Bitewing Radiographs in Proximal Caries Diagnostics and Restorative Treatment of Adults: A Retrospective Study of Electronic Health Records

Abstract

Introduction

The restorative threshold of dentists varies considerably. This retrospective study examined the association between the depth of radiolucency in proximal surfaces on bitewing radiographs and the probability of restorative treatment among Finnish dentists.

Methods

The study population comprised 160 patients aged 18–40 years who underwent routine dental examinations by 16 randomly selected dentists within the Public Dental Services of Kuopio, Finland, in 2022. Data on patient characteristics, caries lesions, radiographic findings, and treatment decisions were manually extracted from the local electronic health record system. The depth of radiolucency was measured for all proximal caries lesions that had been registered by the dentists and were visible on the radiographs. The association between radiolucency depth and treatment decisions was estimated using linear regression analyses with cluster-robust standard errors.

Results

Bitewing radiography had been used for 128 (80%) patients. Most of the 394 caries lesions that had been registered were radiographically confined to the enamel (57%) or outermost third of dentin (27%), whereas 12% were in the middle third of dentin, and 4% extended to the inner third of dentin. Regression analyses revealed that the predicted probability of restorative treatment was 8% for lesions reaching the dentinoenamel junction and 75% for those extending to the border between the outer and middle thirds of dentin.

Conclusions

The probability of restorative treatment increased sharply with the depth of radiolucency and was unexpectedly high for lesions confined to enamel or the outer third of dentin.

Introduction

Dental caries is a multifactorial, dynamic disease in which bacteria produce acidic metabolic byproducts that lead to the demineralization of tooth tissue [1]. Visual-tactile examination, supported by structured classification systems, has long been recommended as the primary method for detecting and classifying caries lesions on accessible surfaces [2]. Bitewing radiographs are commonly used as a complementary diagnostic tool, particularly for assessing proximal surfaces, which are difficult to evaluate using visual-tactile methods alone [2, 3]. However, some discrepancies exist regarding whether a bitewing examination increases the risk of overdiagnosis and overtreatment [2, 4–6]. This discrepancy may arise from the fact that the benefits and harms of radiographic examinations depend on how they are interpreted and how they influence clinical decision-making [7]. Some dentists may rely heavily on radiographic findings in proximal caries diagnosis without confirming the findings clinically which may lead to overdiagnosis and further overtreatment, that is filling therapy of non-cavitated, asymptomatic, and shallow caries lesions.

Over the past 2 decades, treatment strategies for proximal lesions have changed. The threshold for restorative treatment of asymptomatic caries lesions has elevated from dentin caries to cavitated caries lesions, with increasing emphasis on noninvasive treatment of caries [8–12]. Repeated Norwegian surveys [13, 14] have shown that the proportion of dentists who would restore proximal caries lesions confined to enamel in radiographs has decreased from 68% to 7% between 1983 and 2009. However, even in 2009, most Norwegian dentists chose restorative treatment for caries lesions that were limited to the outermost third of dentin in bitewing radiographs, whereas approximately one-third of dentists would opt to restore only when the radiolucency extends to the middle third of dentin [14]. Surveys conducted worldwide indicate that the threshold for restorative treatment is significantly lower in Scandinavia than in the rest of the world [15], probably due to considerable differences in treatment culture and caries burden across countries [16].

However, as Adams et al. [17] noted, “although self-reports may provide information regarding clinicians knowledge of guideline recommendations, they are subject to bias and should not be used as the sole measure of guideline adherence.” Retrospective evaluation of radiographs and treatment decisions allows the evaluation of which caries lesions are restored, and which are not. Investigations on Norwegian and Danish adolescents have shown that the percentage of carious lesions in the enamel and outer half of dentin that were restored declined significantly from the 1970s to the 1990s [18, 19]. Among 15-year-olds in Oslo in 1996, 5% of the caries lesions in the inner enamel half and 48% in the outer dentin half were restored. By contrast, in 1979, these percentages were 47% and 80%, respectively. However, we are unaware of more recent studies on this topic or those focusing on adults. Therefore, this electronic health record-based study examined the association between the depth of radiolucency in proximal surfaces on bitewing radiographs and the probability of dentists choosing restorative treatment in the Public Dental Services (PDS) of Kuopio, Finland.

Methods

Study Design

This was a retrospective, health record-based study. Data were collected from the electronic health record system WinHIT^® (In Net Oy, Lahti, Finland), used by the Public Dental Services (PDS) in Kuopio, Finland. In Finland, PDS are heavily subsidized and accessible to individuals of all ages. Data collection was conducted manually by a Doctor of Dental Surgery (L.O.) and a fifth-year dental student (M.H.M.), both familiar with the electronic health record system WinHIT and local usage conventions in PDS in Kuopio, between March and May 2024. The sample selection and data extraction protocols were collaboratively developed by the research team. To ensure data collection quality and effectiveness, the data for the first few cases were collected by primary data collectors (M.H.M. and L.O.) and experienced research group members (J.L. and E.R.) together to align manual data collection and measurements. J.L. is a specialist dentist in cariology and endodontics, and E.R. is a specialist dentist in dental public health. During the data collection, any unclear or ambiguous cases that arose were discussed and resolved within the research group.

Patient Selection

A two-stage sampling approach was chosen to ensure representation across different dentists and to account for potential clustering of patient characteristics or outcomes within dentists. First, a random sample of 16 dentists was selected from the 70 general dentists working in PDS in Kuopio area in 2022. The age and sex of each dentist were recorded. Because we had to use dentists’ appointment books to identify patients, for each selected dentist, the first ten 18-40-year-old patients who had undergone oral examination were chosen chronologically from their appointment book, starting from January 1, 2022 (i.e., the first 10 eligible patients for each dentist). The inclusion criteria for patients were age between 18 and 40 years and having undergone an oral examination in 2022 in Kuopio PDS. Edentulous patients were excluded. The final sample size of 16 dentists and 10 patients per dentist was determined based on interim analyses, during which confidence intervals from logistic regression models were visually assessed after an initial selection of 10 dentists and 10 patients each.

We excluded patients under 18 and over 40 years of age to make the patient group more homogeneous and to exclude patient groups who seldom undergo a bitewing examination (children, adolescents, and elderly). For each patient, we collected the following individual-level data: year of birth, sex, teeth with dentin caries, missing teeth, filled teeth, surfaces with dentin caries, missing surfaces, filled surfaces, the number of tooth surfaces with active enamel caries, and Community Periodontal Index (CPI) for each sextant. Caries risk assessments, if conducted, are not systematically recorded in electronic health records and are therefore unavailable.

Additionally, we collected specific data on whether bitewing radiographs had been taken during the examination and the date of the previous bitewing radiographs. Other types of intra- or extraoral radiographs were not considered as taken for caries diagnostics and, thus, were not included in the data.

Lesion Selection

We included lesions involving proximal surfaces registered as enamel or dentin caries lesions by the examining dentist and verified their visibility on bitewing radiographs taken during the examination. Lesions were excluded if the tooth had undergone root canal treatment prior to the bitewing radiograph, if the bitewing radiograph was unavailable, if the lesion was not visible in enamel or dentin on the bitewing radiograph, or if the radiograph was unsuitable for diagnostics or measurements. We categorized the caries lesions as primary or secondary based on whether there was visible restoration on the surface. Teeth were classified as molars or premolars.

Radiolucency Measurement

For each eligible lesion, the depth of each radiolucency was measured in millimeters from tooth surface to the radiolucent area closest to the pulp using a precalibrated/default digital measuring tool in Planmeca Romexis software (Planmeca, Helsinki). The thickness of the enamel and the dentin were also measured, allowing us to calculate the depth of radiolucency as a percentage of dentin thickness as shown in Figure 1. The measurements were performed by M.H.M. and L.O. under the supervision of more experienced members of the research group (J.L. and E.R.).

Fig. 1.

a Schematic illustration of a single tooth cross-section, highlighting anatomical layers including enamel, dentin, and pulp, rendered in grayscale. A red line extends horizontally from the left margin of the tooth toward the center, transitioning into a yellow line at the periphery. b Radiographic image displaying three adjacent teeth, with the central tooth marked by a similar red-to-yellow horizontal line, indicating a comparable anatomical or procedural reference point.

Restorative Treatment

For each evaluated surface, the following treatment decisions were identified from the electronic health record system during the treatment period: (a) restoration involving the surface, (b) root canal treatment, (c) extraction, and (d) no invasive treatment. The treatment period was defined as the time between the oral examination in 2022 and the following oral examination if it occurred prior to the start of data collection in March 2024. Based on this, a binary variable was created to represent whether the surface was treated restoratively or noninvasively. Unfortunately, other clinical information regarding the lesions or treatment decisions, such as whether the lesions were cavitated, non-cavitated, or presented as shadows (a large dentin caries lesion creates a discoloration although enamel is intact or microcavitated), was not registered in electronic health records and is thus not available. Furthermore, the clinical and radiological caries status is registered on the same sheet in the electronic health record system and cannot be distinguished from one another in retrospect.

Statistical Analysis

We analyzed the basic characteristics of patient- and surface-level data using frequencies, distributions, and means. The association between the depth of radiolucency in dentin and treatment decisions was estimated using linear regression analysis. These models included binary treatment decision variables (restorative treatment or not) as an outcome and the depth of radiolucency in dentin as an exposure and were adjusted for age, sex, tooth type (molar/premolar), primary/secondary lesion, and dentin thickness. We also used cluster-robust standard errors taking account of the clustering of observations by dentists, patients, and teeth. Then, with these models, the average predicted probability of restorative treatment according to the depth of radiolucency in dentin was estimated. Additionally, we investigated whether this relationship depended on the tooth type, primary/secondary lesion, or dentin thickness by including the interaction term between them and the depth of radiolucency in the dentin.

Results

Patient Demographics and Bitewing Radiographs

Our study included 160 patients who had been examined by 16 dentists. The dentists were 28- to 58-year-olds. There were 11 female dentists and 5 male dentists. Of the patients, 72 (45%) were men and 88 (55%) were women. For all but 3 patients’ bitewing radiographs had been taken from both sides, resulting in 253 bitewing radiographs. Bitewing radiography had been used slightly more if patients had a higher number of filled teeth or surfaces (p = 0.06). Otherwise, there were no significant differences in background factors between patients who had and had not undergone a bitewing examination (Table 1).

Table 1.

Patient background characteristics according to whether a bitewing examination had been performed or not during oral health examination in 2022

Characteristic	Bitewing examination not performed	Bitewing examination performed	p value1
	median (Q1, Q3); n (%)	median (Q1, Q3); n (%)
N	32	128	​
Age	30.5 (26.5, 36.0)	31.0 (27.0, 35.0)	0.93
Women	21 (66%)	67 (52%)	0.18
Teeth with dentin caries	0 (0, 2)	1 (0, 2)	0.25
Missing teeth	2.5 (0, 4)	1.0 (0, 4)	0.54
Filled teeth	3 (1, 5)	4 (1, 8)	0.06
Surfaces with dentine caries	0 (0, 2)	1 (0, 2)	0.20
Missing surfaces	12 (0, 20)	5 (0, 20)	0.60
Filled surfaces	4 (2, 8)	8 (2, 16)	0.06
Teeth with active enamel caries	4 (1, 7)	5 (1, 8)	0.31
Surfaces with active enamel caries	4 (1, 8)	5 (1, 10.5)	0.34
Missing	1	0	​
Year of the previous bitewing examination	2016 (2014, 2018)	2016 (2014, 2018)	0.98
Missing	14	59	​
CPI1	2 (1, 2)	2 (1, 3)	0.98
Missing	1	9	​
CPI2	1 (1, 2)	1 (0, 2)	0.76
Missing	1	7	​
CPI3	2 (1, 2)	2 (1, 3)	0.83
Missing	1	8	​
CPI4	2 (1, 2)	2 (1, 3)	0.35
Missing	1	7	​
CPI5	2 (1, 2)	2 (1, 2)	0.83
Missing	1	7	​
CPI6	2 (1, 2)	2 (1, 3)	0.45
Missing	1	7	​
Bleeding on probing (%)	31 (16, 51)	31 (12, 50)	0.70
Missing	9	43	​

CPIX, Community Periodontal Index in X sextant.

¹Wilcoxon rank sum test; Pearson's χ² test.

Of the 16 dentists in our study, eight had performed bitewing radiography on all 10 patients. Whereas four dentists had performed bitewing radiography in 7 to 8 patients, and another four dentists had performed bitewing radiography in 2 to 5 patients.

Caries Lesion Characteristics

A total of 394 proximal caries lesions were registered by the examining dentists and were measurable on the available radiographs. A slight majority (51%) of the lesions were in molars (n = 199, 51%). Most lesions (n = 361, 92%) were primary caries lesions.

One radiograph was of poor quality, and two lesions could not be measured due to critical overlap with adjacent teeth. In addition, one tooth was too severely damaged to allow measurement. Radiographic assessment revealed that 6% (n = 25) of radiolucencies extended to the outer half of enamel, 51% (n = 201) to the inner half of enamel, 27% (n = 107) to the outer third of dentin, 12% (n = 47) to the middle third, and 4% (n = 14) to the inner third of dentin. In total, 32 (8%) radiolucencies extended to the inner half of the dentin.

The average depth of radiolucency into dentin was 11% in premolars and 15% in molars. The mean enamel thickness was 1.0 mm (SD = 0.24) in premolars and 1.1 mm (SD = 0.27) in molars. Whereas the mean dentin thickness was 2.5 mm (SD = 0.41) in premolars and 3.0 mm (SD = 0.55) in molars.

Provided Treatments

Of the 394 caries lesions, 113 (29%) had received restoration, five teeth had received root canal treatment and restoration, and one tooth had been extracted (Table 2). Data on putative treatments for one lesion were missing. Lesions in the molars had been restored slightly more frequently than those in the premolars (Table 2, p = 0.10). The distribution of registered lesions according to the depth of radiolucency and whether the lesion had been treated restoratively during the treatment period is shown in Figure 2. Five percent (n = 11) of the enamel radiolucencies, 47% (n = 50) of the radiolucencies in the outer dentin third, 83% (n = 39) of the radiolucencies in the middle third, and all (n = 15) radiolucencies in the inner third had been restored. Five root canal treatments had been performed on teeth in which the radiolucency extended to the inner half of the dentin. In addition, 28 caries lesions with radiolucency extending to the inner half of the dentin were restored. Whereas, in the tooth that had been extracted, the radiolucency reached the pulp.

Table 2.

The treatments provided for 394 registered caries lesions

Provided treatment	Premolars (n = 195), n (%)	Molars (n = 199), n (%)
No invasive treatment	146 (75)	133 (67)
Restoration involving the surface	48 (25)	65 (33)
Root canal treatment	2 (1.0)	3 (1.5)
Extracted	0 (0)	1 (0.5)
Missing information	1 (0.5)	0 (0)

Surfaces which were restored and root canal treated are counted twice.

Fig. 2.

The distribution of registered caries lesions according to the depth of radiolucency in dentin (%) and whether the lesion had been treated restoratively during the treatment period. The data were based on 392 lesions which were either restored or had not received invasive treatment during the treatment period.

Association between the Depth of Radiolucency in Dentin and the Restorative Treatment Decision

When adjusted for age, sex, tooth type (molar/premolar), primary/secondary lesion, and dentin thickness, the predicted probability of restorative treatment strongly increased relative to the depth of radiolucency in the dentin (shown in Fig. 3a). When these covariates were held at their observed means or proportions, the analyses indicated that the average predicted probability of a lesion being restored was approximately 8% when the radiolucency was at the dentinoenamel junction, increased to approximately 75% when the radiolucency extended to the middle third of the dentin, and reached 100% when the radiolucency extended to the inner third of the dentin.

Fig. 3.

Predicted probability of restorative treatment by depth of proximal radiolucency in dentin for the total sample (a), molars and premolars (b), and primary and secondary lesions (c). Shaded area represents 95% confidence interval. The models were adjusted for adjusted for age, sex, tooth type (molar/premolar), primary/secondary lesion, and dentine thickness. The data were based on 392 lesions which were either restored or had not received invasive treatment during the treatment period.

The relationship between restorative treatment probability and depth of proximal radiolucency in dentin was similar in premolars and molars (shown in Fig. 3b). Secondary caries lesions were more likely to be restored than primary caries lesions (shown in Fig. 3c).

Discussion

This study included 160 adult patients examined by 16 dentists. Bitewing radiography was performed in four out of five examinations. There were no notable differences in age, sex, caries experience, or periodontal status between the patients who had undergone bitewing examination and those who had not. The dentists had registered caries lesions on 394 proximal surfaces. A slight majority of the caries lesions were in molars, and most of the lesions were primary caries lesions. Most of the radiolucencies associated with the lesions were confined to the enamel or the outermost third of dentin, while one out of six of the radiolucencies extended to the middle third of dentin, and only one in twenty-five into the inner third of dentin. Regression analyses showed that the probability of invasive treatment increased significantly with depth of radiolucency. Regression analyses indicated that the predicted probability of restorative treatment was 8% when the radiolucency reached the dentinoenamel junction and 75% when the radiolucency reached the border of the outer and middle thirds of the dentin. All lesions extending to the inner third of the dentin had been restored, root canal treated, or extracted.

Since non-cavitated lesions seldom progress to cavitated or symptomatic stages in low-caries populations such as the Finnish, it is reasonable to consider cavitation as a threshold to commence restorative treatment [8, 10, 20, 21]. However, many cavitations are not undermining but saucer-shaped, and thus their cleanability and prognosis may be somewhat close to those of non-cavitated caries lesions [7, 22, 23]. Although there is some variability in the findings [24], approximately one-third of the lesions with radiolucencies reaching the outer third of the dentin are cavitated, whereas approximately half of those reaching the middle third represent cavitated caries [7, 25]. Our findings indicate that when available factors (age, sex, tooth type, primary/secondary lesion, and dentin thickness) potentially affecting the treatment approach were controlled for, the predicted probability of restorative treatment was approximately 75% for lesions just reaching the border of the outer and middle third of dentin. Only 5% of caries lesions whose radiolucency was limited to enamel had been restored. Interestingly, 5% is also the proportion of cavitated caries lesions of all approximal radiolucencies limited to enamel [7, 25]. Although we were unable to determine which of the restored lesions were in fact non-cavitated and that there is uncertainty in the cavitation probability by radiographic depth, it seems likely that a considerable number of non-cavitated caries lesions had been restored. Unfortunately, only one randomized controlled trial that has studied patient-centered outcomes of including bitewing radiographs in caries diagnostics, and the trial included 3-6-year-olds limiting the generalization of the results to adult populations such as ours [26]. Thus, the indications and interpretation of bitewing examinations in the adult population remain debatable.

Furthermore, we found wide variation in dentin thickness, which is a concern for assessing the cavitation status of a caries lesion according to radiolucency by dentin portions. Although measuring radiolucencies in millimeters has been possible for decades, the radiolucencies have typically been categorized according to their extent [7, 18, 19, 27]. If a study comparing cavitation status and bitewing radiolucency extent in millimeters, perhaps in conjunction with artificial intelligence, the detection accuracy of cavitated caries lesions based on bitewing radiography might improve [28]. Especially, as the assessment of radiolucency depth varies between dentists [29, 30].

Our findings are in accordance with a survey among Norwegian dentists in 2009, in which two-thirds reported that they would commence restorative treatment before the radiolucency reaches the middle third of the dentin [14]. Moreover, the Norwegian survey found that 7% of dentists would provide restorative treatment to lesions radiographically at the dentinoenamel junction, while the predicted probability for caries lesions with radiolucency at the dentinoenamel junction was 8% in our study.

Compared to earlier studies on adolescents in Norway and Denmark using similar retrospective evaluation of radiographs and restorative treatment decisions as ours, the dentists in our study had a similar threshold for restorative treatment with the Norwegian dentists but lower threshold for restorative treatment than the Danish dentists [18, 19]. Among 15-year-olds in Oslo in 1996, 48% of lesions in the outer dentin half were restored, and among 17-year-olds in Hillerød (Denmark) in 1984–1985, only 6% of lesions in the outer dentin half were restored [18]. In our study, 54% of the lesions in the outer half of the dentin had been restored. The difference between Finland and Denmark is surprising considering that both countries represent low-caries populations [31, 32], where the progression of non-cavitated lesions into cavitated lesions is slow and rare [20], and minimally invasive dentistry is emphasized in dental education and highlighted in continuing education [7, 10]. However, the difference between the threshold for restorative treatment between our study and the Danish study might result partly from the age difference of the participants in the studies or the decade the studies have been performed, but more importantly the differences in the clinical caries assessment protocols used in Denmark and Finland. The Danish caries assessment system emphasizes thorough clinical examination and assessment of lesion activity, which may influence the treatment decisions compared to Finland. In fact, the electronic health record system used in our study does not enable activity status registering for dentin caries lesions. Further studies that incorporate the clinical examination are necessary to investigate the indications the Finnish dentists have for restorative treatment.

Three out of four restorations were performed on caries lesions whose radiolucency was limited to enamel or inner half of the dentin. This relatively low threshold for restorative treatment is surprising because neither the Finnish Current Care Guidelines on filling therapy nor the cariology course book used in three out of four dental schools in Finland (including the one in Kuopio) recommend restorative treatment of proximal caries lesions that do not reach the inner half of the dentin [10, 33]. Perhaps the dentists in our study assessed that many of the detected lesions were active and that they would cavitate or become symptomatic before the next recall examination, thus justifying preemptive restorative treatment. Caries progression may also worry Finnish dentists, especially as only two out of three Finnish adults visit a dentist within a 2-year timeframe [34]. The low dentist examination frequency may also explain our finding that there had been a median of 6 years since the previous bitewing examination. This may have lowered the threshold of the dentists in our study to perform bitewing examinations. In addition, intolerance to uncertainty and fear of patient complaints may have led to the vast use of bitewing examination and commencing invasive treatment at a low threshold. Low threshold for restorative treatment brings excess burden to dentists, low-value treatment to patients, and costs to society [30]. The electronic health record system used in Kuopio does not allow recording the cavitation status of any caries lesions or the activity status of dentin caries lesions. In addition, the widely used ICD-10 classification only categorizes caries lesions as either enamel or dentin caries, which may lead to the misconception that dentin caries per se is an indication for restorative treatment. Hopefully caries classifications in the future electronic health record systems will support evidence-based practice and reasoning required to apply it better.

Another putative explanation for the restorative therapy of caries lesions whose radiolucency did not extend beyond the outer third of dentin is related to the fact that caries lesions extend 0.8 mm further than the radiolucency in the bitewing radiograph [35]. Perhaps, the dentists in our study interpreted the Finnish Current Care Guidelines according to the true caries lesion depth and not the depth of radiolucency. Overall, it appears that following or arresting caries progression was rarely chosen as a treatment strategy when radiolucency extended beyond the outer third of the dentin, even though dental caries is a biofilm-mediated disease, and disturbing the maturation of biofilm on the tooth surface by toothbrushing prevents the progression of caries lesion regardless of the lesion depth [10].

Dentists in Finnish PDS receive additional compensation for various procedures and examinations, excluding bitewing examinations, which lack financial incentives. While restorative treatments are compensated for, the amount is modest compared to other procedures and examinations. Our data do not provide insights into why bitewing radiography has been used in the majority of examinations. However, considering that the number of bitewing radiographs taken in Finland has increased by approximately 20% between 2015 and 2021 [36], it would be important to investigate the patterns and variations in the use of bitewing radiography and their associations with restorative treatment performed on a larger scale in the future.

A strength in a retrospective study of electronic health records such as ours is that all treatment decisions made by the random subset of dentists working for City of Kuopio PDS have been included in the analyses, which reduces the selection bias of participants. However, this type of study protocol is not without limitations. We did not examine the patients clinically and relied only on electronic health records. They did not include the cavitation nor activity status of the lesions, caries risk assessment of the patient, or reliable records on symptoms; therefore, these data could not be included in this study. Therefore, determining the “correct” diagnosis or clinical treatment decision was not possible, and we could only discuss the findings by considering, e.g., cavitation status information from other studies. Combining questionnaire studies and pseudo-patient studies would be critical to determine how dentists decide which caries lesions they consider indicated for filling therapy. Additionally, we were unable to evaluate the depth of the caries lesions in patients who had not undergone a bitewing examination. Therefore, we could not evaluate the relationship between the clinical restorative treatment decision and radiolucency depth or cavitation status of the caries lesions in the patients who had not undergone bitewing examination. The patients were recall examination patients and thus presumably asymptomatic. Therefore, the findings cannot be generalized to acute or symptomatic patients. Moreover, because there was generally a delay of several months between examination and treatment visits, it may be that some of the caries lesions that were initially decided to be monitored only could have been restored due to lesion progression during the delay. This may have resulted in a slight overestimation of the probability of restorative treatment of active shallow caries lesions.

Conclusions

This retrospective study based on Finnish electronic health records indicates that the probability of restorative treatment increases significantly with the depth of radiolucency measured from bitewing radiographs taken during recall dental examinations. The predicted probability of restorative treatment was approximately 8% for lesions with radiolucency reaching the dentinoenamel junction and increased to approximately 75% when the radiolucency extended to the middle third of the dentin.

Statement of Ethics

This study protocol was reviewed and approved by the Wellbeing Services County of North Savo (approval #5553947). A statement from the University of Eastern Finland Committee on Research Ethics was not required as the study was based on medical records in accord to Finnish data protection act (1050/2018, § 31). Written informed consent from participants was not required in accordance with local/national guidelines. Secondary usage of medical records in scientific research is allowed according to Finnish act on secondary use of health and social data (552/2019, § 38). This study was performed according to the Declaration of Helsinki.

Conflict of Interest Statement

The authors have no conflicts of interest to declare.

Funding Sources

This study was not supported by any sponsor or fund.

Author Contributions

M.H.M.: conceptualization; resources; investigation; methodology; writing – original draft. E.R.: conceptualization; supervision; methodology; writing – review and editing; formal analysis; and visualization. L.O.: conceptualization; investigation; methodology; writing – review and editing. K.M.L.: conceptualization; supervision; writing – review and editing. J.L.: conceptualization; supervision; writing – review and editing; methodology; and project administration.

Funding Statement

This study was not supported by any sponsor or fund.

Data Availability Statement

The data that support the findings of this study are available from the Wellbeing Services County of North Savo. Restrictions apply to the availability of these data, which were used under license for this study. The dataset may be requested from the corresponding author after obtaining permission from the Wellbeing Services County of North Savo.