What Communication Patterns Predict Complaints and Safety Events

When a serious adverse event is reviewed, the root-cause analysis almost always arrives at the same place. Somewhere in the chain—between patient and clinician, between team members at shift change, between the concern raised and the response given—communication broke down. This is not a new observation. It has shown up in sentinel event data, malpractice claims, and patient complaint research for decades.

What is less well understood is that these communication failures do not arrive without warning. The patterns that precede complaints and adverse events are recognizable, and they tend to repeat. Quality and patient-experience leaders who can see them earlier—before harm has occurred—are in a fundamentally different position than those who encounter them only in retrospective review.

The link between communication and harm

The evidence for communication as a driver of preventable harm is substantial and consistent across data sources.

The Joint Commission has reported that ineffective communication was identified as a root cause in roughly two-thirds of all sentinel events reported to it over a decade, making it one of the most consistently cited contributing factors—and specifically implicated in handoffs, shift transitions, and surgical time-outs.

On the malpractice side, Candello's national comparative benchmarking database—representing approximately one-third of all U.S. medical professional liability claims—found that communication failures contributed to about 30 percent of claims analyzed in a cohort from 2009 to 2013. A decade later, their 2025 report covering 2014–2024 found that figure had risen to 40 percent—not despite improvements in digital communication technology, but alongside them. A companion analysis of that report tallied more than $1.5 billion in losses tied to communication-related cases over that ten-year span.

These numbers describe the downstream end: cases already filed, events already reviewed, harm already done. The harder and more useful question is what these events look like before they become statistics.

Patterns that recur before complaints and events

Root-cause analyses, complaint investigations, and malpractice case reviews point to a relatively consistent set of communication behaviors that appear upstream of harm. They are not dramatic failures. Most are small and ordinary—which is part of why they persist.

• Unaddressed patient concerns. A patient or family member raises a worry—about symptoms, about a treatment plan, about something they read—and the response closes down the concern rather than engaging it. The concern goes on record nowhere. When something goes wrong later, that prior expression of worry becomes significant.
• Dismissive or minimizing responses. Related to the above, but distinct: the clinician's response is technically accurate but functionally discouraging. The patient learns not to raise concerns again. Escalation thresholds shift upward.
• Failure to confirm understanding. Instructions are delivered; comprehension is assumed. Health literacy research has documented repeatedly that patients cannot reliably reproduce discharge instructions or medication instructions by the time they leave the clinical setting. The gap between what was said and what was understood is where medication errors, readmissions, and missed follow-ups originate.
• Escalation that goes unheard. A concern is raised—by a patient, by a nurse, by a trainee—to someone with authority to act on it. The concern does not receive a documented response. The person who raised it is left uncertain whether it was registered. This pattern appears frequently in serious event reviews.
• Handoff and transition gaps. Incomplete or ambiguous information transfer at shift changes, unit-to-unit transfers, or discharge creates conditions for missed diagnoses and duplicated or omitted treatments. AHRQ's Patient Safety Network identifies handoff failures as a major cause of preventable adverse events, with failures concentrated in the accuracy and completeness of information conveyed and the opportunity given to ask clarifying questions.

These are not exotic failure modes. They occur in ordinary clinical practice, across specialties and settings, and they occur at frequencies that surveillance tools rarely capture in real time.

Why surveys and incident reports miss them

The tools quality teams rely on to understand communication—patient satisfaction surveys, incident reports, grievance logs—share a structural limitation: they are lagging indicators. By the time data is in hand, the encounter that generated it is weeks or months in the past.

Survey response rates are low, and the patients who respond are not representative of those who experienced the most concerning care. A patient who felt dismissed mid-visit is unlikely to return a survey. A family member who raised an unheard concern during a hospitalization may have reported it verbally to staff—who did not document it—and then said nothing more.

Incident reports capture events that were serious enough, and visible enough, for someone to file a report—and where the organizational culture supported doing so. They do not capture the near-misses, the communication failures that resolved without consequence that time, or the patterns that will accumulate across dozens of encounters before manifesting in a harm event or a complaint cluster.

Grievances and formal complaints have predictive value, but only in aggregate and only retrospectively. Hickson and colleagues, in a 2002 study published in JAMA, found that unsolicited patient complaints were significantly and positively associated with malpractice experience: physicians with two or more malpractice suits had nearly six times as many patient complaints as colleagues with none. The relationship was strong enough that complaint counts predicted malpractice outcomes with over 80 percent concordance. But acting on complaints after they accumulate is still acting after the fact.

From lagging indicators to leading indicators

The clinical communication literature suggests that the patterns preceding complaints and harm events are identifiable before they produce harm—they just are not currently being observed in the right ways, at the right time.

Survey data and incident reports will always have a role. But they are, by design, retrospective and sampled. What quality and patient-experience teams need alongside them is something more continuous: signals from the actual flow of clinical communication, not reconstructed from memory or filtered through reporting systems after the fact.

This means paying attention to where communication patterns concentrate. Which service lines generate the most escalation concerns? Where do patients consistently report feeling unheard? Where are handoff failures appearing in near-miss reports? These questions are more tractable than they appear—provided the data exists to answer them.

The operational shift is from asking "what happened?" after a complaint or event to asking "where is risk accumulating?" in real time. That requires a different kind of observation, and different data, than most quality programs currently have access to.

A de-identified, aggregate view

This is the problem Inflect addresses for Quality and Safety teams. The individual coaching that clinicians receive through Inflect generates de-identified, aggregate signal about communication patterns across a department, a service line, or a care setting—without creating a surveillance system or exposing individual clinicians to comparative scrutiny.

The unit of analysis for quality purposes is the pattern, not the person. Where are concerns being met with dismissive responses? Which care transitions are generating the most communication gaps? Where is the escalation pathway functioning, and where is it not? These are questions that patient safety programs need answered, and they are questions that are very difficult to answer from lagging indicators alone.

De-identification is not a workaround—it is the design. Clinicians engage with coaching because it is genuinely useful to them, and because the privacy model is credible. The aggregate insight that flows from that engagement is a byproduct of real clinical development, not a monitoring program.

If you lead quality, patient safety, or patient experience for a health system and want to understand how aggregate communication data could fit into your existing improvement infrastructure, explore how Inflect works for Quality and Safety teams or request a demo to see what the signal looks like in practice.