugethe nature of a person’s experience in terms of pain sever-ity, pain-related functional interference, and the emotionalburden associated with unremitting discomfort [8–10]. Tothat end, the 15-item Profile of Chronic Pain: Screen(PCP:S) [11] allows for quick identification of an individu-al’s multidimensional pain experience. The PCP:S com-prised a severity scale (four items; possible range 0–32),an interference scale (six items; possible range 0–36), andan emotional burden scale (five items; possible range0–25) [11]. The importance of these three dimensions(severity, interference, and emotional burden) has recentlybeen underscored by the Initiative on Methods, Measure-ment, and Pain Assessment in Clinical Trials [12], a teamof pain experts drawn from academia, government, aself-help organization, and the pharmaceutical industry. Itis now understood that chronic pain is a multidimensionalexperience that incorporates numerous factors related toits assessment and treatment, and that it has been widelyunderdiagnosed, and consequently undertreated. Thus,the importance of having a reliable assessment instrumentto use in different cultures justifies the translation of thePCP:S and the validation of a Brazilian Portuguese versionof the PCP:S.

Several assessment instruments target the key dimen-sions of pain noted above, as well as related aspectsof the pain experience and its consequences. Among themost popular assessment devices are the McGill PainQuestionnaire (MPQ) [13], the West Haven–Yale Multidi-mensional Pain Inventory (MPI) [14,15], and a number ofspecific functional disability assessment devices [16].While the West Haven–Yale MPI measures the domains of

interest and possesses a variety of norms, it is too long fora screen, containing more than 60 questions. The MPQhas strong psychometric properties and normative databut does not measure all three areas of interest. Further,the instructions for completing the MPQ pain adjectivessection are complex, and it is recommended that theyshould be read to patients. However, the 15-item PCP-Shas strong psychometric properties, has been used inboth clinical and research settings, has national norms, isself-administered in paper and pencil or online format, andtakes approximately 5–10 minutes to complete.

Thus, the aim of the present study was to cross-culturallyadapt the English version of the PCP:S into Brazilian Por-tuguese, and to then examine the psychometric proper-ties of the translated instrument. We investigated theceiling and floor effects, internal consistency, factor struc-ture, and correlations of the Brazilian Portuguese versionof the Profile of Chronic Pain: Screen (B-PCP:S) withrelevant correlates of chronic pain, such as the level ofcatastrophizing, an important correlate of dysfunctionalchronic pain adaptation [17]. Further, we assessed theability of the B-PCP:S to discriminate among patientsacross specific conditions (working vs not working; intreatment vs not in treatment).

Methods

The protocol was approved by the local ethics committeeof the institution in which the study was developed (Appli-cation No. 1005-55—Postgraduate Research Group atHospital de Clínicas de Porto Alegre). The sample wasrecruited in a tertiary chronic pain clinic, in which allpatients are referred from the primary units. In addition, wealso recruited through newspaper advertisements. Theinclusion criterion was diagnosis of musculoskeletal painby physicians with clinical experience in pain treatment.The diagnoses were established using standard protocolsfor each of one type of pain, that is, the clinical criteria todefine myofascial pain syndrome patients were regionalpain, normal neurologic examination, presence of triggerpoints, taut bands, tender points, and pain characterizedas “dull,” “achy,” or “deep” [18]. The chronic tensionalheadache diagnosis was according to the InternationalHeadache Society [19], and the diagnosis of fibromyalgiawas given according to the American College of Rheuma-tology criteria for fibromyalgia [20]. The exclusion criteriawere inability to understand Brazilian Portuguese, illiteracy,or inability to come to the hospital for evaluations [21].The multiple standardized study phases are presentedin Figure 1.

PCP:S

The PCP:S consists of four questions related to painseverity, six questions related to pain’s interference withfunctioning, and five questions related to emotionalburden. Two of the pain severity items are presentedas numeric rating scales (NRS), with 0 = no pain and10 = unbearable pain. Two of the items are scored on a 0(never) to 6 (daily) scale. The PCP:S instructs patients to

2

Caumo et al.

rate the frequency of any pain, the frequency of severepain, average pain, and greatest pain intensity over thepast 6 months. A pain severity index can be calculated byadding the scores on the pain severity items (four items;possible range 0–30). The six items assessing pain inter-ference on patient function are also presented as NRSscales, with 0 = never interferes to 6 = interferes daily. Theinterference items ask how often pain interferes with suchactivities as hobbies, basic self-care, housework, relationswith others, personal goals, etc. (six items; possible range0–36). Emotional burden items ascertain how often painhas caused the person to feel sad or depressed, anxious,angry, isolated, or to experience reduced enjoyment of life,with 0 = never to 5 = extremely often (five items; possiblescore range 0–25).

Translation

The translation into Brazilian Portuguese and cross-cultural adaptation of the original English version of thePCP:S [11] was carried out in accordance with previouslypublished guidelines [21–23]. Four native Brazilian Portu-

guese speakers (translators 1–4: T1, T2, T3, T4) carriedout independent translations of the PCP:S from English toBrazilian Portuguese. T1 was a professional translator, T2was a clinical psychiatrist, T3 was a physician with painspecialization, and T4 was a linguist. The forward transla-tions were compared with one another and with the origi-nal English version.

Back-Translation

Two native English speakers with Brazilian Portuguese astheir second language carried out the back-translation ofthe Brazilian Portuguese version into English. Both back-translators were considered bilingual, according to thedefinition of Deyo [24]. None of the back-translators werefamiliar with the subject matter of the questionnaire. A thirdbilingual person highlighted any conceptual errors or grossinconsistencies in the content of the translated versions inpreparation for the expert committee meeting. Ten profes-sionals who work with patients with chronic pain assessedthe meaning of the translated questions and the layout ofthe prefinal version of the B-PCP:S questionnaire.

I. Translation

II. Synthesis

III. Back-translation

IV. Expert committee

review

V. Pretesting

VI. To assess the internal consistency, concurrent

validity, and exploratory factor analysis

VII. To assess the capability

of B-PCP: S to discriminating

Two translations (T1 & T2)1. Professional translator (uninformed)2. Linguistic

Synthesize (T1; T2; T3; T4) – into T-1,2,3,41. Professional translator (uninformed)2. Linguistic3. Psychiatrist4. Physician with pain specialization

Two English first languageNaïve outcome measuresWork from T-1,2,3,4 versionCreate back-translation

A third bilingual person highlightedany conceptual errors or gross

inconsistencies in thecontent of the translated versions

Resolve any discrepancies withtranslators’ reports

Ten professionals who work with patient with chronic pain assessed themeaning of the translated questions and the layout of the prefinal version of

the Brazilian Profile of Chronic Pain: Screen. In addition, the expertcommittee (all translators and back-translators one methodologist, one

clinical research scientist and the originator of the English version RuehlmanLS and Karoly P). The task of this expert committee was to ensure semantic

and idiomatic equivalence, and experiential and conceptual equivalence

30 patients with chronic musculoskeletal pain weregiven in the prefinal version of the B-BPI: Theyassessed the understanding of the questions of

the B-PCP: S by 10-cm visual analog scales

414 patients with chronic musculoskeletal paincomprised the sample to assess the final

version of the B-PCP: S

To assess the relationship between the treatmentstatus, we compared the B-PCP: S severity,interference and burden emotional between:

119 subjects who was not working due to painand 199 subjects who was working despite pain.116 subjects under doctor’s care for pain and

122 not under doctor’s care for pain

Figure 1 Flow of the multiple standardized phases of the study [11,27].

3

Brazilian Profile of Chronic Pain: Screen

Expert Committee

An expert committee was formed consisting of all of thetranslators and back-translators, an expert in clinicalresearch methodology, and one clinical research scientist.The task of this expert committee was to ensure semanticand idiomatic equivalence, and experiential and concep-tual equivalence (i.e., to address potential issues specificto the cultures examined) between the Brazilian Portu-guese and the English versions of the questionnaire. Thechanges to the B-PCP:S were based on the consensus ofmembers involved in the translation process. We used theprinciple of translation from the English to the Brazilianversion of the B-PCP:S. This principle underscores thatsentences in different language versions should have thesame meanings and refer to the same underlying concept.Therefore, our goal was to obtain a better idiomatic andconceptual rather than literal equivalence.

Assessment of Psychometric Properties of the FinalVersion of Portuguese Brazilian Version of B-PCP:S

Four hundred fourteen patients were recruited from theoutpatient Pain Clinic and Palliative Care at Clínicas Hos-pital of Porto Alegre. All subjects were recruited from aconvenience sample, and they gave their signed informedconsent to participate. Before beginning the interviews,study participants completed a baseline questionnairethat included questions about sociodemographic vari-ables (age, gender, work status due pain, doctor’s care totreat pain, marital status, and education), and the BrazilianPain Catastrophizing Scale (PCS) [25,26].

Assessment of Pain Catastrophizing

To evaluate convergent validity, we used a version of thePCS adapted for the Brazilian population [25,26]. Itis a self-administered questionnaire consists of 13 itemsthat assess the extent of the patient’s catastrophizingthoughts and behaviors. It comprised three subscales:helplessness, magnification, and rumination. The ques-tionnaire is completed in relation to the patient’s thoughtsand feelings when they are in pain. The total score, whichranges from 0 to 52, is computed by summing all items onthe questionnaire [26].

Statistical Analysis

Floor and ceiling effects were determined in two ways:first, we simply calculated the number of individualsobtaining, respectively, the lowest or highest possiblescores on the severity scale (range 0–30), interferencescale (range 0–36), and emotional burden scale (range0–25), where a limit of 15% of patients should not beexceeded [27,28]; and second, we computed the propor-tion of individuals obtaining a score within the limits of theminimum detectable change (95% confidence interval [CI])at the two ends of the scale (Table 2).

Internal consistency reliability was assessed using Cron-bach’s a (and Cronbach’s a if the item deleted) calculated

for B-PCP:S severity, interference, and emotional burdenof B-PCP:S to compare with the English version.

A confirmatory factor analysis (CFA), with a three-factormodel (severity, interference, and emotional burden) asdescribed by Ruehlman and colleagues [11], was con-ducted. CFA was utilized to investigate whether theestablished dimensionality and factor-loading pattern fitsdata from a new Brazilian sample with chronic pain.M-Plus version 6 was used to analyze the factor struc-ture of B-PCP:S. A weighted least squares estimationprocedure was utilized in a factor structure analysis(M-Plus software, Muthén and Muthén, Los Angeles,CA, USA). Each item was assessed based on the factorcorrelation and t statistic of the parameter estimate.Absolute t values greater than 1.96 were considered sig-nificant at the 0.05 level. These items were retainedbecause they were considered important to the model.In addition, the criteria for removing items were factorloadings below 0.5 and greater than 1.0. The M-Plus 6output was examined for out-of-range values, such asnegative variances or factor correlations greater than 1.Any inadmissible results may have been caused by highmulticollinearity, by outliers, and/or by flaws in the struc-tural equation modeling program. Correlated errorparameters were allowed based on modification indicesand logical reasoning. For example, many items for eachsubscale were similarly worded and measure closelyrelated concepts. One item from each factor wasselected as a marker variable. The chi-square signifi-cance test is used to determine the degree to which aproposed model fits the data. Indices of fit included thecomparative fit index (CFI), the root mean square error ofapproximation (RMSEA), and the standardized rootmean square residual (SRMR). According to recommen-dations by Hu and Bentler [29], a model is deemed toshow adequate fit when a CFI index greater than 0.95 isobserved with an SRMR less than 0.08, or when anRMSEA less than 0.08 is observed in conjunction withan SRMR less than 0.08.

Two approaches to validity assessment were undertaken.The first represents a convergent validity effort in whichscores on the B-PCP:S were correlated with anotherquestionnaire that evaluates the aspects related topain, such as B-PCS, and should result in moderatepositive correlation coefficients, not exceeding 0.7. ThePearson correlation was used to compare the relation-ships between the B-PCP:S severity, interference andemotional burden scales, and catastrophizing as mea-sured by the B-PCS. In the second approach, involvingcriterion group validity, two multivariate analyses of cova-riance (MANCOVA) were conducted to determine theextent to which known groups differed in their scores onthe B-PCP:S scales.

Finally, a stratified-by-gender analysis was used to assessthe correlation between age and education level, and thescores of the dependent variables: B-PCP:S scales (sever-ity, interference, and emotional burden). We employedregression analysis with stepwise forward technique.

4

Caumo et al.

Results

Assessment of the B-PCP:S in Pilot Version

After discussing any discrepancies, the four versions werecombined into one Brazilian Portuguese version. For Item6 (an interference item), the original wording focusedon “responsibilities” at home, but the translationfocuses on “activities” at home as the term “responsibili-ties” at home is not typically employed by BrazilianPortuguese speakers.

This pilot version of the translated PCP:S was completedby a group of 30 subjects of both genders (N = 15women) with chronic musculoskeletal pain. They werealso interviewed in order to collect qualitative data byexploring what subjects thought was meant by eachquestion and the chosen response. They were also askedfor their general comments on the questionnaire. Theirmean � SD of the number of formal schooling was18.93 � 3.99, and the mean age was 33.95 � 10.50. Inaddition, we assessed their understanding of the ques-tions of the B-PCP:S using 10-cm visual analog scales (0meaning completely unclear, and 10 cm indicating com-pletely clear). The global mean � SD of comprehensionof the 15 questions of BPI was 8.41 � 1.90. All thefindings were evaluated by the workgroup (to assess facevalidity), after which two questions were slightly modifiedto achieve the final Brazilian Portuguese version ofthe PCP:S.

Assessment of the Final Version of B-PCP:S

Four hundred fourteen patients were recruited fromthe outpatients of Pain Clinic and Palliative Care at

Clinics Hospital of Porto Alegre (N = 347 women) withchronic musculoskeletal pain, with a mean (�SD) on edu-cation of 11.03 (�5.43) years and a mean age of50.23 � 17.10. In view of the disproportionate number offemales in our sample, we analyzed the mean scores formales and females and compared them. We observedstatistically significant difference between males andfemales (Table 1).

Distribution of the B-PCP:S Scores

There were no missing data for any item. The distributionof scores is shown in Table 2. The scores ranged from theminimum to the maximum for all 15 items. There isminimal ceiling effect of the highest score possible in theB-PCP:S scales. The ceiling effect is 5.5 for the severityscale, 9.9 for the emotional burden scale, and 8.2 for theinterference scale. The floor effect for the lowest possiblescore was 0.2 for the severity scale and 2.9 for the emo-tional burden scale.

Internal Consistency Reliability

To assess the internal consistency reliability of theB-PCP:S, Cronbach’s a coefficients were computed forthe four severity items, six interference items, and fiveemotional burden items. The alpha coefficient was 0.76for the severity scale, 0.88 for the interference scale, and0.87 for the burden emotional scale. The alpha values forthe scales if an item was deleted were comparable to theoverall alpha coefficient for the scales (Table 3), whichindicates that each of the items contributes similarly to theconstruct it is intended to measure.

Table 1 Mean (SD) domain scores of B-PCP:S: For males and females

Range of the ScoresMen (N = 67) Women (N = 347)

PMean (SD) Mean (SD)

Severity 0–32 18.83 � 6.67 22.09 � 5.44 <0.0001Interference 0–36 15.68 � 9.95 22.07 � 9.30 <0.0001Emotional burden 0–25 10.52 � 6.21 15.68 � 6.93 <0.0001Total B-PCP:S 0–93 45.50 � 21.19 59.64 � 18.64 <0.0001

The t test for independent samples was used to compare these two groups (N = 414).B-PCP:S = Brazilian Profile of Chronic Pain: Screen; SD = standard deviation.

Table 2 Mean (SD), median (range), floor and ceiling effects, and tests of normality of distribution(N = 414)

Scale Mean (SD) Median (Range) Ceiling Effect (%) Floor Effect (%) Skewness Kurtosis

Severity 21.56 � 5.77 22.00 (19–26) 5.5 0.2 -0.84 0.64Interference 21.05 � 9.68 22.00 (14–29) 8.2 1.4 -0.19 -0.90Emotional burden 14.76 � 7.22 16.00 (9–20) 9.9 2.9 -0.27 -0.99

SD = standard deviation.

5

Brazilian Profile of Chronic Pain: Screen

CFA

CFA revealed that all items were significantly related totheir specified factors, verifying the hypothesized relation-ships among the item and latent factors. Figure 2 showsthe diagram as well as factor loadings generated for thehypothesized model by Ruehlman and colleagues [11].

The analysis elicited adequate model goodness of fit(Figure 2). Results show consistency in an adequatemodel fit by all of the goodness-of-fit measures(CFI = 0.93; SRMR = 0.048; RMSEA = 0.075; 95% CIrange 0.073, 0.093). Standardized factor loadings arepresented in Figure 2. All loadings differed significantlyfrom zero.

Table 3 Reliability analysis and a values if item deleted* (N = 414)

Pain Severity Items(a = 0.76)

Pain InterferenceItems (a = 0.88)

Burden EmotionalItems (a = 0.87)

Pain worst 0.76 General activity 0.85 Mood 0.82Pain least 0.66 At home 0.85 Anxiety 0.82Pain average 0.65 Relationship to others 0.85 Angry 0.86Pain 0.69 Personal goals 0.85 Lonely 0.84

Self-care 0.87 Enjoy 0.84Concentration 0.86

* Cronbach’s a for individual items represents the internal consistency of the scale with the removal of the item.

Figure 2 Three-factor model of the B-PCP:S with standardized parameter estimates. Confirmatory factoranalysis with parameter estimates for the error variances.

6

Caumo et al.

Convergent Validity

Construct validity of the B-PCP:S was further establishedby examining the pattern of correlations between the totalscore on the severity, interference, and burden emotionalscales, and the criterion measure of catastrophizing.Pearson correlations were performed between the scoreson the B-PCP:S scales, and the helplessness, magnifica-tion, and rumination subscales, and with the total score ofB-PCS (Table 4).

The Brazilian version of B-PCP:S showed evidence ofconvergent validity (Tables 5 and 6). Similar to the origi-

nal validation study of the PCP:S [11], it was expectedthat individuals who were in treatment for their pain andthose who were unable to work due to pain would scorehigher on pain severity, interference, and/or emotionalburden than would subjects who were not in treatmentor were able to work, despite their pain. Age served asa covariate in MANCOVA in which the three subscalesserved as multiple dependent variables. The followinggroups were compared: 1) working despite pain vs notworking due to pain, 2) not receiving treatment for painvs receiving treatment. Each of the validity MANCOVAsinvolved subgroups of targeted respondents, that is,those who are in treatment for pain or those who are not

Table 4 Pearson correlations between B-PCP:S scale and its subscales, and painBrazilian-Catastrophizing Pain Scale and its subscales (N = 414)

B-PCS-Helplessness B-PCS-Magnification B-PCS-Rumination B-PCS-Total Scale

Severity rs = 0.56* rs = 0.48* rs = 0.51* rs = 0.58*Interference rs = 0.59* rs = 0.50* rs = 0.54* rs = 0.57*Emotional burden rs = 0.63* rs = 0.53* rs = 0.50* rs = 0.61*

* P < 0.001.B-PCS = Brazilian Pain Catastrophizing Scale; B-PCP:S = Brazilian Profile of Chronic Pain: Screen.

Table 5 MANCOVA for the treatment status: Evaluation of group differences in severity, interference, andemotional burden with age as a covariate (N = 238)

Under Doctor’s Carefor Pain ([N = 116]Female 97/Male 19)

Not under Doctor’s Carefor Pain ([N = 122]Female 101/Male 21)

FMean SD Mean SD

Severity 25.06 3.52 20.39 6.00 299.06*Interference 31.44 4.82 17.66 7.01 137.02*Emotional burden 21.12 4.85 13.34 6.64 275.26*

Wilks’ Lambda = 0.88; F(3,232) = 138.56; * P < 0.001.MANCOVA = multivariate analysis of covariance; SD = standard deviation.

Table 6 MANCOVA for the work status: Evaluation of group differences in severity, interference, andemotional burden (N = 238)

Not Working Due toPain ([N = 119]Female 102/Male 17)

Working DespitePain ([N = 119]Female = 102/Male 17)

FMean SD Mean SD

Severity 24.87 3.83 20.47 5.98 52.76*Interference 30.98 5.63 17.78 8.12 105.56*Emotional burden 20.63 5.97 13.63 6.55 251.47*

Wilks’ Lambda = 0.52; F (3,234) = 70.91; * P < 0.001.MANCOVA = multivariate analysis of covariance; SD = standard deviation.

7

Brazilian Profile of Chronic Pain: Screen

working due to their pain. We sought to compare thesetargeted subgroups to remaining respondents in thesample. However, discrepancies in sample sizes can beproblematic for MANCOVA, which becomes less robustto violations of assumptions as these discrepanciesincrease [30]. To reduce sample size differences amongthe groups, a subsample of respondents was randomlyselected from the larger “comparison” group prior toeach analysis so as to achieve equal sample sizesacross groups. The sample size for the comparisongroup was chosen to be equal to that of the targetgroup; the comparison group was then selected atrandom from the larger sample.

Significant omnibus tests were followed with univariateanalyses of covariance to examine group differences foreach dependent variable. As shown in Tables 5 and 6,both omnibus tests across the two MANCOVAs weresignificant, indicating that, in each analysis, each pair ofgroups differed significantly on the set of scales of theB-PCP:S. In addition, all univariate follow-up tests weresignificant, indicating, in each case, that the groupsdiffered significantly on each of the three scales of ourinstrument in accordance with our predictions.

Table 5 demonstrates that those under doctor’s care forpain have significantly higher scores than those not undera doctor’s care for pain on each one of B-PCP:S scales(severity, interference, and emotional burden).

Table 6 indicates that those not working due to pain havesignificantly higher scores than those who were workingdespite their pain in each one of B-PCP:S scales (severity,interference, and emotional burden).

Finally, the association between gender, age, educationlevel, and each one of B-PCP:S scales—severity, inter-ference, and emotional burden—was analyzed by regres-sion analysis in a forward stepwise model. The factorwas the gender. In males, only the variable number ofyears of schooling was retained in the model with thedependent variable severity with b coefficient -0.36 (95%CI; -0.73 to -0.16; P = 0.002). Age and schooling werenot retained in the model. Interference and emotionalburden were not associated with the covariates ageand schooling. For females, when the dependent variablewas severity, the b coefficient for schooling was -0.17(95% CI; -0.28 to -0.06; P = 0.002) and 0.13 (95% CI;0.008 to 0.23; P = 0.01) for age. When the dependentvariable was emotional burden, the b coefficient forschooling was -0.18 (95% CI; -0.37 to -0.10; P =0.001). Interference was not associated either with ageor with schooling.

Discussion

The goal of the present research was to translate thePCP:S to a Brazilian Portuguese version, and to evaluateand verify basic psychometric properties. The process oftranslating and back-translating the English PCP:S to Bra-zilian Portuguese version was carried out strictly in

accordance with established guidelines [21], and with theassistance of a diverse panel of experts as well as personswith pain. This process yielded a Brazilian Portugueseversion that is semantically equivalent to the English-language PCP:S, while reflecting cultural variations. Thus,the current version can be used without major difficultyin Portuguese-speaking populations. Further, minimalrespondent burden was observed, with a completion timeof approximately 5–10 minutes.

Having established a Brazilian Portuguese version of thePCP:S, we examined a number of basic psychometricproperties. Minimal floor and ceiling effects were revealedusing the traditional approach to analyzing such effects.Further, the scales demonstrate adequate variation inrange of scores, that is, the scores did not show largeconcentration near upper or lower limits. Kurtosis andskewness tests suggested a normality of the distributionof scores, which support the use of parametric statisticalanalysis. Overall, the B-PCP:S showed satisfactory psy-chometric properties. The test for internal consistency,Cronbach’s a, indicates that the participants showedadequate consistency in their responses (see Table 3) andwere similar to alpha coefficients obtained in previousstudy [31].

CFA of the B-PCP:S using a variety of different goodness-of-fit measures suggested an adequate model fit. TheBrazilian translation shows satisfactory goodness of fitwith three factors, consistent with the model suggested byRuehlman and colleagues [11]. Overall, the items con-tained in each of the three scales should, therefore, remainas proposed by Ruehlman and colleagues [11]. Our CFAsuggests that it is possible to maintain the original struc-ture scale items in the refined Brazilian translation ofthe B-PCP:S.

The convergent validity of B-PCP:S was examined byinvestigating the strength of the relationship between theB-PCP:S scales and the scores for other pain-relatedaspects, such as catastrophic thinking. The correlationcoefficients for these relationships ranged from 0.48 to0.71 (see Table 4). This represents moderate agreement,which confirms that the B-PCP:S assesses a differentconstruct, and it can thus be considered suitable as partof a multidimensional battery of assessments in chronicpatients. The fact that most of the correlation coefficientswere higher than 0.5 allows us to infer that this translatedscale measures a comparable concept, a criterionusually considered satisfactory for establishing constructvalidity [32].

The B-PCP:S version of severity, interference, and emo-tional burden scales showed criterion-group validity todifferentiate between subgroups. This finding indicatesthe capability of all three scales to discriminate in thecomparison involving 1) individuals under doctor care forpain vs not under doctor care for pain, and 2) notworking due to pain vs working despite pain. In sum,these findings provide evidence to support the validity ofB-PCP:S. This result is similar to the findings from a

8

Caumo et al.

previous study involving chronic pain patients recruitedfrom primary care [31]. Establishing the validity of aninstrument is an intensive process. The data obtainedherein provide a solid foundation from which toexplore validity in more depth. For example, it wouldbe useful to evaluate the predictive validity of the toolas well as its sensitivity to change in conjunctionwith treatment.

Gender differences are not uncommon in research in thearea of chronic pain [33]. In the present study, for bothmales and females, years of schooling were associatedwith scores on the severity scale. However, in females,age was correlated with severity and schooling with emo-tional burden. Future research will need to examine thiscomplex network of relationships more carefully, and inparticular to evaluate these associations over time. Ingeneral, the explanation for gender differences in pain hasbeen proposed to be partially attributed to biologicalfactors (hormones, brain-derived neurotrophic factor, etc.)[34] and psychosocial factors. In addition, cultural factorsmay influence gender differences in attitudes and healthbehavior [35].

The study is limited by the nonrandom selection ofpatients from an outpatient pain clinic. Selection bias waspossible, and it is uncertain whether these findings can beextrapolated to chronic pain patients from different set-tings, such as inpatients, those receiving care from theirfamily physician, or those who are not receiving any treat-ment for their pain. However, it is noteworthy that ourresults are consistent with findings observed in the originalEnglish language version, which involved a variety ofdevelopment samples, including primary care, community,and a representative national sample [12]. Future researchcould evaluate the utility of the instrument in a variety ofclinical settings.

In summary, this study demonstrates that the originalEnglish version of the PCP:S successfully adapted toPortuguese as methodologically demonstrated herein.B-PCP:S constitutes a reliable research instrument forevaluating severity, interference, and emotional burdenrelated to pain. It can be conveniently used to assesshealth status as it is completed both easily andquickly. Chronic pain is a consistent predictor of pooroutcomes and warrants attention in treatment; it mightbe useful to incorporate additional outcome measuresinto a standardized assessment battery. This study pro-vides evidence for the validity and reliability of theB-PCP:S. Our results have demonstrated that the psy-chometric properties of the B-PCP:S were satisfactory.Also, the B-PCP:S showed good discriminative proper-ties. The B-PCP:S represents a valuable tool for use inscientific studies and in clinical setting involving patientswith chronic pain in Portuguese-speaking countries.Future studies should investigate whether the B-PCP:Sis sufficiently sensitive to detect a meaningful clinicalchange in chronic pain conditions over time, before andafter a treatment, and its predictive properties of inter-ventions effect.

Acknowledgment

We would like to express our appreciation to Linda S.Ruehlman and Paul Karoly for their permission to use theProfile of Chronic Pain: Screen.

References1 Reid MC, Engles-Horton LL, Weber MB, et al. Use of

opioid medications for chronic noncancer pain syn-dromes in primary care. J Gen Intern Med 2002;17:173–9.

2 Miranda VS, Decarvalho VB, Machado LA, Dias JM.Prevalence of chronic musculoskeletal disorders inelderly Brazilians: A systematic review of the literature.BMC Musculoskelet Disord 2012;29:13–82.

3 Ospina M, Harstall C. Prevalence of chronic pain: Anoverview. Edmonton: Alberta Heritage Foundation forMedical Research, Health Technology Assessment;2002;1–60.

4 Cimmino MA, Ferrone C, Cutolo M. Epidemiology ofchronic musculoskeletal pain. Best Pract Res ClinRheumatol 2011;25:173–83.

5 Kendrick T, Pilling S. Common mental healthdisorders—Identification and pathways to care: NICEclinical guideline. Br J Gen Pract 2012;62:47–9.

6 McQuay HJ, Moore RA, Eccleston C, Morley S, Will-iams AC. Systematic review of outpatient services forchronic pain control. Health Technol Assess 1997;1:i–iv, 1–135.

7 Smith BH, Hopton JL, Chambers WA. Chronic pain inprimary care. Fam Pract 1999;16:475–82.

8 Asmundson GJ, Norton PJ, Norton GR. Beyond pain:The role of fear and avoidance in chronicity. ClinPsychol Rev 1999;19:97–119.

9 Chapman CR, Nakamura Y. A passion of the soul: Anintroduction to pain for consciousness researchers.Conscious Cogn 1999;8:391–422.

10 Von Korff M, Jensen MP, Karoly P. Assessing globalpain severity by self-report in clinical and healthservices research. Spine 2000;25:3140–51.

11 Ruehlman LS, Karoly P, Newton C, Aiken LS. Thedevelopment and preliminary validation of a briefmeasure of chronic pain impact for use in the generalpopulation. Pain 2005;113:82–90.

12 Dworkin RH, Turk DC, Farrar JT, et al. Core outcomemeasures for chronic pain clinical trials: IMMPACTrecommendations. Pain 2005;113:9–19.

13 Melzack R. The McGill Pain Questionnaire: Majorproperties and scoring methods. Pain 1975;1:277–99.

9

Brazilian Profile of Chronic Pain: Screen

14 Jacob MC, Kerns RD. Assessment of the psychoso-cial context of the experience of chronic pain. In: TurkDC, Melzack R, eds. Handbook of Pain Assessment.New York: Guilford; 2001:362–85.

15 Kerns RD, Turk DC, Rudy TE. The West Haven-YaleMultidimensional Pain Inventory (WHYMPI). Pain1985;23:345–56.

16 Battie MC, May L. Physical and occupational therapyassessment approaches. In: Turk DC, Melzack R, eds.Handbook of Pain Assessment (2nd ed.). New York:Guilford; 2001:204–24.

17 Sullivan MJ, Thorn B, Haythornthwaite JA, et al. Theo-retical perspectives on the relation between catastro-phizing and pain. Clin J Pain 2001;17:52–64.

18 Harden RN, Bruehl SP, Gass S, Niemiec C, Barbick B.Signs and symptoms of the myofascial pain syn-drome: A national survey of pain managementproviders. Clin J Pain 2000;16:64–72.

19 Society HCSotIH. The international classification ofheadache disorders. Cephalalgia 2004;24:1–160.

20 Wolfe F, Smythe HA, Yunus MB, et al. The AmericanCollege of Rheumatology 1990 criteria for the classi-fication of fibromyalgia. Report of the MulticenterCriteria Committee. Arthritis Rheum 1990;33:160–72.

21 Beaton DE, Bombardier C, Guillemin F, Ferraz MB.Guidelines for the process of cross-cultural adaptationof self-report measures. Spine 2000;25:3186–91.

22 Guillemin F, Bombardier C, Beaton D. Cross-culturaladaptation of health-related quality of life measures:Literature review and proposed guidelines. J ClinEpidemiol 1993;46:1417–32.

23 Wagner AK, Gandek B, Aaronson NK, et al. Cross-cultural comparisons of the content of SF-36 transla-tions across 10 countries: Results from the IQOLAProject. International Quality of Life Assessment. J ClinEpidemiol 1998;51:925–32.

24 Deyo RA. Pitfalls in measuring the health status ofMexican Americans: Comparative validity of theEnglish and Spanish sickness impact profile. Am JPublic Health 1984;74:569–73.

25 Sullivan MJL, Bishop SR, Pivik J. The Pain Catastro-phizing Scale: Development and validation. PsycholAssess 1995;7:524–32.

26 Sehn FC, Vidor E, Dall-Agnol LP, et al. Adaptationand validation of the Brazilian Portuguese Versionof the Pain Catastrophizing Scale. Pain Med 2012.doi:10.1111/j.1526-4637.2012.01492.x.

27 McHorney CA, Tarlov AR. Individual-patient monitor-ing in clinical practice: Are available health statussurveys adequate? Qual Life Res 1995;4:293–307.

28 Terwee CB, Bot SD, de Boer MR, et al. Quality criteriawere proposed for measurement properties of healthstatus questionnaires. J Clin Epidemiol 2007;60:34–42.

29 Hu L, Bentler PM. Cutoff criteria for fit indexes incovariance structure analysis: Conventional criteria vsnew alternatives. Struct Equat Mod 1999;6:1–55.

30 Tabachnick BG, Fidell LS. Using Multivariate Statistics,4th edition. Needham Heights, MA: Allyn & Bacon;2001.

31 Karoly P, Ruehlman LS, Aiken LS, Todd M, Newton C.Evaluating chronic pain impact among patients inprimary care: Further validation of a brief assessmentinstrument. Pain Med 2006;7:289–98.

32 Terwee CB, van der Slikke RM, van Lummel RC, et al.Self-reported physical functioning was more influ-enced by pain than performance-based physical func-tioning in knee-osteoarthritis patients. J Clin Epidemiol2006;59:724–31.

33 Fillingim RB, King CD, Ribeiro-Dasilva MC, Rahim-Williams B, Riley JL, 3rd. Sex, gender, and pain: Areview of recent clinical and experimental findings.J Pain 2009;10:447–85.

34 Stefani LC, Torres IL, de Souza IC, et al. BDNF as aneffect modifier for gender effects on pain thresholds inhealthy subjects. Neurosci Lett 2012;514:62–6.

35 Ray-Mazumder S. Role of gender, insurance statusand culture in attitudes and health behavior in aUS Chinese student population. Ethn Health 2001;6:197–209.

10

Caumo et al.