Translation, Cross-cultural Adaptation to Brazilian Portuguese, and Analysis of Measurement Properties of the Consultation and Relational Empathy Measure

Abstract

Objective

The purpose of this study was to translate, cross-culturally adapt to Brazilian Portuguese, and analyze the measurement properties of the Consultation and Relational Empathy (CARE) Measure and investigate whether empathy can be a predictor of clinical improvement.

Methods

This psychometric study was divided into 2 stages: the cross-cultural adaptation process included 30 patients, and the evaluation of the measurement properties included 106 patients with chronic musculoskeletal pain. After the third therapy session with the same physical therapist, the following questionnaires were applied to assess internal consistency, construct validity, and ceiling and floor effects: Pain Numerical Rating Scale, Brazilian Portuguese version of the CARE Measure (CARE-Br), MedRisk Instrument for Measuring Patient Satisfaction with Physical Therapy Care, and Global Perceived Effect Scale. To assess reliability and measurement error, the CARE-Br was answered 48 hours later. For prediction analysis of the CARE-Br in relation to clinical improvement, the participants answered the Pain Numerical Rating Scale and Global Perceived Effect Scale 2 months after baseline.

Results

The internal consistency was adequate (Cronbach’s ɑ = 0.88), reliability was substantial (intraclass correlation coefficient = 0.77), measurement error was good (standard error of the measurement = 5.16%), and a moderate correlation was found with the MedRisk Instrument for Measuring Patient Satisfaction with Physical Therapy Care (r = 0.50). A ceiling effect was also found (39.6% of participants). Empathy was not considered a predictor of clinical improvement.

Conclusion

The Brazilian Portuguese version of the CARE Measure is reliable, adequate, and applicable in clinical settings and research in Brazil. However, it is not capable of predicting clinical improvement in patients with chronic musculoskeletal pain.

Introduction

The quality of the relationship between health professionals and patients can affect the outcome of treatment.¹ Studies suggest that good care is able to increase the confidence of the patients in their therapist and consequently in the treatment.^{1., 2., 3., 4.} Specifically, patient satisfaction, adherence, and treatment outcomes have been associated with the therapist–patient relationship.^{5., 6., 7.} Empathy is defined as the ability of the health professional to share the negative and positive feelings of the patients,⁸ is considered an important component in the therapist–patient relationship, and has been identified as a determinant of satisfaction in this relationship.^{9., 10.} The empathy of the health professional is highly valued by patients¹¹ and is associated with fewer malpractice complaints.^{12., 13.}

Studies have been conducted to analyze the influence of empathy on clinical outcomes. A systematic review¹⁴ aimed to investigate whether the therapist–patient relationship is related to treatment outcomes in the context of physical rehabilitation and found that the relationship between the therapist and the patient is positively associated with the result of treatment. Other studies have also shown that the therapist–patient relationship can positively affect the clinical improvement of patients with chronic pain.^{15., 16.} However, a recent systematic review¹⁷ concluded that there is no evidence of a strong relationship between therapeutic alliance and pain relief in patients with musculoskeletal disorders, which remains a question for future research considering the lack of studies regarding this subject.

The therapist–patient relationship, with special emphasis on empathy, is assessed through questionnaires commonly used in medical settings, such as the Consultation and Relational Empathy (CARE) Measure.¹⁸ The CARE Measure is a questionnaire that evaluates the health professional’s empathy regarding the patient.¹⁸ It has been validated internationally and is a replicable, relevant, and practical measure for use in the context of health.^{9., 18., 19.} The CARE Measure has adequate internal consistency and can be considered a complete instrument to assess the health professional’s empathy from the patient’s point of view.²⁰ Consequently, it is useful not only in scientific research, but also as a tool for assessment and clinical follow-up.¹⁹

To date, there are no questionnaires in Brazilian Portuguese that assess therapist empathy and have also followed the guidelines for translation and cross-cultural adaptation and analysis of measurement properties recommended by the literature.^{21., 22., 23.} In 2014, a Brazilian study translated the CARE Measure, but its sample included only 20 patients, and it only evaluated internal consistency and construct validity.²⁴ Thus, it is necessary to translate and conduct a complete evaluation of the questionnaire’s measurement properties according to the guidelines proposed in the literature. It is also worth highlighting the importance of providing a reliable tool to assess the quality of consultations in the aspects of “human” medical care.

Therefore, the primary objective of this study was to translate and cross-culturally adapt the CARE Measure to Brazilian Portuguese and test its measurement properties in a sample of patients with chronic musculoskeletal pain. As a secondary objective, the study investigated whether the therapist’s empathy can be a predictor of clinical improvement, taking into account pain intensity and perceived improvement.

Methods

This study was approved by the research ethics committee of Universidade Cidade de São Paulo (CAAE: 35585114.6.0000.0064), and it followed all ethical recommendations. All participants agreed and signed the informed consent form to take part in the study. The original author of the CARE Measure, Dr. Stewart Mercer, gave permission to use the questionnaire by e-mail.

Sample

The study included 136 patients with chronic musculoskeletal pain undergoing free physical therapy treatment in clinics and public hospitals in the city of São Paulo between January and September 2015. The inclusion criteria were ability to read and write in Portuguese, age 18 years or more, chronic musculoskeletal pain lasting more than 3 months,²⁵ and treatment with the same physical therapist. The study excluded patients who had undergone surgery to the site of pain in the past 6 months, had serious illnesses (eg, tumors or infections), suffered traumatic injury to the painful area in the last 3 months, had cognitive impairment, or were pregnant.

Measurement Instruments

Demographic Questionnaire

Information was collected in relation to age, sex, education, weight, height, and site and duration of pain. The patient’s attendance at therapy was also collected directly from the chart.

The CARE Measure

The CARE Measure assesses the therapist–patient relationship from the patient’s point of view, focusing on the therapist’s empathy. This questionnaire consists of 10 questions, and the total score ranges from 10 to 50 points, with each answer worth 1 to 5 points (poor = 1 point, fair = 2 points, good = 3 points, very good = 4 points, excellent = 5 points). The total score is the sum of the points of all the questions; higher scores indicate greater therapist empathy.¹⁸ Up to 2 “does not apply” responses are allowed,^{19., 26., 27.} in which case they are replaced with the average score for the remaining items. In our study, the necessary calculations were performed as suggested by the official website for the CARE Measure.²⁸

Pain Numerical Rating Scale

The Pain Numerical Rating Scale is an 11-point scale in which the patient indicates the pain intensity from 0 (no pain) to 10 (pain as bad as could be). Patients were asked about pain intensity at 3 times: in the last episode, average pain in the last 2 weeks, and pain at the time of evaluation. This scale is available in Brazilian Portuguese, and the values for its measurement properties are similar to those of the original scale.²⁹

Global Perceived Effect Scale

This scale assesses the perceived improvement by comparing the onset of symptoms to the last few days. It is a numerical 11-point scale, ranging from -5 to +5, with -5 being vastly worse, 0 no change, and +5 completely recovered. Higher scores indicate better recovery from the condition.²⁹ This scale is available in Brazilian Portuguese, and its property measurement values are similar to those of the original version.²⁹

MedRisk Instrument for Measuring Patient Satisfaction With Physical Therapy Care

This questionnaire aims to evaluate the satisfaction of patients undergoing physical therapy treatment.³⁰ It has been adapted to Brazilian Portuguese, and its measurement properties are adequate.³⁰ The MedRisk Instrument for Measuring Patient Satisfaction With Physical Therapy Care consists of 13 questions, each scored on a Likert scale of 1 to 5 points, with 1 point for “strongly disagree” and 5 for “strongly agree.” The total score ranges from 13 to 65 points, and the higher the score is, the higher the patient satisfaction.³⁰

Procedures

The study was conducted in 2 stages: (1) translation and cross-cultural adaptation of the CARE Measure to Brazilian Portuguese in a sample of 30 patients in the pretest of the final version, and (2) assessment of the measurement properties of the Brazilian Portuguese version in a sample of 106 patients with chronic musculoskeletal pain.

The process of translation and cross-cultural adaptation followed the recommendations of the COnsensus-based Standards for the selection of health Measurement INstruments,²¹ which analyzes health questionnaires using checklists. The checklist for translation and cross-cultural adaptation is based on the guidelines described by Beaton et al,²² which divides the process into 5 phases:

• •Initial translation: This phase was carried out by 2 bilingual and independent translators, who translated the CARE Measure from English into Brazilian Portuguese. One of the translators (T1) was from the area of health, and the other (T2) was not.
• •Synthesis of translation: In this phase, a meeting was held with the authors and translators (T1 and T2) to compare the 2 versions and analyze any ambiguous expressions. The aim was to reach a consensus and a single version of the translated questionnaire.
• •Back-translation: Two new bilingual translators (one from the area of health and the other not) carried out the back-translation of the questionnaire into English. This step aimed to confirm whether the translated version represented the original version of the questionnaire.
• •Expert committee review: The fourth phase was the formation of a committee of experts made up of health professionals, translators, back-translators, and authors. They analyzed all versions of the questionnaire, discussed differences in translation, revised translations, and then developed the final version with appropriate cultural equivalence.
• •Pretest of the final version: In this phase, the questionnaire was administered to 30 patients with chronic musculoskeletal pain. After that, they were asked about their comprehension of the questionnaire and any difficulties in completing it.

For the assessment of the measurement properties, the patients answered all of the questionnaires at baseline, that is, in the period between the third and fourth physical therapy sessions. We opted for this period because we believed that would be sufficient time for the patient to get to know the therapist. A telephone assessment was conducted 48 hours after the initial assessment only for the CARE Measure. At baseline, we evaluated internal consistency, construct validity, and floor and ceiling effects; at the 48-hour follow-up, we analyzed reliability and measurement error. There was 1 more follow-up 2 months after the baseline to determine whether empathy could be a predictor of clinical improvement. For this, the patients answered the Pain Numerical Rating Scale and Global Perceived Effect Scale.

Statistical Analysis

In this study, the following hypotheses were tested:

• •The Brazilian Portuguese version of the CARE Measure (CARE-Br) will present an adequate level of internal consistency (Cronbach’s ɑ >0.70).
• •The CARE-Br will present acceptable levels of reliability and measurement error in a test–retest with a 48-hour interval. Reliability values are expected to be moderate to substantial, and measurement error is expected to be good to very good.
• •The CARE-Br will present a moderate to good positive correlation with the MedRisk Instrument for Measuring Patient Satisfaction with Physical Therapy Care (MRPS) because both questionnaires measure similar constructs.

Statistical analysis was performed with the software Statistical Packages for Social Sciences for Windows, version 19 (IBM Corp, Armonk, New York). To analyze whether the physical therapist’s empathy was a predictor of clinical improvement, 2 separate univariate linear regression analyses were performed: one between the total CARE-Br score at baseline and the Pain Numerical Rating Scale and another between the total CARE-Br score at baseline and the Global Perceived Effect Scale applied 2 months after baseline (dependent variables). For this analysis, the linear relationship between the variables and the homoscedasticity were evaluated by the scatter plots, and the normality and independence of residuals were evaluated by the histograms. Values of P ≤ .05 were considered to have association.³¹ The measurement properties of the domain’s reliability (internal consistency, measurement error, and reliability) and validity (construct validity) were tested as follows:

• •Internal consistency was used to assess the homogeneity of the questionnaire and was calculated using Cronbach’s ɑ and Cronbach’s ɑ if an item was deleted. Alpha values were adequate when greater than or equal to 0.70 and less than 0.95.²³
• •Reliability was calculated using type 2,1 intraclass correlation coefficient, and its respective 95% CIs. Smaller intraclass correlation coefficient values are classified as poor (less than 0.40), moderate (between 0.40 and 0.75), substantial (0.75 to 0.90), and excellent (greater than 0.90).²³
• •Measurement error was analyzed using standard error of the measurement (SEM) and smallest detectable change (SDC). The SEM is expressed in the measurement units of the instrument and calculated by dividing the standard deviation of the mean of the differences by the square root of 2 (standard deviation of differences / √2).³² The percentage of SEM related to the total score of the questionnaire is considered an important indicator of agreement and should be interpreted as ≤5% very good, >5% and ≤10% good, >10% and ≤20% doubtful, and >20% negative.²³ The SDC was calculated using the formula SDC = 1.645 × √2 × SEM with 90% CI. It reflects the smallest change detected in a patient’s score. Thus, values above the SDC indicate a change above the error in the patient’s score.²³
• •Construct validity was assessed using Pearson’s correlation. The correlation was measured between the total scores in the CARE-Br and the MRPS questionnaire. When r <0.30, the correlation is considered weak; when r ≥0.30 and <0.60 the correlation is considered moderate; and when r ≥0.60 it is considered good.³³

Ceiling and floor effects were evaluated by frequency analysis of the total scores obtained in the baseline. This effect is considered present when more than 15% of the sample obtained the minimum (floor) or maximum (ceiling) score in the questionnaire.²³

Results

The process of translation and cross-cultural adaptation was carried out with 30 participants. The analysis of internal consistency and ceiling and floor effects was carried out with 106 participants. Measurement error and reliability were analyzed with 104 participants because the evaluations of 2 participants were lost in the 48-hour follow-up. Construct validity was analyzed with 50 participants. The sample size of all analyses is considered appropriate by the guidelines.²¹ Table 1 shows the characteristics of the sample, both for the participants in the translation and cross-cultural adaptation process (n = 30), and the participants in the process of analyzing the measurement properties (n = 106). Some information regarding the patient’s pain and the other questionnaires were not collected from the sample of the translation and cross-cultural adaptation phase because they were not relevant for that phase. Regarding attendance at treatment sessions, we chose to present the average of attendance in percentage because the number of sessions varied widely from location to location. These data were collected from 50 participants, and the average attendance rate was 93.6%.

Table 1.

Characteristics of the Participants

Characteristics	Translation (n = 30)	Baseline (n = 106)
Sex
Female	20 (66.7)	80 (75.5)
Weight (kg)	72.8 (8.9)	76.4 (14)
Height (m)	1.6 (0.1)	1.6 (0.1)
Age (y)	57.9 (15.7)	56.8 (12.5)
Pain duration (mo)	98.3 (109.6)	86.5 (98.8)
Location of pain
Lumbar spine	24 (80)	67 (63.2)
Knee	6 (20)	32 (30.2)
Cervical spine	0 (0)	3 (2.8)
Foot	0 (0)	4 (3.8)
Empathy (10-50 points)	45.7 (4.7)	46.1 (4.8)
Pain in the last episode (0-10 points)	NA	7.6 (2.2)
Pain in the last 2 weeks (0-10 points)	NA	6.5 (2.3)
Pain at the time of evaluation (0-10 points)	NA	4.3 (2.8)
Perceived improvement (-5 to + 5 points)	NA	0.8 (3.7)
Satisfaction (13-65 points)	NA	54.7 (6.2)

NOTE. Continuous variables are expressed as mean (standard deviation) and categorical variables are expressed as n (%).

NA, not applicable.

Translation and Cross-Cultural Adaptation Process

The translation process followed the steps described by the guidelines.²² Questions 4, 6, and 9 were revised by the expert committee because they were translated differently by translators T1 and T2. Thus, there were some cross-cultural adaptations. Appendix A shows the different translations and the consensus version. These changes were made so that the target audience could understand the questionnaire better. In the pretest of the final version, the participants answered the questionnaire without difficulty, making it possible to generate the final version of the questionnaire (Appendix B).

Analysis of the Measurement Properties

The values of the analysis of the measurement properties are presented in Table 2. The CARE-Br showed adequate values for internal consistency and Cronbach’s ɑ if an item was deleted. Its reliability was rated as substantial, and the measurement error was rated as good. The correlation with the MRPS questionnaire was used to assess construct validity, which was classified as moderate. Regarding the ceiling and floor effects, a ceiling effect was found, with 39.6% of participants reaching the maximum value of the questionnaire.

Table 2.

Measurement Properties of the CARE-Br

Measurement Properties	Values
Internal consistency
Cronbach’s ɑ (Cronbach’s ɑ if an item was deleted)	0.88 (0.86-0.88)
Reliability: ICC2,1 (95% CI)	0.77 (0.67-0.83)
Measurement error
SEM (points, %)	2.58 (5.16)
SDC (points)	6.00
Construct validity, r (p)
MRPS	0.50 (<0.001)

CARE-Br, Brazilian Portuguese version of the Consultation and Relational Empathy Measure; ICC, intraclass correlation coefficient; MRPS, MedRisk Instrument for Measuring Patient Satisfaction with Physical Therapy Care; SDC, smallest detectable change; SEM, standard error of measurement.

Linear Regression Analysis

The linear regression analysis is shown in Table 3. Two univariate linear regressions were performed: one between the CARE-Br and the Pain Numerical Rating Scale and the other between the CARE-Br and the Global Perceived Effect Scale. These analyses showed that the physical therapist’s empathy was not a predictor of improvement of pain intensity and perceived improvement (P > .05).

Table 3.

Results of the Linear Regression Analysis Using Empathy as the Independent Variable

Dependent Variables	β Coefficient	95% CI	P Value
Pain Numerical Rating Scale	-0.018	-0.319 to 0.267	.860
Global Perceived Effect Scale	0.168	-0.052 to 0.675	.092

CI, confidence interval.

Discussion

The results showed that the CARE-Br has adequate internal consistency, reliability, and measurement error according to the guidelines.²³ Regarding construct validity, the questionnaire showed a moderate correlation with the MRPS, but with a ceiling effect.

The results of the measurement properties of the present study and the other studies on the translation of the CARE Measure are shown in Table 4. These other studies also found adequate values of internal consistency.^{26., 27., 34.} They also analyzed internal consistency using Cronbach’s ɑ, and the values were 0.98 in the Japanese study,²⁶ 0.96 in the Chinese study,²⁷ and 0.72 in the Croatian study.³⁴ The CARE-Br showed a moderate correlation with the MRPS questionnaire, which assesses satisfaction. The Japanese²⁶ and Chinese²⁷ studies also examined construct validity. Both correlated the CARE Measure with a questionnaire assessing patient satisfaction (Spearman’s ρ) and the results were considered statistically significant with good and moderate correlation (r = 0.74 in the Japanese study²⁶ and r = 0.59 in the Chinese study).²⁷ None of these studies^{26., 27., 34.} analyzed measurement error, reliability, or ceiling and floor effects. A previous Brazilian study²⁴ that translated the CARE Measure and assessed some measurement properties also found adequate internal consistency (Cronbach’s ɑ = 0.87) and a weak and moderate correlation with questionnaires that assess empathy. This study²⁴ did not include the sample size required to assess internal consistency and construct validity and used non-validated questionnaires to assess construct validity. Thus, our study performed a more complete assessment of the measurement properties of the CARE Measure considering the guidelines.^{21., 22., 23.}

Table 4.

Measurement Properties of the Translated Versions of the CARE

	CARE Measure Translations
	Mercer et al18 Original (English)	Aomatsu et al26 (Japanese)	Fung et al27 (Mandarin)	Hanževački et al34 (Croatian)	This Study (Brazilian Portuguese)
Internal consistency
Cronbach’s ɑ	0.93	0.98	0.96	0.72	0.88
Reliability
Reliability, ICC2,1 (95% CI)	-	-	-	-	0.77 (0.67-0.83)
Standard error of measurement (points, %)	-	-	-	-	2.58 (5.16)

NOTE. A dash ( - ) indicates data not presented by the authors.

CARE, Consultation and Relational Empathy; CI, confidence interval; ICC, intraclass correlation coefficient.

All studies considered the CARE Measure a suitable and reliable questionnaire.^{26., 27., 34.} None of these studies performed a complete analysis of the reliability and ceiling and floor effects, which is essential to determine whether a questionnaire is reliable and suitable for application in all its contexts.^{21., 23.} The present study found a ceiling effect in 39.6% of the participants.

The original study of the development of the CARE Measure analyzed internal consistency, construct validity, and ceiling and floor effects.¹⁸ The internal consistency analysis showed a Cronbach’s ɑ value of 0.93. The original study also found a ceiling effect. In an attempt to reduce this ceiling effect, 2 pilot studies were conducted before reaching the final version. A ceiling effect was found in both pilot studies and in the final version. The ceiling effect was 42% in the first pilot study, 27% in the second, and 26% in the final version,¹⁸ that is, the changes were not sufficient to reduce the ceiling effect to an acceptable value according to the guidelines.

A Brazilian study that analyzed the measurement properties of a questionnaire on patient satisfaction also found a ceiling effect. The authors stated that the ceiling effect appears to be common in studies that assess the measurement properties of patient satisfaction questionnaires.³⁰ Although not the same construct, empathy also seems to follow the same trend. A possible explanation for this effect may be that the constructs do not evaluate clinical outcomes directly and involve a number of factors regarding the service itself. Another factor may be that part of our sample was composed of patients who were receiving free care as part of clinical studies, therefore their therapists gave them special attention because their presence was necessary for the completion of these studies.

Our secondary objective was to investigate whether the therapist’s empathy assessed by the CARE-Br would be a predictor of improvement in pain intensity and in the patients’ perceived improvement. The result was that empathy was not able to predict the patient’s clinical improvement. This finding is not consistent with other studies that investigated whether the therapist’s relationship or alliance with their patients influences outcomes such as pain intensity and perceived improvement.^{14., 15., 16., 35.} However, most of these studies did not assess prediction but a correlation of therapeutic alliance with the patient’s pain, and this analysis was performed only at 1 time and in only 1 session.^{14., 15., 16.} Only 1 study assessed prediction, and the results showed that the therapist–patient relationship is a predictor of perceived improvement.³⁵ However, a factor that may compromise the discussion of these results is that these studies did not use questionnaires that specifically measure empathy, which remains an area of deficiency in the literature and particularly in physical rehabilitation studies.

Limitations

This study did not control variables that could influence pain improvement and the therapist’s relationship with the patient, such as use of medication or previous experiences in relation to satisfaction with physical therapy treatment. Another limitation is that we did not deepen the analysis of the factorial structure of the CARE-Br because this analysis requires a larger sample size (200 patients or more).³⁶ A future research with a larger sample should be performed to confirm the unidimensionality of the CARE-Br. However, regarding the process of translation, cross-cultural adaptation, and analysis of measurement properties, this study followed all current guidelines, confirming the reliability of the results. Thus, we recommend that researchers invest in longitudinal studies that assess the relationship of physical therapists with patients.

Conclusion

The CARE-Br can be considered adequate and reliable for use both in future research and in clinical practice. However, it cannot be used as a predictor of the clinical improvement in patients with chronic musculoskeletal pain.

Funding Sources and Conflicts of Interest

Funding was received from FAPESP process number 2014/13461-9 (São Paulo Research Foundation/Fundação de Amparo à Pesquisa do Estado de São Paulo). No conflicts of interest were reported for this study.

Practical Applications

• •The CARE Measure evaluates the professional’s empathy regarding the patient.
• •The Brazilian Portuguese CARE Measure is reliable and adequate.
• •Empathy cannot be a predictor of a patient's clinical improvement.

Alt-text: Unlabelled Box

Contributorship Information

• Concept development (provided idea for the research): A.C.T.M., F.R.C.F., C.M.N.C.

• Design (planned the methods to generate the results): A.C.T.M., F.F.F., C.M.N.C.

• Supervision (provided oversight, responsible for organization and implementation, writing of the manuscript): A.C.T.M., F.R.C.F., F.F.F., C.M.N.C.

• Data collection/processing (responsible for experiments, patient management, organization, or reporting data): A.C.T.M., F.F.F.

• Analysis/interpretation (responsible for statistical analysis, evaluation, and presentation of the results): A.C.T.M., C.M.N.C.

• Literature search (performed the literature search): A.C.T.M.

• Writing (responsible for writing a substantive part of the manuscript): A.C.T.M.

• Critical review (revised manuscript for intellectual content, this does not relate to spelling and grammar checking): A.C.T.M., F.R.C.F., F.F.F., C.M.N.C.

Appendix A. Modifications performed in the process of translation and crosscultural adaptation of the CARE-Br

Appendix B. Final version of the CARE-Br measure