Do I Need a Filling? What the Evidence Says

Reviewed 2 September 2026. Twenty-six records, each pulled and re-checked against Europe PMC before publication: one registry study of 111,771 teeth, four clinical trials, three umbrella or overview reviews of systematic reviews, six diagnostic-accuracy systematic reviews and meta-analyses, two guideline-level papers, four cross-sectional validation studies, three surveys of clinical decision-making and four reviews of non-operative and micro-invasive care. Every number below is printed in the abstract of the record cited beside it; where the abstract says the evidence is weak, that is written here too. Written for a person staring at a dentist’s treatment plan at the pharmacy counter and for clinicians who have to write the note that justifies it. Not medical or dental advice.

The short answers

• Pain is not the test. In a school survey of 363 children aged 6 to 10, 28.2% had at least one tooth with exposed pulp, and pulp involvement was associated with worse oral-health-related quality of life (OR 1.49, 95% CI 1.22 to 1.81) (Fonseca et al., Clin Oral Investig 2026-02-01, PMID 41729310). Deep, cavitated disease can be painless for a long time; a sensitive white spot can be harmless. If you are waiting for pain as the signal to book, you are using the worst available instrument.
• Not every dark spot or every “cavity” on a chart means a filling. In a retrospective study of 160 adults examined by 16 randomly selected dentists, most of the 394 proximal lesions registered were radiographically confined to enamel (57%) or the outer third of dentin (27%), yet the predicted probability of restorative treatment for a lesion reaching the dentino-enamel junction was 8% and for one at the boundary between outer and middle dentin 75% — the authors call the latter “unexpectedly high” for lesions still in enamel or outer dentin (Muhonen et al., Caries Res 2025-11-01, PMID 41196837). Read that sentence twice: the number that decides a drill is often the depth on an X-ray, and thresholds differ between dentists more than patients assume.
• The same image plus a bit of clinical context changes the plan. In a cross-sectional reading study of 212 bitewings assessed twice, two weeks apart — once with the radiograph alone and once with the radiograph plus the Caries Associated with Restorations or Sealants score and lesion activity — the distribution of decisions about whether a restoration was needed changed significantly (p < 0.001), with more decisions favouring intervention when clinical information was added; agreement of scores was moderate to good (κ = 0.749) (Maydana et al., Braz Dent J 2024-01-01, PMID 39699493). A filling decision is a judgment about a process, not a reading of a picture.
• The tools are better than folklore, worse than marketing. Unaided visual examination with ICDAS criteria showed specificity 100% and sensitivity 92% against histology on extracted teeth, while adding 6x magnification produced sensitivity 100% but specificity 50% — more false positives (Rekha et al., Int J Clin Pediatr Dent 2023-09-01, PMID 38078031). Quantitative light-induced fluorescence reached AUC 0.94 to 0.98 for incipient occlusal lesions and pooled in vivo sensitivity/specificity 0.86/0.82 occlusally and 0.74/0.82 approximally (Lee et al., Sci Rep 2025-11-01, PMID 41238662). Optical coherence tomography detected early approximal lesions better than digital radiography in vitro (sensitivity 0.66 to 0.91 vs 0.46 to 0.82; specificity 0.67 to 0.92 vs 0.33 to 0.58) (Strumpski et al., Caries Res 2025-01-01, PMID 39999820). And an artefact, not a cavity, explains some of the anxiety: triangular-shaped radiolucencies were present in 26.44% of maxillary molars in the pooled estimate and led to false-positive diagnoses of caries or fracture in about 13% of observations (heterogeneity I² > 90%, certainty low to very low) (Dioguardi et al., Oral Radiol 2026-04-01, PMID 41331196).
• Artificial intelligence is not the referee yet. Across 19 studies, AI models showed consistently higher sensitivity than human examiners for early enamel lesions (0.56 to 0.96 vs 0.21 to 0.83), but 12 of the 19 studies were judged at high risk of bias in at least one domain, and the authors note the performance reflects agreement with expert radiographic interpretation “rather than validation against biological truth” (Banihashem Rad et al., Caries Res 2026-06-01, PMID 42313715). An umbrella review of 17 systematic reviews found pooled sensitivity 0.76 to 0.94 and specificity 0.85 to 0.91 and concluded explicitly that current evidence does not support AI as a stand-alone diagnostic tool (Bittar et al., J Dent 2026-07-01, PMID 42537896). A separate meta-analysis that could use only 20 of the 137 primary studies inside 14 reviews reported pooled sensitivity 0.85 (95% CI 0.83 to 0.93), specificity 0.90 (0.85 to 0.95), AUC 0.86, with positive post-test probability 79% and negative 6% (Arzani et al., PLoS One 2025-01-01, PMID 40802701).
• There is a middle column between “watch it” and “drill it”, and it has trials behind it. A randomised trial in adults with non-cavitated occlusal lesions (36 participants, 64 lesions) found a bioactive sealant reached 93.75% success at 24 months versus 65.62% for a fluoride-releasing sealant, RR 0.18 (95% CI 0.044 to 0.756; p = 0.019) (Abdelsalam et al., Sci Rep 2025-12-01, PMID 41381619). A triple-blind randomised placebo-controlled trial of silver diamine fluoride on initial approximal lesions in adults found more than 90% of lesions arrested in both arms at 12 months, with the SDF arm better only at 6 months by digital subtraction radiography (p = 0.05) and no difference at 12 months (Jaradat et al., J Conserv Dent Endod 2025-09-01, PMID 40964644). Resin infiltration for non-cavitated occlusal lesions rests on eight studies of which only two are randomised trials, with clinical certainty graded low (Miranda et al., J Clin Med 2026-02-01, PMID 41682994).
• When drilling does happen, less is now the guideline-supported default. The EFCD-ESE-ORCA S3-level guideline concludes that evidence supports selective or stepwise caries removal over non-selective removal to reduce pulp exposure, that routine cavity liners show no consistent benefit and are not routinely recommended, that both direct pulp capping and pulpotomy work in teeth without irreversible pulpitis, and that hydraulic calcium silicate cements outperform calcium hydroxide — with certainty ranging from very low to moderate (Schwendicke et al., Caries Res 2026-04-01, PMID 42018467). An umbrella review of seven systematic reviews found no single removal technique superior across all outcomes and rated overall certainty low to very low (Aldakhil et al., J Dent 2026-10-01, PMID 42309376).
• The interval you keep is itself a treatment. In a randomised trial of 224 preschool children at high caries risk, those on an 8-month recall were 2.5 times more likely to develop new cavitated lesions over 30 months than those reviewed every 4 months (Cordeschi et al., Int J Paediatr Dent 2026-01-01, PMID 40984027).

What “you need a filling” is actually a shorthand for

Clinically, a restoration is justified when four things line up, and every one of them is separately contestable. The lesion has progressed far enough into dentin that the enamel surface no longer holds; the surface is not cleansable, so biofilm stays on it; the process is active rather than arrested; and the tooth is worth restoring rather than monitoring or replacing. The classification systems used in the research above exist precisely to make those four judgements explicit. ICDAS describes what the examiner sees (code 1 to 6 for visual change and cavitation, 7 for a post-restoration or sealed surface), and its management companion sorts the finding into preventive, non-operative or operative need — in the validation work across 3,076 children in three countries, a short version of the index agreed with the full version on treatment need with kappa above 0.833 in all dentitions, and areas under the ROC of more than 90% in primary, 89% in permanent and 86% in mixed dentition, with the weakest performance exactly where patients care most: telling non-operative from operative need (ElSalhy et al., Community Dent Oral Epidemiol 2022-12-01, PMID 34897763). A separate cross-sectional validation in 1,200 schoolchildren found the short version had 99.6% sensitivity and 98.6% specificity for moderate and advanced lesions, with early non-cavitated lesions slightly under-reported (Jain et al., Cureus 2025-09-01, PMID 41058820).

The practical translation for you: “a small cavity” is not one thing. A stain in a groove with an intact surface, a white or brown spot where mineral is being lost but the surface is still closed, a lesion that has caved in and catches a floss strand, and a tooth with a night-time ache are four different biological states, and in the published guidance the first two are addressed with non-operative or micro-invasive strategies, chosen on the activity and depth of the lesion rather than on the presence of a hole (Santos et al., J Clin Med 2026-08-01, PMID 42652888) (Todorova et al., J Clin Med 2026-02-01, PMID 41753221). The reason a single word covers all four is that the treatment most people remember from their own childhood — drill, fill — works for the last two and is over-used for the first two.

How the decision really gets made, and how much it varies

The Finnish register study is the most honest number in this whole article, because it measured dentists as they actually work rather than as guidelines describe them: 160 patients aged 18 to 40, 16 randomly selected dentists in public care, bitewings taken for 128 of the 160 (80%), every proximal lesion the dentists registered measured for radiolucency depth, and the treatment decision modelled with cluster-robust standard errors. The result is a steep curve — 8% probability of restoration for a lesion reaching the dentino-enamel junction, 75% once the radiolucency crosses the outer-to-middle-third boundary — and the authors’ own conclusion is that probability was unexpectedly high for lesions still confined to enamel or outer dentin (Muhonen et al., Caries Res 2025-11-01, PMID 41196837).

Variation goes the other way too. Among 442 Thai dentists surveyed with case-based questions, agreement with the ICCMS recommendations for occlusal lesions in primary teeth was 39.6% at code 2, 53.9% at code 3 and 42.5% at code 4; in permanent teeth 34.4%, 46.8% and 39.6%. For approximal lesions the pattern inverted: agreement was 69.6% at RA1, 78.0% at RA2, then collapsed to 12.4% at RA3 — a stage where many dentists treated a lesion the guideline would not — and 99.5% at RB4. Years since graduation, postgraduate training and practice type predicted the choices (Ramayasinpong et al., Int J Paediatr Dent 2025-07-01, PMID 39890991).

So when you ask “do I need a filling”, three things are being decided before the anaesthetic: how deep the lesion is judged to be, what the examiner thinks the activity and cleansability mean, and where that particular dentist learned to draw the line. This is why the same second molar can be sealed by one clinician and drilled by another, and why “get a second opinion” is not an insult to your dentist but a rational response to measured variation (Maydana et al., Braz Dent J 2024-01-01, PMID 39699493) (Ramayasinpong et al., Int J Paediatr Dent 2025-07-01, PMID 39890991).

How good is the examining, actually

Method	Performance as reported	What it means at the chair
Visual ICDAS, unaided (100 extracted teeth, histology as reference) (Rekha et al., Int J Clin Pediatr Dent 2023-09-01, PMID 38078031)	Specificity 100%, sensitivity 92%; inter-examiner κ 0.695	What an experienced eye calls cavitated is almost always cavitated
The same with 6x magnification (Rekha et al., Int J Clin Pediatr Dent 2023-09-01, PMID 38078031)	Sensitivity 100%, specificity 50%; correlation with histology slightly lower than unaided	Zooming in manufactures false positives on occlusal surfaces
Short ICDAS-II for treatment need (3,076 children, three countries) (ElSalhy et al., Community Dent Oral Epidemiol 2022-12-01, PMID 34897763)	κ > 0.833 versus full ICDAS; AUC > 0.90 primary, 0.89 permanent, 0.86 mixed; worst at the non-operative versus operative boundary	Good at “does this child need drilling at all”, weaker at “which tooth”
Short ICDAS-II in 1,200 schoolchildren (Jain et al., Cureus 2025-09-01, PMID 41058820)	Sensitivity 99.6%, specificity 98.6%, weighted κ 99.0% for moderate to advanced lesions; early non-cavitated lesions slightly under-reported	Advanced disease gets caught; early disease is under-counted by design
Laser fluorescence versus ICDAS on smooth surfaces (60 patients, 161 non-cavitated lesions, 32 sound teeth) (Shaalan et al., J Conserv Dent 2023-03-01, PMID 37205891)	Overall accuracy 84.45% (sensitivity 87.58%, specificity 96.87%); accuracy 74.15% when only the first visual change counted; 100% at code 2	Devices replicate the visual exam; they do not outrank it
Quantitative light-induced fluorescence, 17 studies (Lee et al., Sci Rep 2025-11-01, PMID 41238662)	In vivo AUC 0.94 to 0.98 for incipient occlusal lesions; pooled sensitivity/specificity 0.86/0.82 occlusal, 0.74/0.82 approximal	Useful for early occlusal lesions; weaker between teeth
Optical coherence tomography vs bitewing radiography, in vitro (Strumpski et al., Caries Res 2025-01-01, PMID 39999820)	Sensitivity 0.66 to 0.91 vs 0.46 to 0.82; specificity 0.67 to 0.92 vs 0.33 to 0.58; moderate to substantial repeatability	A promising research tool for early proximal lesions, not yet standard care
Radiographic artefacts, systematic review with meta-analysis (5 studies of 640 reports) (Dioguardi et al., Oral Radiol 2026-04-01, PMID 41331196)	Non-carious triangular radiolucency in 26.44% of maxillary molars (270/1,021); false-positive diagnoses in ~13% of readings (60/464); I² > 90%, low to very low certainty	Some “cavities” are optical effects; correlation with the clinical exam is mandatory
Dentists’ ability with ICDAS (318 respondents from 50 clinics and 393 students) (Al Dhubayb et al., Clin Cosmet Investig Dent 2021-01-01, PMID 34526823)	62.6% knew the system; all groups struggled with activity assessment at code 2; detection ability score 38.7% for general practitioners, 53.8% and 57.6% for senior students	Early detection is genuinely hard, and training changes it

What pain tells you, and what it hides

Two findings from school surveys set the frame. First, severity of impact tracks the stage, not the symptom: in 643 caregiver-child pairs with children aged 0 to 5, caries was present in 72.6%, at least one oral-health impact was reported for 48.2% of children, children with extensive lesions (ICCMS stage C) had significantly worse quality-of-life scores than caries-free children, while moderate lesions (stage B) only showed an effect when present on both anterior and posterior teeth (RR 1.54, p = 0.045) (Díaz et al., Braz Oral Res 2025-01-01, PMID 41379130). Second, by the time the pulp is involved, quality of life is already worse: 28.2% of 6- to 10-year-olds had at least one pulp-exposed tooth, and pulp exposure was associated with impact on the child’s reported quality of life (OR 1.49, 95% CI 1.22 to 1.81), with additional impact where maternal schooling was lower (Fonseca et al., Clin Oral Investig 2026-02-01, PMID 41729310).

Pain that does change the plan is a specific pattern: lingering pain after hot or cold, pain that wakes the patient, pain on biting that persists after the stimulus is removed. Those are pulp questions more than enamel questions, which is why the deep-caries guideline ties treatment to pulpal diagnosis rather than to how dark the spot looks, and why it separates teeth “without irreversible pulpitis” (where capping or pulpotomy works) from teeth with signs of it (where pulpotomy is only an alternative to pulpectomy) (Schwendicke et al., Caries Res 2026-04-01, PMID 42018467). If the question in your head is actually “filling or root canal”, the survival data are on our root canal versus extraction page; the short version is that keeping and replacing land within about a percentage point of each other at ten years, so the deciding variables are restorability and cost, not the myth that a root canal always saves the tooth.

The middle column: what non-drilling options score

The menu these reviews describe is wider than the four options above — fluoride therapy, remineralising agents such as casein phosphopeptide-amorphous calcium phosphate, biomimetic peptides, sealants, infiltration, chemo-mechanical caries removal, air and ultrasonic abrasion, erbium lasers — and its own conclusion is that minimally invasive dentistry is a philosophy of preserving structure rather than a guarantee, limited in practice by cost, equipment and how technique-sensitive it is (Todorova et al., J Clin Med 2026-02-01, PMID 41753221).

Sealing an initial lesion rather than cutting it. In the randomised trial in adults, both sealants held up, and the difference was large: 93.75% versus 65.62% success at 24 months, RR 0.18 (95% CI 0.0437 to 0.756), p = 0.019, with failures in the control arm driven by loss of retention or progression; the authors frame this as supporting the recommendation to seal initial non-cavitated lesions, with the caveat that the sample was 36 participants (Abdelsalam et al., Sci Rep 2025-12-01, PMID 41381619).

Infiltrating a proximal lesion rather than filling it. The evidence-based review of resin infiltration concludes that the strongest support is for non-cavitated proximal lesions in permanent teeth, particularly enamel and selected outer-dentin lesions, where long-term studies show reduced radiographic progression, and states explicitly that infiltration is a selective micro-invasive option guided by lesion assessment and caries risk “rather than a universal substitute for prevention or restoration” (Santos et al., J Clin Med 2026-08-01, PMID 42652888). On occlusal surfaces the same idea rests on far less: eight studies, only two of them randomised, certainty low (Miranda et al., J Clin Med 2026-02-01, PMID 41682994).

Arresting an open lesion chemically. In the systematic review of non-restorative cavity control for dentinal caries in primary teeth, nine randomised trials were included, most showing that 38% silver diamine fluoride arrests dentinal caries even with a 30-second application, one reporting higher survival for the Hall crown than for SDF, and acceptance rated high by children and parents despite staining (Alwusaybie et al., Cureus 2026-01-01, PMID 41625031). The adult randomised trial of SDF on initial approximal lesions is the useful counterweight: more than 90% of lesions arrested in both the SDF and placebo arms at 12 months, so the drug effect was confined to the first 6 months (Jaradat et al., J Conserv Dent Endod 2025-09-01, PMID 40964644). When someone sells you “SDF dissolves cavities”, the trials describe arrest, not disappearance — and in low-progression environments placebo catches up.

Leaving infected dentin under a temporary seal and coming back. Stepwise removal is supported for deep lesions to protect the pulp, and a Swedish registry study of 111,771 teeth treated with stepwise removal reported a treatment success rate above 60% at 110 months, with better prognosis for premolars, patients under 40 and those with low or medium caries risk (Malmberg et al., Caries Res 2026-03-01, PMID 41843704). The umbrella review of seven systematic reviews tempers all of this: no technique was superior across all outcomes, selective removal was favoured when exposure risk was high but avoiding exposure did not ensure pulp health, post-operative pain was slightly lower with non-selective than stepwise removal, and certainty was low to very low (Aldakhil et al., J Dent 2026-10-01, PMID 42309376).

Whole-programme evidence. The Danish Nexø model, evaluated over 25 years of national register data, replaced “drill every lesion that appears” with structured prevention and individually dosed recall intervals built from four risk criteria; the analysis reports a significant fall in mean decayed/missing/filled surfaces relative to national data and concludes the shift was cost-effective even with limited sealant use (Ekstrand et al., Children (Basel) 2026-03-01, PMID 41897144). That is the population version of your question: the reason a lesion never became a filling somewhere else in the world is scheduling and fluoride, not luck.

When drilling is the right answer

• The surface has actually caved in and cannot be kept clean: the review of micro-invasive pathways puts management decisions on cavitation status, activity and cleansability, not on the colour in the mirror (Santos et al., J Clin Med 2026-08-01, PMID 42652888) (Todorova et al., J Clin Med 2026-02-01, PMID 41753221).
• Radiolucency is deep: in the register study, probability of restoration jumps to 75% at the outer-to-middle dentin boundary, and this is the point where most clinicians stop talking about non-operative options (Muhonen et al., Caries Res 2025-11-01, PMID 41196837).
• The lesion is in dentine and the patient’s caries risk is high and modifiable factors are already controlled: sealing or infiltrating a lesion you cannot otherwise keep clean in a child who still gets new cavities every year is a different risk calculation than in the low-risk adult, which is exactly what the risk-dosed recall studies formalise (Cordeschi et al., Int J Paediatr Dent 2026-01-01, PMID 40984027) (Ekstrand et al., Children (Basel) 2026-03-01, PMID 41897144).
• Pulp symptoms are present: then the question has changed from restoration to pulp therapy, and the guideline is explicit that capping or pulpotomy options apply only where irreversible pulpitis is not diagnosed (Schwendicke et al., Caries Res 2026-04-01, PMID 42018467).
• The existing restoration has failed and decay has re-opened around it: adding the clinical score to the radiograph significantly changed the decision towards intervention in the reading study, and this is the case where the X-ray alone is least trustworthy (Maydana et al., Braz Dent J 2024-01-01, PMID 39699493).

How to read this like a clinician

• Record the four variables separately — depth, cavitation, activity, cleansability — because the billing code follows the fourth, not the first, and the justification for not drilling is exactly that record (Santos et al., J Clin Med 2026-08-01, PMID 42652888).
• Quote the lesion-level numbers, not the tooth-level reassurance: in the Finnish data 57% of registered proximal lesions were enamel-confined, and an 8% treatment probability at the dentino-enamel junction is a legitimate chart entry, not a delay of care (Muhonen et al., Caries Res 2025-11-01, PMID 41196837).
• Do not let the radiograph decide alone: adding CARS and activity to the same images moved decisions significantly in one direction; you are choosing a threshold, so document that you chose one (Maydana et al., Braz Dent J 2024-01-01, PMID 39699493).
• Screen for artefacts before drilling a first premolar or molar “radiolucency”: non-carious triangular radiolucency is present in about a quarter of maxillary molars, and false positives were reported in 13% of readings in the pooled analysis of low-certainty evidence (Dioguardi et al., Oral Radiol 2026-04-01, PMID 41331196).
• Use devices as tie-breakers with known weaknesses: QLF is strong occlusally and mediocre approximally (0.86/0.82 vs 0.74/0.82), DIAGNOdent replicated rather than beat the visual exam on smooth surfaces (Lee et al., Sci Rep 2025-11-01, PMID 41238662) (Shaalan et al., J Conserv Dent 2023-03-01, PMID 37205891).
• Treat AI outputs as flags, not diagnoses: stage-specific sensitivity for early enamel lesions is higher than human readers but the reference standard is expert agreement, and 12 of 19 studies had a high risk of bias in at least one domain (Banihashem Rad et al., Caries Res 2026-06-01, PMID 42313715) (Bittar et al., J Dent 2026-07-01, PMID 42537896).
• Default to selective or stepwise removal in deep lesions and skip routine liners, since that is where the S3 guideline found support, while remembering the certainty band is very low to moderate (Schwendicke et al., Caries Res 2026-04-01, PMID 42018467).
• Dose the recall: 4 versus 8 months in high-risk preschoolers produced a 2.5-fold difference in new cavitated lesions over 30 months (Cordeschi et al., Int J Paediatr Dent 2026-01-01, PMID 40984027).

Evidence at a glance

Question	Estimate as printed	Design and certainty
Does radiographic depth predict drilling?	8% probability at the dentino-enamel junction, 75% at the outer/middle dentin boundary; 57% of 394 lesions enamel-confined	Retrospective register of 160 patients, 16 dentists, cluster-robust regression (Muhonen et al., Caries Res 2025-11-01, PMID 41196837)
Does adding clinical data change the plan?	Distribution of restorative decisions differed significantly (p < 0.001); agreement κ 0.749	Crossover reading of 212 bitewings with two-week washout (Maydana et al., Braz Dent J 2024-01-01, PMID 39699493)
Do dentists follow classification-based guidance?	Agreement 34.4% to 53.9% for occlusal codes; 12.4% at approximal RA3; 99.5% at RB4	Web survey of 442 dentists, case-based; association with training and practice type (Ramayasinpong et al., Int J Paediatr Dent 2025-07-01, PMID 39890991)
How accurate is visual diagnosis?	Specificity 100%, sensitivity 92% unaided; specificity 50% with magnification	100 extracted teeth, histological reference; in vitro (Rekha et al., Int J Clin Pediatr Dent 2023-09-01, PMID 38078031)
How accurate is adjunctive fluorescence?	AUC 0.94-0.98 incipient occlusal; 0.86/0.82 occlusal and 0.74/0.82 approximal sensitivity/specificity	17 studies, QUADAS-2, in vitro and in vivo mixed (Lee et al., Sci Rep 2025-11-01, PMID 41238662)
Can AI replace the reader?	Sensitivity 0.56-0.96 (AI) vs 0.21-0.83 (human) for early enamel; pooled sens 0.76-0.94, spec 0.85-0.91 in the umbrella; not supported as stand-alone	19-study systematic review and 17-review umbrella, both limited by bias and heterogeneity (Banihashem Rad et al., Caries Res 2026-06-01, PMID 42313715) (Bittar et al., J Dent 2026-07-01, PMID 42537896)
Seal or drill an initial occlusal lesion?	93.75% vs 65.62% success at 24 months; RR 0.18 (0.0437-0.756), p = 0.019	RCT, 36 participants, 64 lesions; small sample (Abdelsalam et al., Sci Rep 2025-12-01, PMID 41381619)
SDF for adult initial approximal lesions?	>90% arrested in both arms at 12 months; SDF better only at 6 months (p = 0.05)	Triple-blind randomised placebo-controlled trial with three progression endpoints (Jaradat et al., J Conserv Dent Endod 2025-09-01, PMID 40964644)
Which caries-removal technique?	No technique superior across outcomes; selective favoured for exposure risk; pain lower with non-selective than stepwise	Umbrella of 7 SR/MA, AMSTAR-2 and GRADE, certainty low to very low (Aldakhil et al., J Dent 2026-10-01, PMID 42309376)
Does stepwise removal hold up long term?	Success >60% at 110 months; better for premolars, under 40, low or medium risk	Registry of 111,771 teeth, Kaplan-Meier (Malmberg et al., Caries Res 2026-03-01, PMID 41843704)
Do cavity liners help?	No consistent clinical benefit; not routinely recommended	S3-level guideline (EFCD-ESE-ORCA), certainty very low to moderate (Schwendicke et al., Caries Res 2026-04-01, PMID 42018467)
Which recall interval?	2.5 times more new cavitated lesions at 8 months vs 4 months over 30 months	RCT, 224 high-risk preschool children (Cordeschi et al., Int J Paediatr Dent 2026-01-01, PMID 40984027)

What the evidence does not support

• That a filled tooth is a treated cavity. None of the sources above compares “fill everything visible” with “select and monitor” as a whole-population strategy inside the same trial; what exists are depth-threshold observations and technique comparisons (Muhonen et al., Caries Res 2025-11-01, PMID 41196837) (Aldakhil et al., J Dent 2026-10-01, PMID 42309376).
• That pain, or its absence, decides anything. Pulp exposure was found in 28.2% of examined schoolchildren with an associated quality-of-life impact, and stage C lesions reduced quality of life regardless of location — pain is not what the evidence uses as a threshold (Fonseca et al., Clin Oral Investig 2026-02-01, PMID 41729310) (Díaz et al., Braz Oral Res 2025-01-01, PMID 41379130).
• That an X-ray is a verdict. Radiographic artefacts produced false-positive readings in about 13% of observations, and the same images generated different decisions depending on what clinical information accompanied them (Dioguardi et al., Oral Radiol 2026-04-01, PMID 41331196) (Maydana et al., Braz Dent J 2024-01-01, PMID 39699493).
• That a device settles it. DIAGNOdent and QLF reproduce or modestly extend visual findings; OCT was validated on 54 extracted surfaces; and AI performance is measured against human agreement rather than histology (Shaalan et al., J Conserv Dent 2023-03-01, PMID 37205891) (Lee et al., Sci Rep 2025-11-01, PMID 41238662) (Strumpski et al., Caries Res 2025-01-01, PMID 39999820) (Banihashem Rad et al., Caries Res 2026-06-01, PMID 42313715).
• That silver diamine fluoride, resin infiltration or a sealant is a substitute for a filling in any cavitated, non-cleansable tooth: the reviews describe them as options for specific stages, and one SDF trial shows placebo matching the drug by 12 months (Jaradat et al., J Conserv Dent Endod 2025-09-01, PMID 40964644) (Miranda et al., J Clin Med 2026-02-01, PMID 41682994) (Santos et al., J Clin Med 2026-08-01, PMID 42652888).
• That AI screening or “digital dentistry” can diagnose your cavities from a photo. The umbrella review’s wording is direct: current evidence does not support stand-alone use, and implementation should stay in decision-support contexts until prospective validation shows an impact on decisions and outcomes (Bittar et al., J Dent 2026-07-01, PMID 42537896).
• That a “mineralising” toothpaste reverses a cavitated lesion. The remineralisation literature concerns closed surfaces and early lesions; once the surface has collapsed, the trials above discuss sealing, infiltration, arrest or restoration, not paste (Santos et al., J Clin Med 2026-08-01, PMID 42652888) (Todorova et al., J Clin Med 2026-02-01, PMID 41753221).

Cost, coverage and the Utah part

Utah’s public children’s dental benefit contains a rule that reads like an answer to “can I just keep watching this tooth”: silver (metal) fillings are allowed once every two years for each surface of a tooth, and if a filling on the same surface has to be replaced within those two years, it is generally not covered unless there is a specific medical reason; tooth-coloured composite fillings are covered for front teeth, and for back teeth coverage depends on the plan, with amalgam the standard choice for back teeth unless the tooth-coloured option is medically necessary or specifically allowed — subject to the same two-year limit per surface (InsureKidsNow summary of benefits for Utah Medicaid, data as of 4 February 2026; see Additional documents). Two clinical consequences follow from that paragraph, and both are worth saying out loud in a treatment plan: a restoration placed earlier than necessary consumes the plan’s replacement allowance for the tooth that may genuinely need work in year three, and the “watch and keep it clean” option is not a way of saving the payer money but a decision the classification systems were built to justify.

The same document lists the alternatives that do have trial support at the earlier stages: silver diamine fluoride once every six months per tooth on baby teeth as a non-invasive alternative to fillings, fluoride varnish up to four times per calendar year, sealants once every two years per sound first or second permanent molar or premolar, and space maintainers once per side of the mouth per lifetime for a baby tooth lost too early — the last one relevant when the “extraction instead of a filling” option is on the table, which our baby teeth page covers. Prior authorisation is not required for fillings, pulpotomy on infected baby teeth, or simple and surgical extractions when the tooth cannot be saved by a filling or root canal; it is required for metal and metal-porcelain crowns on permanent teeth, for periodontal scaling and root planing, and for all orthodontic treatment. Nothing here replaces your actual benefit booklet: the frequency limits above are per tooth surface, and plans differ.

A plan, if you want one

• Ask which of the four things you have. Words to use: is the surface closed or cavitated; is the lesion active or arrested; can it be kept clean; how deep does the radiograph put it. Those are the variables the research actually measures, and an answer in those terms can be checked (ElSalhy et al., Community Dent Oral Epidemiol 2022-12-01, PMID 34897763) (Muhonen et al., Caries Res 2025-11-01, PMID 41196837).
• Ask for the number, not the adjective. “ICDAS 3, upper right six, radiolucency in the outer third of dentin” is a sentence you can bring somewhere else for a second opinion; “a bit of decay” is not (Ramayasinpong et al., Int J Paediatr Dent 2025-07-01, PMID 39890991).
• If the answer is “small and not cavitated”, ask what the alternative is. Sealing, infiltration, fluoride varnish and a shorter review interval all have evidence in this article; “we’ll just see” without a date is the version to push back on (Abdelsalam et al., Sci Rep 2025-12-01, PMID 41381619) (Santos et al., J Clin Med 2026-08-01, PMID 42652888) (Cordeschi et al., Int J Paediatr Dent 2026-01-01, PMID 40984027). For the topical piece, see our fluoride varnish and sealant reviews, and the tap-water fluoride page for what the household baseline looks like.
• If the answer is “into dentin”, ask which removal technique and why. Selective or stepwise removal is what the guideline supports for deep lesions, routine liners are not, and the material for any pulp capping is a documented choice (hydraulic calcium silicate over calcium hydroxide) (Schwendicke et al., Caries Res 2026-04-01, PMID 42018467) (Aldakhil et al., J Dent 2026-10-01, PMID 42309376).
• If a white spot is the whole finding — after braces, or as the first sign of demineralisation — the question is infiltration versus varnish rather than a filling; our page on white spots lays out that comparison.
• Set the review date before leaving. High risk is measured in intervals, not in exhortations (Cordeschi et al., Int J Paediatr Dent 2026-01-01, PMID 40984027) (Ekstrand et al., Children (Basel) 2026-03-01, PMID 41897144).

Frequently asked questions

Do I need a filling if I have no pain? Possibly yes, possibly no — pain is not the variable the evidence uses. Exposed pulps were found in 28.2% of 6- to 10-year-olds in a school survey where quality of life was worse for those children, and stage C lesions reduced quality of life regardless of location, so the absence of pain says little about depth (Fonseca et al., Clin Oral Investig 2026-02-01, PMID 41729310) (Díaz et al., Braz Oral Res 2025-01-01, PMID 41379130). What decides is cavitation, cleansability and depth (Santos et al., J Clin Med 2026-08-01, PMID 42652888) (Muhonen et al., Caries Res 2025-11-01, PMID 41196837).

Does a small cavity need a filling? In the trials above, non-cavitated lesions were successfully sealed (93.75% success at 24 months with a bioactive sealant) or managed by infiltration, and in adults more than 90% of untreated initial approximal lesions were arrested at 12 months in the placebo arm of a randomised trial — so “small” is exactly where the non-drilling options live (Abdelsalam et al., Sci Rep 2025-12-01, PMID 41381619) (Santos et al., J Clin Med 2026-08-01, PMID 42652888) (Jaradat et al., J Conserv Dent Endod 2025-09-01, PMID 40964644).

Can a cavity heal without a filling? A closed lesion can remineralise and an open lesion can be arrested; neither becomes a virgin tooth. The micro-invasive literature speaks of progression rates and arrest, not of regrowth of lost structure (Santos et al., J Clin Med 2026-08-01, PMID 42652888) (Alwusaybie et al., Cureus 2026-01-01, PMID 41625031).

Do I need a filling or a crown? That question is answered by how much sound wall is left and whether the pulp is involved, not by how the tooth looks: the guideline ties pulp capping and pulpotomy to the absence of irreversible pulpitis, and the registry data on stepwise treatment show that prognosis differs by tooth type and age (Schwendicke et al., Caries Res 2026-04-01, PMID 42018467) (Malmberg et al., Caries Res 2026-03-01, PMID 41843704). Material choice itself is argued on our amalgam versus composite page.

Is a black line in a groove a cavity? Not necessarily — staining without cavitation is precisely the state that the classification systems separate from a restorable lesion, and radiographic artefacts alone produced false-positive readings in about 13% of observations in the pooled review (Dioguardi et al., Oral Radiol 2026-04-01, PMID 41331196) (Rekha et al., Int J Clin Pediatr Dent 2023-09-01, PMID 38078031). The distinguishing test is whether the surface is intact and cleansable, which a probe-and-visual exam describes better than a photograph.

Why did two dentists tell me different things? Because the published agreement with classification-based guidance ranges from 12.4% to 99.5% depending on lesion stage, and treatment probability itself varies with the depth threshold the examiner uses (Ramayasinpong et al., Int J Paediatr Dent 2025-07-01, PMID 39890991) (Muhonen et al., Caries Res 2025-11-01, PMID 41196837). Ask which stage and depth the recommendation is based on, and you convert a disagreement into a comparison.

Should I just get it filled to be safe? The two-year per-surface replacement limit described above means an early restoration can use up the allowance for a tooth that will genuinely need one; the guideline-level evidence supports less invasive removal and skipping liners when treatment does happen; and a registry of 111,771 teeth shows that even deliberately conservative deep-lesion treatment holds up in more than 60% of cases at nine years (Schwendicke et al., Caries Res 2026-04-01, PMID 42018467) (Malmberg et al., Caries Res 2026-03-01, PMID 41843704) (Aldakhil et al., J Dent 2026-10-01, PMID 42309376). “Safe” here means monitored, not drilled.

Glossary: kitchen words ↔ chart words

What you say at home	What is in the notes	How studies measure it
“a small cavity”	Initial or moderate lesion; ICDAS 1 to 3; non-cavitated	Visual codes plus radiographic depth; AUC against classification-based treatment need (ElSalhy et al., Community Dent Oral Epidemiol 2022-12-01, PMID 34897763) (Jain et al., Cureus 2025-09-01, PMID 41058820)
“the hole”	Cavitation; ICDAS 4 to 6; micro-cavitation on histology	Explorer-and-wedge criteria; histological reference in extraction studies (Rekha et al., Int J Clin Pediatr Dent 2023-09-01, PMID 38078031)
“it’s still active”	Active lesion; plaque-covered, matte, soft; against arrested shiny and hard	Activity scores inside ICCMS; dentist agreement measured by case vignettes (Ramayasinpong et al., Int J Paediatr Dent 2025-07-01, PMID 39890991)
“how deep is it?”	Radiolucency in enamel, outer, middle or inner third of dentin (E1 to D4)	Bitewing depth grading, cluster-robust regression against treatment decision (Muhonen et al., Caries Res 2025-11-01, PMID 41196837)
“between the teeth”	Proximal or approximal lesion; CARS code if next to a restoration	Paired bitewing reading with two-week washout; subtraction radiography (Maydana et al., Braz Dent J 2024-01-01, PMID 39699493) (Jaradat et al., J Conserv Dent Endod 2025-09-01, PMID 40964644)
“seal instead of fill”	Micro-invasive management: sealant or resin infiltration	Retention and progression at 12 and 24 months in randomised designs (Abdelsalam et al., Sci Rep 2025-12-01, PMID 41381619) (Miranda et al., J Clin Med 2026-02-01, PMID 41682994)
“the medicine that stops it”	Silver diamine fluoride; non-restorative cavity control	Arrest rates as binary outcomes; survival versus Hall crown (Jaradat et al., J Conserv Dent Endod 2025-09-01, PMID 40964644) (Alwusaybie et al., Cureus 2026-01-01, PMID 41625031)
“leave some decay on purpose”	Selective or stepwise caries removal	Pulp exposure, post-operative pain, restorative success in SR/MA and registry data (Aldakhil et al., J Dent 2026-10-01, PMID 42309376) (Malmberg et al., Caries Res 2026-03-01, PMID 41843704)
“the nerve is involved”	Pulp exposure; irreversible pulpitis; need for pulp therapy	Prevalence of exposed pulp in school surveys; OR for quality of life (Fonseca et al., Clin Oral Investig 2026-02-01, PMID 41729310) (Schwendicke et al., Caries Res 2026-04-01, PMID 42018467)
“come back in six months”	Recall interval dosed by risk	New cavitated lesions per interval arm in randomised trials; risk-score-driven interval rules (Cordeschi et al., Int J Paediatr Dent 2026-01-01, PMID 40984027) (Ekstrand et al., Children (Basel) 2026-03-01, PMID 41897144)

Related reading on this site: what to do when a tooth chips or breaks, what to do about a toothache at night, what to eat and drink in the first days and how long to wait before smoking or vaping again.

How this page was built, and what it cannot tell you

Method: candidate records were found in Europe PMC using queries that match what people type, and every bibliographic entry used here — authors, journal, volume, pages, DOI, PMID, open-access flag, citation count — was re-fetched on the day of publication and is reproduced below with its DOI. Numbers come from the retrieved abstract text; where an abstract reports an interval, the interval is reproduced. The single figure was drawn from one source reporting one unit, so it cannot be read as a mixture of studies. Coverage statements were read from the CMS InsureKidsNow benefit summary for Utah Medicaid (data as of 4 February 2026), and each claim in this article was checked against the text of that document. Two limitations are structural rather than accidental: most diagnostic studies use histology on extracted teeth or expert agreement on images as the reference standard, which is not the same as following a lesion over time (Rekha et al., Int J Clin Pediatr Dent 2023-09-01, PMID 38078031) (Banihashem Rad et al., Caries Res 2026-06-01, PMID 42313715); and the treatment-decision literature is dominated by cross-sectional and register designs, which cannot tell you what a given lesion would have done if left alone (Muhonen et al., Caries Res 2025-11-01, PMID 41196837) (Ramayasinpong et al., Int J Paediatr Dent 2025-07-01, PMID 39890991).

What this page cannot tell you: whether the spot on your tooth is one of the 57% that were enamel-confined or one of the 4% that reached the inner third of dentin. That judgement needs a cleaned, dried tooth, a probe, a set of bitewings and someone who has calibrated their thresholds — and, where the answer is not obvious, a plan for watching it that says when and on what terms (ElSalhy et al., Community Dent Oral Epidemiol 2022-12-01, PMID 34897763) (Muhonen et al., Caries Res 2025-11-01, PMID 41196837). It also cannot tell you the price of your restoration: the numbers in the coverage section are benefit rules published by the state, not fees, and private estimates vary by operator, not by ZIP code.

Sources

Peer-reviewed evidence

• Al Dhubayb S, Al Sultan M, Al Sudairi S, Hakami F, Al Sweleh FS. Ability of Dentists and Students to Detect Caries by Using the International Caries Detection and Assessment System. Clin Cosmet Investig Dent 2021-01-01;13():379-387. doi:10.2147/ccide.s324396 · PMID 34526823 · PMCID PMC8436776 · open access · cited by 8 (Europe PMC)
• ElSalhy M, Alsumait A, Behzadi S, Lai H, Almerich-Silla JM, Piovesan C, Flores-Mir C, Amin M. A proposed community caries index of treatment need derived from the international caries detection and assessment system. Community Dent Oral Epidemiol 2022-12-01;50(6):522-528. doi:10.1111/cdoe.12706 · PMID 34897763 · cited by 2 (Europe PMC)
• Shaalan OO. DIAGNOdent versus International Caries Detection and Assessment System in detection of incipient carious lesions: A diagnostic accuracy study. J Conserv Dent 2023-03-01;26(2):199-206. doi:10.4103/jcd.jcd_575_22 · PMID 37205891 · PMCID PMC10190091 · open access · cited by 13 (Europe PMC)
• Rekha GS, Chandrasekhar R, Vinay C, Uloopi KS, RojaRamya KS, Chaitanya P. In Vitro Evaluation of Reliability and Validity of International Caries Detection and Assessment System II Coding for Occlusal Caries using Magnification. Int J Clin Pediatr Dent 2023-09-01;16(Suppl 2):138-141. doi:10.5005/jp-journals-10005-2584 · PMID 38078031 · PMCID PMC10701173 · open access · cited by 4 (Europe PMC)
• Maydana GS, Romero VHD, Signori C, Uehara JLS, Sande FHV, Cenci MS, Montagner AF, CaCIA collaborative group. Impact of using bitewing radiographs alone or in combination with clinical information on treatment decisions. Braz Dent J 2024-01-01;35():e246005. doi:10.1590/0103-644020246005 · PMID 39699493 · PMCID PMC11654006 · open access
• Ramayasinpong K, Nakornchai S, Jirarattanasopha V. Decision Making on Caries Management in Children and Adolescents Among Thai Dentists. Int J Paediatr Dent 2025-07-01;35(4):857-868. doi:10.1111/ipd.13297 · PMID 39890991 · PMCID PMC12159579 · open access · cited by 1 (Europe PMC)
• Strumpski M, Schneider H, Rüger C, Schmidt J, Schulz-Kornas E, Haak R. Validity and Reliability of Intraoral Optical Coherence Tomography and Bitewing Radiography for Detecting Approximal Carious Lesions. Caries Res 2025-01-01;59(6):476-489. doi:10.1159/000544789 · PMID 39999820 · PMCID PMC12688330 · open access · cited by 3 (Europe PMC)
• Arzani S, Karimi A, Iranmanesh P, Yazdi M, Sabeti MA, Nekoofar MH, Kolahi J, Bang H, Dummer PMH. Examining the diagnostic accuracy of artificial intelligence for detecting dental caries across a range of imaging modalities: An umbrella review with meta-analysis. PLoS One 2025-01-01;20(8):e0329986. doi:10.1371/journal.pone.0329986 · PMID 40802701 · PMCID PMC12349118 · open access · cited by 6 (Europe PMC)
• Jaradat M, Kolker J, Owais A, Guzman-Armstrong S, Haes A, Anamali S, Comnick C, Zeng E, Kanellis M. Silver diamine fluoride effectiveness in arresting initial approximal caries lesions in adults: A triple-blinded, randomized controlled clinical trial. J Conserv Dent Endod 2025-09-01;28(9):892-898. doi:10.4103/jcde.jcde_83_25 · PMID 40964644 · PMCID PMC12440336 · open access
• Cordeschi T, Besseler M, Olegário IC, Ekstrand K, Bakhshandeh A, Berti G, Abanto J, Bönecker M. Effectiveness of a 4-Month vs. 8-Month Recall Interval for High-Caries-Risk Preschool Children: A 30-Month Clinical Trial. Int J Paediatr Dent 2026-01-01;36(1):105-112. doi:10.1111/ipd.70032 · PMID 40984027
• Jain K, Gupta BD, Gupta R, Bhargava A, Chaudhary SK, Chaudhary H. Validation of International Caries Detection and Assessment System (ICDAS)-II Short Version and Community Caries Index for Assessing Caries and Treatment Needs in Six- to 15-Year-Old Children in Ghaziabad: A Cross-Sectional Study. Cureus 2025-09-01;17(9):e91718. doi:10.7759/cureus.91718 · PMID 41058820 · PMCID PMC12498487 · open access
• Muhonen MH, Raittio E, Olkkonen L, Leinonen KM, Leinonen J. Bitewing Radiographs in Proximal Caries Diagnostics and Restorative Treatment of Adults: A Retrospective Study of Electronic Health Records. Caries Res 2025-11-01;:1-11. doi:10.1159/000549383 · PMID 41196837 · PMCID PMC12707913 · open access
• Lee ES, Hwang J, Cheng L, Jung HI, Kim BI. Diagnostic accuracy of quantitative light-induced fluorescence in detecting caries of various types and locations: a systematic review and meta-analysis. Sci Rep 2025-11-01;15(1):39905. doi:10.1038/s41598-025-23631-6 · PMID 41238662 · PMCID PMC12618513 · open access · cited by 4 (Europe PMC)
• Dioguardi M, Guerra C, Sovereto D, Martella A, Zhurakivska K, Tisci A, Esperouz F, Bizzoca ME, Sanesi L, Mastrangelo F, Lo Muzio L, Cirillo N, Ciavarella D, Ballini A. Radiographic artifacts in the diagnosis of dental caries: systematic review with meta-analysis. Oral Radiol 2026-04-01;42(2):281-296. doi:10.1007/s11282-025-00879-2 · PMID 41331196 · PMCID PMC12988984 · open access · cited by 1 (Europe PMC)
• Díaz S, Abanto J, Boneckër M, Lara JS, Paiva SM, Olegário IC. Impact of clinical stage and location of dental caries lesions on oral health-related quality of life. Braz Oral Res 2025-01-01;39():e125. doi:10.1590/1807-3107bor-2025.vol39.125 · PMID 41379130 · PMCID PMC12683720 · open access
• Abdelsalam AF, Shaalan O. 24-month efficacy of bioactive versus fluoride sealant for non-cavitated occlusal caries in adults: a randomized clinical trial. Sci Rep 2025-12-01;15(1):43738. doi:10.1038/s41598-025-29730-8 · PMID 41381619 · PMCID PMC12700883 · open access · cited by 1 (Europe PMC)
• Alwusaybie MM, Madiraju GS, Alhasan HA, Alshaks MA, Alhafi MH, Alkadem A, Alnasser RM, Alshamari S, Alhamoud MM, Alhulaybi RS, Almulhim AN. Effectiveness of the Non-restorative Cavity Control (NRCC) Approach in Managing Dentinal Caries in Primary Teeth: A Systematic Review. Cureus 2026-01-01;18(1):e102458. doi:10.7759/cureus.102458 · PMID 41625031 · PMCID PMC12854532 · open access
• Miranda SB, Lins RBE, Silveira JAMD, Leal CFC, Souza CMC, Moura HS, Guimarães LG, Prosini P, Montes MAJR. Efficacy of Resin Infiltrants in Non-Cavitated Occlusal Carious Lesions: A Systematic Review. J Clin Med 2026-02-01;15(3):1310. doi:10.3390/jcm15031310 · PMID 41682994 · PMCID PMC12897895 · open access
• Fonseca PG, Ramos-Jorge ML, de Carvalho BO, da Consolação Soares ME, Barbosa Fernandes I. The impact of pulp disease on oral health-related quality of life in children aged 6 to 10 years. Clin Oral Investig 2026-02-01;30(3):89. doi:10.1007/s00784-026-06770-6 · PMID 41729310 · PMCID PMC12929309 · open access
• Todorova V. Non-Operative, Micro- and Minimally Invasive Methods for Caries Treatment-A Narrative Review. J Clin Med 2026-02-01;15(4):1534. doi:10.3390/jcm15041534 · PMID 41753221 · PMCID PMC12941922 · open access · cited by 1 (Europe PMC)
• Malmberg J, Hedenbjörk-Lager A. Long-Term Treatment Success Rate after Stepwise Caries Removal: A Registry Study. Caries Res 2026-03-01;:1-12. doi:10.1159/000551560 · PMID 41843704 · PMCID PMC13171086 · open access
• Ekstrand KR, Christiansen MEC. The Nexø Method-Clinical Evidence for the Paradigm Shift in Caries Management for Children and Adolescents in Denmark Being Cost-Effective. Children (Basel) 2026-03-01;13(3):432. doi:10.3390/children13030432 · PMID 41897144 · PMCID PMC13025739 · open access
• Schwendicke F, Kosan E, Banerjee A, Baysan A, Bjørndal L, Ceballos L, Duncan HF, Herbst S, Neuhaus KW, O Apos Connell AC, Paris S, Dujic H. Deep Caries Management: EFCD-ESE-ORCA S3-Level Clinical Practice Guideline. Caries Res 2026-04-01;:1-19. doi:10.1159/000551659 · PMID 42018467 · PMCID PMC13102370 · open access · cited by 1 (Europe PMC)
• Aldakhil S, Mubarak M, BaHammam F, Vasconcellos A. Effectiveness of selective, non-selective, and stepwise caries removal techniques in permanent teeth: An umbrella review of systematic reviews and meta-analyses. J Dent 2026-10-01;173():106844. doi:10.1016/j.jdent.2026.106844 · PMID 42309376
• Banihashem Rad SA, Azami N, Campus G, Hugo FN, Esteves-Oliveira M. AI-Driven Decision Thresholds in Cariology: A Systematic Review of Lesion Stage Detection on Bitewing Radiographs. Caries Res 2026-06-01;:1-22. doi:10.1159/000552757 · PMID 42313715
• Bittar A, Dascanio R, Özcan M. Artificial intelligence for dental caries detection: An umbrella review. J Dent 2026-07-01;175():106944. doi:10.1016/j.jdent.2026.106944 · PMID 42537896
• Santos GC, Santos MJMC. Resin Infiltration in the Conservative Management of Incipient Dental Caries: An Evidence-Based Review. J Clin Med 2026-08-01;15(16):6485. doi:10.3390/jcm15166485 · PMID 42652888

Additional documents

• InsureKidsNow (CMS). Summary of Benefits Report for Utah, Medicaid — dental (data as of 04 Feb 2026, print date 03/02/2026) — fillings: silver (metal) fillings are allowed once every two years for each surface of a tooth, and if a filling on the same surface has to be replaced within those two years it is generally not covered unless there is a specific medical reason; tooth-coloured composite fillings are covered for front teeth, and for back teeth coverage depends on the member’s plan and eligibility, with silver (amalgam) the standard choice for back teeth unless a tooth-coloured filling is medically necessary or specifically allowed by the plan, subject to the same two-year limit per tooth surface (PDF)
• InsureKidsNow (CMS), same report — the non-invasive options at the earlier stages: silver diamine fluoride once every six months per tooth on baby teeth as a non-invasive alternative to traditional fillings; fluoride treatments including varnish up to four times per calendar year (below age 5 a physician may apply it during a well-child exam, from age 5 it must be done in a dental office); sealants once every two years per tooth on first and second permanent molars and premolars provided the tooth has no existing decay or fillings; space maintainers for children who lose a baby tooth too early, once every three years and limited to one per side of the mouth per lifetime, not covered for front teeth or to replace lost permanent teeth (PDF)
• InsureKidsNow (CMS), same report — when care goes deeper: a pulpotomy (the document’s “mini” root canal for baby teeth) is covered for infected baby teeth without prior authorisation, once per tooth, and not for teeth already loose or too damaged to be repaired with a filling or crown; root canals on permanent teeth do not require prior authorisation; simple and surgical extractions do not require prior authorisation and are covered when a tooth cannot be saved by a filling or root canal; stainless steel crowns are allowed once every two years per tooth, while metal and metal-porcelain crowns on permanent teeth require prior authorisation and are limited to once every five years per tooth; nitrous oxide analgesia is listed as a non-covered service (PDF)

Bibliographic records above were retrieved and verified programmatically from Europe PMC (authors, journal, volume, pages, DOI, PMID, citation count). Coverage, guideline and regulatory statements are quoted from the cited public documents with their dates; verify against the current version before relying on them. This article is not medical or dental advice.