Tooth survival following pulpectomy in primary molars with apical periodontitis and preoperative factors associated with failure: a retrospective cohort study Authors: Sirirak Satimanon, Veeritta Yimcharoen, Papimon Chompu-inwai, Natchalee Srimaneekarn, Sawanya Prutthithaworn Published: 2026 ABSTRACT BackgroundLong-term survival data for pulpectomy in primary molars with apical periodontitis (AP) remain limited. This retrospective cohort study aimed to evaluate the tooth survival following pulpectomy in primary molars with AP and to identify factors associated with failure.MethodsDental records of healthy children aged 3–6 years who underwent pulpectomy in at least one primary molar between 2019 and 2021 at Mahidol University were reviewed. Teeth were included when preoperative radiographs showed findings consistent with apical periodontitis. Failure was defined as the presence of clinical symptoms or radiographic progression involving displacement of the succedaneous tooth bud. Tooth survival was estimated using the Kaplan–Meier method, and restricted mean survival time (RMST) was calculated using a 48-month truncation time. Factors associated with failure were analysed using multivariable Cox proportional hazards regression with robust cluster variance estimation. Sensitivity analysis was performed using an inverse probability of censoring weighting-adjusted Fine–Gray model to account for documented competing events and potential informative censoring.ResultsA total of 326 primary molars from 242 children were included. Over the observation period, 47 molars (14.4%) failed. The Kaplan–Meier estimated survival probability was 100% at 12 months, 95.4% at 24 months, and 52.02% at 48 months. The RMST at 48 months was 42.77 months. In the primary multivariable Cox model, older age at treatment (HR 1.05, 95% CI 1.01–1.09), preoperative sinus tract (HR 4.98, 95% CI 1.99–12.45), and root resorption involving one-third to two-thirds of the root length (HR 2.96, 95% CI 1.01–8.69) were associated with higher failure hazard. Sensitivity analysis supported the robustness of the association with preoperative sinus tract, whereas the association with age was attenuated and the association with advanced root resorption was less stable. Treatment-level variables, including visit regimen, irrigant type, obturation material, obturation quality, and operator training level, were not associated with failure.ConclusionsWithin the limitations of this retrospective single-centre study and the adapted retention-focused outcome framework, pulpectomy in primary molars with AP showed a 48-month Kaplan–Meier survival probability of 52.02% and an RMST of 42.77 months. Preoperative sinus tract showed the most consistent prognostic association with failure, while age remained directionally consistent but less precise, and advanced root resorption should be interpreted cautiously. These prognostic findings should be considered hypothesis-generating and warrant confirmation in larger prospective cohorts.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12903-026-08617-8. Introduction The two most common reasons for the child’s first dental visit were caries and tooth pain [1]. Pulpectomy is the recommended treatment for primary teeth with irreversible pulpitis when their retention provides functional benefits [2]. The prognosis of pulpectomy is influenced by multiple interrelated factors, including preoperative radiolucency [3–10], root resorption [3, 5, 9], and the child’s age at the time of treatment [6, 8, 9]. Severe early childhood caries (S-ECC) is a global oral health problem [11]. In S-ECC, bacterial complexity and interspecies interactions increase with disease progression [12], raising the risk of rapid pulpal and periradicular involvement and the development of apical periodontitis (AP) lesions [13]. AP is an inflammatory response to bacteria and their by-products within the root canal system [14]. The bacteria in AP are predominantly multispecies anaerobic communities organised into thick, multilayered biofilm structures adhering to the dentinal walls [15, 16]. Thorough debridement and disinfection are required to eliminate established infections [14–17]. In paediatric dentistry, pulpectomy commonly uses simplified chemo-mechanical protocols that are generally less intensive than those used in permanent-tooth endodontics and may not completely eliminate persistent biofilm within complex root canal anatomy. Additionally, a more pronounced hazard ratio was observed in the presence of periapical/furcation radiolucency compared to widening of the periodontal ligament space [5, 9], implying that the risk of pulpectomy failure may increase with the extent of the lesion. Nevertheless, the relationship between the severity of periradicular lesions and long-term pulpectomy outcomes remains insufficiently defined. Pulpectomy outcomes in primary teeth have been reported to be time-dependent, with survival generally declining over follow-up periods [3–10]. However, long-term survival data specifically for primary molars remain limited. Several reports have combined anterior and posterior teeth in their analyses [4, 5, 9], thereby limiting the applicability of their findings to molar-specific clinical decision-making. Primary molars possess complex anatomy that differs substantially from that of incisors, including multiple canals and accessory anatomy [18], which may influence treatment efficacy. Moreover, long-term pulpectomy success is more significant in primary molars than in incisors because their premature loss adversely affects space maintenance [19]. Molar-specific survival data are therefore essential for accurate clinical guidance and prognostic expectations. Additionally, the outcome assessment criteria varied across studies [3–10]. Based on the health and disease concepts proposed by Kvist and Hofmann (2023) [20], these endpoints can be categorised into the “ideal model” [3, 4, 6–10] and the “praxis concept” [5]. The ideal model is biologically oriented (curative-focused), prioritising radiographic healing and biological success, whereas the praxis concept is clinically oriented (retention-focused), viewing health and disease as a spectrum [20]. Although the ideal model is important for confirming treatment efficacy, it does not consistently translate into clinical decision-making. Clinical decisions are fundamentally guided by symptom control, functional outcomes, child cooperation, and exfoliation timing. In the context of primary dentition, preserving teeth maintains arch integrity, function, and sufficient space for the eruption of succedaneous teeth [19]. In the present study, an adapted retention-focused endpoint was used to reflect clinical decision-making in primary dentition, where symptom control, function, and successor safety may influence management alongside radiographic findings. This endpoint should not be interpreted as directly equivalent to the biologically oriented success criteria used in many previous pulpectomy studies. Hence, this retrospective cohort study focused on primary molars with radiographic evidence of apical periodontitis and aimed to evaluate tooth survival following pulpectomy and to identify factors associated with failure. Methods Ethical approval and sample size estimation This retrospective cohort study was approved by the Institutional Review Board, Faculty of Dentistry and Faculty of Pharmacy, Mahidol University, Bangkok, Thailand (CoA.No.MU-DT/PY-IRB2023/080.2612). The infinite population proportion was used to estimate the association between factors and the survival outcome with a 5% margin of error and a significance level of 0.05, and, considering the 74.5% success rate of pulpectomy reported by Coll J.A. and Sadrian R. [3], a total sample of 289 teeth was required. This study was reported according to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines [21]. Study population and eligibility criteria Dental records of children aged 3 to 6 years who received pulpectomy in at least one primary molar between 2019 and 2021 at the Paediatric Dental Clinics, Mahidol University, Bangkok, Thailand, were reviewed. Eligible cases included healthy children with no known medical or developmental disorders whose primary molars were affected by caries. Teeth were included when preoperative radiographs demonstrated findings consistent with AP, and complete dental records were available for all variables of interest. Preoperative radiographic findings were classified according to the extent of the radiographic lesion as loss of lamina dura, furcation radiolucency, or periradicular radiolucency to reflect the degree of periradicular inflammatory involvement and corresponding bacterial burden [15, 22]. Representative radiographic examples of these preoperative categories are provided in Supplementary Figure S1. The variables were categorised as tooth-level variables (child age at time of treatment and tooth type), preoperative disease-burden variables (tooth mobility, soft tissue findings, extent of root resorption, and radiographic findings), and treatment-level variables (visit regimen, irrigant type, obturation material, and operator training level). Inclusion also required documented clinical signs and symptoms at follow-up visits, high-quality immediate postoperative and follow-up radiographs, and a minimum mandatory clinical and radiographic follow-up period of six months. Teeth were excluded if the dental record was illegible, a non-adhesive interim material was retained for longer than six months, or leakage or failure of the definitive coronal restoration was documented. Pulpectomy treatment approach All pulpectomy procedures and follow-up were performed in a teaching clinic setting under faculty supervision. In our practice, systemic antibiotics for pulpal- and periapical-related conditions are prescribed in accordance with current guidelines [23, 24]. The clinical procedures followed the protocol recommended in the annually updated guidelines and evidence-based practice published between 2019 and 2021 [25–27]. Treatments were carried out under rubber dam isolation by both postgraduate and undergraduate dental students. Working length was established using a combined radiographic–tactile technique. Canal preparation was performed with hand stainless-steel K-files, with enlargement to a master apical file size of ISO 30–35. Canal irrigation was conducted using either 1% sodium hypochlorite (NaOCl) or normal saline solution (NSS). Canals were obturated with calcium hydroxide–iodoform paste (Vitapex®) or zinc oxide eugenol (ZOE), and intermediate restorative material (IRM®) was subsequently placed over the obturation material. Treatment was completed using either a single-visit or two-visit regimen. In cases managed over two visits, a non-setting calcium hydroxide paste (Ultracal®) was placed as the inter-appointment intracanal medicament. All treated teeth subsequently received definitive stainless steel crowns (SSCs). The treated teeth underwent scheduled clinical and radiographic evaluations during routine periodic recall visits according to the patient’s caries risk [28]. Outcome assessment The failure criteria were modified from Songvejkasem et al. [5]. The presence of clinical symptoms (pain, swelling, abscess, or fistula) or radiographic progression with displacement of the succedaneous tooth bud relative to previous radiographs was defined as the failure event. Time-to-event (T) was calculated from the date of pulpectomy completion to the date on which the failure event was observed. For censored cases, follow-up ended on the last documented date the tooth was present in the dental record. Teeth documented as exfoliated without evidence of pulpectomy failure, teeth extracted for reasons unrelated to pulpectomy failure, and teeth lost to follow-up were censored at the date of the last available assessment in the primary survival analysis. Loss to follow-up or unknown outcome status was defined as the absence of subsequent clinical or radiographic records after the last documented assessment, with no documented failure, exfoliation, non-failure extraction, or completion of the 5-year observation period. Teeth present throughout the observation period were censored at the 5-year follow-up. Examiner calibration Calibration was performed for radiographic assessment of the level of root resorption, the extent of periradicular lesions, the quality of root canal obturation, and the position and follicular status of the successor tooth. A 10% of the sample was randomly selected for inter- and intra-examiner calibration among two board-certified paediatric dentists (faculty members with 11 and 10 years of experience, respectively) and a third-year paediatric dental resident. Radiographs were independently evaluated by the three examiners, with disagreements resolved through consensus discussion. A second assessment was performed after a two-week interval to determine intra-examiner consistency. Dental records were reviewed by a third-year paediatric dental resident and re-evaluated after a two-week interval to assess intra-examiner reliability. Agreement was evaluated using Cohen’s kappa statistic. Inter- and intra-examiner agreement demonstrated substantial to almost perfect reliability, with kappa values ranging from 0.720 to 0.901 and 0.754 to 0.936, respectively, across all calibration topics. Chart and radiographic review The calibrated third-year paediatric dental resident reviewed preoperative radiographs and longitudinal dental records. Immediate postoperative and follow-up radiographs obtained over the 5-year observation period were independently evaluated by the two board-certified paediatric dentistry faculty assessors who were blinded to clinical information and baseline radiographs. Statistical analysis All data were analysed using IBM® SPSS® (Version 30.0; IBM Corp., Armonk, NY, USA) and Stata/SE (Version 17.0; StataCorp LLC, College Station, TX, USA). Descriptive statistics were computed for all variables. Categorical variables were summarised as frequencies and percentages, and follow-up duration was summarised as median and interquartile range. Tooth survival was estimated using the Kaplan–Meier method, and restricted mean survival time (RMST) was calculated at 48 months using a 48-month truncation time because the number of teeth remaining at risk was limited beyond this point. Restricted mean time lost (RMTL) was calculated as 48 months minus the 48-month RMST. Factors influencing survival outcomes were analysed using Cox proportional regression. Variables with a univariate p-value < 0.10 were considered for the final multivariable model. Robust cluster variance estimation using the Huber–White sandwich estimator was applied to account for multiple teeth within the same child. Child age at treatment was analysed as a continuous variable. The final model included 47 failure events and five model parameters, yielding an events-per-parameter ratio of approximately 9.4. Model stability was assessed using bootstrap 95% confidence intervals (CIs). The proportional hazards assumption was assessed using Schoenfeld residual-based tests and log-minus-log plots. Sensitivity analysis was performed using an inverse probability of censoring weighting (IPCW)-adjusted Fine–Gray model, with documented non-failure extractions treated as competing events. Results were reported as hazard ratios (HRs) or subdistribution hazard ratios (sHRs) with 95% CIs. Statistical significance for all tests was set at p < 0.05. A step-by-step description of the analytical workflow is provided in Supplementary Table S1. Results A total of 326 primary molars from 242 patient records were included in the final analysis (Fig. 1). Among the 242 children, 184 contributed 1 primary molar, 38 contributed 2, 15 contributed 3, 4 contributed 4, and 1 contributed 5. The mean age at pulpectomy treatment was 4.94 ± 0.74 years. Out of 326 treated primary molars, 47 (14.4%) experienced failure, while 279 (85.6%) were censored. Censored teeth were categorised as follows: 4 teeth that remained event-free at study end (administrative censoring; 1.43%), 20 teeth extracted for reasons unrelated to pulpectomy failure (competing events; 7.17%), and 255 teeth lost to follow-up or with unknown status before failure (true censoring; 91.40%). Follow-up duration by outcome status and censoring reason is summarised in Supplementary Table S2.Fig. 1Flowchart of the study Flowchart of the study Baseline characteristics of the study sample Baseline characteristics and factors associated with failure are summarised in Table 1. Mandibular teeth were more frequently affected than maxillary teeth, with mandibular second molars representing the largest proportion (42.64%). Most teeth showed no preoperative mobility (90.80%) and a normal soft-tissue appearance (86.91%) at baseline. Regarding treatment characteristics, all teeth were treated in a teaching clinic, with 94.48% performed by postgraduate students and 5.52% by undergraduate students. Two-visit pulpectomy was more frequently performed (57.98%) than single-visit treatment (42.02%). The predominant irrigant and obturation material were 1% sodium hypochlorite (90.80%) and Vitapex® (89.88%), respectively. In this cohort, no systemic antibiotic prescriptions were recorded.Table 1Cox regression analysis of factors associated with pulpectomy failureFactorTotalN (%)Failuresn (%)Univariate Cox regressionaMultivariate Cox regressionbHR (95% CI)p-valuecHR (95% CI)p-valuecTooth-level variables Child age at time of treatment, yearsd326 (100)47 (14.42)1.98 (1.29,3.05)0.0021.05 (1.01,1.09)0.009 3 – < 4 years55 (16.87)2 (3.63)1 4 – < 5 years114 (34.97)18 (15.79)5.17 (1.21,22.16)0.027 5–6 years157 (48.16)27 (17.20)6.67 (1.63,27.29)0.008 Tooth type Maxillary first molar58 (17.79)11(18.97)1.53 (0.72,3.20)0.269 Maxillary second molar51 (15.64)5 (9.80)0.61 (0.23,1.59)0.314 Mandibulary first molar78 (23.93)11 (14.10)0.92 (0.44,1.91)0.824 Mandibulary second molar139 (42.64)20 (14.39)1Preoperative disease-burden variables Preoperative tooth mobility Absence296 (90.80)43 (14.53)1 Presence30 (9.20)4 (13.33)0.73 (0.28,1.88)0.510 Preoperative soft tissue appearance Normal283 (86.91)36 (12.72)11 Gingival or vestibular Swelling35 (10.74)7 (20.00)1.52 (0.67,3.4)0.3161.81 (0.84,3.92)0.133 Sinus tract8 (2.45)4 (50.00)5.86 (2.07,16.59) < 0.0014.98 (1.99,12.45)0.001 Preoperative root resorption No resorption120 (36.81)11 (9.17)11 < 1/3 of root length179 (54.91)29 (16.20)1.80 (0.89,3.66)0.0991.39 (0.66,2.94)0.385 1/3–2/3 of root length27 (8.28)7 (25.93)3.70 (1.31,10.51)0.0142.96 (1.01,8.69)0.048 Preoperative radiographic findings Loss of lamina dura172 (52.76)20 (11.63)1 Furcation radiolucency85 (26.07)14 (16.47)1.29 (0.63,2.57)0.464 Periradicular radiolucency69 (21.17)13 (18.84)1.43 (0.75,2.75)0.280Treatment-level variables Treatment visit regimen Single-visit137 (42.02)21 (15.33)1 Two-visit189 (57.98)26 (13.76)1.13 (0.63,2.03)0.681 Irrigant type NSS30 (9.20)5 (16.67)1 1% NaOCl296 (90.80)42 (14.19)0.83 (0.33,2.09)0.689 Obturation material Vitapex®293 (89.88)44 (15.02)1 ZOE33 (10.12)3 (9.09)0.51 (0.15,1.75)0.287Operator training level Postgraduate student308 (94.48)42 (13.64)1 Undergraduate student18 (5.52)5 (27.78)2.09 (0.84,5.25)0.115Abbreviations: HR hazard ratio, 95% CI 95% confidence interval, NaOCl sodium hypochlorite, NSS normal saline solution, ZOE zinc oxide eugenolaUnivariate screening of factorsbMultivariable Cox proportional hazards regression with robust cluster variance estimation (Huber–White sandwich estimator) was applied to variables with p < 0.10 in the univariate analysiscBold p-values indicate statistical significance (p < 0.05)dChild age at the time of treatment was analysed as a continuous variable in the multivariate model Table 1Cox regression analysis of factors associated with pulpectomy failureFactorTotalN (%)Failuresn (%)Univariate Cox regressionaMultivariate Cox regressionbHR (95% CI)p-valuecHR (95% CI)p-valuecTooth-level variables Child age at time of treatment, yearsd326 (100)47 (14.42)1.98 (1.29,3.05)0.0021.05 (1.01,1.09)0.009 3 – < 4 years55 (16.87)2 (3.63)1 4 – < 5 years114 (34.97)18 (15.79)5.17 (1.21,22.16)0.027 5–6 years157 (48.16)27 (17.20)6.67 (1.63,27.29)0.008 Tooth type Maxillary first molar58 (17.79)11(18.97)1.53 (0.72,3.20)0.269 Maxillary second molar51 (15.64)5 (9.80)0.61 (0.23,1.59)0.314 Mandibulary first molar78 (23.93)11 (14.10)0.92 (0.44,1.91)0.824 Mandibulary second molar139 (42.64)20 (14.39)1Preoperative disease-burden variables Preoperative tooth mobility Absence296 (90.80)43 (14.53)1 Presence30 (9.20)4 (13.33)0.73 (0.28,1.88)0.510 Preoperative soft tissue appearance Normal283 (86.91)36 (12.72)11 Gingival or vestibular Swelling35 (10.74)7 (20.00)1.52 (0.67,3.4)0.3161.81 (0.84,3.92)0.133 Sinus tract8 (2.45)4 (50.00)5.86 (2.07,16.59) < 0.0014.98 (1.99,12.45)0.001 Preoperative root resorption No resorption120 (36.81)11 (9.17)11 < 1/3 of root length179 (54.91)29 (16.20)1.80 (0.89,3.66)0.0991.39 (0.66,2.94)0.385 1/3–2/3 of root length27 (8.28)7 (25.93)3.70 (1.31,10.51)0.0142.96 (1.01,8.69)0.048 Preoperative radiographic findings Loss of lamina dura172 (52.76)20 (11.63)1 Furcation radiolucency85 (26.07)14 (16.47)1.29 (0.63,2.57)0.464 Periradicular radiolucency69 (21.17)13 (18.84)1.43 (0.75,2.75)0.280Treatment-level variables Treatment visit regimen Single-visit137 (42.02)21 (15.33)1 Two-visit189 (57.98)26 (13.76)1.13 (0.63,2.03)0.681 Irrigant type NSS30 (9.20)5 (16.67)1 1% NaOCl296 (90.80)42 (14.19)0.83 (0.33,2.09)0.689 Obturation material Vitapex®293 (89.88)44 (15.02)1 ZOE33 (10.12)3 (9.09)0.51 (0.15,1.75)0.287Operator training level Postgraduate student308 (94.48)42 (13.64)1 Undergraduate student18 (5.52)5 (27.78)2.09 (0.84,5.25)0.115Abbreviations: HR hazard ratio, 95% CI 95% confidence interval, NaOCl sodium hypochlorite, NSS normal saline solution, ZOE zinc oxide eugenolaUnivariate screening of factorsbMultivariable Cox proportional hazards regression with robust cluster variance estimation (Huber–White sandwich estimator) was applied to variables with p < 0.10 in the univariate analysiscBold p-values indicate statistical significance (p < 0.05)dChild age at the time of treatment was analysed as a continuous variable in the multivariate model Cox regression analysis of factors associated with pulpectomy failure Abbreviations: HR hazard ratio, 95% CI 95% confidence interval, NaOCl sodium hypochlorite, NSS normal saline solution, ZOE zinc oxide eugenol aUnivariate screening of factors bMultivariable Cox proportional hazards regression with robust cluster variance estimation (Huber–White sandwich estimator) was applied to variables with p < 0.10 in the univariate analysis cBold p-values indicate statistical significance (p < 0.05) dChild age at the time of treatment was analysed as a continuous variable in the multivariate model Survival analysis During the 60-month observation period, a total of 47 teeth met the failure criteria, and no retreatment procedures were documented for any primary molars in the cohort. The KM analysis demonstrated high survival at early follow-up, followed by a gradual decline over time (Fig. 2). The number of teeth at risk decreased from 326 at baseline to 290 at 12 months, 217 at 24 months, 125 at 36 months, and 43 at 48 months, with only 4 teeth remaining at risk at 60 months. Due to the limited number of teeth remaining at risk beyond 48 months, survival estimates beyond this time should be interpreted cautiously. Therefore, survival was primarily summarised at 48 months. At this time point, the Kaplan–Meier estimated survival probability was 52.02% (SE 9.47; 95% CI: 32.4–68.5). The RMST at 48 months was 42.77 months (SE 0.61; 95% CI: 41.58–43.96), corresponding to an RMTL of 5.23 months (SE 0.61; 95% CI: 4.04–6.42).Fig. 2Kaplan–Meier survival curves of pulpectomy treatment in primary molars with apical periodontitis Kaplan–Meier survival curves of pulpectomy treatment in primary molars with apical periodontitis Cumulative survival estimates were calculated at 6-month intervals up to 48 months. Cumulative survival remained at 100% up to 12 months (95% CI: 100.0–100.0), reflecting the absence of symptomatic failure during early follow-up, then declined to 98.6% (95% CI: 97.3–99.9) at 18 months, 95.4% (95% CI: 92.8–98.0) at 24 months, 90.5% (95% CI: 86.6–94.4) at 30 months, 86.6% (95% CI: 81.9–91.3) at 36 months, and 78.9% (95% CI: 72.6–85.3) at 42 months. At 48 months, the cumulative survival rate was 72.1% (95% CI: 63.9–80.3), based on 43 teeth remaining at risk. Factors associated with failure In the multivariable Cox regression analysis, child age at the time of treatment and the presence of a preoperative sinus tract were statistically significant positive associations with failure (Table 1). Age at treatment was analysed as a continuous variable, with each 1-year increase in age corresponding to a hazard ratio of 1.05 (CI 1.01–1.09, p = 0.009). KM survival curves stratified by age at treatment showed that survival probability began to decline 1 year after treatment, with a more pronounced decline from 36 months onward, particularly in older age groups (Fig. 3).Fig. 3Kaplan–Meier survival curves of pulpectomy treatment in primary molars with apical periodontitis, stratified by age at treatment Kaplan–Meier survival curves of pulpectomy treatment in primary molars with apical periodontitis, stratified by age at treatment Among preoperative disease-burden variables, primary molars with preoperative sinus tracts showed a strong positive association with failure compared with normal soft tissue (HR 4.98; CI 1.99–12.45, p = 0.001), whereas the presence of preoperative gingival or vestibular swelling was not associated with failure. Root resorption involving one-third to two-thirds of the root length also showed a significant association with a hazard of failure (HR 2.96; CI 1.01–8.69, p = 0.048). However, these estimates should be interpreted cautiously because the low event counts in the sinus tract (n = 4) and in the root resorption subgroups (one-third to two-thirds, n = 7) resulted in wider confidence intervals. Treatment-level variables, including the visit regimen, irrigant type, obturation material, and operator training level, were not significantly associated with failure. Model robustness and proportional hazards diagnostics are summarised in Supplementary Table S3 and Supplementary Figure S2. Bootstrap assessment supported the stability of the child-age estimate but indicated wider uncertainty for categorical predictors, particularly sinus tract and root resorption involving one-third to two-thirds of the root length. Schoenfeld residual-based tests and graphical diagnostics did not indicate a meaningful violation of the proportional hazards assumption. Sensitivity analysis The comparison between the unweighted Kaplan–Meier estimate, plotted as cumulative incidence, and the IPCW-adjusted Fine–Gray cumulative incidence function did not indicate a qualitatively different failure pattern (Supplementary Figure S3). However, the IPCW-adjusted Fine–Gray estimate suggested a modestly higher absolute probability of failure than the unweighted Kaplan–Meier analysis, indicating that substantial loss to follow-up or unknown outcome status may have led to some underestimation of failure risk in the primary analysis. At 48 months, the Kaplan–Meier estimated survival probability was 52.02% (SE 9.47), with an RMST of 42.77 months (SE 0.61) and an RMTL of 5.23 months (SE 0.61), while the IPCW-adjusted Fine–Gray cumulative incidence of failure was 43.07% (SE 8.32) (Supplementary Table S4). In the IPCW-adjusted Fine–Gray sensitivity analysis, which accounted for documented non-failure extractions as competing events and potential informative censoring, the direction of the main associations was generally consistent with the primary multivariable Cox model (Supplementary Table S5). Child age remained positively associated with failure, although the association was attenuated from statistical significance in the primary model to borderline significance in the sensitivity analysis (primary Cox: HR 1.05, 95% CI 1.01–1.09; p = 0.009; IPCW-adjusted Fine–Gray: sHR 1.04, 95% CI 1.00–1.08; p = 0.052). Preoperative sinus tract remained a strong predictor in both analyses (primary Cox: HR 4.98, 95% CI 1.99–12.45; p = 0.001; IPCW-adjusted Fine–Gray: sHR 4.36, 95% CI 1.94–9.77; p < 0.001). In contrast, root resorption involving one-third to two-thirds of the root length was attenuated and was no longer statistically significant in the sensitivity analysis (primary Cox: HR 2.96, 95% CI 1.01–8.69; p = 0.048; IPCW-adjusted Fine–Gray: sHR 2.04, 95% CI 0.57–7.36; p = 0.274). Reasons for failure Among 47 failed primary molars, 34 (72.3%) demonstrated clinical failure, including pain, pathological mobility, swelling, abscess formation, or tenderness to percussion, and 13 (27.7%) exhibited radiographic failure, characterised by progression of periapical or furcation pathology with displacement or misdirection of the succedaneous tooth bud relative to previous radiographs. All failed teeth underwent extraction, and none of the subsequently erupted succedaneous teeth exhibited qualitative or quantitative developmental defects in the subset of erupted successors available for evaluation. Discussion Although AP has been shown to significantly increase the risk of pulpectomy failure in primary molars [3–10], pulpectomy treatment remains a viable treatment for infection control and tooth retention in young children [19]. In this retrospective cohort study, KM analysis of tooth survival following pulpectomy in primary molars with AP showed complete survival at 12 months and remaining above 95% at 24 months, followed by a gradual decline over time. Because of the limited number of teeth remaining at risk beyond 48 months, survival estimates beyond this point were interpreted cautiously. At 48 months, the Kaplan–Meier estimated survival probability was 52.02%, with an RMST of 42.77 months. In the primary Cox analysis, child age at treatment and preoperative sinus tract showed positive prognostic associations with failure, whereas the association between advanced root resorption and failure was less stable after sensitivity analysis. Each additional year of age increased the hazard of failure by an estimated 5%, while a preoperative sinus tract at the level of one-third to two-thirds resulted in a 4.98 and 2.96-fold increase in the hazard, respectively. In contrast, the other preoperative disease-burden variables and treatment-level variables showed no significant association with outcomes. In this study, the survival rate was comparable to those reported in the literature, regardless of preoperative AP [6–8]. This is likely attributed to differences in the evaluation criteria used. Many studies define treatment success using an “ideal model” [20] based on radiographic healing or lesion stability [3, 6–9]. In contrast, this study adapted a survival analysis framework from Songvejkasem et al. [5], prioritising tooth retention over strict biological resolution, in line with a core tenet of the 'praxis concept'. The ‘praxis concept’ and the ‘ideal model’ represent fundamentally divergent views of health and disease in endodontics [20], they yield differing estimates of survival and failure over time. Periapical healing after root canal treatment (RCT) is a prolonged process [17]. Although thorough chemo-mechanical debridement and preservation of coronal and apical integrity may support healing [17], lesion persistence is influenced by the balance between microbial virulence and host immune responses at the periradicular interface [17, 22], which may remain subclinical for months or even years before complete resolution [17]. The dynamic nature of this healing process allows lesions to improve, stabilise, or worsen over time [29]. However, the intervention is indicated only when the expected benefits clearly outweigh the associated risks and costs [20]. Pulpectomy procedures utilise more simplified debridement and obturation protocols than RCT. Furthermore, the morphological complexity of primary root canals and the resorptive behaviour of obturation materials limit the efficacy of debridement and compromise canal integrity, respectively [6, 8, 9]. In the present study, an adapted retention-focused outcome framework was used to reflect treatment decisions in primary dentition. The framework was modified from Songvejkasem et al. [5] by including tooth bud displacement as a criterion for failure, which is often a precursor to subsequent orthodontic complications [30, 31]. Among 47 teeth classified as failures, 34 (72.3%) showed clinical failure, and 13 (27.7%) showed radiographic deviation of succedaneous teeth. In the subset of erupted successors available for evaluation, neither qualitative nor quantitative developmental enamel defects were observed. Previous studies [3, 19, 31, 32] have reported developmental enamel defects (DEDs) and successor malposition following pulpal infection or pulpectomy in primary teeth, with prevalence ranging from 11.1% to 57% [19, 32], and 36.6% [31], respectively. While the prevalences of DEDs and ectopic eruption following pulpectomy have been reported by JA Coll and R Sadrian [3], at 18.7% and 21.6%, respectively. Despite comparable cohorts focused on teeth exhibiting AP, the rate of successor malposition found in our study (27.7%) was slightly higher. The apparent discrepancies between those reports [3, 19, 31, 32] and the present findings may be due to differences in outcome definition, timing of assessment, and methods used to detect eruption disturbance. In the present study, serial radiographic follow-up allowed early identification of eruption-path deviations before clinical ectopic eruption became evident. However, because successor outcomes were assessed retrospectively within the available follow-up records, these findings should be interpreted cautiously and should not be taken as definitive evidence that pulpectomy has no effect on permanent successor development. Across the variables representing disease burden, preoperative sinus tract, and root resorption affecting one-third to two-thirds of the root, were associated predictors of failure, increasing the hazard by 4.98 (95% CI 1.99–12.45) and 2.96 (95% CI 1.01–8.69), respectively. A sinus tract usually signifies a chronic, draining stage of AP, whereas soft tissue swelling indicates an acute, pressurised state [15, 22]. Whereas root resorption is a natural process during primary tooth exfoliation, root resorption involving one-third to two-thirds of the root length is plausibly attributed to pathological rather than physiological resorption, considering the age of children in this cohort [33]. Our results suggest that the presence of a preoperative sinus tract or root resorption front extending to one-third to two-thirds may reflect a more advanced or persistent infectious condition and was associated with a higher hazard of failure in this cohort. This interpretation is biologically plausible and consistent with a previous report [3, 5, 8, 9]. However, this finding should be interpreted cautiously because the sinus-tract and the root resorption affecting one-third to two-thirds subgroups were small, with limited failure events and wider uncertainty. In addition, apical diagnoses in this retrospective study were based primarily on radiographic findings rather than formal AAE diagnostic classification [34]. Therefore, these results may not be interpreted as demonstrating a definitive prognostic difference between chronic and acute apical periodontitis. The child’s age at the time of treatment is a well-established predictive risk factor for pulpectomy failure [5, 6, 8, 9]. However, age has been categorised inconsistently, ranging from binary 36-month thresholds [5] and one-year intervals [6, 8] to broad multi-year groupings [9]. Utilising one-year age bands provides a clinically practical representation, as it captures the year-over-year failure risk following developmental milestones. For this cohort, each additional year of treatment was associated with a 5% higher hazard of failure (HR 1.05, 95% CI 1.01,1.09). In chronic apical periodontitis, lateral canals and apical ramifications frequently harbour resistant biofilms [15]. Increasing age is associated with changes in root canal anatomical complexity [6, 8, 9], which hinder effective canal debridement [6, 8, 9] and compromise obturation quality [9]. Furthermore, increasing age diminishes healing capacity [6, 8]. Collectively, these factors reduce the long-term integrity of the apical seal and facilitate persistent infection, thereby increasing the likelihood of post-treatment failure. Given that root canal morphological intricacy is an inherent feature across all primary molar types [18], tooth type did not emerge as a significant predictor of pulpectomy failure in this cohort. While the study [4] analysing combined anterior and posterior teeth found no age-related impact on AP resolution, our findings suggest that the unique anatomical challenges of primary molars make age a critical determinant of pulpectomy success. Within the retention-focused endpoint used in this study, pulpectomy in primary molars with apical periodontitis showed an estimated 48-month KM survival probability of 52.02%, with an RMST of 42.77 months. In the primary analysis, increasing age, the presence of preoperative sinus tracts, and the root resorption affecting one-third to two-thirds of the root increased the risk of failure. However, model diagnostics and sensitivity analyses indicated that these estimates differed in stability. Preoperative sinus tract showed the most consistent association across the primary Cox and IPCW-adjusted Fine–Gray models, whereas child age remained directionally consistent but less strongly supported, and advanced root resorption was attenuated and lost statistical significance. These findings support cautious prognostic counselling, meticulous chemo-mechanical debridement and disinfection, and closer follow-up for teeth presenting with higher-risk features. The outcome assessment used in this study was based on an adapted retention-focused endpoint and therefore differs from more commonly used biologically oriented success criteria. Several limitations should be considered when interpreting the present findings. This retrospective design relied mainly on radiographic evidence of AP rather than comprehensive AAE diagnostic criteria [34], the cohort may not capture the full clinical spectrum of AP. In addition, the survival analysis was vulnerable to limitations related to censoring and competing events. A substantial proportion of non-failure observations were classified as censored because of loss to follow-up or unknown status. Although an IPCW-adjusted Fine–Gray sensitivity analysis was performed to assess the potential influence of informative censoring and documented competing events, residual bias cannot be excluded. Documented non-failure extractions were treated as competing events in the sensitivity analysis; however, exfoliation could not be analysed as a separate competing event because it was not consistently distinguishable from loss to follow-up or unknown outcome status in the dataset. Furthermore, only 47 failures were observed, resulting in an event-per-parameter ratio below the conventional preferred threshold [35]. Bootstrap assessment indicated wider uncertainty for categorical predictors with small subgroup counts, particularly sinus tract and advanced root resorption. Therefore, these estimates should be interpreted as cautious prognostic associations. All cases were managed at a single institution's teaching clinic. Furthermore, the absence of documentation explaining the rationale behind the treatment regimen, whether due to disease severity or administrative time constraints, limits the generalisability of our findings regarding the specific therapeutic efficacy of the inter-appointment medicament in managing periradicular lesions. Therefore, generalisability may be limited because this was a single-centre university teaching clinic cohort managed under institution-specific protocols, with most teeth treated using Vitapex®. These methodological constraints underscore the need for prospective, multi-centre studies with standardised diagnostic protocols and longer, systematic follow-up. Future research should prioritise larger sample sizes to improve precision of clustered multivariable estimates and a more detailed exploration of how diverse debridement and instrumentation protocols influence survival in complex apical environments. Conclusion Within the limitations of this retrospective single-centre cohort and the adapted retention-focused endpoint used, primary molars with apical periodontitis showed a 48-month Kaplan–Meier survival probability of 52.02% after pulpectomy, with an RMST of 42.77 months. In the primary analysis, older age at treatment and the presence of a preoperative sinus tract were associated with a higher hazard of failure. Sensitivity analysis accounting for competing events and potential informative censoring supported the robustness of the association with preoperative sinus tract, while the association with age remained directionally consistent but less precise. The association between advanced root resorption and failure was less stable and should be interpreted cautiously. The prognostic factor findings should be considered hypothesis-generating, indicating potential signals that warrant confirmation in larger prospective cohorts. Substantial late censoring and sparse subgroup events limit the precision of these estimates, underscoring the need for prospective studies to further investigate the prognosis of pulpectomy in primary molars with apical periodontitis. Supplementary Information Additional file 1: Supplementary Figure S1 Example radiographs illustrating representative preoperative radiographic categories of apical periodontitis and root resorption in the mandibular second molars. Additional file 2: Supplementary Table S1 Statistical analysis steps. Additional file 3: Supplementary Table S2 Follow-up duration by outcome status and censoring reason. Additional file 4: Supplementary Table S3 Bootstrap robustness check and proportional hazards diagnostics for the final multivariable Cox model. Additional file 5: Supplementary Figure S2 Proportional hazards diagnostic plots for the final multivariable Cox model. Additional file 6: Supplementary Figure S3 Comparison of unweighted Kaplan–Meier and IPCW-adjusted Fine–Gray cumulative incidence estimates for pulpectomy failure. Additional file 7: Supplementary Table S4 Comparison of 48-month Kaplan–Meier survival estimates and IPCW-adjusted Fine–Gray cumulative incidence estimates for pulpectomy failure. Additional file 8: Supplementary Table S5 Comparison of primary Cox regression and IPCW-adjusted Fine–Gray sensitivity analysis for factors associated with pulpectomy failure. Additional file 1: Supplementary Figure S1 Example radiographs illustrating representative preoperative radiographic categories of apical periodontitis and root resorption in the mandibular second molars. Additional file 2: Supplementary Table S1 Statistical analysis steps. Additional file 3: Supplementary Table S2 Follow-up duration by outcome status and censoring reason. Additional file 4: Supplementary Table S3 Bootstrap robustness check and proportional hazards diagnostics for the final multivariable Cox model. Additional file 5: Supplementary Figure S2 Proportional hazards diagnostic plots for the final multivariable Cox model. Additional file 6: Supplementary Figure S3 Comparison of unweighted Kaplan–Meier and IPCW-adjusted Fine–Gray cumulative incidence estimates for pulpectomy failure. Additional file 7: Supplementary Table S4 Comparison of 48-month Kaplan–Meier survival estimates and IPCW-adjusted Fine–Gray cumulative incidence estimates for pulpectomy failure. Additional file 8: Supplementary Table S5 Comparison of primary Cox regression and IPCW-adjusted Fine–Gray sensitivity analysis for factors associated with pulpectomy failure.