Is the Restorative Emergence Angle a Cause or a Geometric Proxy? Placement Depth, Diameter Mismatch, and Peri-Implantitis Risk
- Paper type
- Methodological study (theoretical, with a causal-inference framework)
- Status
- Original Work · Working Paper
- Keywords:
- emergence angle · peri-implantitis · causal inference · directed acyclic graph · supracrestal tissue height · confounding · implant prosthodontics
Contents
Abstract
Background. A restorative emergence angle above about 30° is widely cited as a risk indicator for peri-implantitis, based mainly on cross-sectional radiographic studies. Later replications conflict and the proposed threshold has drifted. The angle is not an independent design choice: for a given profile it is fixed by the crown width at the end of the measurement line, the implant platform diameter, and the vertical distance available between the platform and that point. That distance depends on implant depth and supracrestal tissue height, which are themselves plausible causes or markers of peri-implant disease.
Objective and hypothesis. We propose that part of the published angle-disease association is confounded by placement depth and tissue height, that part may be mediated by the angle, and that cross-sectional radiographs cannot separate these components. A direct effect of the angle remains biologically plausible and is not excluded.
Proposed approach. We state the straight-profile geometry, tan θ = (Wc − Dp)/(2H), define each term and its measurement, and describe a directed acyclic graph with the adjustment sets it implies. Because the angle is a deterministic function of its geometric parents, we argue that its effect is identifiable only through an intervention that changes the transmucosal profile while holding platform diameter and height fixed. We specify a four-layer design: standardized re-measurement of existing cohorts with stratification by height, target-trial emulation, negative-control exposures, and a randomized recontouring trial against sham polishing. An illustrative simulation under stated assumptions, not empirical data, shows how an inert angle can yield a crude odds ratio of about 1.5 that disappears after adjustment for a perfectly measured height but persists when height is measured with modest reliability.
Expected contribution. The manuscript narrows an existing observation, the geometric dependence of angle on tissue height, into a testable causal structure with numbered predictions and refuting results.
Conclusions. The geometric dependence has been described before; what is missing is an explicit causal treatment. The angle should be regarded as an unresolved candidate cause until a design that breaks its link to depth and tissue height is completed.
1. Introduction
Clinicians restoring bone-level implants are often advised to keep the restorative emergence angle at or below roughly 30°. The advice traces to cross-sectional radiographic analyses in which larger angles were associated with a higher prevalence of peri-implantitis [1, 2]. The recommendation has since entered clinical teaching and the narrative literature [3], while systematic syntheses have reported either a limited and inconsistent evidence base or no significant difference [4–6]. Cohort studies of the same exposure have been mixed. Some found no difference in bone change between wide and narrow angles [7, 8], and one recent report found that over-contoured tissue-level implants lost less bone than others [9].
This heterogeneity is usually framed as a measurement or population problem. A different explanation deserves explicit treatment. The emergence angle is not set by the clinician in isolation. It is a geometric consequence of three quantities: how wide the restoration is where the measurement ends, how wide the implant platform is, and how much vertical room exists between the two. When vertical room is small, a crown of ordinary width cannot join a narrow platform without a steep flare. Vertical room in turn depends on how deep the platform was placed relative to the bone crest and on the height of the supracrestal tissue. Both of these have their own literature as determinants of marginal bone stability and mucosal health (Section 2.4).
This observation is not new. A short communication has already derived the tissue height needed to achieve a given contour angle from the implant and crown dimensions [10], and a retrospective study found that measured angles rose with the ratio of space width to implant diameter and fell with placement depth [11]. A cone-beam study likewise found wider angles associated with lower mucosal height [12]. What has not been done, as far as our searches revealed, is to treat the dependence formally as a causal structure: to specify which paths connect angle, geometry and disease, what must be conditioned on, what must not be, and what a design capable of separating the paths would look like. We found no application of causal diagrams to the emergence angle question in PubMed or Europe PMC.
The purpose of this manuscript is therefore narrow. It does not claim that the angle is harmless, nor that it is a pure proxy. It asks what the cross-sectional association can and cannot show, specifies the geometry and the causal graph, quantifies by illustration how confounding and measurement error could produce a crude association without any angle effect, and proposes a design that could separate the angle from depth. All computations are illustrative, run under assumptions we state, and are not empirical data. No patient data were analyzed.
2. Background and critical appraisal of the evidence
Origin of the 30° rule
Katafuchi and colleagues analyzed radiographs from 96 patients with 225 implants, of which 168 implants in 83 patients met inclusion criteria, at a mean follow-up of 10.9 years. They divided implants into bone-level and tissue-level groups, measured the restoration emergence angle and profile, and diagnosed peri-implantitis from probing depth and radiographic bone loss. Within the bone-level group, the prevalence of peri-implantitis was higher with angles above 30° (31.3% versus 15.1% in the notes compiled for this project), while no such relation was found for tissue-level implants [1]. Yi and colleagues retrospectively analyzed 169 patients with 349 implants, measuring bone loss, emergence angle, emergence profile, crown-implant ratio and splinting on radiographs taken 1 and 5 years after prosthesis insertion, and used multivariable generalized estimating equations [2]. Both studies are cross-sectional or retrospective radiographic designs, and both measured the angle in a two-dimensional mesio-distal projection.
Supporting and contradicting studies
Table 1 collects the pivotal studies with their design, size, main source of bias, and an operational evidence level. Supportive recent work includes a cross-sectional study of 332 implants at a mean of 11.2 years, in which crown angles above 30° at bone-level implants and above 48° at tissue-level implants were consistently associated with peri-implant disease [3]. Further supportive reports come from a population-sampled analysis using a threshold of 45° [13], a study of 260 restorations [14], and a study of 417 implants placed with static computer-assisted guidance [15].
Contradicting evidence is at least as extensive. A five-year cross-sectional study of 67 platform-switched bone-level implants found no influence of cervical crown contour on marginal bone loss [8]. A three-year retrospective study of 312 implants compared groups above and at or below 30° and reported similar bone change [7]. Studies of external-hexagon implants [16], of patients compliant with supportive periodontal therapy [17], and of restorations with mean angles of 40° to 50° [18] likewise found no association, and a recent report on posterior prostheses also found no significant association between measured angles and bone stability [11]. One retrospective analysis of 200 tissue-level implants found that angles of 40° or more were accompanied by less, not more, bone change [9].
Two systematic syntheses reached cautious conclusions. One identified only three eligible studies and concluded that causality remained open [4]. A meta-analysis of four studies found a difference that was not statistically significant [5]. A later American Academy of Periodontology and Academy of Osseointegration review addressed prosthetic design and bone loss more broadly [6].
Preclinical work gives the mechanistic argument its strongest support. In a dog study, wide restorative angles increased marginal bone loss and impaired junctional epithelium integrity [19]; a second histological analysis found inflammation varying with angle, with effects pronounced beyond roughly 40° to 60° [20]. These angles exceed the 30° clinical threshold, so the experimental data do not locate the threshold in humans.
| Study | Design and size | Main finding (as reported in abstract or project notes) | Main bias or limitation | Level |
|---|---|---|---|---|
| Katafuchi 2018 [1] | Cross-sectional radiographic, 168 implants | Bone-level >30°: higher peri-implantitis prevalence; none at tissue-level | Cross-sectional; 2-D angle; depth and tissue height not evident in abstract | C |
| Yi 2020 [2] | Retrospective, 349 implants, GEE | Angle examined with other prosthetic features | Radiographic definitions; limited geometry covariates | C |
| Raabe 2026 [3] | Cross-sectional, 332 implants, 11 years | Angle >30° (BL), >48° (TL) associated with disease; diameter also associated | Prevalent-case design; many covariates | C |
| Hentenaar 2020 [8] | Cross-sectional, 67 implants, 5 years | No contour effect on bone loss | Small; platform-switched only | C |
| Lops 2022 [7] | Retrospective, 312 implants, 3 years | Similar bone change at >30° and ≤30° | Short follow-up | C |
| Dukka 2026 [9] | Retrospective, 200 tissue-level implants | Over-contoured had less bone change | Radiographic; tissue-level only | C |
| Chang 2026 [11] | Retrospective, 89 implants | Angle rose with space-to-diameter ratio, fell with depth; no association with bone stability | Small; bone outcome only | C |
| Puisys 2023 [10] | Short communication, geometric derivation | Required tissue height follows from implant and crown dimensions | No outcome data | D |
| Rungtanakiat 2024 [12] | Cross-sectional, 103 implants, CBCT and scans | All angles negatively associated with mucosa height | Cross-sectional | C |
| Han 2025 [14] | Retrospective, 260 restorations | Association in the 0–2 mm zone, not 2–3 mm | One system | C |
| Strauss 2024 [19] | Preclinical | Wide angle increased bone loss | Angles beyond clinical range | C |
Evidence levels are operational categories for this manuscript, not GRADE: A strong, B moderate, C preliminary, D mechanistic or theoretical, E speculative hypothesis.
What the evidence says about the geometry
Three observations in the existing literature bear directly on the proxy hypothesis. First, a derivation based on the Pythagorean theorem has related the vertical height of the supracrestal complex to the contour angle, and its authors argue that a contour of 30° or less cannot be achieved without adequate vertical height [10]. Second, in a study of 103 bone-level implants measured with cone-beam computed tomography and intraoral scans, all three angles measured (mucosal emergence, deep, total) were negatively associated with mucosa height, and reduced mucosal height and width were associated with more signs of inflammation [12]. Third, in the retrospective study of 89 implants cited above, the angle was positively correlated with space-to-diameter ratio and negatively correlated with implant placement depth [11].
The zone at which the angle is measured also matters. In a study of CAD-CAM abutments, the profile angle correlated with bone loss in the 0 to 1 mm and 1 to 2 mm ranges from the platform but not in the 2 to 3 mm range, and cut-offs differed by site [14]. A narrative review has argued that the importance of abutment height depends on retention type, because screw-retained restorations place a mobile interface at a distance from bone that abutment height controls [21]. Other recent reviews emphasize emergence profile and angle as modifiable design features [22] and outline the evidence for design elements of the implant supracrestal complex [23]. A search of the clinical-trial registry for the angle returned no relevant interventional trial.
Evidence that depth and tissue height themselves matter
If depth and tissue height had no relation to disease, they could not confound. The available evidence is mixed but not empty. In a two-year clinical trial, bone remodeling and marginal bone loss depended mainly on abutment height, with crestal position and keratinized mucosa width also contributing in regression models [24]. In a seven-year randomized trial comparing 1-mm and 3-mm abutments at subcrestally placed implants, marginal bone change favored the taller abutment, while differences in peri-implantitis were not statistically significant [25]. A study of initial soft tissue thickness examined its influence on peri-implant bone remodeling [26], and a recent review identified crestal soft tissue thickness and the dimension of the mucosal attachment as factors worth assessing in the susceptibility to peri-implantitis [27]. In one randomized trial of buccal emergence profile design, the restorative angle in the test group was lower than in the control group, but the outcome was gingival margin recession and the sample was restricted to sites with adequate initial soft tissue thickness [28].
These studies make the vertical dimension at the platform a plausible, independently acting determinant of bone and mucosal outcomes, though most outcomes are marginal bone change rather than disease case definitions.
3. Hypothesis and theoretical framework
Formal statement
We state the hypothesis in three parts. H1 (determination). For a given restoration profile, the measured emergence angle is largely determined by three geometric quantities that are not themselves restorative contour choices: the platform diameter, the restoration width at the end of the measurement line, and the vertical height between platform and that point. H2 (confounding or mediation). Vertical height depends on implant depth and baseline supracrestal tissue height, and these have their own plausible effects on bone and mucosa. A crude association between angle and peri-implantitis therefore combines (a) paths from depth and tissue height that do not pass through the angle, (b) paths through the angle, and (c) paths from the site and patient that influence both crown width and disease. H3 (non-identification). Cross-sectional radiographs, in which depth and tissue height are unmeasured or measured with error, cannot distinguish these paths.
H1 and H2 are largely geometric and partly documented [10–12]. H3 is the methodological claim on which this manuscript rests.
Geometry
Consider a restoration whose profile between the implant platform and the end of the measurement line is approximated by a straight line, and which flares symmetrically. The angle θ between that line and the implant long axis satisfies
tanθ=((W_c−D_p)/2)/H
where θ is expressed in degrees from the implant axis; Wc (mm) is the width of the restoration at the point where the measurement line ends; Dp (mm) is the platform diameter at the level where the line begins; and H (mm) is the vertical distance along the implant axis from the platform to that end point. Wc and Dp are measured perpendicular to the axis in the same radiographic or digital plane. The end point is the mucosal exit of the restoration in some protocols and the crown’s height of contour in others, so Wc and H must be stated together with the landmark used. For one-sided measurement (for example mesial only), the numerator is the lateral offset δ on that side, and tan θ = δ/H.
If the platform lies at a signed depth s (mm, positive when subcrestal) below the bone crest and the supracrestal tissue extends a height S (mm) above the crest, and the measurement ends at the mucosal margin, then
H=S+s.
This is the vertical room available. Rearranging gives the maximum crown width compatible with a target angle θ*,
W_(c,max)=D_p+2Htanθ^*
and the minimum height for a given crown,
H_(min)=(W_c−D_p)/(2tanθ^*).
The sensitivity of the angle to height, in radians per millimeter, is
dθ/dH=−(δ/H^2)/(1+(δ/H)^2), δ=(W_c−D_p)/2.
The negative sign states that the angle rises as vertical room falls, steeply when H is small. A real profile is rarely straight. The angle computed from the straight line is a secant angle, and convex or concave profiles produce tangent angles at the mucosal exit that are respectively larger or smaller. We denote the departure from straightness by a profile-shape term κ, the one quantity in the angle that the clinician and laboratory can change without altering Wc at the height of contour, Dp, or H.
Illustrative computation under stated assumptions, not empirical data
We computed θ from the straight-profile equation for crown widths of 7, 9 and 11 mm, platform diameters of 3.5, 4.1, 4.8 and 5.5 mm, and heights of 1.5 to 5 mm (script geometry_sim.py in the paper folder). These ranges are chosen to be plausible for posterior sites; they are not measured values. Table 2 shows the 9-mm case. For Wc = 9 mm and Dp = 4.1 mm, the script returned θ = 50.8° at H = 2 mm, 39.2° at H = 3 mm and 31.5° at H = 4 mm. The minimum height for θ ≤ 30° was 4.24 mm for that combination, 2.51 mm for Wc = 7 mm with Dp = 4.1 mm, and 5.98 mm for Wc = 11 mm with Dp = 4.1 mm. In this arithmetic, each millimeter of lost height at moderate H costs roughly 8° to 12° of angle.
| Platform diameter Dp (mm) | H = 1.5 mm | H = 2 mm | H = 3 mm | H = 4 mm | H = 5 mm | Hmin for 30° (mm) |
|---|---|---|---|---|---|---|
| 3.5 | 61.4 | 54.0 | 42.5 | 34.5 | 28.8 | 4.76 |
| 4.1 | 58.5 | 50.8 | 39.2 | 31.5 | 26.1 | 4.24 |
| 4.8 | 54.5 | 46.4 | 35.0 | 27.7 | 22.8 | 3.64 |
| 5.5 | 49.4 | 41.2 | 30.3 | 23.6 | 19.3 | 3.03 |
The table makes the proxy argument concrete. A single narrow implant under a molar-width crown passes 30° unless the platform is deep or the tissue is tall. A threshold on θ is therefore partly a threshold on H and on the mismatch Wc − Dp.
The causal graph
Figure 1 shows the directed acyclic graph we propose, with each arrow stated in words.
Nodes. Site (tooth position, mesio-distal space, bone width) determines the crown width Wc, the platform diameter Dp chosen, and the achievable depth s. Patient factors (periodontitis susceptibility, plaque control, mucosal phenotype) influence baseline supracrestal tissue height S0 and the outcome Y, peri-implant mucositis or peri-implantitis. Depth s and baseline tissue height S0 together determine the height H. The angle θ is a deterministic function of Wc, Dp, H and the profile term κ. The outcome Y is influenced by θ (the contested path), and by s, S0, Dp, Site and Patient directly.
Direct paths from the geometric parents. Depth and tissue height have their own routes to Y: crestal remodeling after placement, mucosal dimension and its inflammatory consequences [12, 24, 27]. Platform diameter may also act directly. In a cross-sectional study, larger diameter was associated with more peri-implantitis [3], even though a larger platform lowers θ at fixed crown width and height. Any model that treats Dp only as a lever on the angle is therefore too simple.
Time ordering. The angle is fixed when the prosthesis is delivered, whereas tissue height can change afterward. In a cross-sectional study, wider angles were associated with less mucosal height [12], which could reflect geometry, remodeling under the restoration, or both. If so, mature tissue height S1 is a mediator of the angle’s effect and must not be used as a confounder. Only the baseline value S0, measured at the healing abutment or the first prosthetic appointment, belongs in the confounder set.
Measurement and the outcome. If H is measured at follow-up using the bone crest as the reference, bone loss changes the apparent depth. The reference then depends on the outcome, and adjustment would condition on a consequence of Y. We therefore require baseline radiographs or scans.
Estimands and adjustment sets
Because θ has several parents, “the effect of θ” is not one question. Reducing the angle by deeper placement, by a narrower crown, by a wider platform, or by reshaping the profile are different interventions with different downstream effects, and the potential-outcome framework requires that the intervention be well defined. We distinguish three estimands.
E1, total effect of vertical room. The effect on Y of changing H through s or S0, for example by deeper placement or soft tissue augmentation. This is the policy-relevant quantity for the surgical side. Adjustment set: Site and Patient factors. The angle is not conditioned on, because it is on the causal path.
E2, controlled direct effect of profile at fixed geometry. The effect on Y of changing the transmucosal profile (the term κ, or equivalently the effective restoration width at the mucosal exit) while holding Dp, s and S0 fixed. This is what a recontouring or prosthesis-modification intervention does. Adjustment set for observational data: Site, Patient, Dp, s and S0. Mature tissue height S1 and follow-up bone level are excluded as mediator and outcome-contaminated variable.
E3, effect of implant selection. The effect of changing Dp at a given crown width, operating through θ and through direct paths. Adjustment set: Site and Patient factors; s and S0 if the aim is to isolate the mismatch.
Adjusting for H in a model of θ and Y is appropriate for E2 but would remove part of the effect of interest for E1. Presenting the coefficients of all covariates in one model as if each were an effect of its own is the Table 2 fallacy described in epidemiology [29]. Adjustment for a mediator in estimating a total effect is a form of overadjustment [30]. DAG-based selection of confounders is the recommended safeguard against both [31, 32].
Determinism and the non-identification result
Under the straight-profile model, θ is an exact function of Wc, Dp and H. If all three are measured and included, θ adds no information beyond them: the estimand E2 is then not identified from data in which κ does not vary. In real data, the residual variation of the measured angle after conditioning on its geometric parents has two sources, the true profile shape κ and measurement error. If measurement error dominates, the coefficient for θ is biased toward the null and the conclusion “no direct effect” is uninformative. This yields a design requirement. E2 can be estimated only where κ varies independently of the parents, for example in a randomized recontouring comparison, or in observational data where angle is measured by a method that is more precise than the residual variation it must resolve. A reliability study of the angle belongs in any plan.
Illustrative computation under stated assumptions, not empirical data
We simulated cross-sectional samples of 3,000 sites in 300 replications, with H ~ normal(3.0, 1.0) mm truncated to 1 to 6 mm, Wc ~ normal(8.0, 1.5) mm truncated to 5.5 to 12 mm, and Dp drawn from 3.5, 4.1, 4.8 and 5.5 mm with probabilities 0.15, 0.45, 0.30 and 0.10. The angle followed the straight-profile equation. The outcome was binary with a logistic model. In scenario A, the log-odds depended on H only (0.5 per millimeter of lost height) and the angle was inert. In scenario B, both acted (0.5 per millimeter and 0.04 per degree above 30°). In scenario C, only the angle acted (0.06 per degree). We estimated the odds ratio for θ > 30° with and without adjustment for H, and repeated the analysis with H observed with reliability 0.6 (measurement error added so that the reliability ratio equals 0.6). One draw had 52% of sites above 30°, a correlation between θ and H of −0.61, and a correlation between θ and the mismatch Wc − Dp of 0.76.
| Scenario | H reliability | Crude OR | OR adjusted for H |
|---|---|---|---|
| A: H acts, angle inert | 1.0 | 1.55 (1.32–1.82) | 1.00 (0.84–1.19) |
| A: H acts, angle inert | 0.6 | 1.54 (1.30–1.82) | 1.23 (1.03–1.48) |
| B: both act | 1.0 | 3.58 (3.05–4.20) | 2.07 (1.73–2.49) |
| B: both act | 0.6 | 3.59 (3.00–4.36) | 2.71 (2.27–3.19) |
| C: angle only | 1.0 | 3.62 (3.08–4.25) | 2.82 (2.37–3.39) |
| C: angle only | 0.6 | 3.60 (3.10–4.20) | 3.17 (2.67–3.84) |
Three features matter. In scenario A, a crude odds ratio of about 1.5 arises with no angle effect and vanishes when H is adjusted for exactly. With a reliability of 0.6, adjustment leaves an odds ratio of 1.23 whose interval excludes 1, so residual confounding can masquerade as a small angle effect. In scenario C, adjustment for H attenuates the odds ratio even though the angle is the only cause, because part of its variation is explained by H. Attenuation after adjustment therefore does not by itself show that the angle is inert. The magnitudes depend wholly on the assumed parameters.
Numbered predictions and refuting results
P1. In unselected cohorts, the straight-profile prediction from Wc, Dp and H will account for most of the variance in measured θ. Refuted if the squared correlation is low (we propose below 0.5 as a pre-specified criterion), which would indicate that profile shape and measurement dominate and that the proxy argument is weak.
P2. The crude association of θ > 30° with bleeding on probing or peri-implantitis will be closer to the null after adjustment for baseline H or its components s and S0. Refuted if the adjusted estimate equals or exceeds the crude one in cohorts where H is reliably measured.
P3. Within narrow strata of H (for example 1-mm bands), the association between θ and the outcome will be weaker than in the pooled sample. Refuted if the stratum-specific associations equal the pooled one.
P4. An angle measured at a landmark at least 3 mm coronal to the platform, which shares the geometric parents but is distant from the supracrestal bone, will show an association with the outcome similar to that of the local angle if the proxy hypothesis holds, and a clearly weaker one if the effect is local. This is a negative-control exposure of imperfect specificity, because the two angles are correlated; its interpretation is conditional on the reported zone-specific pattern [14].
P5. In a randomized comparison at over-contoured bone-level restorations, recontouring will reduce bleeding on probing relative to sham polishing only if a direct effect of profile exists. Refuted (for the direct-effect hypothesis) if outcomes are equal within a pre-specified margin.
P6. The randomized estimate will lie closer to the H-adjusted observational estimate than to the crude estimate. Refuted if it matches the crude estimate, which would imply no confounding by H.
P7. If diameter acts only through the angle, larger platforms under a crown of fixed width will show less disease. The cross-sectional finding of more peri-implantitis with larger diameter [3] contradicts this; a direct diameter pathway must be modeled. Refuted if in matched analyses diameter shows the expected protective association with disease.
4. Proposed methodology
The design is staged so that each layer addresses a weakness of the one before. No layer has been run. Numerical inputs below are assumptions for planning, not findings.
Layer A: standardized re-measurement of existing cohorts
Population. Cohorts of bone-level implants restored with single crowns or short splinted prostheses, with a baseline radiograph or scan at prosthesis delivery and a follow-up examination with probing at one or more years. Cohorts that provided the original associations [1, 2], and those with null results [7, 8], are natural candidates if images are available and ethical permission allows re-analysis.
Measurements. (i) Restoration width Wc and platform diameter Dp at stated landmarks; (ii) H, defined as the vertical distance from the platform to the end of the measurement line, with its components s (signed depth relative to the bone crest) and baseline tissue height S0; (iii) the angle at three zones, the deep zone (0 to 2 mm from the platform), the zone to the mucosal exit, and the crown’s total contour, since zone-specific patterns have been reported [14] and since three-dimensional measurement has been combined with intraoral scanning [12]; (iv) profile type (convex, straight, concave); (v) outcomes, bleeding on probing, probing depth, and radiographic bone change, with a stated case definition of peri-implantitis [33].
Reliability. Two calibrated raters measure each site twice, at least two weeks apart, blind to outcome. We report intraclass correlation coefficients and Bland-Altman limits for θ, H, Wc and Dp. This step directly addresses the non-identification problem of Section 3.5: if the limits of agreement for θ are wider than the residual spread of θ given its geometric parents, E2 cannot be estimated from that cohort.
Analysis. Site-level logistic or ordinal models with generalized estimating equations clustered by patient. Model 1: θ alone. Model 2: adds Site and Patient confounders. Model 3: adds s, S0 and Dp (the E2 set). Models for E1 omit θ. We predefine strata of H in 1-mm bands, with the number of bands limited by the sample. We report the change in the log odds ratio for θ between Model 1 and Model 3 rather than interpreting any single coefficient, in line with warnings against reading all coefficients of a mutually adjusted model as causal effects [29]. A mediation analysis for the path s → H → θ → Y is secondary, with the stated caveat that its identifying assumptions include no unmeasured confounding of the angle-outcome relation and that the profile term κ must vary independently of the parents.
Unmeasured confounding. We calculate the E-value for the adjusted estimate [34], that is, the minimum strength of association an unmeasured confounder would need with both angle and outcome to explain the estimate away.
Layer B: target trial emulation
Where cohorts have enough baseline information, we specify the target trial first [35, 36]. Eligibility: patients with a bone-level implant restored with an over-contoured crown (angle above a stated threshold, for example 40°) and with baseline H recorded. Strategies: recontour the transmucosal profile to below 30° at fixed H and Dp, versus leave the restoration unchanged. Time zero: the date of the modification decision, so that groups are aligned at the start of follow-up. Outcome: bleeding on probing and incident bone loss at 3 years. Confounders: Site, Patient, s, S0, Dp, baseline bleeding and baseline bone level. The emulation answers E2 only if modification practice varies for reasons unrelated to disease risk, which is doubtful: clinicians recontour crowns that are already inflamed. A 30-month prospective case series combining debridement with prosthesis modification at mucositis sites had no unmodified comparison group [37]. The emulation is therefore a step toward the randomized layer, not a substitute.
Layer C: negative controls
Two negative controls can detect confounding by shared geometry. First, a negative-control exposure: an angle computed at a reference point at least 3 mm coronal to the platform. It shares Wc, Dp and the global dependence on H, but it lies away from the supracrestal bone and, on the reported zone-specific pattern, should carry little local effect [14]. If it predicts disease as well as the local angle, the association is more consistent with a geometric proxy. Because the two angles are correlated, the test supports a relative statement only. Second, a negative-control outcome: a disease measure not plausibly influenced by the transmucosal profile but sharing the patient-level risk factors, for example probing depth at the adjacent natural tooth. If crown-level angle at the implant predicts that outcome, patient-level confounding is likely.
Layer D: randomized recontouring trial
Design. A two-arm, parallel-group, assessor-blinded, randomized trial in patients with at least one bone-level implant restored with a crown whose measured angle exceeds a pre-specified value, for example 40°, and with peri-implant mucositis or bleeding on probing at baseline. Participants are randomized to prosthesis recontouring to an angle below 30° (re-fabrication or chairside modification, with a concave or straight transmucosal profile) or to sham polishing of the same surfaces with no change in contour. Both arms receive identical professional debridement and oral hygiene instruction. Randomization is stratified by baseline H (below 3 mm versus 3 mm or more) and by implant system. The stratifier makes the trial informative about effect modification by vertical room, and prevents the arms from being unbalanced on the main confounder by chance.
Feasibility constraint. At low H, a reduction to 30° requires narrowing the restoration at the mucosal exit, which changes Wc and may produce an undercontoured crown with open embrasures. The trial therefore tests recontouring as clinically achievable, which is a form of the E2 intervention but cannot hold every geometric parent fixed. This limits interpretation and is stated rather than hidden. A registered interventional trial of this specific comparison was not found in our single registry query; the registry search was narrow and should be repeated before protocol finalization.
Outcomes. Primary: the proportion of sites with bleeding on probing at 12 months. Secondary: probing depth, suppuration, radiographic bone change at 3 years, patient-reported outcomes, plaque index, crown retention and technical complications.
Blinding and ethics. The assessor is blinded to allocation. Ethical approval, consent and registration are required and have not been obtained; this is a protocol outline, not a registered trial. The sham arm is offered recontouring at study end.
Sample size reasoning
For a two-arm comparison of means with two-sided α = 0.05 and 80% power,
n_(per arm)=(2 (z_(1−α/2)+z_(1−β))^2 σ^2)/(Δ^2)
where σ is the standard deviation of the outcome and Δ the difference to detect. With assumed values σ = 0.6 mm of bone change and Δ = 0.3 mm, the calculation (run in the paper folder) gives 62.8, rounded up to 63 per arm, and 75 per arm after 15% attrition. For a binary primary outcome, with an assumed bleeding prevalence of 0.50 under sham and 0.30 under recontouring, the standard two-proportion formula gives 93 per arm and 110 per arm after attrition. These values are planning assumptions, not estimates from the literature. One implant per patient, or a cluster adjustment with design effect 1 + (m − 1)ρ where m is implants per patient and ρ the intracluster correlation, is required if several implants per patient are analyzed. A stratified analysis powered to detect effect modification by H would require substantially more participants, and the trial as outlined would be underpowered for that.
Bias control
Selection: consecutive recruitment with screening logs. Information: blinded assessment and duplicate angle measurement. Confounding: randomization in Layer D and the Section 3.4 adjustment sets in Layers A and B. Attrition: intention-to-treat analysis with missing-data sensitivity analyses.
5. Competing explanations and how the design discriminates among them
Four explanations account for a crude angle-disease association. Table 1 and Section 3 give their evidence base; here we state what each predicts under the design.
Direct causal effect of the angle. A wide transmucosal profile may impede cleaning, compress the mucosa, or disturb the junctional epithelium; the preclinical studies are consistent with this [19, 20], and a narrative review argues that modifying the profile can resolve inflammation [22]. This hypothesis predicts persistence of the association within strata of H (P3), a stronger local than distant angle (P4), and a randomized benefit of recontouring (P5). It does not predict equality of the crude and adjusted estimates.
Confounding by vertical room. If depth and tissue height act through routes that do not involve the angle, the association attenuates with adjustment (P2) and within strata (P3), the distant angle performs about as well as the local one (P4), and recontouring gives no benefit beyond sham (P5, P6).
Mediation of vertical room by the angle. If shallow placement harms through the angle, the crude association reflects a true effect of H, and E1 is large while E2 is also non-zero. The randomized recontouring benefit (P5) would be present but smaller than the crude association implies (P6). Adjustment for H in this scenario removes part of the real effect, as the scenario C rows of Table 3 illustrate.
Site-level confounding by position. Large crowns on narrow platforms are typical of molar sites, where access for hygiene, splinting and load differ from premolar sites. This predicts that the association weakens after adjusting for tooth type and that the negative-control outcome at the adjacent tooth is also associated with the angle.
The design discriminates only partly. Stratification cannot separate mediation from direct effect without independent variation in κ. The trial can, but only over the achievable range of angle reduction. A null trial would reject a direct effect within that range and leave the observational association attributable to confounding; the combination of layers is more informative than any one.
6. Clinical and engineering implications
These implications are conditional on the predictions holding and should not be applied clinically on the basis of this manuscript.
If P2 and P3 are borne out, the 30° rule would be better understood as a flag for inadequate vertical room or excessive mismatch than as a stand-alone treatment target. Prevention would then lie in treatment planning: placing the platform deeper where biology allows, choosing platform diameters in relation to the planned crown, preparing or augmenting tissue height, and avoiding wide crowns on narrow platforms. A randomized trial of abutment height found that vertical dimension at the platform can influence bone change [25], which fits this view, although that trial did not target the angle.
If P5 holds, recontouring remains a reasonable intervention for established mucositis, as suggested by observational and review evidence [23], and the threshold would be treated as a design guide rather than a proven cut-off. If P5 fails but P2 holds, recontouring a crown without addressing vertical room would be expected to give little benefit.
Studies of the angle could report H, s, S0, Wc, Dp, the landmark and measurement reliability, so readers can judge whether an association is separable from geometry. Design software could display the vertical room needed for a target angle at planning, as the earlier derivation suggests [10].
7. Limitations
Geometry. The straight-profile equation is a simplification. Real profiles are curved, platform-switched abutments add a step, and angles measured on radiographs are projections that capture only mesial and distal surfaces; buccal and lingual geometry is unmeasured. The equation also assumes a symmetric flare and a single, consistently defined landmark. The published angles use different landmarks and zones [14], so H and Wc are protocol-specific, and a method that works for one definition may not transfer.
Causal graph. The graph encodes assumptions, not facts. Arrows we omitted, for example from plaque-retention behavior to crown design, or from the clinician’s skill to both depth and contour, could open further backdoor paths. Depth and supracrestal tissue height are treated as separate nodes, although in practice the surgeon chooses depth in response to tissue height. The disease mechanism itself is debated: a narrative review sets biofilm-driven and prosthetically driven explanations against one another [38], and the graph does not adjudicate between them.
Simulation. The computations in Section 3 use arbitrary distributions and effect sizes. They show that a pattern is possible, not that it occurs. The reliability of 0.6 is an unverified illustrative value.
Evidence base. Evidence was appraised from PubMed abstracts only; full texts, supplementary tables and adjustment sets of the cited studies were not examined. Several statements about what the original studies did or did not adjust for are therefore uncertain, and a study that did adjust for depth would weaken the confounding argument. Embase, Scopus and Web of Science were not accessible, and the search used simple queries because several Boolean combinations returned nothing.
Design. The randomized layer cannot hold every geometric parent fixed, as Section 4.4 notes. The sample size is computed for a main effect only. Bleeding on probing is a variable outcome, and three-year radiographic bone change is small relative to measurement error.
Scope. The paper concerns bone-level posterior single crowns or short splints. Tissue-level implants, where reports differ [3, 9], and anterior esthetic sites [28] are outside scope.
8. Conclusions
The emergence angle is geometrically bound to implant depth, tissue height and the mismatch between crown and platform, as earlier work has shown [10, 12]. The central point is what follows from that dependence for causal inference. Cross-sectional radiographs cannot separate confounding by vertical room from a direct or mediated effect of the angle; an exact geometric relation makes the angle’s own effect identifiable only where the profile shape varies independently of its parents, and measurement error can mask that variation. We propose a graph, three estimands with adjustment sets, seven refutable predictions, and a four-layer design ending in a randomized recontouring trial. Until such a design is run, the 30° figure is best treated as a marker of risk of uncertain causal status.
Cite this paper
Jain S, Gupta A. Is the Restorative Emergence Angle a Cause or a Geometric Proxy? Placement Depth, Diameter Mismatch, and Peri-Implantitis Risk. Working paper. Center for Implant Dentistry; 2026. Available from: https://bayareaimplantdentistry.com/clinician-research-and-publications/emergence-angle-geometric-proxy
References
- Katafuchi M, Weinstein BF, Leroux BG, Chen YW, Daubert DM. Restoration contour is a risk indicator for peri-implantitis: A cross-sectional radiographic analysis. J Clin Periodontol. 2018;45(2):225-232. doi:10.1111/jcpe.12829 (opens in a new tab). PMID: 28985447 (opens in a new tab).
- Yi Y, Koo KT, Schwarz F, Ben Amara H, Heo SJ. Association of prosthetic features and peri-implantitis: A cross-sectional study. J Clin Periodontol. 2020;47(3):392-403. doi:10.1111/jcpe.13251 (opens in a new tab). PMID: 31912511 (opens in a new tab).
- Raabe C, Fonseca M, Chappuis V, Avila-Ortiz G, Rodrigues DM, Galindo-Moreno P, et al. Association of Transmucosal Emergence Geometry and Peri-Implant Diseases Prevalence Around Bone- and Tissue-Level Implants: A Cross-Sectional Study. Clin Oral Implants Res. 2026;37(7):883-895. doi:10.1111/clr.70131 (opens in a new tab). PMID: 41966820 (opens in a new tab).
- Soulami S, Slot DE, van der Weijden F. Implant-abutment emergence angle and profile in relation to peri-implantitis: A systematic review. Clin Exp Dent Res. 2022;8(4):795-806. doi:10.1002/cre2.594 (opens in a new tab). PMID: 35713938 (opens in a new tab).
- Atieh MA, Shah M, Ameen M, Tawse-Smith A, Alsabeeha NHM. Influence of implant restorative emergence angle and contour on peri-implant marginal bone loss: A systematic review and meta-analysis. Clin Implant Dent Relat Res. 2023;25(5):840-852. doi:10.1111/cid.13214 (opens in a new tab). PMID: 37183357 (opens in a new tab).
- Lin GH, Lee E, Barootchi S, Rosen PS, Curtis D, Kan J, et al. The influence of prosthetic designs on peri-implant bone loss: An AO/AAP systematic review and meta-analysis. J Periodontol. 2025;96(6):634-651. doi:10.1002/JPER.24-0144 (opens in a new tab). PMID: 40489290 (opens in a new tab).
- Lops D, Romeo E, Stocchero M, Palazzolo A, Manfredi B, Sbricoli L. Marginal Bone Maintenance and Different Prosthetic Emergence Angles: A 3-Year Retrospective Study. J Clin Med. 2022;11(7). doi:10.3390/jcm11072014 (opens in a new tab). PMID: 35407622 (opens in a new tab).
- Hentenaar DF, De Waal YC, Van Winkelhoff AJ, Raghoebar GM, Meijer HJ. Influence of Cervical Crown Contour on Marginal Bone Loss Around Platform-Switched Bone-Level Implants: A 5-Year Cross-Sectional Study. Int J Prosthodont. 2020;33(4):373-379. doi:10.11607/ijp.6365 (opens in a new tab). PMID: 32639696 (opens in a new tab).
- Dukka HB, Alhachache S, Imbrogno M, Mallela D, AlAsaad N, Hill M, et al. Impact of over-contoured restorations on the marginal bone around tissue-level implants: A retrospective radiographic analysis. J Periodontol. 2026. doi:10.1002/jper.70111 (opens in a new tab). PMID: 41789944 (opens in a new tab).
- Puisys A, Janda M, Auzbikaviciute V, Gallucci GO, Mattheos N. Contour angle and peri-implant tissue height: Two interrelated features of the implant supracrestal complex. Clin Exp Dent Res. 2023;9(3):418-424. doi:10.1002/cre2.731 (opens in a new tab). PMID: 36988518 (opens in a new tab).
- Chang CC, Chang J, Osborne A, Al Hatem AMS, Chien EC, Wismeijer D, et al. Emergence angles of posterior implant-supported fixed dental prostheses and their effect on peri-implant bone loss: A retrospective clinical study. J Prosthet Dent. 2026;135(2):340.e1-340.e10. doi:10.1016/j.prosdent.2025.09.011 (opens in a new tab). PMID: 41047319 (opens in a new tab).
- Rungtanakiat P, Thitaphanich N, Janda M, Strauss FJ, Arksornnukit M, Mattheos N. Association of Peri-Implant Mucosa Dimensions With Emergence Profile Angles of the Implant Prosthesis. Clin Exp Dent Res. 2024;10(4):e939. doi:10.1002/cre2.939 (opens in a new tab). PMID: 39039934 (opens in a new tab).
- Mauland EK, Bunæs DF, Lie SA, Verket A. Association Between Local Factors and Implant-Related Complications: A Population-Sampled Cross-Sectional Analysis. J Clin Periodontol. 2026. doi:10.1111/jcpe.70209 (opens in a new tab). PMID: 42869496 (opens in a new tab).
- Han JW, Han JW, Pyo SW, Kim S. Impact of profile angle of CAD-CAM abutment on the marginal bone loss of implant-supported single-tooth posterior restorations. J Prosthet Dent. 2025;134(4):1180-1186. doi:10.1016/j.prosdent.2023.11.025 (opens in a new tab). PMID: 38129258 (opens in a new tab).
- Sirirattanagool P, Asavanamuang P, Jain S, Tavelli L, Finkelman M, Chen YW, et al. Prosthetic factors influencing the prevalence of peri-implant diseases and marginal bone loss in static computer-assisted implant sites: A cross-sectional study. J Periodontol. 2026;97(1):47-61. doi:10.1002/jper.11387 (opens in a new tab). PMID: 40820487 (opens in a new tab).
- Volp Junior LC, Matarazzo F, Dias DR, de Oliveira RP, Sábio S, Araújo MG. The effect of the interproximal contour of single external hexagon implant restorations on the prevalence of peri-implantitis: A retrospective study. J Prosthodont. 2024;33(7):655-662. doi:10.1111/jopr.13835 (opens in a new tab). PMID: 38487989 (opens in a new tab).
- Chou YH, Hsiao WL, Chen CJ, Lin YC, Liu PF, Hu KF, et al. Association between prosthesis contour and peri-implantitis in patients compliant with supportive periodontal therapy: A retrospective cohort study. J Prosthodont Res. 2025;69(1):68-75. doi:10.2186/jpr.JPR_D_23_00236 (opens in a new tab). PMID: 39198202 (opens in a new tab).
- Dellacasa P, Schiavon L, Menini M, Pesce P, Tomasi C, Bressan E. Prosthetic Emergence Angles and Implant Outcomes: A Retrospective Study With at Least 5 Years of Follow-Up. Clin Implant Dent Relat Res. 2026;28(2):e70127. doi:10.1111/cid.70127 (opens in a new tab). PMID: 41736707 (opens in a new tab).
- Strauss FJ, Park JY, Lee JS, Schiavon L, Smirani R, Hitz S, et al. Wide Restorative Emergence Angle Increases Marginal Bone Loss and Impairs Integrity of the Junctional Epithelium of the Implant Supracrestal Complex: A Preclinical Study. J Clin Periodontol. 2024;51(12):1677-1687. doi:10.1111/jcpe.14070 (opens in a new tab). PMID: 39385502 (opens in a new tab).
- Strauss FJ, Monje A, Park JY, Lee JS, Hitz S, Jung RE, et al. Peri-Implant Inflammation Varies With Different Restorative Emergence Angles: An Histological Evaluation. J Esthet Restor Dent. 2026;38(5):1009-1016. doi:10.1111/jerd.70098 (opens in a new tab). PMID: 41527488 (opens in a new tab).
- Kim S, Mattheos N, Strauss FJ. Abutment height matters - but not always: the role of prosthetic retention type in peri-implant marginal bone loss. J Adv Prosthodont. 2026;18(4):392-406. doi:10.4047/jap.2026.18.4.392 (opens in a new tab). PMID: 42688413 (opens in a new tab).
- Pirc M, Esquivel J, Patrizzi A, Jung RE, Strauss FJ. Emergence Profile Angle Matters-Restoring Peri-Implant Health by Adjusting Prosthetics-A Narrative Review. J Esthet Restor Dent. 2026;38(6):1188-1196. doi:10.1111/jerd.70109 (opens in a new tab). PMID: 41618482 (opens in a new tab).
- Mattheos N, Janda M, Acharya A, Pekarski S, Larsson C. Impact of design elements of the implant supracrestal complex (ISC) on the risk of peri-implant mucositis and peri-implantitis: A critical review. Clin Oral Implants Res. 2021;32 Suppl 21:181-202. doi:10.1111/clr.13823 (opens in a new tab). PMID: 34642979 (opens in a new tab).
- Quispe-López N, Guadilla Y, Gómez-Polo C, López-Valverde N, Flores-Fraile J, Montero J. The influence of implant depth, abutment height and mucosal phenotype on peri‑implant bone levels: A 2-year clinical trial. J Dent. 2024;148:105264. doi:10.1016/j.jdent.2024.105264 (opens in a new tab). PMID: 39053878 (opens in a new tab).
- Muñoz M, Pardo G, Vilarrasa J, Babiano Á, Ruíz-Magaz V, Nart J. Long-Term Impact of 1-mm Versus 3-mm Definitive Abutments on Marginal Bone Loss and Peri-Implant Disease: A 7-Year Randomised Clinical Trial. J Clin Periodontol. 2025;52(11):1595-1604. doi:10.1111/jcpe.70009 (opens in a new tab). PMID: 40827015 (opens in a new tab).
- Vervaeke S, Dierens M, Besseler J, De Bruyn H. The influence of initial soft tissue thickness on peri-implant bone remodeling. Clin Implant Dent Relat Res. 2014;16(2):238-47. doi:10.1111/j.1708-8208.2012.00474.x (opens in a new tab). PMID: 22758656 (opens in a new tab).
- Monje A, Jung RE, Valles C, Naenni N. Emerging risk factors influencing the occurrence of peri-implantitis. Periodontol 2000. 2026. doi:10.1111/prd.70031 (opens in a new tab). PMID: 41881525 (opens in a new tab).
- Wang J, Tang Y, Qiu L, Yu H. Influence of buccal emergence profile designs on peri-implant tissues: A randomized controlled trial. Clin Implant Dent Relat Res. 2022;24(3):329-338. doi:10.1111/cid.13088 (opens in a new tab). PMID: 35389559 (opens in a new tab).
- Westreich D, Greenland S. The table 2 fallacy: presenting and interpreting confounder and modifier coefficients. Am J Epidemiol. 2013;177(4):292-8. doi:10.1093/aje/kws412 (opens in a new tab). PMID: 23371353 (opens in a new tab).
- Schisterman EF, Cole SR, Platt RW. Overadjustment bias and unnecessary adjustment in epidemiologic studies. Epidemiology. 2009;20(4):488-95. doi:10.1097/EDE.0b013e3181a819a1 (opens in a new tab). PMID: 19525685 (opens in a new tab).
- Tennant PWG, Murray EJ, Arnold KF, Berrie L, Fox MP, Gadd SC, et al. Use of directed acyclic graphs (DAGs) to identify confounders in applied health research: review and recommendations. Int J Epidemiol. 2021;50(2):620-632. doi:10.1093/ije/dyaa213 (opens in a new tab). PMID: 33330936 (opens in a new tab).
- Greenland S, Pearl J, Robins JM. Causal diagrams for epidemiologic research. Epidemiology. 1999;10(1):37-48. PMID: 9888278 (opens in a new tab).
- Schwarz F, Derks J, Monje A, Wang HL. Peri-implantitis. J Clin Periodontol. 2018;45 Suppl 20:S246-S266. doi:10.1111/jcpe.12954 (opens in a new tab). PMID: 29926484 (opens in a new tab).
- VanderWeele TJ, Ding P. Sensitivity Analysis in Observational Research: Introducing the E-Value. Ann Intern Med. 2017;167(4):268-274. doi:10.7326/M16-2607 (opens in a new tab). PMID: 28693043 (opens in a new tab).
- Hernán MA, Robins JM. Using Big Data to Emulate a Target Trial When a Randomized Trial Is Not Available. Am J Epidemiol. 2016;183(8):758-64. doi:10.1093/aje/kwv254 (opens in a new tab). PMID: 26994063 (opens in a new tab).
- Hernán MA, Wang W, Leaf DE. Target Trial Emulation: A Framework for Causal Inference From Observational Data. JAMA. 2022;328(24):2446-2447. doi:10.1001/jama.2022.21383 (opens in a new tab). PMID: 36508210 (opens in a new tab).
- de Tapia B, Bonnin M, Valles C, Mozas C, Herrera D, Sanz M, et al. Clinical outcomes and associated factors in the treatment of peri-implant mucositis, combining mechanical debridement and prosthesis modification: A 30-month follow-up prospective case series. J Clin Periodontol. 2022;49(12):1357-1365. doi:10.1111/jcpe.13711 (opens in a new tab). PMID: 36054620 (opens in a new tab).
- Roshan Arbaaz B, Peter S, Kangaiyan S, Shiamala J, Maheswaran T, Arunkumar VR. Peri-Implantitis: Biofilm-Induced Disease or Prosthetically Driven Complication?. Cureus. 2026;18(7):e113497. doi:10.7759/cureus.113497 (opens in a new tab). PMID: 42662765 (opens in a new tab).

