To the Editor:
After reviewing 11 recent empirical studies of the decline in empathy during medical (and dental) education, Colliver and colleagues1 expressed concerns about the conceptualization of empathy, the validity of self-report measures of empathy, and the clinical significance of the findings in the studies. As the developers of the Jefferson Scale of Physician Empathy (JSPE), which was used in 5 of the 11 studies, we are compelled to respond to these concerns.
First, we certainly agree that empathy is an elusive concept. In fact, we systematically addressed its conceptualization when we were developing the scale and offered the following definition of empathy in the context of patient care: “… a predominantly cognitive attribute that involves an understanding of patients' experiences, concerns, and perspectives combined with a capacity to communicate this understanding and an intention to help.”2–4 With regard to the confusion about empathy and sympathy, we differentiated empathy, which is a predominantly cognitive attribute, from sympathy, which is a predominantly affective attribute, and described their different outcomes.2,3
Second, the authors' concern about the validity of the JSPE's scores does not recognize the extensive literature on its psychometrics. Because of the strong evidence in support of its validity and reliability, the JSPE has been broadly used by researchers in the United States and abroad, where it has been translated into 38 languages. In addition, a relationship between physicians' JSPE scores and scores on the Jefferson Scale of Patient Perception of Physician Empathy has been reported.5 Furthermore, we recently completed another validity study that revealed a strong link between physicians' scores on the JSPE and the clinical outcomes of their diabetic patients.6
Finally, to address the clinical, or practical, significance of the findings, Colliver and colleagues transformed the scores in each study back to the units of the original Likert scales for the instruments used. However, they failed to recognize that this conversion does not produce what is referred to as “scale-free” measures, because the three instruments used in the studies used three different types of Likert scales of four, seven, or nine points. Thus, the transformed scores are not comparable. The widely accepted solution to the issue is to calculate the effect size of differences, which yields a “scale-free” and operationally defined index.7,8 The effect sizes for declining empathy in our studies ranged from 0.29 to 0.64.3,9 The authors failed to use effect sizes and well-known meta-analytic techniques to base their conclusion on firmer ground.
Critical review of the literature on important issues has the potential to make valuable contributions when the criticism is well founded. If not, the purpose can be better served by designing new studies or replicating previous ones to improve their shortcomings.10
We firmly stand behind our concluding remarks that erosion of empathy observed in our studies is not only statistically significant but also is of practical importance that must not be ignored.
Mohammadreza Hojat, PhD
Research professor of psychiatry and human behavior and director, Jefferson Longitudinal Study, Center for Research in Medical Education and Health Care, Jefferson Medical College, Philadelphia, Pennsylvania; firstname.lastname@example.org.
Joseph S. Gonnella, MD
Dean emeritus, distinguished professor of medicine, and founder and director, Center for Research in Medical Education and Health Care, Jefferson Medical College, Philadelphia, Pennsylvania.
Jon Veloski, MS
Director, Medical Education Division, Center for Research in Medical Education and Health Care, Jefferson Medical College, Philadelphia, Pennsylvania.