Introduction

Empathy (E) is a trait with cognitive and emotional components. One of the instruments employed to measure empathy is the Jefferson Scale of Physician Empathy (JSE) (Delgado-Bolton et al., 2016), which has three components: Compassionate Care (CC), Taking Patient's Perspective (TPP), and ‘Walking in Patient’s Shoes’ (WIPS). One of the key properties of these components is that they interact dialectically (Díaz-Narváez et al., 2017). In consequence, conducting separate studies on each of these components is problematic given the unitary nature of this concept, which must be taken into account when diagnosing empathy.

The development of empathy in humans has occurred and occurs under the influence of processes associated with evolution (Decety, 2011) and ontogeny (Díaz-Narváez et al., 2017). There is also a complex interaction between these components, but ontogeny nowadays predominates over the evolutionary factor. In consequence, the neuronal structure (biological substrate) that supports empathy processes is determined by genetic information that remains relatively constant over generations, unless external factors modify the neuronal structure and thus affect how genetic information associated with empathy is expressed. Therefore, family factors such as the mother-child relationship (Stone et al., 2015), complex social networks, psychological factors (Löffler-Stastka et al., 2017), moral factors (Decety and Cowell, 2014), and stress (Durán et al., 2017), among others, gain more relevance in the constitution of empathy.

This attribute pays a major role in the dentist-patient therapeutic relationship, clinical treatment, and treatment adherence. Indeed, dentist associations have stated that increasing dentistry students' empathy levels is among the key goals of clinical training (Kalyan et al., 2017). The relevance of empathy has prompted several important studies (Waldrop et al., 2016), many of which have involved three factors: students' gender, progress in their program, and their chosen specialty in dentistry (Decety and Fotopoulou, 2015). Regarding the behavior of empathy in higher education, it has been suggested that medical students, as well as dentistry students, display a so-called ‘decline’ in their third year (Hojat et al., 2009;Nunes et al., 2011; González-Martínez et al., 2015;Quince et al., 2016). Studies on Latin American dentistry students have shown that the declination model is not the only one that exists (Alcorta-Garza et al., 2005;González-Martínez et al., 2015; Durán et al., 2017). A similar situation has been observed with respect to gender, with women not always behaving more empathetically than men (Alcorta-Garza et al., 2005; González-Martínez et al., 2015; Durán et al., 2017). In consequence, declining empathy and gender differences are still an unresolved matter, at least in Latin America (González-Martínez et al., 2015; Durán et al., 2017).

The existence of a single empathetic behavior pattern would make it easier to study empathy and deliver useful interventions in teaching learning processes, but the information presented above shows that Latin America lacks standardized models that reflect how empathetic behavior changes as students progress in their medical education programs and across genders. In consequence, several authors in Latin America have stated that intervention planning requires a thorough prior diagnosis of the concrete empathetic situation (González-Martínez et al., 2015; Durán et al., 2017).

Authors performing empathy measurements in Latin America tend to employ the Jefferson Scale of Medical Empathy; however, the psychometric properties of this instrument have been scarcely studied in our region. If this study managed to describe the properties of this scale, it could provide a solid basis for refuting or confirming the findings of Latin American researchers regarding the variability of empathy (and its dimensions) throughout health programs, across genders, and among students within the same country and across countries. As the scientific evidence shows, the psychometric properties of the JSE have been shown to be consistent across several studies and there are no theoretical or empirical reasons for the results of these measurements to differ in dentistry students in Latin America. However, due to the aforementioned variability, it is necessary to estimate invariance by gender and try to explore cutoff scores to establish comparisons of empathy and its dimensions between genders, among students in different entering classes of a university program, and among dentistry faculties in one country or in several countries. In this context, the aim of this study was to measure and evaluate the psychometric properties of the JSE-S, its underlying components, and its gender invariance, to estimate a regional norm, and determine cutoff scores in dentistry students in Latin America.

Methods

Participants

The sample comprised first- to fifth-year students attending 18 dental schools in seven countries of Latin America: Dominican Republic (Universidad Central de Este), Costa Rica (Universidad de Costa Rica, Universidad Latinoamericana de Ciencia y Tecnología), Panama (Universidad de Panamá, Universidad Latina), Colombia (Universidad Metropolitana, Universidad Libre, Universidad San Martín de Barranquilla, Universidad de Cartagena, Universidad de Magdalena), Argentina (Universidad Católica de Córdoba), Peru (Universidad Nacional Mayor de San Marcos, Universidad Cayetano Heredia), and Chile (Universidad de Antofagasta, Universidad de Concepción, Universidad Finis Terrae, Universidad del Desarrollo, Universidad Andrés Bello). Total n= 4407. Stratified samples were obtained by gender and entering class in each population analyzed. Data collection was carried out between July 2013 and August 2017.

Instrument

The Jefferson Scale of Empathy student version (JSE-S) was used in each of the participating countries, following the criteria set out by Alcorta-Garza et al. (2005). The instrument contains 20 Likert-type items on a seven-point scale (1: Strongly disagree, 7: Strongly agree), which allows measuring three empathy factors: Taking Patient's Perspective, Compassionate Care and Walking in Patient's Shoes.

Procedure

The scale was answered by students participating in the classroom or clinic, using only a confidential measurement by a neutral operator, after signing the informed consent. The study is bioethically governed by the Declaration of Helsinki and approved by the Research Ethics Committee of the German Development University and Clinic with CAS-UDD code 2011-64 in Santiago, Chile.

Data analysis

Item analysis. In order to differentiate students according to their level of empathy, we deemed it relevant to evaluate the discrimination power of the items of the JSE-S. To do this, we studied item-test correlation and generated a discrimination index (corrected item-total score correlation).

Factor analysis. Confirmatory factor analysis (CFA) was used to confirm the latent structure of the JSE-S. We specified a model composed of 20 items and three latent variables: ‘Taking Patient’s Perspective’ (10 items), ‘Compassionate Care’ (7 items), and ‘Walking in Patient’s Shoes’ (3 items) (Hojat et al., 2018). The model was estimated with weighted least square mean and variance adjusted (WLSMV), while fit was evaluated using multiple indexes: RMSEA, Tucker Lewis Index (TLI), Comparative Fit Index (CFI), and weighted root mean square residual (WRMR). For the CFI and TLI, values ≥0.95 were accepted (Hu and Bentler, 1999). For the RMSEA, values of 0.08 indicate good fit (Chen et al., 2008). As for the WRMR, values <1.0 are acceptable and indicate that the model was adequately specified (DiStefano et al., 2018). Since the chi-square index is sensitive to sample size and can be significant in large samples even if the fit is adequate (Barrett, 2007), we used the ratio of χ2/df, with values <3.0 being deemed acceptable (Kline, 2010).

Reliability. Reliability was determined using Cronbach's , the Spearman-Brown prediction formula, and McDonald’s ω. Reliability was estimated for the full scale and for each of its dimensions. A value of 0.80 was deemed adequate for the alpha coefficient (George and Mallery, 2003).

Factor invariance analysis. invariance tests were conducted to explore the degree to which the latent structure of the JSE-S is equivalent for men and women (Millsap and Yun-Tein, 2004). To do this, we imposed successive restrictions on various parameters of the latent structure of the JSE-S. This procedure yielded four invariance models: configural (same number of factors in both groups), metric (equivalent factor loadings in the groups), scalar (equivalent means or intercepts across groups), and strict (equivalent residuals in the groups). Since these are nested models, they can be compared to one another. To do this, we used Comparative Fit Index (CFI) variations. A new level of invariance was accepted if the CFI difference was ≤0.01 (Cheung and Rensvold, 2002).

Latin American norms for the JSE-S. As of this writing, there are no norms for the JSE-S in Latin America. Because they were collected at several Latin American faculties of dentistry, the data used in this study make it possible to construct norm tables for empathy scores. To do this, we estimated the percentile associated with each score obtained by the students examined in the population (AERA, 2014). This procedure was performed both for the full sample and for the subsamples of men and women.

Cluster analysis. The data were standardized and a hierarchical cluster analysis for cases was performed to explore cutoff points. Clusters were generated with the group-linking method (centroid grouping) and interval measurements, while the distance between clusters was estimated with the squared Euclidean method. The mean and Huber's M-estimator were determined in each cluster. The comparison between means was conducted through a two-factor (model III) analysis of variance (ANOVA). The level of significance employed was α<0.05 and β≤0.20.

Results

Sample characteristics

A total of 5033 students from the 18 faculties were eligible to participate, with 87.56% of them taking part in the study. The study was aimed at obtaining information about at least 80% of the total population of enrollees at the time it was conducted. Of the 4407 students in the full sample, 2830 (64.2%) of the participants were women and 1577 (35.8%) were men. The descriptive statistics of the full sample are presented in Table I.

Item analysis

Item-test correlations ranged from 0.14 to 0.56 (median= 0.41). All were positive and significant (p<0.05). The discrimination index ranged from 0.57 to 1.36 (median= 1.05). This suggests that the discrimination power of all items of the JSE-S is adequate.

Confirmatory factor analysis (CFA)

The model was successfully identified with 167 degrees of freedom. A total of 143 parameters were estimated (including cutoff scores for the response categories, six per question except for those used to identify the participants). Factor loadings >0.389 were obtained (Table II) for all items, except for item P18, belonging to the Compassionate Care factor. The three-factor model fit the data reasonably well, with RMSEA= 0.05, CFI= 0.957 and TLI= 0.951 (exceeding 0.95), the WRMR= 1.629 (exceeding 1), and the ratio χ²/df reaching 6.5.

Reliability

We estimated the reliability of the JSE-S considering the full sample studied. Cronbach's alpha reached 0.76, 0.79, and 0.59 for Compassionate Care, Taking Patient's Perspective, and Walking in Patient's Shoes, respectively. For the overall scale, we estimated the α and Spearman-Brown coefficients at 0.81, with McDonald’s ω= 0.87.

Invariance analysis

Table III presents the results of the invariance analysis for men and women. The analysis reveals configural, metric, and scalar invariance. That is, the scale preserves the same number of latent factors in both groups (configural invariance) while also maintaining factor loadings (metric invariance) and mean item scores (scalar invariance).

Normative sample

In the cluster analysis performed, we identified three clearly defined clusters in Empathy and the dimensions ‘Compassionate Care’ and ‘Taking Patient's Perspective’, but we only found two in the dimension ‘Walking in Patient’s Shoes’. The results of the comparison of means between the clusters in empathy and its components are presented in Table IV. According to the comparison, such clusters are adequately differentiated and allows establishing three ordinal categories of empathy and their dimensions.

Finally, Table V contains the results of the estimation of the percentiles observed along with the minimum and maximum values in each of the empathy clusters and in each of its components. Table VI presents the established norm for the total sample and for the sample segmented by gender.

Discussion

The item analysis conducted suggests that the discrimination power of all the individual items of the JSE-S is adequate. The CFA revealed that the one-factor model did not properly fit the data and that the two-factor model displayed adequate indexes, even though the ratio χ2/df was higher than in the three-factor model. Consistent with the literature reported, we decided to keep the three-factor solution (Hojat et al., 2018), which is a good fit to the data. It should be noted that there is a small degree of correlation between errors in items 9 and 10 (r= 0.16), items 9 and 17 (r= 0.22), items 15 and 20 (r= 0.19), items 16 and 17 (r= 0.15), and items 7 and 8 (r= 0.18), which slightly improve the model's goodness of fit indexes when included (χ2/df= 5.94, GFI= 0.972, CFI= 0.956, RMR= 0.076, RMSEA= 0.033). This suggests that the items involved may be very close to each other or that similar meanings are attached to them when being assessed.

Reliability levels were satisfactory in general, especially for the overall scale. These results are consistent with those obtained in research conducted in the same populations included in this study (González-Martínez et al., 2015;Durán et al., 2017;Díaz-Narváez et al., 2018; Fortich-Mesa and Díaz-Narváez, 2018; Varela et al., 2018).

The invariance results obtained suggest that the scale measures the same construct in the same way in both men and women; therefore, it is possible to perform valid comparisons based on JSE-S scores. This finding supports the validity of comparisons between the empathy levels of men and women. There is a clear tendency to state that women are more empathetic than men (Fields et al., 2011;Nunes et al., 2011). However, a study conducted in Latin America revealed that all three possible arrangements are manifested: more empathy in women than in men, more in men than women, and no differences (neither statistical nor in terms of absolute values) (Díaz-Narváez et al., 2015). Therefore, the distribution of empathy levels has been a matter of controversy (Harenski et al., 2008; Mestre et al., 2009). There are no sufficient empirical-theoretical grounds to account for this variability. Research seems to indicate that women and men manifest empathy differently and that their neuronal response also differs (Harenski et al., 2008;Mestre et al., 2009; Michalska et al., 2013;Díaz-Narváez et al., 2015). Affective empathy seems to be more automatic and makes it possible to understand other people's emotions quickly and accurately. From a neurological point of view, this empathy involves the mirror neuron system as well as the limbic system, the anterior insula, and the anterior cingulate cortex. In contrast, cognitive empathy involves a different form of mentalization and includes the prefrontal, temporal, and cingulate cortices, areas such as the prefrontal ventromedial cortex, the temporoparietal junction, the temporal medial area and lobe, and Brodmann areas 10 and 12. Several approaches to explaining the differences observed between women and men exist. One of them, the evolutionary perspective (Decety, 2011), holds that women have a more developed sympathetic system as a result of child-rearing, which results in features such as pain inhibition in the presence of strangers, strong reactions when protecting children, and the construction of deep emotional bonds with them, all of which appear to correlate positively with high oxytocin levels. In contrast, the cultural approach holds that the family, along with society, teach women and men to express their emotions differently and that empathy influences emotional health through culture (Michalska et al., 2013), while also correlating with altruistic behavior and inhibiting antisocial and aggressive behavior (Carlo et al., 2003). Despite their differences, these approaches should not be regarded as contradictory.

The results for the normative sample make it possible to adopt a percentage-based distribution system to group people according to the value of the empathy levels reached and compare different populations. However, these data do not make it possible to place people in specific reference groups such as high, middle, or low; therefore, they are not sufficient for establishing whether a population is more empathetic than another. The cluster analysis conducted makes it possible to compare and classify, simultaneously, empathy levels (and the level of its components) across student populations labeled high, mid, and low, including hierarchical percentiles within each classification (cluster 1, 2, and 3 respectively), except for the component ‘Walking in Patient’s Shoes’, which only yielded two clusters (1: high values; 2: low values). But the values observed (which can be employed as cutoff values) can be used for populations with one common characteristic. For instance, in this study, the shared trait of this population is that it comprises dentistry students in Latin America.

Conclusion

Regarding the psychometric properties of the JSE-S, our findings, based on a sample of Latin American dentistry students, are consistent with analyses conducted in populations of dentistry students and other medical students around the world. This constitutes another element that confirms the latent structure of the instrument, thus revealing its construct validity. The values of the norm table of observed empathy may constitute a point of reference to gain an overall understanding of the empathetic behavior of these students in Latin America. However, if we also take into account the result of the cutoff analysis, it will become possible to establish comparisons between populations and determine which values could be classified as ‘higher than…’, ‘equal to…’ (if they fall into the same percentile), or ‘lower than...’. The reliability coefficients calculated for the overall scale and the subscales display acceptable or good values, which indicates that the measure is reliable when estimated using a sample of dentistry students. The invariance analysis made it possible to conduct a valid comparison between genders. This supports the view that men and women conceptualize empathy in the same way and provides empirical evidence against the universality of the hypothesis that women are necessarily more empathetic than men. In general, this study shows that the JSE-S is a valid instrument and a reliable measure for Latin American dentistry students, with adequate discrimination power and gender invariance.

THE JEFFERSON SCALE OF EMPATHY: MEASUREMENT PROPERTIES, UNDERLYING COMPONENTS, NORMS, AND CUTOFF POINTS IN DENTISTRY STUDENTS IN LATIN AMERICA. A MULTICENTRIC STUDY

Alcorta-Garza A, González-Guerrero JF, Tavitas-Herrera SE, Rodríguez-Lara FJ, Hojat M (2005) Validación de la escala de empatía médica de Jefferson en estudiantes de medicina mexicanos. Salud Mental, 28: 57-63.

AERA (2014) Standards for Educational and Psychological Testing 2014. American Educational Research Association, American Psychological Association, National Council on Measurement in Education. Washington, D.C., USA.

Barrett P (2007) Structural equation modeling: Adjudging model fit. Personal. Indiv. Diff. 42: 815-824. https://doi.org/10.1016/j.paid.2006.09.018

Carlo G, Hausmann A, Christiansen S, Randall BA (2003) Sociocognitive and behavioral correlates of a measure of prosocial tendencies for adolescents. J. Early Adolesc. 23: 107-134. https://doi.org/10.1177/ 0272431602239132

Chen F, Curran PJ, Bollen KA, Kirby J, Paxton P (2008) An empirical evaluation of the use of fixed cutoff points in RMSEA test statistic in structural equation models. Sociol. Meth. Res. 36: 462–494. https://doi.org/10.1177/0049124108314720

Cheung GW, Rensvold RB (2002) Evaluating goodness-of-fit indexes for testing measurement invariance. Struct. Equat. Model. 9: 233-255. https://doi.org/10.1207/S15328007SEM0902_5

Decety J (2011) The neuroevolution of empathy. Ann. NY Acad. Sci. 1231: 35-45. https://doi.org/10.1111/j.1749-6632.2011.06027.x

Decety J, Cowell JM (2014) The complex relation between morality and empathy. Trends Cognit. Sci. 18: 337-339.

Decety J, Fotopoulou A (2015) Why empathy has a beneficial impact on others in medicine: unifying theories. Front. Behav. Neurosci. 8: 1-11.

Delgado-Bolton R, San-Martín M, Alcorta-Garza A, Vivanco L (2016) Empatía médica en médicos que realizan el programa de formación médica especializada. Estudio comparativo intercultural en España. Atención Primaria 48: 565-571.

Díaz-Narváez VP, Erazo AM, Bilbao JL, González F, Padilla M, Howard M, Silva MG, Bullen M, Gutiérrez F, Varela de Villalba T, Salcedo M, Huberman J, Carrasco D, Utsman R (2015). Empathy gender in dental students in Latin America: An exploratory and cross-sectional study. Health 7: 1527-1535. https://doi.org/10.4236 / health.2015.711166

Díaz-Narváez VP, Calzadilla-Núñez A, Alonso LM, Torres-Martínez PA, Cervantes-Mendoza, M, Fajardo-Ramos E (2017) Empathy and ontogeny: A conceptual approach. West Indian Med. J. 66(3): 1-4. https://doi.org/10.7727/wimj.2016.344

Díaz-Narváez VP, Amezaga-Avitia AC, Sarabia-Álvarez PA, Lagos-Elgueta M, Saavedra-Madrid M, Silva-Reyes P, Padilla M, Rodríguez-Hopp MP (2018) Chilean dentistry students, levels of empathy and empathic erosion: Necessary evaluation before a planned intervention: Levels of empathy, evaluation and intervention. Saudi Dental J. 30: 117-124. https://doi.org/10.1016/j.sdentj.2017.11.004

DiStefano C, Liu J, Jiang N, Shi D (2018) Examination of the weighted root mean square residual: Evidence for trustworthiness? Struct. Equat. Model. 25: 453-466. https://doi.org/10.1080/10705511.2017.1390394

Durán E, Padilla M, Utsman R, Reyes-Reyes A, Calzadilla-Núñez A, Díaz-Narváez V (2017) Analysis of the relationship between empathy and family functioning in dentistry students of the Latin American. 9^th Annu. Int. Conf. on Education and New Learning Technologies. EDULEARN, July 2017. Barcelona, Spain.

Fields S, Mahan P, Tillman P, Harris J, Maxwell K, Hojat M (2011) Measuring empathy in healthcare professions students using the Jefferson Scale of Physician Empathy: health provider-student version. J. Interprofess. Care 25: 287-293.

Fortich-Mesa N, Díaz-Narváez VP (2018) Empathetic attitude in dentistry students at Corporación Universitaria Rafael Núñez in Cartagena. Cienc. Salud 16: 129-143.

George D, Mallery P (2003) SPSS for Windows step by step: A simple guide and reference. 11.0 Update. Allyn & Bacon. Boston, MA, USA.

González-Martínez FD, Díaz-Narváez VP, Arrieta-Vergara K, Díaz-Cárdenas S, Tirado-Amador LR, Madera-Anaya M (2015) Distribution of empathetic orientation among dentistry students on Cartagena, Colombia. Rev. Salud Públ. 17: 401-415. https://doi.org/10.15446/rsap.v17n3.37591

Harenski CL, Antonenko O, Shane MS, Kiehl KA (2008) Gender differences in neural mechanisms underlying moral sensitivity. Soc. Cognit. Affect. Neurosci. 3: 313-321. https://doi.org/10.1093/scan/nsn026

Hojat M, Vergare MJ, Maxwell K, Brainard G, Herrine SK, Isenberg GA, Veloski J, Gonnella JS (2009) The devil is in the third year: A longitudinal study of erosion of empathy in medical school. Acad. Med. 84: 1182-1191.

Hojat M, DeSantis J, Shannon SC, Mortensen LH, Speicher MR, Bragan L, LaNoue M, Calabrese LH (2018) The Jefferson Scale of Empathy: A nationwide study of measurement properties, underlying components, latent, variable structure, and national norms in medical students. Adv. Health Sci. Educ. 23: 899-920.

Hu L, Bentler PM (1999) Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Struct. Equat. Model. 6: 1-55. https://doi.org/10.1080/10705519909540118

Kalyan VS, Manjula S, Padma TM, Pratap KVNR, Vineela P, Varma S (2017) Assessment of empathy among clinical dental student in a teaching dental institution in Telangana State, India. J. Indian Assoc. Public Health Dentist. 15: 162-165.

Kline RB (2010) Principles and Practice of Structural Equation Modeling. Guilford Press. New York, NY, USA. 422 pp.

Löffler-Stastka H, Datz F, Parth K, Preusche I, Bukowski X, Seidman C (2017) Empathy in Psychoanalysis and Medical Education – what can we learn from each other? BMC Med. Educ. 17(74): 1-6. https://doi.org/10.1186/s12909-017-0907-2

Mestre MV, Samper P, Frías MD, Tur AM (2009) Are women more empathetic than men? A longitudinal study in adolescence. Span. J. Psychol. 12: 76-83. https://doi.org/10.1017/S1138741600001499

Michalska KJ, Kinzler KD, Decety J (2013) Age-related sex differences in explicit measures of empathy do not predict brain responses across childhood and adolescence. Devel. Cognit. Neurosci. 3: 22-32. https://doi.org/10.1016/j.dcn.2012.08.001

Millsap R, Yun-Tein J (2004) Assessing factorial invariance in ordered-categorical measures. Multivar. Behav. Res. 39: 479-515. https://doi.org/10.1207/S15327906MBR3903_4

Nunes P, Williams S, Sa B, Stevenson K (2011) A study of empathy decline in students from five health disciplines during their first year of training. Int. J. Med. Educ. 2: 12-17.

Quince TA, Kinnersley P, Hales J, da Silva A, Moriarty H, Thiemann P, Hyde S, Brimicombe J, Wood D, Barclay M, Benson J (2016) Empathy among undergraduate medical students: A multicentre crosssectional comparison of students beginning and approaching the end of their course. BMC Med. Educ. 16(92): 1-10. https://doi.org/10.1186/s12909-016-0603-7

Stone LL, Janssens JM, Vermulst AA, van Der Maten M, Engels RC, Otten R (2015) The strengths and difficulties questionnaire: psychometric properties of the parent and teacher version in children aged 4-7. BMC Psychol. 3(4): 1-12. https://doi.org/10.1186/s40359-015-0061-8

Varela T, Ulloque MJ, Villalba S, Villalba R, Díaz-Narváez VP (2018) Levels of empathy in dentistry students: Measurement and comparison in two academic periods. SaludUninorte 34: 641-651.

Waldrop D, Nochajski T, Davis EL, Fabiano J, Goldberg L (2016) Empathy in dentistry: How attitudes and interaction with older adults make a difference. Gerontol. Geriatr. Educ. 37: 359-380.

Notas de autor

victor.diaz@unab.cl