Clinical outcomes of a Locator-bilateral segmental bar concept versus a conventional rigid bar concept for maxillary implant-supported overdentures: a retrospective cohort study

Abstract

Background

This retrospective cohort study evaluated a Locator-bilateral segmental bar concept (LBSB) for maxillary implant-supported overdentures and compared it with a conventional rigid bar concept (CRB) in terms of implant survival, peri-implant conditions, marginal bone loss (MBL), complications, and numeric rating scale (NRS)-based satisfaction.

Methods

Consecutive patients treated between January 2013 and December 2025 were reviewed. Eligible patients had been treated using either the LBSB or CRB concept. Outcomes included implant survival, MBL, modified sulcus bleeding index (mSBI), modified plaque index (mPI), time-to-first complication, NRS-based overall patient satisfaction (primary outcome), and seven domain-specific satisfaction items. Between-group associations were assessed using regression models adjusted for follow-up duration; complications were analyzed using Kaplan-Meier methods and Firth-penalized Cox models.

Results

Nineteen patients were included (LBSB, n = 10; CRB, n = 9), with mean follow-up durations of 5.10 ± 3.07 and 5.67 ± 3.54 years, respectively. Implant and prosthesis survival were 100% in both groups. In adjusted analyses, the LBSB group showed lower MBL than the CRB group (adjusted mean difference, − 0.310 mm; p = 0.007), whereas mSBI and mPI did not differ significantly. Kaplan-Meier curves for any complication were comparable between groups (log-rank p = 0.860), while the LBSB group showed a lower estimated hazard of relining (HR, 0.056; Holm-adjusted p = 0.019). Overall satisfaction scores appeared higher in the LBSB group, with an adjusted mean difference of 0.983 (p = 0.031).

Conclusions

Within the limitations of this small non-randomized retrospective cohort, the LBSB concept was associated with lower MBL, an estimated lower hazard of relining, and higher NRS-based satisfaction compared with the CRB concept. These findings should be interpreted as exploratory, hypothesis-generating concept-level associations, rather than as evidence of clinical superiority or causal effects of any individual design component. Confirmation in adequately powered prospective studies is warranted.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12903-026-08764-y.

Background

Tooth loss remains a major global health burden, particularly among aging populations, and substantially impairs quality of life [1]. Edentulism reduces masticatory efficiency and may contribute to nutritional deficiencies and related systemic consequences [2].

It is also associated with compromised facial support, impaired speech, and reduced self-esteem, leading to aesthetic dissatisfaction and psychological distress [3].

Dental implant therapy has transformed the management of edentulism and generally provides better outcomes than conventional complete dentures [4–6]. Fixed implant-supported prostheses offer high stability and favorable functional and aesthetic performance [7]. Their use, however, depends on adequate bone volume and the patient’s ability to maintain meticulous oral hygiene [8, 9]. Furthermore, high treatment costs and mechanical complications remain challenges that may limit access for a broader patient population [10].

Implant-supported overdentures offer a flexible alternative, particularly for older adults. Compared with conventional complete dentures, they improve retention and stability while facilitating daily hygiene, but usually require sufficient inter-arch space to accommodate attachment components and prosthetic materials [11]. Four-implant bar-supported maxillary overdentures have been described as a viable treatment option with a growing body of supporting evidence [5, 12, 13]. By splinting implants with a rigid bar superstructure, this approach may contribute to improved denture stability and remains a recognized treatment option in clinical practice [13]. However, bar-supported designs require sufficient restorative space for the bar-attachment complex and adequate denture base thickness [11]. In addition, cantilever effects may occur, potentially resulting in less favorable stress distribution and increasing the biomechanical burden on implants. Clinically, these factors may lead to higher maintenance demands, including attachment wear, screw loosening, and the need for denture base relining or repair [14].

Although current evidence supports the use of maxillary implant-supported overdentures, the optimization of treatment protocols to balance restorative space, retention, hygiene access, and long-term maintenance has not been sufficiently characterized.

Against this background, we evaluated an alternative concept for maxillary implant-supported overdentures combining anterior Locator attachments with bilaterally segmented bars. We compared this concept with conventional rigid bar overdentures by assessing implant survival, marginal bone loss, peri-implant soft-tissue indices, complications, and patient-reported satisfaction. The two concepts differed simultaneously in implant number, implant distribution, and retention design; therefore, this study was designed to compare the overall clinical performance of two concepts rather than to evaluate the independent effect of any single component. Our working hypothesis was that the LBSB concept and the CRB concept would show differences in clinical, maintenance-related, and patient-reported outcomes at the concept level.

Methods

Study design and setting

This retrospective cohort study was conducted at Beijing Stomatological Hospital, Capital Medical University, and approved by the institutional Ethics Committee (Approval No. A-2025-02). The study was conducted in accordance with the Declaration of Helsinki. We retrospectively reviewed consecutive patients treated between January 2013 and December 2025, with follow-up censored on December 31, 2025. This study was reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement; the completed checklist is provided in Additional file 1.

Participants

Eligible patients had complete maxillary edentulism or no more than two remaining natural maxillary teeth confined to the posterior region. Patients were excluded if systemic or local conditions were considered likely to compromise surgical or prosthetic outcomes (e.g., Sjögren’s syndrome, oral mucosal disorders, temporomandibular joint disorders, or uncontrolled systemic disease). A total of 112 patients were screened for eligibility. Of these, 86 were excluded because they had 3 or more remaining maxillary teeth (n = 18), had received maxillary implant-supported fixed dentures (n = 25), had received mandibular implant-supported overdentures (n = 36), or had systemic/local conditions likely to compromise outcomes (n = 7). The remaining 26 patients met the initial eligibility criteria and underwent medical record review. Seven patients were excluded for insufficient follow-up, which was defined as either a follow-up duration of less than 12 months after definitive prosthesis delivery or failure to complete the scheduled annual recall, resulting in incomplete records for final clinical, radiographic, complication-related, or patient-reported outcome assessments. Ultimately, 19 consecutive eligible patients were included in the final analysis, comprising 10 patients in the LBSB group and 9 patients in the CRB group.

As this was a retrospective study including all consecutive eligible patients, no formal a priori sample size calculation or power analysis was performed. The final sample size was determined by the number of eligible patients available during the study period. Given the small sample size, inferential analyses were considered exploratory, with emphasis placed on effect estimates and confidence intervals rather than definitive hypothesis testing.

Treatment allocation and clinical feasibility criteria

Treatment allocation was non-randomized and based on clinical and radiographic assessment. The treating clinicians first determined whether maxillary implant-supported overdenture treatment was clinically appropriate and whether both study concepts were technically feasible. The clinical feasibility criteria included the absence of systemic or local conditions likely to compromise surgical or prosthetic outcomes; sufficient maxillary bone volume to permit the planned implant number and an acceptable anteroposterior spread; adequate interarch restorative space for the intended attachment/bar components; ridge and mucosal conditions compatible with overdenture support and maintenance; an occlusal relationship compatible with an implant-supported overdenture; and the patient’s ability to maintain oral hygiene and attend supportive maintenance.

Patients were included only when both treatment concepts were considered clinically feasible at baseline. The clinicians then explained the basic features, implant number and distribution, retention design and treatment procedures of each concept, and the final treatment choice was made by the patient. Given the retrospective nature of the study, the specific reasons for each patient’s choice were not systematically documented in the medical records. Therefore, unmeasured patient-related factors may have influenced treatment allocation.

Clinical procedures and prosthetic workflow

Implant surgery was performed using Straumann implants (Switzerland), including bone-level cylindrical or tapered implants with diameters of 3.3–4.8 mm and lengths of 8–12 mm. No concomitant sinus floor elevation, guided bone regeneration, or other ridge augmentation procedures were performed. A delayed loading protocol was applied in all cases before definitive prosthetic rehabilitation. Two treatment concepts were used: the Locator-bilateral segmental bar concept (LBSB group) and the conventional rigid bar concept (CRB group). All bar superstructures were fabricated from either cast metal or zirconia. In both groups, definitive maxillary overdentures were reinforced with a metal framework and designed with a reduced-coverage palatal plate. All patients had a complete-arch removable mandibular prosthesis as the opposing dentition, and the definitive maxillary overdenture was designed and adjusted to achieve bilaterally balanced occlusion.

In terms of design details, in the LBSB group, five to six implants were distributed along the anteroposterior dimension to provide broader prosthetic support. Locator attachments were placed anteriorly, while the posterior implants were splinted with bilateral segmental bars. No retentive clips were incorporated into the denture base.

Instead, the bars directly contacted the intaglio surface of the metal framework. During surveying, lateral cast tilting was used to establish an oblique path of insertion and removal by altering the surveyed buccolingual undercut distribution in the posterior region, thereby enhancing resistance to denture dislodgement [15]. In the CRB group, four implants were placed to maximize the anteroposterior spread and were rigidly splinted with a bar superstructure. Given that the groups differed in implant number, implant distribution, and retention design, this study should be interpreted as a comparison of two treatment concepts rather than an assessment of the isolated effect of any single factor. Therefore, no attempt was made to estimate component-specific effects.

All participants were enrolled in a standardized supportive maintenance program after definitive prosthesis delivery. Recall visits were scheduled annually, with additional visits arranged when patients reported discomfort, reduced retention, prosthesis instability, or other prosthetic problems. At each visit, peri-implant tissues, oral hygiene, prosthesis fit, retention, occlusion, attachment wear, and the integrity of the bar/Locator components were assessed. Professional cleaning and individualized oral hygiene reinforcement were provided when indicated. Maintenance procedures were recorded and used to identify maintenance-related complications and determine their dates of occurrence.

Outcomes

Implant survival was defined as the implant remaining in situ at the follow-up visit.

Implant success was defined according to Buser-based criteria [16] as an implant remaining in situ without persistent implant-related complaints, recurrent peri-implant infection with suppuration, clinical mobility, continuous peri-implant radiolucency, or loss of prosthetic restorability. In line with published consensus terminology [17], prosthesis survival was operationally assessed as the definitive overdenture remaining in clinical service at the last follow-up visit, irrespective of whether maintenance or repair had been required. Prosthesis success was assessed more stringently as the definitive overdenture remaining in service without any recorded biological, mechanical, or maintenance-related complication requiring clinical or laboratory intervention.

Marginal bone loss (MBL, mm) was assessed on intraoral periapical radiographs obtained at definitive prosthesis delivery and follow-up visits. Radiographs were obtained using the department’s standardized long-cone paralleling protocol with a film/sensor holder; however, because of the retrospective study design, no individualized bite registration or customized positioning stent was used to reproduce projection geometry across serial examinations. Radiographs were included only when the implant shoulder, implant body/threads, and mesial and distal crestal bone levels were clearly visible without severe distortion or overlap. Each radiograph was calibrated using the known implant length recorded in the surgical chart. MBL was measured from the implant shoulder to the most coronal bone-to-implant contact on the mesial and distal aspects, corrected for magnification, and averaged to obtain implant-level MBL.

Measurements were performed by a single examiner using de-identified, coded radiographs, without access to treatment allocation. Intra-examiner repeatability was assessed using a two-way mixed-effects, absolute-agreement, single-measure intraclass correlation coefficient (ICC). All radiographic measurements were repeated by the same examiner after a 2-week interval, and the mean of the two measurements was used for analysis. Given the small sample size and clustering of implants within patients, primary analyses were conducted at the patient level by averaging implant-level MBL values within each patient.

Peri-implant bleeding was evaluated at the last follow-up visit using the modified sulcus bleeding index (mSBI) as described by Mombelli et al. [18]. A Williams periodontal probe was gently passed along the peri-implant sulcus, and bleeding at the gingival margin within 30 s was recorded on a 0–3 scale. Peri-implant plaque accumulation was assessed using the modified plaque index (mPI) according to Mombelli’s criteria [18], also scored on a 0–3 scale. Site-level scores were first averaged for each implant, and the resulting implant-level means were then averaged across all implants within the same patient. Follow-up duration was recorded for each patient and treated as a covariate.

Complications were assessed as time-to-first-event outcomes using longitudinal clinical records from scheduled annual recalls and unscheduled maintenance visits, including records of clinical or laboratory interventions. Five complication types were evaluated: hyperplasia, sore spot, base fracture, tooth fracture, and relining.

Complications were recorded when clinical or laboratory intervention was required, including adjustment for sore spots, tissue management for hyperplasia, repair of denture base or tooth fractures, or chairside/laboratory relining for loss of fit. A composite endpoint termed “any complication” was defined as the first occurrence of any of these events. Time-to-first complication was calculated from definitive prosthesis delivery to the first event; patients without the event of interest were censored at their last documented follow-up.

Patient-reported satisfaction was assessed once at the final follow-up visit using an 11-point numeric rating scale (NRS) ranging from 0 to 10, with 0 indicating “not satisfied at all” and 10 indicating “completely satisfied”. Overall satisfaction was recorded as a global rating, and seven domain-specific items were assessed: masticatory function, esthetics, foreign body sensation, pronunciation, ease of denture handling, stability, and ease of oral hygiene. Standardized verbal instructions were provided, and when clarification was required, only the NRS anchors were repeated without further interpretation. The NRS was used as a simple, low-burden clinical measure to capture patients’ perceived satisfaction with the prosthesis and specific functional domains. Its 0–10 scoring format provides clinically interpretable information in routine follow-up settings and allows between-group differences to be expressed directly in familiar score units. Nevertheless, the NRS does not capture the broader multidimensional constructs assessed by validated PROMs or OHRQoL instruments, such as the OHIP-14, which were not routinely collected in this retrospective cohort. Accordingly, these scores should be interpreted as NRS-based satisfaction data rather than as validated multidimensional patient-reported outcomes.

Statistical analysis

For MBL, mSBI and mPI, the association between treatment group and the outcome was evaluated using multivariable linear regression, with follow-up duration included as a covariate. Potential effect modification by follow-up duration was assessed by additionally fitting an interaction term between treatment group and mean-centered follow-up duration (Group × FU_c). Regression coefficients are reported with 95% confidence intervals (CIs). To address potential heteroscedasticity and small-sample bias, heteroscedasticity-consistent (HC3) robust standard errors were used as sensitivity analyses. Adjusted mean MBL (estimated marginal means) at the mean follow-up duration was derived from the fitted model and presented with 95% CIs.

Kaplan-Meier methods were used to estimate time-to-first complication (complication-free survival). Between-group differences were assessed using log-rank tests for the composite endpoint (any complication); for analyses of the five individual complication types, exact log-rank tests were used due to sparse event counts. Effect sizes were estimated using Firth-penalized Cox proportional hazards models to mitigate small-sample bias and potential monotone-likelihood issues in sparse-event settings, and are reported as hazard ratios (HRs) with 95% confidence intervals (CIs) for LBSB relative to CRB. To account for multiple testing across the five individual complication types, Holm-adjusted p values were calculated.

Overall satisfaction was analyzed as the primary patient-reported outcome using linear regression adjusted for follow-up duration, with inference based on HC3 robust standard errors. The seven domain-specific satisfaction items were analyzed as secondary exploratory outcomes using the same regression framework, with Holm adjustment applied across these seven tests. Although NRS scores are ordinal integer values, the 0–10 scale provides 11 ordered response levels and was treated as approximately continuous to estimate adjusted mean differences, which were considered clinically interpretable for between-group comparisons. Alternative ordinal modeling approaches, such as ordinal logistic regression, could also be considered. However, given the small sample size, sparse distribution of responses across NRS categories, and the primary aim of estimating adjusted mean differences in clinically interpretable score units, linear regression with HC3 robust standard errors was selected as the main analytical approach.

Statistical analyses were performed using IBM SPSS Statistics (version 27.0) and R (version 4.5.2) via RStudio. All tests were two-sided, and statistical significance was set at p < 0.05. All p values were reported to three decimal places, except for values below 0.001, which were reported as p < 0.001. No missing data were identified for the analyzed variables; therefore, no imputation or other missing-data procedures were applied.

Multiplicity was addressed by outcome family. Holm adjustment was applied separately to the five complication-specific tests and to the seven domain-specific satisfaction tests. Overall satisfaction was analyzed separately as the primary patient-reported outcome and was not included in the satisfaction-domain adjustment. MBL, mSBI, and mPI were analyzed as distinct clinical outcomes rather than as a single family of repeated tests; therefore, no Holm adjustment was applied across these outcomes. Given the exploratory nature of this small retrospective cohort, p values for secondary outcomes were interpreted cautiously, with emphasis placed on effect estimates and 95% confidence intervals rather than on dichotomous statistical significance.

Model diagnostics for linear regression were assessed using standardized residuals-versus-fitted plots, normal probability plots of residuals, standardized residuals, Cook’s distance, leverage values and collinearity diagnostics. The Shapiro-Wilk test was used as a supplementary assessment of residual normality rather than as the sole diagnostic criterion. HC3 robust standard errors were used where appropriate to support regression-based inference or sensitivity analyses in this small-sample setting. In determining model complexity, events-per-variable (EPV) considerations were explicitly taken into account for time-to-event analyses. Because the number of events was limited, particularly for individual complication outcomes, Cox models were restricted to treatment group as the sole explanatory variable and Firth penalization was used to reduce small-sample and sparse-event bias. For continuous or approximately continuous outcomes, including MBL, mSBI, mPI, and NRS-based satisfaction scores, EPV was not directly applicable because these analyses did not involve event counts. Nevertheless, given the limited number of patients, model complexity was intentionally restricted by fitting parsimonious models that included only treatment group and follow-up duration. Importantly, these approaches do not increase statistical power, overcome the limited sample size, or eliminate uncertainty due to sparse events; therefore, all model-based estimates were interpreted as exploratory rather than definitive.

Bias control

All consecutive patients treated during the study period were screened using predefined eligibility and follow-up criteria, and only patients for whom both treatment concepts were clinically feasible at baseline were included. Clinical care and follow-up were standardized through consistent surgical and prosthetic protocols, use of implants from the same manufacturer, and a common maintenance program.

Radiographic MBL measurements were performed on de-identified, coded radiographs by an examiner without access to clinical group information; radiographs were calibrated using the known implant length, and measurements were repeated after a 2-week interval to assess intra-examiner repeatability. Patient-reported satisfaction was collected using standardized, non-leading instructions. Patient-level analyses, adjustment for follow-up duration, and restricted model complexity were used to reduce analytical bias.

Results

Study population and baseline characteristics

The participant flowchart is shown in Fig. 1. A total of 19 patients were included and categorized according to the treatment concept: Locator-bilateral segmental bar concept (LBSB group) (Fig. 2A, B) and the conventional rigid bar concept (CRB group). Baseline demographic and clinical characteristics are summarized in Table 1.

Fig. 1.

Flow diagram of patient selection and reasons for exclusion. Insufficient follow-up was defined as a follow-up duration of less than 12 months after definitive prosthesis delivery or failure to complete the scheduled annual recall. LBSB, Locator-bilateral segmental bar concept; CRB, conventional rigid bar concept

Fig. 2.

Intraoral view of an implant-supported overdenture in the LBSB group. A Anterior Locator attachments and bilateral posterior segmental bars. B Tissue-surface view of the denture base. LBSB, Locator-bilateral segmental bar concept

Table 1.

Characteristics of included patients at study entry

Characteristic	Overall (n = 19)	LBSB (n = 10)	CRB (n = 9)
Male sex, n (%)	13 (68.42)	7 (70.00)	6 (66.67)
Smoker, n (%)	2 (10.53)	1 (10.00)	1 (11.11)
Age (years), mean ± SD	62.89 ± 8.67	64.30 ± 9.62	61.33 ± 7.75
Follow-up (years), mean ± SD	5.37 ± 3.22	5.10 ± 3.07	5.67 ± 3.54

Data are presented as mean ± SD for continuous variables and n (%) for categorical variables. All implants were from the same manufacturer (Straumann, Switzerland). No patient had bruxism, osteoporosis, uncontrolled diabetes, untreated periodontitis, or concomitant sinus floor elevation at baseline. All patients had an opposing complete-arch removable prosthesis

Mean age was 64.30 ± 9.62 years in the LBSB group and 61.33 ± 7.75 years in the CRB group; mean follow-up was 5.10 ± 3.07 and 5.67 ± 3.54 years, respectively. All implants were from the same manufacturer (Straumann, Switzerland). Smoking status appeared similar between groups; no patient had uncontrolled diabetes, osteoporosis, concomitant sinus floor elevation, untreated periodontitis, or bruxism, and all opposing prostheses were complete-arch removable prostheses. No missing data were identified for the variables included in the analyses.

Survival and success of implants and prostheses

A total of 93 implants were evaluated, including 36 in the CRB group and 57 in the LBSB group. Implant survival and success were 100% in both groups, with no implant loss during follow-up. Prosthesis survival was also 100%, as all definitive overdentures remained in clinical service at the last follow-up visit. Prosthesis success was 3/9 (33.3%) in the CRB group and 4/10 (40.0%) in the LBSB group (Table 2).

Table 2.

Implant- and prosthesis-level survival and success rates

Outcome	CRB	LBSB	Overall
Implant survival	36/36 (100%)	57/57 (100%)	93/93 (100%)
Implant success	36/36 (100%)	57/57 (100%)	93/93 (100%)
Prosthesis survival	9/9 (100%)	10/10 (100%)	19/19 (100%)
Prosthesis success	3/9 (33.3%)	4/10 (40.0%)	7/19 (36.8%)

Values are presented as the number of implants or prostheses meeting the specified criterion/the total number assessed (%)

Marginal bone loss

Intra-examiner reliability for repeated implant-level MBL measurements was excellent, with a single-measure ICC of 0.943 (95% CI, 0.915 to 0.962). Mean MBL was higher in the CRB group than in the LBSB group (1.530 ± 0.579 vs. 1.130 ± 0.457 mm). In the multivariable linear regression model adjusting for follow-up duration, longer follow-up was associated with greater MBL (B = 0.154 mm/year, 95% CI 0.120 to 0.187; p < 0.001; Fig. 3A), and the LBSB group was associated with lower MBL than the CRB group (B = − 0.310 mm, 95% CI − 0.522 to − 0.098; p = 0.007; Table 3; Fig. 3B). There was no evidence of a group-by-time interaction after adding an interaction term with follow-up duration mean-centered at 5.37 years (p = 0.430). Sensitivity analysis using HC3 robust standard errors yielded similar results (Additional file 2).

Fig. 3.

Adjusted mean marginal bone loss by group and follow-up duration. A Patient-level marginal bone loss plotted against follow-up duration. Each point represents one patient. Fitted linear trends are shown by group, with shaded areas indicating 95% confidence bands. B Adjusted mean marginal bone loss in the CRB group (n = 9) and LBSB group (n = 10), estimated from a linear regression model adjusted for follow-up duration. Error bars represent 95% confidence intervals. MBL, marginal bone loss; LBSB, Locator-bilateral segmental bar concept; CRB, conventional rigid bar concept

Table 3.

Multivariable linear regression analyses for marginal bone loss (MBL, mm)

Predictor	B	95% CI	p
Follow-up duration (years)	0.154	0.120 to 0.187	< 0.001
Group (LBSB vs. CRB)	−0.310	−0.522 to − 0.098	0.007

With CRB coded as 0 and LBSB as 1, the group coefficient (B) corresponds to the adjusted mean difference (LBSB − CRB)

 CI Confidence interval, B Unstandardized regression coefficient

Peri-implant bleeding and plaque indices

In multivariable linear regression models adjusting for follow-up duration, group assignment was not associated with either mSBI (B = − 0.037, 95% CI − 0.133 to 0.060; p = 0.429) or mPI (B = − 0.020, 95% CI − 0.149 to 0.109; p = 0.749; Table 4; Fig. 4A, B). Longer follow-up duration was associated with higher mSBI and higher mPI (Fig. 4A, B). Sensitivity analyses using HC3 robust standard errors yielded similar conclusions (Additional file 3).

Table 4.

Multivariable linear regression analyses for peri-implant soft-tissue indices

Outcome	Group (LBSB vs. CRB)			Follow-up duration (years)
	B	95% CI	p	B	95% CI	p
mSBI	−0.037	−0.133 to 0.060	0.429	0.052	0.036 to 0.067	< 0.001
mPI	−0.020	−0.149 to 0.109	0.749	0.065	0.045 to 0.086	< 0.001

With CRB coded as 0 and LBSB as 1, the group coefficient (B) corresponds to the adjusted mean difference (LBSB − CRB)

CI Confidence interval, B Unstandardized regression coefficient, mSBI Modified sulcus bleeding index, mPI Modified plaque index

Fig. 4.

Peri-implant bleeding and plaque indices by group and follow-up duration. A Modified sulcus bleeding index and (B) modified plaque index plotted against follow-up duration in the CRB group (n = 9) and LBSB group (n = 10). Each point represents one patient. Lines represent fitted linear trends by group, and shaded areas indicate 95% confidence bands. mSBI, modified sulcus bleeding index; mPI, modified plaque index; LBSB, Locator-bilateral segmental bar concept; CRB, conventional rigid bar concept

Overall and individual complications

Over the follow-up period, the composite endpoint (any complication) occurred in 6/9 patients in the CRB group and 6/10 patients in the LBSB group (Table 5). Kaplan-Meier analysis and the Firth-penalized Cox proportional hazards model showed no significant between-group difference in the risk of any complication (log-rank p = 0.860; Table 5; Fig. 5). Consistently, the Firth-penalized Cox model yielded an HR of 0.899 (95% CI, 0.283 to 2.965; p = 0.855; Table 5).

Table 5.

Time-to-first complication analyses comparing CRB and LBSB groups

Outcome	Events CRB (n/N)	Events LBSB (n/N)	Exact log-rank p	Holm-adjusted p (log-rank)	Firth Cox HR (95% CI)*	p (Firth)	Holm-adjusted p (Firth)
Any complication	6/9	6/10	0.860†	-	0.899 (0.283 to 2.965)	0.855	-
Hyperplasia	3/9	3/10	0.968	1.000	1.030 (0.212 to 4.970)	0.971	1.000
Sore spot	5/9	1/10	0.182	0.547	0.346 (0.034 to 1.907)	0.232	0.697
Base fracture	0/9	0/10	NA	NA	NA	NA	NA
Tooth fracture	1/9	1/10	0.650	1.000	0.625 (0.051 to 7.695)	0.685	1.000
Relining	6/9	1/10	< 0.001	< 0.001	0.056 (< 0.001 to 0.483)	0.005	0.019

*Firth-penalized Cox proportional hazards model. †The composite endpoint was compared using the conventional log-rank test; for individual complications, exact log-rank tests were applied. NA, no events occurred in either group; therefore, between-group comparison and hazard ratio estimation were not feasible.

Fig. 5.

Kaplan-Meier curves for time-to-first occurrence of any complication. Curves show complication-free survival for the CRB group (n = 9) and LBSB group (n = 10). Time was calculated from definitive prosthesis delivery to the first occurrence of any complication. Patients without complications were censored at their last follow-up visit and are indicated by “+”. The p value was calculated using the log-rank test. Numbers at risk are shown below the plot. LBSB, Locator-bilateral segmental bar concept; CRB, conventional rigid bar concept

When complications were analyzed individually, relining was more frequent and occurred earlier in the CRB group. Consistently, both Kaplan-Meier analysis and the Firth-penalized Cox model suggested a lower hazard of relining in the LBSB group compared with the CRB group (exact log-rank p < 0.001; HR = 0.056, 95% CI < 0.001 to 0.483; p = 0.005; Table 5 and Additional file 4), and the finding remained significant after Holm adjustment. No statistically significant between-group differences were observed for hyperplasia or tooth fracture. For sore spot, although the between-group comparison did not reach statistical significance, events were numerically more common in the CRB group (5/9) than in the LBSB group (1/10). No base fractures were observed in either group.

NRS-based patient-reported satisfaction

Mean overall satisfaction was 8.300 ± 0.640 in the LBSB group and 7.110 ± 0.737 in the CRB group at the final follow-up visit (Table 6). Adjusted mean difference was estimated using linear regression controlling for follow-up duration, with inference based on HC3 robust standard errors. In the adjusted model, overall satisfaction scores were higher in the LBSB group than in the CRB group (p = 0.031; Table 6; Fig. 6). For the seven domain-specific satisfaction outcomes, adjusted model estimates suggested higher scores for masticatory function and stability in the LBSB group (Fig. 7). These associations remained statistically significant after Holm adjustment (p = 0.019 and p = 0.030, respectively). No statistically significant between-group differences were observed for esthetics, foreign body sensation, pronunciation, ease of handling the dentures, or ease of oral hygiene after Holm adjustment (Fig. 7).

Table 6.

Multivariable linear regression results for overall satisfaction using HC3 robust standard errors

Outcome	Mean ± SD (CRB)	Mean ± SD (LBSB)	Adjusted mean difference (LBSB − CRB)*	95% CI*	p*
Overall satisfaction	7.110 ± 0.737	8.300 ± 0.640	0.983	0.103 to 1.864	0.031

*Adjusted for follow-up duration using multivariable linear regression. Positive values indicate higher scores in the LBSB group than in the CRB group. Confidence intervals and p values were based on HC3 robust standard errors.

Fig. 6.

Adjusted mean overall satisfaction by group and follow-up duration. Adjusted mean overall satisfaction at the final follow-up visit in the CRB group (n = 9) and LBSB group (n = 10). Estimates were obtained from a linear regression model adjusted for follow-up duration. Error bars represent 95% confidence intervals based on HC3 robust standard errors. Satisfaction was assessed using an 11-point numeric rating scale from 0 to 10. LBSB, Locator-bilateral segmental bar concept; CRB, conventional rigid bar concept; NRS, numeric rating scale

Fig. 7.

Adjusted group effects for seven domain-specific satisfaction outcomes. Forest plot showing adjusted mean differences between the LBSB group (n = 10) and CRB group (n = 9), calculated as LBSB minus CRB. Estimates were obtained from linear regression models adjusted for follow-up duration (years). Error bars represent 95% confidence intervals based on HC3 robust standard errors. The p values were adjusted using the Holm method across the seven domain-specific outcomes. Positive values indicate higher satisfaction scores in the LBSB group. LBSB, Locator-bilateral segmental bar concept; CRB, conventional rigid bar concept; NRS, numeric rating scale

Regression diagnostics

Regression diagnostics did not suggest major violations of model assumptions. Residual plots showed no clear non-linearity or marked heteroscedasticity, and no severe outliers or dominant influential observations were identified. Collinearity diagnostics indicated no relevant multicollinearity.

Discussion

Implant-supported overdentures (ISODs) are widely used for the rehabilitation of maxillary edentulism; however, treatment planning remains challenging because it requires a balance among restorative space, retention, hygiene accessibility, and long-term maintenance [11, 19, 20]. In this retrospective cohort, we compared a Locator-bilateral segmental bar concept with a conventional rigid bar concept for maxillary implant-supported overdentures. Both concepts showed 100% implant and prosthesis survival, while the LBSB concept was associated with lower marginal bone loss, higher NRS-based patient satisfaction, and a lower estimated hazard of relining. However, these findings should be interpreted cautiously because of the small sample size, non-randomized allocation, and concept-level differences between groups.

Recent consensus guidelines recognize both unsplinted stud attachments and bar-splinted designs as treatment options for maxillary implant-supported overdentures, emphasizing that prosthetic design should be individualized according to implant position, prosthetic space, retention, stability, and maintenance requirements [21–23].

From biomechanical and maintenance perspectives, different attachment systems may influence overdenture performance by affecting splinting, prosthetic support, rotational movement, and wear of retentive components [11, 24]. Conventional full-arch rigid bars may provide cross-arch splinting and favorable stability, but require adequate restorative space and may complicate hygiene procedures and prosthetic maintenance [23, 25]. In contrast, solitary Locator attachments are low-profile and relatively easy to clean; however, their retentive inserts may wear over time and require replacement, and posterior support may be limited when implants are not splinted [24, 26]. Based on these considerations, the LBSB concept was developed to combine anterior low-profile Locator retention with bilateral posterior segmental bar support, aiming to reduce posterior denture rotation and improve load sharing without creating a continuous full-arch bar. Together, these features may partly explain the lower MBL, higher perceived stability, and lower estimated hazard of relining observed in the LBSB group; however, this interpretation remains hypothesis-generating because no direct biomechanical assessment was performed.

Both groups showed 100% implant and prosthesis survival, which was broadly consistent with the favorable survival profile reported in previous studies of maxillary implant-supported overdentures. In long-term randomized trials by Slot et al. evaluating maxillary bar overdentures supported by four or six implants, 10-year implant survival ranged from 96.1% to 100% [13, 27]. Onclin et al. evaluated implant and overdenture survival using Kaplan-Meier curves and log-rank tests in a 5-year randomized trial; the four-implant bar-retained group showed 96.3% implant survival and 95.0% overdenture survival [28]. In the present cohort, no implant or prosthesis loss occurred; therefore, formal Kaplan-Meier or Cox modeling for these endpoints would not yield informative between-group estimates.

Peri-implant soft-tissue indices (mSBI and mPI) were low and did not differ meaningfully between groups after adjustment for follow-up duration, suggesting that both treatment concepts were associated with similarly acceptable peri-implant conditions under standardized supportive care [19, 20]. One possible explanation is that both concepts incorporated bar designs with soft-tissue clearance, which may have facilitated oral hygiene procedures and contributed to plaque control [25]. In addition, plaque and bleeding indices are influenced by patients’ oral hygiene and maintenance protocols [18]; therefore, the absence of between-group differences may also reflect the structured recall and repeated hygiene reinforcement implemented in this cohort. This is consistent with evidence indicating that prosthesis contour, cleaning access, and ongoing maintenance may play an important role in peri-implant soft-tissue outcomes for removable implant prostheses [11, 25].

Marginal bone loss (MBL) was limited in both groups, with mean values of 1.130 ± 0.457 mm in the LBSB group and 1.530 ± 0.579 mm in the CRB group. In a 5-year randomized trial of maxillary four-implant overdentures retained by bars or solitary attachments, Onclin et al. used standardized longitudinal radiographic assessment and reported mean 5-year marginal bone-level changes of − 0.99 ± 0.96 mm and − 1.41 ± 1.38 mm, respectively [28]; the values observed in the present cohort were broadly comparable in magnitude. Long-term randomized trials by Slot et al. also indicate that bar-supported maxillary overdentures are associated with limited marginal bone loss over time [13, 27]. However, in the present study, the LBSB concept was associated with lower MBL than the CRB concept. This finding may be related to the overall configuration of the LBSB concept, which is intended to promote a more favorable biomechanical load distribution and reduce denture rotation. These features might have promoted more favorable load transfer to the implants and peri-implant bone, thereby providing a possible explanation for the lower peri-implant bone resorption observed in the LBSB group [24]. Nevertheless, because no direct biomechanical assessment was performed, this explanation remains hypothetical.

For patient-reported satisfaction at the final follow-up visit, the LBSB group showed higher NRS-based scores for masticatory function, stability, and overall satisfaction.

These findings may be partly attributable to the broader anteroposterior implant distribution and anterior Locator attachments of the LBSB concept, which could have improved perceived prosthetic support and stability [29]. However, this interpretation remains speculative, as individual design components could not be evaluated separately.

Notably, foreign body sensation did not differ significantly between the LBSB and CRB groups, suggesting that the additional implants and anterior stud attachments used in the LBSB concept might not be associated with greater oral discomfort. This observation is broadly consistent with previous clinical evidence indicating that patient-perceived oral function may be influenced more by prosthesis design features such as palatal coverage than by the number of attachments [30]. The NRS provides a clinically intuitive measure of perceived prosthesis-related satisfaction in familiar 0–10 score units, but it cannot replace validated multidimensional PROMs for assessing broader oral health-related quality of life. Thus, these findings should be interpreted as differences in NRS-based satisfaction rather than as validated multidimensional PROMs or OHRQoL outcomes. Because satisfaction was assessed only once at the final follow-up visit, the resulting scores reflect cross-sectional ratings of current satisfaction rather than longitudinal changes. Confirmation in prospective studies using validated PROMs is warranted.

For complication outcomes, implant and prosthesis survival were 100% in both groups, whereas prosthesis success was substantially lower, indicating that implant-supported overdentures may still require repeated clinical or laboratory intervention. The composite endpoint of any complication did not differ between groups, suggesting that the overall burden of clinically recorded maintenance events was comparable between the two treatment concepts.

Previous evidence suggests that attachment design may influence the occurrence and pattern of complications in maxillary implant-supported overdentures. A 5-year randomized trial reported lower complication rates with bar-retained maxillary overdentures than with solitary attachments [28], and recent retrospective evidence similarly suggests that unsplinted attachment systems are associated with higher overall prosthetic complication rates, particularly deformation of retentive elements [26]. The LBSB concept, however, was designed based on a hybrid biomechanical rationale that combines anterior Locator retention with posterior bar support. Therefore, the lower estimated hazard of relining in the LBSB group may reflect improved posterior support and reduced tissue-supported denture movement. This interpretation is also consistent with the numerically lower frequency of sore spots in the LBSB group, although this difference did not reach statistical significance. Nevertheless, the relining finding should be regarded as exploratory because only seven relining events were observed and the estimate remained imprecise.

Several limitations should be acknowledged. First, this was a single-center retrospective study with a small sample size, which limited statistical power, precision of the estimates, and external validity. Although consecutive eligible patients were included and parsimonious models with small-sample corrections were used, sparse events for some complication outcomes resulted in wide confidence intervals and limited the feasibility of more fully adjusted analyses. In addition, all patients were treated in a specialist-care setting, received implants from the same manufacturer, and participated in a structured maintenance program; thus, the findings may not be directly generalizable to broader clinical settings, other implant systems, or patients with irregular follow-up.

Complications managed outside the treating center or not reported by patients may also have been missed. Furthermore, although propensity score methods could improve between-group comparability in larger retrospective cohorts with more comprehensive baseline covariate data, the small sample size and limited covariate information in the present study made a robust propensity-based analysis unreliable. Therefore, the findings should be interpreted as exploratory associations rather than causal effects.

Second, treatment allocation was non-randomized and partly guided by patient preference following clinical assessment, with no systematic documentation of the reasons for patients’ choices. Therefore, selection bias, residual confounding, and unmeasured baseline factors cannot be excluded. Moreover, the two groups differed simultaneously in implant number, implant distribution, and retention design; consequently, the observed between-group differences should be interpreted as concept-level associations rather than effects attributable to any individual component. The proposed biomechanical explanations remain hypothetical because no direct biomechanical assessment was performed.

Third, despite the use of calibrated periapical radiographs and a standardized long-cone paralleling protocol, the absence of individualized positioning stents or bite-registration devices may have introduced minor projection-related measurement bias. Moreover, bone-level data are inherently time-dependent and may be influenced by multiple covariates; therefore, future studies with larger samples and standardized serial radiographs should apply more comprehensive longitudinal or time-dependent methods for MBL analysis. In addition, patient satisfaction was assessed using a 0–10 NRS at the final follow-up visit, rather than through repeated assessments with validated multidimensional PROMs or OHRQoL instruments. This single-time-point assessment may have introduced recall bias, as patients’ ratings could have been influenced by their recollection of previous prosthetic problems, maintenance events, or recent clinical experiences. Response bias also cannot be excluded, as satisfaction scores were self-reported during clinical follow-up and may have been influenced by social desirability, the clinical assessment setting, or patients’ interactions with the treating team. Overall, these findings should be interpreted as cross-sectional NRS-based satisfaction data rather than validated multidimensional patient-reported outcomes.

Conclusions

Within the limitations of this small non-randomized retrospective cohort, the LBSB concept was associated with lower MBL, an estimated lower hazard of relining, and higher NRS-based satisfaction compared with the CRB concept. Because implant number, implant distribution, and retention design differed simultaneously between concepts, these findings should be interpreted as exploratory, hypothesis-generating concept-level associations rather than evidence of clinical superiority or causal effects of any individual component. The generalizability of the findings is limited to similarly selected specialist-care populations receiving structured maintenance.

Adequately powered prospective studies, ideally incorporating standardized longitudinal radiographic assessment, validated patient-reported outcome measures, and direct biomechanical evaluation, are needed to confirm these observations.

Abbreviations

CRB

Conventional rigid bar

CI

Confidence interval

HC3

Heteroscedasticity-consistent robust standard errors

HR

Hazard ratio

ICC

Intraclass correlation coefficient

ISOD

Implant-supported overdenture

LBSB

Locator-bilateral segmental bar concept

MBL

Marginal bone loss

mPI

Modified plaque index

mSBI

Modified sulcus bleeding index

NRS

Numeric rating scale

SD

Standard deviation

STROBE

Strengthening the Reporting of Observational Studies in Epidemiology

Authors' contributions

JZ, MC and JF contributed to the study conception and design. JZ collected the data and performed the statistical analyses. MC assisted with data interpretation and methodological design. JF supervised the study and provided critical revision of the manuscript. JZ and JF drafted the manuscript. All authors read and approved the final manuscript.

Funding

Not applicable.

Data availability

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

This retrospective study was approved by the Ethics Committee of Beijing Stomatological Hospital, Capital Medical University (Approval No. A-2025-02). Written informed consent was obtained from all patients.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.