How much orthodontic harm could we avoid?

Patient harm is an uncomfortable subject. Nevertheless, if we want to improve orthodontic care, we need to study what goes wrong rather than concentrating only on successful treatment. This blog post is on the second part of a thought-provoking study by a Danish team. Their headline finding was striking: 70% of the assessed harm was avoidable.

I have previously posted about the first part of the study.

I have previously posted about the first part of the study.

What did they ask?

They did this study to:

1. Identify the factors contributing to patient harm;
2. Estimate whether the harm was avoidable;
3. Explore the relationship between avoidability and the number and type of contributory factors; and
4. Suggest ways of reducing future incidents.

What did they do?

The authors examined 125 orthodontic claims approved for compensation by the Danish Dental Compensation Association from September 2019 to August 2024. They aimed to identify the factors contributing to harm and estimate how much of it might have been avoided.

An experienced orthodontist reviewed the clinical records, radiographs, photographs, correspondence and compensation decisions for each claim.

The reviewer classified possible contributory factors using the Eindhoven Classification Model. These included:

• human factors, such as monitoring, verification and clinical knowledge;
• organisational factors, such as protocols, management priorities and workplace culture;
• technical factors involving appliances or materials; and
• patient-related factors.

The same reviewer then rated the avoidability of each incident on a six-point scale, from completely unavoidable to completely avoidable.

Finally, the authors grouped the cases into low-, moderate- and high-avoidability categories. They used ChatGPT to help develop qualitative descriptions of the patterns and possible mitigation strategies. The authors state that the AI did not review patient records, identify contributory factors, or determine whether harm was avoidable.

What did they find?

The most common categories of compensated harm were:

• Gingival or other soft-tissue harm: 17%;
• Apical root resorption leading to tooth loss: 15%;
• Worsening of the occlusion following treatment: 11%;
• Cervical root resorption leading to tooth loss: 10%; and
• Damage to a tooth crown: 7%.

Just over half of the harm-causing events occurred during active orthodontic treatment. Delayed diagnosis accounted for 22%, while retention-related incidents accounted for 10%.

The authors identified a median of five contributory factors per case. Monitoring and verification problems were the most common, occurring in 56% and 54% of cases, respectively. They recorded deficiencies in clinical knowledge and professional qualifications in approximately 42%.

Organisational factors were also common. These included management priorities, workplace culture, inadequate protocols, and ineffective knowledge transfer between team members.

Patient-related factors were identified in 45% of the compensated cases. This does not necessarily mean that patients “caused” the harm. It indicates that factors such as biological variation or cooperation may have formed part of a more complicated chain of events.

Finally, they concluded that 82 incidents were probably, highly or completely avoidable. They reported this as 70.1% of the claims assessed for avoidability.

Cases considered highly avoidable tended to contain more contributory factors. In other words, serious, avoidable harm often appeared to stem from several failures interacting rather than from a single isolated mistake. This is similar to other “never” events that arise from a series of unfortunate events. 

Their conclusions were

“70% of the compensated cases were avoidable and most harms were mutltifactorial in nature.”

What did I think?

This is an important paper because patient safety receives surprisingly little attention in orthodontic research. We devote considerable effort to comparing appliances and measuring small differences in treatment efficiency. Avoiding tooth loss, periodontal damage or a missed diagnosis is likely to be far more important to patients.

The use of a national, no-fault compensation scheme is also a strength. Patients did not need to establish malpractice before applying, which may reduce some barriers found in litigation-based datasets. The investigators reviewed detailed clinical material rather than relying only on brief incident reports.

The central message is both plausible and useful. Harm often develops when monitoring, verification, communication and clinical systems fail together. Simply identifying and blaming the last clinician in the chain is unlikely to prevent the next incident.

Limitations

As with all studies, there are some limitations that influence our certainty about the findings.

Most importantly, these 125 cases were not a random sample of orthodontic treatment. They were approved compensation claims. They are consequently likely to represent unusually severe or readily identifiable harm. We cannot use this study to estimate the risk of being harmed during orthodontic treatment, nor can we conclude that 70% of all orthodontic complications are avoidable.

Additionally, only one investigator performed both the incident analysis and the avoidability assessment. Even though this was an experienced reviewer using a structured system, this is still a structured judgement. The authors did not test agreement with a second assessor or repeat the ratings to assess the reliability of the ratings.

Finally, I was not sure whether the AI-generated interpretations and recommendations were helpful. The important work was the human review and classification of the cases. Suggestions generated from those classifications remain suggestions; the study did not test whether checklists, protocols or additional training actually reduce orthodontic harm.

What can we learn from this paper?

I took from this paper that we should:

• monitor treatment actively;
• verify diagnoses, treatment decisions and critical procedures;
• arrange appropriate radiographic review when there is a risk of root resorption or damage associated with impacted teeth;
• establish clear routes for communication and referral;
• document the reasons for important decisions;
• use significant incidents as opportunities for team learning; and
• develop systems that make safe actions easier and unsafe omissions more difficult.

This may be obvious to many readers of my blog. But patient safety requires competent clinicians working within practices that support communication, verification, reflection and timely action.

Final thoughts

This was an interesting and worthwhile study of a neglected area.

The findings suggest that, among a selected group of compensated incidents, many harms were considered avoidable and usually involved several unfortunate events.

I think the 70% estimate is only a rough estimate. However, the wider conclusion is convincing: monitoring, verification and good clinical systems are important.