Tooth extractions and the prevalence of postoperative complications in individuals with and without diabetes: a retrospective cross-sectional study

Abstract

Aims

Individuals with diabetes are at increased risk for periodontitis and tooth loss, yet the primary causes of tooth extractions and associated postoperative complications in this population remain underexplored. This retrospective cross-sectional study investigated the reasons for tooth extractions and the prevalence of postoperative complications among individuals with and without diabetes.

Methods

Patient- and tooth-related data between 2022 and 2024 were collected from participants undergoing permanent tooth extractions, excluding third molar and orthodontic extractions. Descriptive statistics and structural equation modeling (SEM) assessed the effects of age, diabetes, and smoking on the number of remaining teeth, mediated by periodontitis.

Results

Of 490 participants, 100 reported type 2 diabetes (mean HbA1c: 7.5%). The diabetic group exhibited more periodontitis and retained significantly fewer teeth after extractions (p < 0.01). 90% of extractions in both groups were due to nonperiodontal conditions (caries or subsequent endodontic infections). Post-extraction complications were rare (2%) and did not differ between the two groups, but antibiotics were prescribed significantly more often for the diabetic group. The SEM revealed that age and smoking were significant negative predictors for the remaining teeth number (p < 0.001), while diabetes and periodontitis were not (p > 0.05).

Conclusions

Tooth extractions in individuals with diabetes were largely due to nonperiodontal infections rather than periodontitis. The higher prescription of antibiotics in this group reflects clinical concern about postoperative infection risk. These findings underscore the importance of caries prevention and cautious use of antibiotics to reduce postoperative complications among individuals with diabetes.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12903-026-09217-2.

Introduction

Odontogenic infections, which are caused by bacteria naturally residing in dental plaque around teeth, are common and opportunistic in nature. Under certain conditions, these nonpathogenic microorganisms can proliferate and lead to oral diseases such as dental caries or periodontal diseases [1]. Tooth loss associated with odontogenic infections remains a significant public health concern due to its impact on individuals’ quality of life, resulting from discomfort, impaired function, poor dietary intake, and esthetic concerns [2]. Although the prevalence of tooth loss has decreased over the decades, it continues to affect vulnerable populations, particularly older adults and those with chronic health conditions [2]. Of chronic health conditions, the risk for tooth loss among individuals with diabetes is reported to be significantly higher than those without diabetes [3].

Aging populations are susceptible to oral and systemic health deterioration due to increased prevalence of chronic diseases and medication intake, which can lead to xerostomia and compromised oral hygiene [4]. Older age is significantly correlated with the development of type 2 diabetes, and the prevalence, incidence, and mortality rates of type 2 diabetes depend on age [5]. Diabetes mellitus is a chronic systemic disease that significantly impacts both general and oral health. Subclinical chronic inflammation is a common feature in the course of diabetes and is associated with many complications [6]. Periodontitis, one of the odontogenic infections, is the sixth complication of diabetes [7].

The relationship between diabetes and oral health, especially periodontitis, has been extensively investigated [8]. Diabetes is a risk factor for periodontitis, and poorly controlled diabetes increases the risk of developing periodontitis [9]. Studies have also investigated the association between diabetes and tooth loss, but findings have been inconsistent. A systematic review reported that no significant association between type 2 diabetes and tooth loss among cohort studies and found a positive association between them only in cross-sectional studies [10].

Tooth loss is closely associated with tooth extractions. Clinical decisions regarding tooth extractions are based on the extent of odontogenic infection and the prognosis of the affected teeth [11]. Assessing glycemic control in individuals with diabetes is important prior to dental surgical procedures since it is generally believed that infection and delayed wound healing after surgical dental treatments frequently occur in patients with poorly controlled diabetes [12]. However, few studies have examined the primary reasons of tooth extractions and post-operative complications following tooth extractions in individuals with diabetes.

The purpose of this retrospective cross-sectional study was to investigate the primary reasons for tooth extractions and the prevalence of post-extraction complications among individuals with diabetes compared to those without diabetes in a university-based clinical setting.

Methods

The study report followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines for observational studies from the Enhancing the Quality and Transparency of Health Research [13].

Study participants

The inclusion criteria were participants aged ≥ 20 years who had permanent teeth extracted in the oral surgery clinic at the University of Maryland School of Dentistry (UMSOD). The exclusion criteria were participants who had only third molar and/or orthodontic tooth extractions and marijuana smokers. The screening population was retrieved from the electronic health record (EHR) database using the following Code on Dental Procedures and Nomenclature (CDT) code: simple extraction of a single tooth (D7140). Potential participants were screened chronologically starting from January 2, 2022, until the number of participants reached the sample size based on the power calculation.

Six hundred sixteen potential participants were initially screened and further evaluated with the exclusion criteria. 107 participants with third molar extractions and 5 participants with orthodontic extractions were excluded. 14 marijuana smokers were also excluded because the effects of marijuana smoking might be potentially different from those of tobacco smoking [14]. The final study population included 490 participants.

Data collection

The following data from the 490 participants was collected: demographic information (age, gender, and medical/social history), date of the last extraction, periodontal diagnosis (gingivitis or periodontitis), documented reasons for tooth extractions, the number of remaining teeth (excluding the third molars) after tooth extractions, post-extraction antibiotic prescription (yes or no), and postoperative complication(s) after extractions (yes or no). Postoperative complications were defined as post-extraction infections, bleeding, pain, or dry socket documented in the EHR with ensuing managements/treatments.

The age of participants was calculated by subtracting the date of birth from the date of the last extraction recorded in the EHR. With respect to medical and social history, diabetes and smoking status were collected. Diabetes status was determined based on self-reported medical history in the EHR. To minimize diabetes misclassification, only individuals with reported type 2 diabetes were included. For participants with type 2 diabetes, self-reported hemoglobin A1C (HbA1c) and/or blood glucose levels were recorded when available. Smoking status was categorized as smoker or nonsmoker. Smoker indicated a participant who was smoking at the time of extractions; nonsmoker indicated an individual was not smoking, which included a former- and a never-smoker. Therefore, the past smoking effects were incorporated in the nonsmoking group.

Radiographs, periodontal charts, and visit notes were reviewed to determine periodontal diagnosis. In this study, gingivitis was defined as gingival inflammation without bone loss [15], and periodontitis was defined as gingival inflammation with bone loss. Gingival inflammation was determined by probing pocket depths with bleeding on probing. Bone loss was determined if the linear distance between the most coronal bone level and the cementoenamel junction on radiographs was greater than 2 mm [16].

Radiographs and visit notes at the extraction procedures were reviewed to determine the primary reason for tooth extractions. The reasons for tooth extractions were classified into periodontal or nonperiodontal categories. Figure 1 provides examples of extractions due to periodontal and nonperiodontal reasons. The nonperiodontal reason included caries, endodontic infections (pulpitis, necrotic pulp, apical periodontitis, apical abscess, or failed root canal therapy), and prosthodontic conditions (a lack of coronal tooth structure, insufficient interocclusal space for prostheses, or strategic extraction). The periodontal reason was assigned when tooth extractions were performed because of periodontal bone loss alone or severe periodontal bone loss (more than 50% bone loss) combined with insignificant caries or manageable endodontic lesions; endodontic-periodontal combined lesions and occlusal trauma were also included in the periodontal reason.

Fig. 1.

Examples of tooth extraction due to periodontal or nonperiodontal reasons. Yellow arrows indicate teeth extracted. (a) The mandibular right first and second molars were extracted because of endodontic conditions. (b) The mandibular left second molar was extracted because of dental caries. (c) The maxillary right lateral incisor was extracted due to a prosthodontic condition. (d) All teeth were extracted because of caries. (e) The maxillary left second premolar was extracted because of severe periodontal bone loss (periodontal reason). (f) All teeth were extracted because of periodontal reason

Statistical analysis

A post hoc power analysis comparing the distribution of reasons for tooth extractions between the two groups indicated that the study had 85% power to detect an effect size of Cohen’s w = 0.15 at a significance level of 0.05. The data were expressed as mean ± standard deviation or count. Chi-square or Wilcoxon rank-sum tests were performed to compare variables between the groups with and without diabetes.

Binomial logistic regression was conducted to estimate the probability of prescribing post-extraction antibiotics based on the HbA1c levels and periodontal diagnosis (gingivitis versus periodontitis). While the full sample size was 490, HbA1c values were available for 79. Missing HbA1c data were retained in the model using listwise deletion for HbA1c predictor; therefore, the HbA1c coefficient reflected estimates based on available cases. Binomial logistic regression was performed to estimate the probability of periodontitis based on age, diabetes, smoking, and the interaction of diabetes and smoking without missing data.

For a modeling approach regarding the number of remaining teeth, a structural equation model (SEM) was developed, guided by a directed acyclic graph (DAG) to inform the hypothesized relationships among predictors (age, diabetes, and smoking), mediated by periodontitis without missing data. All analyses were conducted using software programs, jamovi [17] and R [18]; p < 0.05 was considered significant.

Results

The study population included 100 individuals with diabetes (20.4%). Among 79 individuals with documented HbA1c levels, the HbA1c level (%) ranged from 5.1 to 14 (32.24─129.51 mmol/mol) with a mean of 7.5 (58.47 mmol/mol). The casual blood glucose level (mg/dL) ranged from 80 to 533 with a mean of 166.5.

Table 1 summarizes the comparisons between the groups with diabetes and without diabetes. The mean age of the diabetic group at the time of tooth extractions was significantly older than that of the nondiabetic group (62 versus 51; p < 0.01). Before tooth extractions, the group with diabetes exhibited significantly more periodontitis than the group without diabetes (93% versus 68%; p < 0.01). The diabetic group retained significantly fewer teeth than the nondiabetic group after tooth extractions (16 versus 19; p < 0.01).

Table 1.

Characteristics in the groups with diabetes and without diabetes. Mean ± standard deviation. Endo, endodontic infection; Prostho, prosthodontic conditions; Endo-Perio, endodontic-periodontal combined lesions

Parameter		Group with diabetes (n = 100)	Group without diabetes (n = 390)	Total (N = 490)	p value
Gender	Female	44 (44%)	211 (54%)	255 (52%)	0.07 1
	Male	56 (56%)	179 (46%)	235 (48%)
Age (years) at the extraction(s)		62 ± 13	51 ± 16	54 ± 16	< 0.01 2
Smoking	Nonsmoker	81 (81%)	281 (72%)	362 (74%)	0.07 1
	Smoker	19 (19%)	109 (28%)	128 (26%)
Periodontal diagnosis	Gingivitis	7 (7%)	125 (32%)	132 (27%)	< 0.01 1
	Periodontitis	93 (93%)	265 (68%)	358 (73%)
The number of remaining teeth		16 ± 8	19 ± 8	18 ± 8	< 0.01 2
Reason for extraction	Nonperiodontal   Caries   Endo   Prostho	86 (86%) 58 (58%) 17 (17%) 11 (11%)	356 (91%) 252 (64.6%) 71 (18%) 33 (8.5%)	442 (90.2%) 310 (63%) 88 (18%) 44 (9%)	0.11 1
	Periodontal   Endo-Perio   Severe bone loss   Occlusal trauma	14 (14%) 4 (4%) 10 (10%) 0 (0%)	34 (9%) 7 (2%) 25 (6%) 2 (1%)	48 (9.8%) 11 (2.2%) 35 (7.1%) 2 (0.4%)
Post-extraction antibiotics	No	86 (86%)	371 (95%)	457 (93%)	< 0.01 1
	Yes	14 (14%)	19 (5%)	33 (7%)
Postoperative complications	No	98 (98%)	384 (98.5%)	482 (98%)	0.75 1
	Yes	2 (2%)	6 (1.5%)	8 (2%)
HbA1c level (%), n = 79		7.5 ± 1.6
Blood glucose level (mg/dL), n = 67		167 ± 85

¹Chi-squared test, ²Wilcoxon rank-sum test

The nonperiodontal reasons, such as caries, endodontic infections, and prosthodontic conditions, were the most common causes for tooth extractions, accounting for 90.2% of cases in the study population (Table 1). 14% of individuals with diabetes lost their teeth due to periodontal conditions, while 9% of individuals without diabetes did; no significant difference in the distribution of reasons for tooth extractions between the two groups was observed (p = 0.11).

The complications after tooth extractions were few; 8 participants (1.6%) experienced post-extraction complications with no different prevalence between the two groups. However, prescribing immediate post-extraction antibiotics was significantly higher in the diabetic group than in the nondiabetic group (14% versus 5%; p = 0.001). A binomial logistic regression model for prescribing antibiotics with respect to HbA1 levels and periodontal diagnosis was statistically significant with a moderate explanatory power (Fig. 2; χ² = 9.39, degree of freedom (df) = 2, p = 0.009, McFadden’s R² = 0.14). The HbA1c levels were significantly associated with antibiotics prescription. As the HbA1c levels increased, the odds of prescribing antibiotics increased (the odds ratio (OR) = 1.51, 95% confidence interval (CI) [1.04─2.18], p = 0.029). Patients with periodontitis were less likely to receive post-extraction antibiotics compared to ones with gingivitis, but this negative estimate was not statistically significant (OR = 0.16, 95% CI [0.026─1.03], p = 0.053).

Fig. 2.

Binomial logistic regression for the probability of prescribing post-extraction antibiotics (χ² = 9.39, degree of freedom = 2, p = 0.009, McFadden’s R² = 0.14) based on the HbA1c levels (p = 0.029) and periodontal diagnosis (p = 0.053). P = probability

Figure 3 shows the plots from the binomial logistic model for the probability of having periodontitis based on age, diabetes, smoking, and the interaction of diabetes and smoking (N = 490, χ² = 203, df = 4, p < 0.001, McFadden’s R² = 0.36). As age increased, the odds of having periodontitis increased (OR = 1.12, 95% CI [1.1 to 1.15], p < 0.001). Smoking significantly increased the probability of having periodontitis (OR = 3.2, 95% CI [1.7 to 5.96], p < 0.001). Individuals with diabetes were 3.67 times more likely to have periodontitis compared to ones without diabetes (OR = 3.67, 95% CI [1.28 to 10.55], p = 0.016). However, the interaction of diabetes and smoking did not significantly affect the probability of periodontitis (OR = 0.37, 95% CI [0.05 to 2.75], p = 0.33).

Fig. 3.

Binomial logistic regression for the probability of having periodontitis (χ² = 203, degree of freedom = 4, p < 0.001, McFadden’s R² = 0.36) with respect to age (p < 0.001), diabetes (p = 0.02), smoking (p < 0.001), and interaction of diabetes and smoking (p = 0.33). S, smoker; NS, nonsmoker

Figure 4 shows a DAG with direct effect (β) values, which guided the SEM for the number of remaining teeth. Since the interaction of diabetes and smoking on periodontitis was not significant (Fig. 3), it was excluded from the SEM to avoid unnecessary complexity. The final SEM showed a perfect fit (df = 0) as all possible paths among the variables were included (Table 2). Age (direct effect = 0.56, p < 0.0001), diabetes (direct effect = 0.09, p = 0.003), and smoking (direct effect = 0.14, p < 0.0001) were significantly associated with an increased risk of periodontitis; age (direct effect = -0.41, p < 0.0001) and smoking (direct effect = -0.25, p < 0.0001) had significant negative effects on tooth retention. Diabetes (direct effect = -0.06, p = 0.16) and periodontitis (direct effect = 0.01, p = 0.81) had no significant direct effects on tooth retention. There were no significant indirect effects of age, diabetes, and smoking on the number of remaining teeth through periodontitis (Table 2).

Fig. 4.

A directed acyclic graph (DAG) for the number of remaining teeth with direct effect (β) values. The red font indicates a significant effect

Table 2.

Parameter estimates from the structural equation model (SEM) for the number of remaining teeth number with respect to age, diabetes, smoking, mediated by periodontitis stage

Regressions for the number of remaining teeth
Outcome	Predictor	Estimate	Standard Error	z value	p value
Periodontitis	Age	0.02	0.001	15.76	< 0.0001
	Diabetes	0.10	0.033	3.19	0.003
	Smoking	0.14	0.038	3.84	< 0.0001
The number of remaining teeth	Age	-0.20	0.024	-8.3	< 0.0001
	Diabetes	-1.21	0.856	-1.4	0.16
	Smoking	-4.48	0.785	-5.7	< 0.0001
	Periodontitis	0.20	0.834	0.24	0.81

Indirect Effect
Predictor	Mediator	Outcome	Indirect Effect (β)		p value
Age	Periodontitis	The number of remaining teeth	0.006		0.81
Diabetes			0.001		0.82
Smoking			0.002		0.81

Discussion

Dental caries and periodontal disease are the leading causes of tooth extractions, with their relative impacts varying by age and socioeconomic status [19]. A systematic review reported that up to 55% of extractions were due to caries, up to 38% due to periodontitis, and up to 19% due to endodontic infection [20]. In adults aged ≥ 55 years, periodontal disease accounts for nearly 38% of extractions, followed by caries (30%) and fractures (18%) [21]. As caries commonly leads to coronal tooth fracture and/or subsequent endodontic infections, caries and endodontic infections together represent the most common reasons for tooth extraction [22, 23].

Our study found that approximately 10% of tooth extractions were conducted because of periodontal reasons. Although 93% of participants with diabetes were diagnosed with periodontitis, only 14% of extractions in this group were mainly due to periodontal reasons. Among the 137 participants aged ≥ 65 years, 17% of tooth extractions were attributed to periodontal reasons. Therefore, the primary reason for tooth extractions was still nonperiodontal reasons in both the elderly and diabetic groups.

Reported postoperative complication rates following simple tooth extractions range from 1 to 11% and may be higher in medically complex patients [24, 25]. While odontogenic infections, either from nonperiodontal or periodontal origins, are the common reasons for tooth extractions, the postoperative complications were few (2%) in this study. Postoperative complications in this study might be underreported because follow‑ups after simple extractions were not routinely performed at the institution and complications managed outside the institution might not have been recorded. However, complications requiring immediate clinical intervention would be most likely to be documented.

Post-extraction antibiotics are not routinely prescribed because odontogenic infections clear up by themselves after tooth extractions [26]. In this study, however, immediate post-extraction antibiotics were prescribed with a significantly higher rate in the diabetic group compared to the nondiabetic group (14% versus 5%). Dental providers were more likely to prescribe antibiotics for individuals with diabetes who reported elevated HbA1c levels (Fig. 2). This may reflect clinicians’ concern about increased infection risk in individuals with poorly controlled diabetes [12]. These preemptive antibiotic prescriptions might help minimize complications following tooth extractions in the diabetic group. Post-extraction complications occurred at similar rates in both groups. All 8 postoperative complications occurred in patients without immediate post-extraction antibiotics prescribed. Systemic antibiotics were administered for 6 postoperative infections and no treatment was rendered for 2 dry sockets.

Age, diabetes, and smoking were associated with a significantly increased probability of periodontitis, but no synergistic effect of diabetes and smoking on the prevalence of periodontitis was observed in this study (Fig. 3). This finding may suggest that the effects of diabetes and smoking are not additive beyond their individual influences, although this interpretation is limited by the small number of individuals with both conditions.

Although individuals with diabetes exhibited a higher prevalence of periodontitis, periodontitis showed no significant direct effect on tooth retention in this extraction-based sample. Age and smoking were significantly associated with fewer remaining teeth, but these negative effects were not mediated through periodontitis (Table 2). Inadequate periodontal support is generally associated with an increased risk of tooth loss, but the main reasons for tooth extractions in this study were nonperiodontal infections, such as caries and subsequent endodontic infections. These conditions are not directly influenced by periodontitis, which may explain why periodontitis did not significantly affect the number of remaining teeth or mediate the relationship between those predictors and tooth retention at the individual level.

Previous cross-sectional studies found the significant association between type 2 diabetes and tooth loss [10]. However, this study did not find such association. In the SEM, diabetes had no significant direct effect on the number of remaining teeth (Fig. 4). The effect of diabetes on teeth retention might be surpassed by age because age significantly influences overall oral health by increasing dental caries and periodontitis [4]. Additionally, the relationship between diabetes, dental caries, and tooth loss is complex. Some studies have reported that individuals with diabetes have also shown an increased risk of caries and have had a greater number of teeth with restorations [3, 27]. However, the greater number of tooth extractions in individuals with diabetes may be more closely associated with the overconsumption of sugar-sweetened beverages rather than diabetes directly increasing the prevalence of caries [3].

The generalizability of the study findings is limited due to the following limitations. First, the study was conducted at a university-based oral surgery clinic. Considering our geographic location, the study population may largely seek emergency care rather than comprehensive dental care. Consequently, patients may present with more advanced disease or complex medical histories, which may limit generalizability to community dental settings. Second, socioeconomic status and access to medical or dental care could not be included because income-related information or primary care provider details are not routinely collected in dental clinical settings. These factors may influence systemic and oral diseases as well as tooth retention, possibly confounding the observed associations directly or indirectly. Third, behavioral factors, such as oral hygiene practices, dietary sugar intake, or fluoride exposure, were not available in this retrospective dataset and may confound the association between diabetes and caries‑related tooth loss. Finally, as a cross-sectional study, this study could not capture the duration of diabetes and changes in glycemic control over time. Future studies should adopt a longitudinal study design to observe changes in these aspects.

Conclusion

Caries and endodontic infections were the leading causes of tooth loss in individuals with and without diabetes. Although post-extraction infections were rare, the higher rate of antibiotic prescriptions for individuals with diabetes, especially with elevated HbA1c levels, was observed. The prevention of caries and subsequent endodontic infections is a critical component of comprehensive dental care for individuals with diabetes. Interprofessional care is essential to reduce infection-related complications undergoing dental surgical procedures in this population.

Authors’ contributions

All authors have made substantial contributions to conception and design of the study. **SO** : Conceptualization, Methodology, Formal analysis, Writing - Original Draft, Writing – Review & Editing, Visualization, Supervision, Project administration; **EZ** : Methodology, Investigation, Writing - Original Draft, Writing – Review & Editing; **JS** : Methodology, Investigation, Writing – Original Draft.

Funding

The study was funded by the In-Spire Grant from the University of Maryland, School of Dentistry.

Data availability

The deidentified data of this study are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

This study was designed and performed in accordance with the principles presented in the Helsinki Declaration as revised in 2013 for biomedical research involving human subjects. This study was approved as the exempt protocol by the Institutional Review Board (IRB) at University of Maryland, Baltimore (HP-00108394). Obtaining informed consent to participate was waived by the IRB because the study was conducted retrospectively and there were no direct interactions with the participants.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.