The Empathic Communication Coding System, Explained

Empathy sounds like the least measurable thing in a clinical encounter—until you notice that it usually has a specific structure. A patient says something that opens a door (“I’m scared this won’t get better”), and the clinician either walks through it or doesn’t. The Empathic Communication Coding System is built on exactly that observation: it treats empathy as a two-step sequence—an opening created by the patient, and a response graded on how far the clinician steps through it (Bylund & Makoul, 2002, 2005).

What is the Empathic Communication Coding System?

The Empathic Communication Coding System (ECCS) is an observational method for coding empathy in medical encounters, developed by Carma Bylund and Gregory Makoul (Bylund & Makoul, 2002, 2005). Rather than asking patients or clinicians to rate empathy after the fact, it codes what happened in the conversation itself, turn by turn.

Its defining move is to split empathy into two steps. First, the patient creates an empathic opportunity—a statement that invites an empathic reply. Second, the provider responds, and that response is scored for how much empathy it conveys. Because the unit of analysis is the opportunity-and-response pair, the ECCS measures empathy as something interactional: not a personality trait, but a thing a clinician does (or fails to do) in reply to a specific opening.

What is an empathic opportunity?

An empathic opportunity is a patient statement that gives the clinician an opening to respond with empathy, and the ECCS sorts them into three types (Bylund & Makoul, 2002). The idea itself comes from Suchman and colleagues, whose 1997 model described how patients offer clues to their emotional concerns and how physicians can acknowledge or bypass them (Suchman et al., 1997).

The three opportunity types are straightforward to recognize in a transcript. An emotion opportunity is a statement that expresses a feeling, directly (“I’m frightened”) or indirectly (“I don’t know how I’ll cope”). A progress opportunity reports a personal achievement or improvement (“I finally managed to walk to the shops”).

A challenge opportunity describes a hardship or difficulty (“the night shifts are wrecking me”). Each is a patient-created moment where an empathic response is possible—the raw material the second step then grades.

The 0–6 empathic response levels

The provider’s response to each opportunity is rated on a seven-point empathic response scale, from denial to shared feeling (Bylund & Makoul, 2005). The ladder is the heart of the system, and it is precise enough to code reliably from a transcript.

Level	Label	What the provider does
0	Denial	Ignores or actively denies the patient's opportunity
1	Perfunctory recognition	An automatic or backchannel response ('uh-huh', 'I see')
2	Implicit recognition	Responds to a secondary aspect, not the central issue
3	Acknowledgment	Explicitly acknowledges the central issue
4	Acknowledgment with pursuit	Acknowledges the issue and follows up with a question or comment
5	Confirmation	Confirms or validates the patient's expressed emotion
6	Shared feeling or experience	Explicitly shares the patient's emotion or a similar experience

Read down the ladder and the logic is one of increasing engagement: from ignoring the opening, through mechanically noting it, to acknowledging it, pursuing it, validating the feeling, and finally sharing it. The rungs are behavioral, not attitudinal, which is what lets two coders agree on where a response sits.

Why isn’t level 6 “better” than level 5?

A subtle but important feature of the ECCS is that level 6 (shared feeling) is not treated as superior to level 5 (confirmation)—the two are conceptually different, not ranked. Sharing a feeling or a similar experience is a distinct move from confirming the patient’s emotion, and neither is automatically the “right” response.

This matters for coding and for interpretation. A coder should not assume higher is always better and inflate borderline responses toward 6. An analyst should not read a high average as unambiguously “more empathic care.”

Context decides which response serves the patient. Sometimes validation (5) is exactly right; at other times a clinician volunteering their own experience (6) can shift focus away from the patient. The scale measures the type and depth of engagement, not a single quality gradient that always points up.

Do clinicians respond to the openings patients offer?

The uncomfortable finding behind this whole line of work is that empathic opportunities are frequently offered and frequently missed. Suchman and colleagues framed the problem precisely: patients often signal emotional concerns only indirectly, and clinicians can either invite them to expand or steer the conversation back to biomedical facts (Suchman et al., 1997).

How large is that gap? Studies that code empathic opportunities put it starkly.

In one analysis of lung-cancer consultations, physicians responded empathically to just 10% of the 384 opportunities identified (Morse et al., 2008). Other studies put the rate higher—around 22% among oncologists, and 38% in surgery versus 21% in primary care (Pollak et al., 2007; Levinson et al., 2000). Those studies used empathic-opportunity schemes descended from Suchman’s model rather than the ECCS itself, and the lung-cancer figure is specific to oncology, but the direction is consistent: most openings go unmet.

That is the reason a coding system like the ECCS exists. If empathy were reliably present, there would be nothing to measure; it is the gap between openings offered and openings taken that makes the response level worth coding. The same pattern shows up in the OPTION shared decision-making scale, where observed scores are low because clinicians involve patients far less than they assume. And the stakes are real: empathic communication is associated with small but measurable improvements in patient pain, anxiety, and satisfaction (SMD ≈ −0.18; Howick et al., 2018), so measuring it turns a soft ideal into an actionable finding.

How reliable is empathy coding?

Empathy coding sits in the middle of the reliability range: harder than surface features, easier than open-ended interpretation. When Bylund and Makoul validated the ECCS on 100 encounters and 249 empathic opportunities, coders rating provider responses on the 0–6 scale reached a Cohen’s κ of 0.73 (Bylund & Makoul, 2002). That is acceptable agreement, but below the near-perfect levels that surface-feature schemes reach.

Deciding whether a provider’s reply is an acknowledgment (3) or an acknowledgment with pursuit (4) is more judgment-dependent than, say, classifying a question type, so the ECCS depends on a clear manual and trained coders to reach dependable agreement.

The practical implication is to treat inter-rater reliability as something to establish, not assume, for any empathy-coding project—calibrate coders on shared examples before coding in earnest, and compute agreement on the ordinal 0–6 scale rather than eyeballing it. Because the response level is ordinal, a plain percent-agreement figure understates disagreement between adjacent levels; the Cohen’s kappa and inter-rater reliability guide explains why a weighted kappa is the right coefficient here. Prompt-type schemes like the NICHD forensic protocol reach far higher agreement precisely because their categories are more surface-level than empathy.

How do you code empathy from a transcript?

To code the ECCS, tag the patient’s empathic opportunity, then rate the provider’s response to it on the 0–6 scale—instance-mode annotation applied to a paired exchange. The opportunity type is marked in the patient’s turn; the response level is coded on the clinician’s reply.

Consider a short synthetic exchange:

Patient: No matter what I do, the numbers won’t come down, and honestly I’m scared I’ll end up like my dad did. [challenge + emotion opportunity]

Clinician: That fear makes real sense given your family history—tell me more about what worries you most. [level 4: acknowledgment with pursuit]

The patient’s turn is both a challenge (the numbers won’t come down) and an emotion (fear), and the clinician’s reply acknowledges the fear and pursues it, a level 4. Tagging the opportunity and rating the response keeps the score anchored to the exact words, so a second coder can check the call. That is what makes the ECCS coding scheme reproducible enough to compute agreement across raters, and it pairs naturally with coding the same consultation for shared decision-making using the OPTION scale.

On data handling: consultation transcripts are sensitive clinical records, so the sane default is de-identified text and a privacy-first setup. Tagaroo supports a browser-side anonymous mode, so transcript content can stay local rather than being uploaded—worth checking against your ethics approval and information-governance rules before any real encounter data touches a tool.

Common mistakes when coding the ECCS

The recurring errors come from treating an interactional code like a rating of the clinician’s character:

• Coding empathy without an opportunity. The ECCS rates responses to patient-created openings. If there is no opportunity, there is nothing to rate—don’t score warmth in the abstract.
• Assuming higher is always better. Level 6 is not superior to level 5. Code the response that occurred, not the one you think would have been ideal.
• Using percent agreement on the 0–6 scale. Adjacent-level disagreement is common; use a weighted kappa so a 3-versus-4 split isn’t scored the same as a 0-versus-6 split.
• Confusing the ECCS with self-report empathy. It codes observed behavior in an encounter, not a questionnaire score—keep the two kinds of measure separate.

The practical upshot: the Empathic Communication Coding System works because it turned an ineffable quality into a two-step, observable sequence—an opening and a response. Tag the opportunity, rate the response on its own terms rather than as more-is-better, and let the exact words carry the score.

References

• Bylund, C. L., & Makoul, G. (2002). Empathic communication and gender in the physician-patient encounter. Patient Education and Counseling, 48(3), 207–216. doi:10.1016/S0738-3991(02)00173-8
• Bylund, C. L., & Makoul, G. (2005). Examining empathy in medical encounters: an observational study using the Empathic Communication Coding System. Health Communication, 18(2), 123–140. doi:10.1207/s15327027hc1802_2
• Suchman, A. L., Markakis, K., Beckman, H. B., & Frankel, R. (1997). A model of empathic communication in the medical interview. JAMA, 277(8), 678–682. doi:10.1001/jama.1997.03540320082047
• Morse, D. S., Edwardsen, E. A., & Gordon, H. S. (2008). Missed opportunities for interval empathy in lung cancer communication. Archives of Internal Medicine, 168(17), 1853–1858. doi:10.1001/archinte.168.17.1853
• Pollak, K. I., Arnold, R. M., Jeffreys, A. S., et al. (2007). Oncologist communication about emotion during visits with patients with advanced cancer. Journal of Clinical Oncology, 25(36), 5748–5752. doi:10.1200/JCO.2007.12.4180
• Levinson, W., Gorawara-Bhat, R., & Lamb, J. (2000). A study of patient clues and physician responses in primary care and surgical settings. JAMA, 284(8), 1021–1027. doi:10.1001/jama.284.8.1021
• Howick, J., Moscrop, A., Mebius, A., et al. (2018). Effects of empathic and positive communication in healthcare consultations: a systematic review and meta-analysis. Journal of the Royal Society of Medicine, 111(7), 240–252. doi:10.1177/0141076818769477

If you code empathy from clinical encounters, Tagaroo turns the Empathic Communication Coding System into a guided, evidence-anchored annotation workflow—with inter-rater reliability computed as your coders work.