Orthodontic Extraction Decisions Diverge Even When the Case Is the Same
Orthodontists reviewing identical standardized cases reached different conclusions with striking frequency. In a case-based survey of 104 orthodontists, the same clinical records produced 2,080 treatment decisions, yet overall agreement on whether to extract teeth was only fair, with Fleiss’ κ of 0.303. The finding matters because extraction planning remains one of the most consequential judgment calls in orthodontic treatment, shaping space management, facial balance, and long-term occlusal stability.
The cohort was built to remove as much ambiguity as possible. Each participant assessed 20 standardized cases with complete diagnostic records and chose extraction or non-extraction treatment while identifying which clinical factors influenced the decision. Even under those controlled conditions, decisions remained highly variable, especially in borderline presentations where the clinical signal is less blunt and the tolerance for risk differs from one clinician to another.
Soft-Tissue Profile and Crowding Drove Decisions More Than Measured Cephalometrics
Profile concerns, crowding severity, and incisor inclination were the most frequently cited factors behind extraction recommendations. Those are familiar orthodontic pressure points: crowding creates immediate arch-length problems, incisor inclination affects lip support and facial esthetics, and profile concerns often sit at the intersection of anatomy and patient preference. In practical terms, those factors can outweigh a single numeric threshold because they influence how the face will look and function after treatment, not just how the teeth line up on paper.
The study also examined six case-level cephalometric measurements in a cross-classified mixed-effects model. None showed a statistically significant association with extraction recommendations after adjustment for differences between orthodontists and cases, with all p values above 0.05. In other words, once the analysis accounted for who was making the decision and which case was being judged, the measured skeletal variables did not independently push clinicians toward extraction in a statistically demonstrable way.
Borderline cases exposed the real fault line in orthodontic planning
The strongest disagreement clustered in borderline cases, where the same diagnostic data can support more than one defensible plan. One clinician may prioritize profile preservation and space relief; another may favor non-extraction mechanics and accept a less aggressive alignment strategy. This is where training, clinical philosophy, and comfort with uncertainty can matter as much as the numbers on a cephalometric tracing.
The random-effects estimates reinforce that point. The estimated random-intercept standard deviations were 0.451 for orthodontist and 1.262 for case, indicating meaningful variation attributable both to the treating clinician and to the specific scenario being judged. Case-level differences were particularly influential, which is not surprising in orthodontics: a borderline Class I malocclusion with moderate crowding is not clinically equivalent to a case with similar crowding but a more favorable soft-tissue profile.
Why Cephalometric Certainty Did Not Translate Into Extraction Consensus
Cephalometric analysis remains central to orthodontic diagnosis, but it does not function as a stand-alone rule engine. Facial esthetics, crowding distribution, dental compensation, and the clinician’s interpretation of risk all enter the decision before a treatment plan is finalized. The absence of significant adjusted associations for the six cephalometric measurements suggests that these variables, at least in this sample, were not strong enough to unify extraction decisions across clinicians once broader judgment was taken into account.
That does not mean the measurements are irrelevant. It means the measured anatomy did not behave like a single decisive switch in the hands of different orthodontists. In a field where treatment planning blends biomechanics with facial esthetics and patient-centered goals, a decision may be internally logical without being widely shared by other experts reviewing the same records.
The study design also places a boundary around how far the results can be extended. Only 20 unique cases were used, and the investigators explicitly noted uncertainty in the estimates. A limited set of cases is a real constraint when the outcome depends on the interaction between clinician judgment and case nuance, because a small case library can amplify the influence of unusually difficult or unusually straightforward scenarios.
Standardization Reduced Noise, but It Did Not Eliminate Clinical Subjectivity
The most important contribution of this work is not that orthodontists disagree; it is that disagreement persisted even after the usual sources of clinical noise were stripped away. Standardized records, complete diagnostic data, and a structured survey format should have narrowed the field. Instead, the responses still scattered widely, suggesting that extraction planning in orthodontics is shaped by interpretive thresholds that are not fully captured by the common diagnostic variables studied here.
That has direct implications for training and for case conferences. Residents and specialists may benefit from explicit discussion of how much weight they assign to profile concerns, crowding severity, and incisor inclination, especially when the cephalometric profile is equivocal. A shared vocabulary for borderline cases could reduce unnecessary variation, even if it cannot remove discretion from the process entirely.
The authors received no specific funding for the work, and the study was approved by the Research Ethics Committee of the College of Dentistry, University of Baghdad. Participants provided informed consent, and the anonymized records were handled under standard ethical safeguards.
Extraction planning will remain a judgment-heavy area of orthodontics, but this study shows where the uncertainty lives: not in the availability of data, but in the way clinicians weigh the same data differently when the case sits near the threshold for intervention.