A complete guide to verifying oral-health claims: an evidence map of popular practices, self-treatment and common myths

A complete guide to verifying oral-health claims: an evidence map of popular practices, self-treatment and common myths

TL;DR

To verify an oral-health claim, ask four things first: who is saying it, the level of the evidence, whether there is a control group, and who paid for it [Fn1][Fn3][Fn5]; a mechanism that sounds plausible is not the same as clinical effectiveness [Fn10].

(Measured on the zh-Hant original with Python `len()` after the [Fn] markers were removed: 54 characters including punctuation, 46 characters excluding punctuation, both within the ≤60-character threshold; see item 6 of the “Self-check record”.)

Introduction

This article is general oral-health education based on international literature. It does not address any country's insurance or regulations; consult local rules for care pathways and costs.

This article does not tell you “whom to believe”. It gives you a procedure for judging things yourself.

The position of this article has to be stated first, because it governs how every later section should be read: the result of verification does not come in only two boxes, “true” and “false”. The eight groups of claims handled here end up in five different boxes — some are “there is a little effect but the evidence is very weak”, some are “neither the safety nor the efficacy side has been substantiated”, some are “the practice itself bypasses clinical supervision”, some are “a historical theory has been refuted and the modern evidence points the other way”, and one is “between the public alarm and the official position there is a genuine scientific tension, not a simple rumour”. Flattening all eight groups into “they are all false” is just as wrong as saying that all of them “have a point”.

This article also deliberately does two things not at all: it names no product, brand, institution or individual; and it does not write “patients are being overtreated” as an accusation against any clinician. What the literature can support is “this phenomenon has been recorded within a study design”; it cannot support “the clinician you saw is like that” — that boundary is itself part of the discipline of verification.

1. The verification framework: five reusable questions

1-1 Why a framework rather than a “list of myths”

A list of myths goes out of date; a procedure does not. And the structure of information sources has itself already changed: a cross-sectional questionnaire study carried out by the faculty of dentistry at Damascus University [Fn11] (301 questionnaires collected, 291 taken into the analysis [Fn11a]) found that 50.2% of respondents said they sometimes feel that the advice on social-media sites is in fact marketing for a company or a product, or advertising for doctors [Fn12]; in the same survey, 89.7% of respondents visit the dentist when facing a medical problem, while 10.3% follow the advice on social-media sites [Fn13].

These two figures have to be read carefully: they come from a self-selected sample recruited through social media in a study run by a single institution [Fn11], and they cannot represent any other country or any global proportion (“Syria” as a country label is general geographical knowledge inferred from the institution that ran the study, not a property of the study written down in the original — and that very distinction is what the first of the questions below exists to handle). This article cites it only to make one point — “does this piece of information have a commercial motive behind it” is already a question ordinary members of the public ask for themselves, and the framework below turns that instinct into five operable questions.

1-2 Question one: who is saying it (the level of the source)

The key element of evidence-based medicine is the hierarchical system that categorises evidence [Fn1]; clinicians use that classification to identify the highest level of evidence able to answer a clinical question [Fn2a]. And different types of research study answer different kinds of clinical question [Fn2] — a point that is often overlooked: a well-designed laboratory study cannot answer “what will happen to the patient after they use it”.

(The nature of this basis itself has to be labelled clearly: it comes from a critical appraisal of “level-of-evidence” classification systems (see the title of #W9 in the source list). What this article takes from it is the consensus background stated in its opening [Fn1][Fn2][Fn2a], not its critical conclusions about the individual classification systems — using a critical paper as an authoritative endorsement of the very thing it criticises is exactly the reading this article is trying to teach readers to avoid.)

In the source list of this article you will see a basis tier marked on every entry (official agency statement / professional-organisation consensus guideline / peer-reviewed literature). That is not decoration; it is what lets you judge for yourself “how much weight this sentence carries”.

1-3 Question two: which level the evidence sits at (for one and the same topic, the highest level available may be very low)

For some topics the highest level of evidence available is simply very low — that is not laziness on the author's part, it is that the topic itself does not yet have high-quality research. The honest way to write it is to label the level, rather than to write a case report as though it were a general rule.

There are two typical instances in this article: first, “gluing a tooth back yourself with super glue”, where the direct evidence retrievable in the literature is a single case report from 2006 [Fn53]; second, NICO (see below), whose systematic review states plainly that all studies were observational in their design [Fn4], while a scoping review records that the evidence in this field consists predominantly of low- to moderate-quality retrospective case series, with no randomised controlled trials [Fn88].

1-4 Question three: whether there is a control group (and what the comparison is against)

“Effective” is always effective relative to some object of comparison. The same intervention compared with “doing nothing”, with “the conventional practice”, and with “the most effective current practice” will yield three different conclusions — the section on oil pulling below is a living example.

Conversely, the absence of a control group is a structural weakness, not a detail. After collating 41 studies, the scoping review of NICO lists “absence of control groups, non-standardised diagnostic criteria, limited histological evidence, and interpretive bias” as recurrent methodological limitations [Fn3][Fn87].

1-5 Question four: who paid for it (interests)

This one has direct, cross-disciplinary quantitative evidence. A Cochrane methodology review (including 75 papers) compared drug and device studies “sponsored by the manufacturer” with those “sponsored by other sources”: industry-sponsored studies more often had efficacy results favourable to the sponsor (RR 1.27, 95% CI 1.17 to 1.37) [Fn5], and more often had conclusions favourable to the sponsor (RR 1.34, 95% CI 1.19 to 1.51) [Fn6]; while within industry-sponsored studies, there was less agreement between the results and the conclusions [Fn7]. The authors of that review concluded that this industry bias cannot be explained by standard risk-of-bias assessments [Fn8].

This has to be read precisely: what it states is a shift in the overall distribution, not that “industry-sponsored studies are all untrustworthy”. What it gives you to do is this — when you see a conclusion favourable to some product, go and look at its funding source and its disclosure-of-interests field, and take that information into the weighting.

In the field of oral health, interests have a further, more upstream form. The second paper in the Lancet oral health Series points out that the sugar industry is influencing dental research, oral health policy and professional organisations through its well developed corporate strategies [Fn9], and argues for the development of clearer and more transparent conflict-of-interest policies and procedures [Fn10a].

This part of the framework applies to this article too, and it has to be applied evenhandedly in both directions. The source used to support the passage on “baking soda dentifrice” (#W3) was published in a supplement issue of a dental journal, and its PubMed record carries no conflict-of-interest statement — these are verifiable bibliographic facts, on the basis of which this article labels that entry's confidence as low and writes that qualification out in the main text (see §2-2).

The same yardstick has to be applied to the entry that points the other way: for the source used to support the passage on “occupational exposure during amalgam removal” (#W10), all four authors' affiliations as registered on PubMed are private dental practices, with no academic or occupational-health institution named, and the funding section of its PMC full text states that the study was funded by the authors themselves (its conflict-of-interest statement, for its part, records that the authors declare no competing interests). These are likewise verifiable bibliographic facts, not accusations against the authors; on that basis this article labels that entry as low confidence and restricts what it is used for (see §2-5 and F13). Check the funding of only the side you dislike and the framework stops working.

1-6 Question five: a mechanism that sounds plausible is not the same as clinical effectiveness

This is the one that is easiest to lose, because mechanistic explanations read very persuasively. Three levels have to be kept apart:

1. The mechanism itself may not have been independently validated. The NICO scoping review writes it out directly: the proposed mechanisms (such as RANTES/CCL5 overexpression) lack independent validation [Fn10].
2. A mechanism holding does not mean clinical outcomes will change. Internet advertisements for charcoal dentifrices included unsubstantiated therapeutic claims — antibacterial, antifungal, antiviral and “oral detoxification” — as well as potentially misleading product assertions [Fn24]; each of those words has a biological story that can be told, but that review's conclusion is that the available clinical and laboratory data are insufficient to substantiate their safety and efficacy claims [Fn25].
3. An unvalidated diagnostic tool makes a mechanistic account unfalsifiable. The NICO scoping review records that diagnostic tools remain unvalidated, while surgical interventions continue on the basis of unclear criteria [Fn78].

The portable reading: when you hear “because X, therefore Y”, ask “is there a clinical study that measured Y directly”. If there is not, that is a hypothesis, not a conclusion.

1-7 Five kinds of verification result (the classification map of this article)

2. The evidence map, claim by claim

Every section below uses the same three-part form: what the claim is / what the evidence says / where the risk lies.

2-1 Oil pulling

What the claim is: that holding edible oil (most often sesame oil or coconut oil) in the mouth and swishing it for a period of time improves oral health.

What the evidence says: this is the group with the fullest evidence in this article, and also the one most easily halved by each side.

A meta-analysis included 25 randomised controlled trials involving 1,184 participants [Fn14], and the authors' conclusion falls into two sentences that point in opposite directions yet hold at the same time: oil pulling probably has a benefit in improving gingival health [Fn15]; while in reducing the amount of plaque, chlorhexidine remained superior to oil pulling [Fn16]. The same paper also states plainly that the overall quality of the body of evidence was very low [Fn17]. In addition, among the 25 trials included, more than half of the trials had participants with no reported oral health issues [Fn14a] — note that the unit the original counts in is “trials” and not “participants”, and this article does not convert it into a number of people; the claim that holds is this: more than half of the trials in this body of evidence were done in populations with no reported oral health problems, so extrapolating to people with pre-existing disease requires reservation [Fn14a].

Another systematic review, which included 31 studies [Fn20a], reports that oil pulling with sesame or coconut oil showed moderate reductions in microbial load and improved gingival health, though the findings varied compared with chlorhexidine mouthwash [Fn18]. The overall conclusion of that review is that traditional oral hygiene practices may offer benefits similar to conventional methods [Fn21], but that methodological limitations and variability among studies require the findings to be interpreted cautiously [Fn19]. That conclusion is an umbrella one, not specific to oil pulling: the traditional practices examined in that same review include several classes, among them miswak and oil pulling [Fn20b], and the one that obtained the stronger result — “statistically significantly reduced plaque accumulation and gingival inflammation, performing comparably to or better than conventional toothbrushes” — was miswak [Fn21a], whereas what oil pulling obtained was the weaker descriptive result quoted in the previous sentence [Fn18]. Taking the umbrella conclusion as oil pulling's own report card is another form of level upgrading.

A common citation error has to be dismantled here. The sentence in that review's abstract about “non-inferiority for short-term plaque / gingivitis control” appears in its hypothesis section (the original reads `We hypothesized non-inferiority for short-term...`) [Fn20]; it is not its research conclusion. Writing an author's hypothesis as the author's conclusion is the commonest form of upgrading in literature citation. This article therefore uses only its conclusion sentence [Fn21] and its descriptive result sentence [Fn18].

Where the risk lies: neither review positions oil pulling as a replacement for routine oral hygiene; one of them states explicitly that further thorough research is required to confirm its efficacy and its integration into modern oral care [Fn19a]. In practice the risk does not lie in the oil itself, but in substitution — swapping a practice with higher effectiveness for one whose evidence level is very low.

Box: A (effective but very weakly so). The accurate one-sentence version is: “there is a little directional evidence, of very low quality, and on a particular measure it loses to an existing practice.”

2-2 Whitening with activated charcoal and with baking soda

What the claim is: that brushing with activated-charcoal powder or an activated-charcoal dentifrice, or with a dentifrice based on baking soda (sodium bicarbonate), achieves whitening; some claims additionally include antibacterial action and “detoxification”.

What the evidence says — for the activated-charcoal group, it is “neither side has been substantiated”:

A literature review identified 118 potentially eligible articles [Fn22a], of which 13 reported brushing the teeth with raw charcoal or soot, yet none of those studies met its inclusion criteria [Fn22]; a further 3 reported deleterious outcomes (increased caries, enamel abrasion, non-quantified negative impact) [Fn23]. The same results section also records findings pointing the other way, and listing them alongside is only fair: 2 studies offered non-specific caries reductions [Fn23a], and 1 study indicated only that brushing with raw charcoal had no adverse effects on oral hygiene [Fn23b]. But not one of those 13 met that review's inclusion criteria [Fn22] — so results in all three directions are equally unusable as evidence, and that is precisely what “neither side has been substantiated” means. The authors' examination of internet advertisements found that they included unsubstantiated therapeutic claims, such as antibacterial, antifungal, antiviral and oral detoxification, as well as potentially misleading product assertions [Fn24]. The conclusion of that review is that the available clinical and laboratory data are insufficient to substantiate the safety and efficacy claims of charcoal and charcoal-based dentifrices [Fn25]; and that dental clinicians should advise their patients to be cautious about charcoal dentifrices carrying unproven claims of efficacy and safety [Fn26].

Note the precise shape of that sentence: it is that neither the safety nor the efficacy has been substantiated, not “the efficacy is in doubt but it is safe”. In addition, that review's search was closed at February 2017 [Fn25a]; this article has not separately searched for whether any new systematic review of the “human safety” of activated charcoal has appeared since, so it must not be inferred that “more complete safety research has become available in recent years”.

The baking soda group is different, but the qualifications have to be marked clearly:

First separate the two mechanisms — the chemical route bleaches with peroxides, while the mechanical route uses abrasives (found in prophylactic pastes and dentifrices) to remove stains and thereby produce a whitening effect [Fn27]. The two address different sources of discolouration and cannot substitute for each other.

A literature review, having collated in-vitro and clinical studies, states that the evidence available in the literature indicates that baking soda-based dentifrices are effective and safe for tooth stain removal and consequently whitening [Fn28]; the same paper records that baking soda has acid-buffering capacities, is antibacterial at high concentrations, and has a relatively lower abrasivity [Fn29].

This entry has to be read with three qualifications: (1) that paper is a literature review, not a systematic review or a meta-analysis; (2) it was published in a supplement issue, and its PubMed record carries no conflict-of-interest statement — under the framework of §1-5, that is information to be taken into the weighting; (3) “relatively lower abrasivity” is a relative comparison, not an absolute value. In fact, an in-vitro study assigned bovine dentine specimens to six groups for testing [Fn31a] (three groups of regular toothpaste, one reference slurry, two whitening toothpastes), and after 10,000 brushing strokes measured them with a non-contact profilometer; the relative dentine abrasion values across all six groups combined ranged from 26 to 166 [Fn31] — note that this is the range across all the tested groups, and not the range for the “whitening toothpaste” class; hanging it on a single class is taking the union for a subset. The same paper has a further result, pointing the other way and often skipped: the abrasion value of the whitening toothpaste containing hydrogen peroxide was not high compared with that of the regular toothpaste [Fn31b]. The authors' conclusion is that the abrasion values of whitening toothpaste could vary depending on the mechanism and the ingredients of the whitening agents [Fn30]. In other words, “whitening toothpaste” is not a homogeneous class, and the class as a whole cannot be judged safe or unsafe; nor can in-vitro specimen data be equated directly with the result of long-term use in a person's mouth.

Where the risk lies: the deleterious outcomes already recorded include enamel abrasion and increased caries [Fn23]. The more structural risk is “using a product in place of a diagnosis” — discolouration of the teeth has more than one source [Fn27], and choosing the wrong route is not merely ineffective, it may also delay a problem that genuinely needs attention.

Box: activated charcoal = B (neither safety nor efficacy substantiated); baking soda = the evidence leans positive, but its level and its disclosure of interests are limited.

Adjacent specific questions (each has its own canonical card and domain article; this article does not expand them)

- The classification of tooth-whitening methods, the procedures and the limits of each — see canonical card KM-DENTAL-19 (in production) and domain article P10, tooth whitening and cosmetic dentistry. This article handles only “should this claim be believed”; it does not compare whitening methods.

2-3 DIY and mail-order orthodontics (the direct-to-consumer orthodontic model)

What the claim is: that clear aligners or orthodontic appliances bought online can complete orthodontic treatment without routine in-person clinical supervision; the selling points are usually price and convenience.

What the evidence says: the definition of this model already states the point at issue — it allows patients to undergo care without routine in-person clinical supervision [Fn32a].

A qualitative study retrieved 721 public posts and, after screening, included 165 for thematic analysis [Fn33], deriving seven themes: dissatisfaction with treatment outcomes, communication difficulties and limited customer support, billing issues and financial frustration, physical discomfort requiring self-management, uncertainty in treatment progress, peer photographs used as visual evidence, and cost-driven decision-making with perceived limitations in care [Fn35a]; overall, negative sentiment predominated across the posts [Fn35]. The authors conclude that reliance on self-monitoring and peer advice may delay recognition of complications and increase risks [Fn34]; and that as this model spreads, concerns remain regarding patient safety, the adequacy of clinical oversight, treatment outcomes and corporate practices [Fn32].

Another cross-sectional questionnaire study, conducted in Spain, analysed 101 valid responses [Fn39]: the majority of respondents reported that their expectations were not met and that they would not recommend the model [Fn36]; cost was the primary motivation for choosing this treatment, followed by convenience [Fn37]; and many users experienced complications, leading to dental consultations [Fn38]. The authors conclude that the model offers affordability and convenience but raises concerns regarding treatment quality and supervision [Fn40], highlighting the need for greater patient education and regulatory oversight [Fn40a].

The limits of these two pieces of evidence have to be stated plainly: one is a qualitative analysis of social-media posts (a self-selected sample, in which negative experiences are more likely to be posted), the other a single-country cross-sectional questionnaire of 101 people [Fn39]. Neither can yield “what the complication rate of this model is”, nor can either be compared directly for effectiveness against supervised orthodontics — this article's search obtained no such controlled study, so it provides no rate of any kind.

Where cost sits here (composition and variables only, no amounts): cost is the primary motivation for users choosing this model [Fn37], while the finance-related experiences recorded in the same body of research include billing issues and financial frustration [Fn35b], as well as the need for additional dental consultations [Fn38]. The portable verification move is this: ask whether “clinical diagnosis and supervision of the process are included in the quoted price”, “who is responsible for handling complications if they occur, and whether that is charged separately”, and “what the arrangement is if the treatment does not turn out as expected” — treat these as items in the composition of the cost — they are variables, not add-on clauses. In the Spanish study, the differences between providers appeared precisely in discomfort levels, the availability of in-person care, and whether additional consultations were needed [Fn40b].

Where the risk lies: besides the complications and delayed recognition above [Fn34][Fn38], there is an earlier and more easily underestimated class of self-treatment. A case report records a 9-year-old boy who developed severe acute periodontitis of the maxillary central incisors because he placed an elastic band himself (to close the gap between his front teeth) and the band migrated apically [Fn41]; despite periodontal and surgical intervention, those two maxillary central incisors were finally extracted [Fn42]. The author's recommendation is that treating the diastema between the anterior teeth with elastic gap bands and without fixed orthodontic appliances should be avoided [Fn43]. This is a single case report and does not represent an incidence rate, but it shows that a practice that “looks like just pulling the teeth a bit closer together” can produce an irreversible result.

Box: C (the practice itself bypasses clinical supervision).

Adjacent specific questions (there is a dedicated domain article; this article does not expand them)

- The classification of orthodontic methods, indications, treatment course and supervision mechanisms — see domain article P08, orthodontics; the logic of how orthodontic costs are composed is in canonical card KM-DENTAL-08 (in production). This article handles only the verification question “can what you buy online replace it”.

2-4 Temporary self-filling, and gluing teeth or dentures back with super glue

What the claim is: that when a tooth comes out, breaks, or a denture works loose, you can first glue it back with cyanoacrylate (“super glue”) or a home-made material, and see a dentist afterwards.

What the evidence says: this group is handled by stating first the evidence level of the correct pathway, and only then the consequences of the wrong substitute — because the former sits at a far higher level. (This article does not expand the first-aid steps themselves: what to do at the moment a tooth is knocked out is a question-level specific issue, covered by canonical card KM-DENTAL-30, “A tooth has suddenly come out — what now?” (in production); this section handles only “which level the evidence on this sits at”.)

Avulsion of permanent teeth is one of the most serious dental injuries [Fn44]; the consensus guidelines of the International Association of Dental Traumatology state plainly that prompt and correct emergency management is essential for attaining the optimal outcome after this injury [Fn45]. The purpose of those guidelines is to provide clinicians with the most widely accepted and scientifically plausible approaches for the immediate or urgent care of avulsed permanent teeth [Fn46a]. The same guidelines also mark their own limits honestly: the association states that it does not, and cannot, warrant favourable outcomes from adherence to the guidelines [Fn46] — that an official guideline is willing to write that sentence down is itself a positive signal of credibility.

A systematic review including 19 cross-sectional studies, 5,752 participants and 10 countries then shows how large the knowledge gap is [Fn50]: the prognosis of an avulsed tooth depends heavily on immediate and appropriate first-aid management, including prompt reimplantation or correct temporary storage of the tooth [Fn47]; yet understanding of the key steps — including the optimal replantation window, the appropriate storage medium and correct handling technique — is generally inadequate across most populations [Fn48] (the specific content of each of those three items belongs to KM-DENTAL-30, and under the non-conflict red line this article only links down to it rather than restating it). The same review records that awareness that an avulsed permanent tooth can be replanted ranged very widely, from 23% among parents to over 80% among dental professionals [Fn49a]; in non-dental populations, only 2% to 37% indicated willingness or ability to attempt immediate replantation [Fn49], while misconceptions regarding inappropriate storage materials, delayed referral and handling of the root surface were prevalent [Fn51]. That same review also notes that studies which included educational interventions demonstrated marked improvements in awareness and decision-making [Fn51a] — which is precisely the reason this article exists.

As for the super-glue route, the direct evidence retrievable sits at the level of a single case report: a patient fabricated 3 prostheses for herself out of the self-polymerising acrylic resin used for nail thickening, and cemented them into her mouth with super glue [Fn52]; the author's warning is that patients must be warned not to attempt self-treatment for aesthetics with self-fabricated prostheses, because severe adverse and irreversible hard and soft tissue reactions may occur [Fn53][Fn54].

An evidence gap that has to be marked honestly: this article's search obtained no systematic review or cohort study designed around “patients treating a broken tooth, a lost tooth or a loose denture themselves with super glue” — self-treatment behaviour of this kind is inherently hard to study prospectively. The case report above can therefore serve only as an individual warning, not as a claim about incidence or statistical significance; the authority for the correct pathway is carried by the consensus guidelines [Fn45][Fn46a] and the systematic review [Fn47][Fn48]. By the same token, for “filling a hole yourself with an over-the-counter temporary filling material” this article likewise obtained no citable clinical study, and therefore makes no judgement at all as to whether it works or does not work; it notes only that this shares one structural problem with the self-treatment above: handling a defect of unknown cause without a diagnosis.

Where the risk lies: the time window is a key variable [Fn48], and any self-attempt that consumes time compresses that window; as for the self-gluing route, what the source gives is a preventive warning — severe and irreversible hard and soft tissue adverse reactions may occur with self-treatment using self-fabricated prostheses [Fn53][Fn54] (that case report does not record the actual tissue outcome for that patient, so this article does not write it as “has been recorded as severe and irreversible”).

Adjacent specific questions (each has its own canonical card and domain article; this article does not expand them)

- What to do at the moment a tooth is knocked out (first-aid steps, time-window figures, the order of preference for storage media, follow-up) — see canonical card KM-DENTAL-30 (A tooth has suddenly come out — what now?, in production). This article handles only “which level the evidence sits at, for the correct pathway and for self-gluing respectively”.

- Choice of filling material, indications and longevity — see canonical cards KM-DENTAL-12 and KM-DENTAL-16 (both in production) and domain article P04, caries and fillings.

- Reading toothache and trauma at the time and knowing when to seek care — see canonical card KM-DENTAL-33 (in production) and domain article P13, symptom triage and seeking care.

2-5 The amalgam-toxicity thesis (“silver fillings”)

What the claim is: that amalgam fillings contain mercury, poison the person carrying them, and should be actively removed and replaced.

What the evidence says: this group cannot be written off as “pure rumour”, because two real but different axes exist at the same time — individual health risk and environmental and occupational exposure policy. Conflating the two is the commonest misreading on this issue, and it goes wrong in both directions.

Axis one: individual health risk for the general population. The official patient-information page of the US Food and Drug Administration gives a direct answer to “should I have my fillings taken out”: if your fillings are in good condition and there is no decay beneath the filling, the agency does not recommend that you have your amalgam fillings removed or replaced [Fn55]; removing intact amalgam fillings results in unnecessary loss of healthy tooth structure and exposes you to a temporary increase in mercury vapour released during the removal process [Fn56]. The same page carries a further sentence that must be read whole: although the available evidence does not show that exposure to mercury from dental amalgam will lead to adverse health effects in the general population, exposure to mercury may pose a greater health risk in the groups of people listed on that page [Fn57a] — the original is a concessive sentence introduced by “although”, and lifting out the first half to stand alone as an independent conclusion is exactly the reading §1-6 asks readers to catch (the list of those groups is in “who needs a separate discussion” later in this section [Fn57]). The policy and position statements of the International Association for Dental Research point the same way: current scientific evidence supports the safety of dental amalgam for the general population, excluding individuals with allergies to amalgam components or severe renal impairment [Fn59]; and the association confirms that mercury bound in the alloy poses a negligible health risk under normal conditions of use [Fn60].

Axis two: policy and the environment. The same association statement, while confirming safety, explicitly supports a responsible phase-down strategy aligned with the Minamata Convention on Mercury [Fn61]. The background section of a 2026 Cochrane overview of systematic reviews likewise states that there are concerns about toxicity to human health and the environment [Fn62], and that the Minamata Convention on Mercury recommends a phase-down of amalgam use in dentistry [Fn63]. “Safe” and “phase down” are not in contradiction here — the former is about the clinical risk to an individual patient, the latter about environmental and occupational exposure across the whole life cycle of the material.

Axis three: occupational exposure is a different population. An occupational-health study measured the particulate generated when dental amalgam is removed with a high-speed drill: the particulate generated by that procedure volatilises significant amounts of mercury vapour [Fn64], at levels that frequently exceed the safety thresholds of several jurisdictions and bodies [Fn65a]; the authors characterise this as a significant, under-recognised source of localised exposure to mercury vapour [Fn65]. This entry has to be read with its qualifications, and its funding checked once by the reader under the framework of §1-5: it is a single study, and what it measures is occupational exposure among dental staff, not health outcomes in patients; all four authors' affiliations as registered on PubMed are private dental practices, and the funding section of the PMC full text states that the study was funded by the authors themselves (its conflict-of-interest statement records no competing interests) — this article discloses those bibliographic facts for the same reason as it discloses them for the baking soda entry in §2-2. It cannot be used to claim “a patient who has amalgam removed will be poisoned”. But it in fact points the same way as the official position — both state that “the act of removal itself generates mercury vapour” [Fn56][Fn64], so having intact fillings taken out without a clinical reason is manufacturing a known exposure event out of nothing.

Who needs a separate discussion: the official page also lists the groups who may be more susceptible to potential adverse effects generally associated with mercury (pregnant women and their developing fetuses; women who are planning to become pregnant; nursing women and their newborns and infants; children, especially those younger than six years of age; people with pre-existing neurological disease; people with impaired kidney function; and people with known heightened sensitivity to mercury or to the other components of dental amalgam — silver, copper, tin) [Fn57]; and it states plainly that if you are allergic or otherwise sensitive to any of the metals in dental amalgam, you should not have amalgam fillings [Fn58]. The exclusion clause in the association statement points to the same set of people [Fn59]. This part is “what to discuss with your dentist”, not a checklist for judging it yourself.

Where the risk lies: risk exists in both directions. The risks of actively removing an intact filling have been listed explicitly by the official page (loss of healthy tooth structure plus a temporary increase in mercury-vapour exposure) [Fn56]; while belonging to one of the susceptible groups above and not discussing it with a dentist is equally a risk [Fn57][Fn58].

Box: E (a genuine scientific tension exists; this is not a simple rumour). The accurate one-sentence version is: “the individual risk to the general population is assessed as negligible, and active removal is not recommended; at the same time, a global phase-down policy grounded in environmental and occupational exposure genuinely does exist — both of these things are true.”

Adjacent specific questions (not expanded here, to avoid duplicating existing cards)

- Comparison of the clinical performance of the various filling materials (failure rates, secondary caries, post-operative discomfort and the limits on what can be inferred from those figures) — see canonical cards KM-DENTAL-12 and KM-DENTAL-16 (both in production) and domain article P04, caries and fillings; this article deliberately does not restate that set of figures. The Cochrane overview cited above itself emphasises that caries prevention is critical to effective and sustainable oral health [Fn63a].

2-6 The root-canal-causes-cancer thesis and the focal-infection theory

What the claim is: that a tooth which has had root canal treatment becomes a chronic source of infection, causing disease in distant organs and even cancer, and that root-treated teeth should therefore be extracted.

What the evidence says: one honest sentence first — this article obtained no cohort study or systematic review that directly investigates the association between root canal treatment and cancer incidence. It therefore does not claim that “research has established that root canal treatment does not cause cancer”, and it provides no epidemiological figures on the matter. What it can give is the historical root of this alarm and the direction of modern mechanistic evidence; that chain of argument is itself solid, but it is not an exhaustive exclusion.

The historical root: the “era of dental focal infection” from the end of the 19th to the beginning of the 20th century resulted in widespread implementation of tooth extraction and limited the progress of endodontics [Fn66]; the theory proposed that bacteria and toxins entrapped in dentinal tubules could disseminate systemically to remote body parts, resulting in many types of degenerative systemic disease [Fn66a]. An interdisciplinary consensus document records that systemic diseases were at the time attributed to focal infections such as caries and periodontal disease, and that dentists were held responsible for these diseases due to the spread of oral infections [Fn70]. That theory was eventually refuted because it rested only on anecdotal evidence [Fn67].

The modern position is “more complex”, not “back to the old theory”: as knowledge of the interrelationship between the oral microbiota and the host immune response has evolved, the focal infection theory has been modified in various ways [Fn71a]; the relationship between oral and systemic health appears to be more complex than that suggested by the classical theory of focal infections [Fn71]. That interdisciplinary working group (covering dentistry, otorhinolaryngology, cardiology, orthopaedics, oncology and diabetology) devised scoring algorithms for odontogenic inflammatory diseases and systemic risks, and developed standardised general procedures [Fn70a] — that is, modern practice brings oral foci into a standardised assessment pathway; it neither rejects the old theory wholesale nor accepts it wholesale.

Modern evidence pointing the other way: the conclusion of a narrative review from 2022 reads that convincing evidence supports that successful root canal treatment has a beneficial impact on systemic health by reducing the inflammatory burden, thereby dismissing the misconceptions of the focal infection theory [Fn68]. A 2026 systematic review and meta-analysis (registered with PROSPERO) included 20 clinical studies, of which 18 reported a reduction in inflammatory biomarkers after endodontic treatment [Fn72]; the meta-analysis showed a significant reduction in hs-CRP [Fn73], while no significant changes were found in the other biomarkers [Fn74]. The timing is a condition of that result holding, and cannot be omitted: the original assessed two time points, 1 month and 6 months after treatment, and the one that reached significance was 6 months after treatment [Fn73a].

The limits have to be stated alongside: the certainty of the evidence in the latter was classified as low under GRADE [Fn75]; and the former likewise marks honestly that further high-quality research is required to support and establish the benefits of endodontic treatment on systemic health [Fn69]. Moreover, that same narrative review also records that there is emerging evidence that bacteraemia and low-grade systemic inflammation associated with apical periodontitis may negatively impact systemic health [Fn68a] — in other words, it is untreated apical lesions that the literature is concerned about, which is exactly the opposite direction from “treated teeth are harmful”.

Where the risk lies: the real risk of this claim is not in the idea but in the action — its logical end point is extracting teeth that could have been kept. The historical consequence of the focal-infection era has been recorded explicitly as “widespread implementation of tooth extraction” [Fn66].

Box: D (a historical theory has been refuted and the modern evidence points the other way).

Adjacent specific questions (there is a dedicated canonical card and domain article; this article does not expand them)

- What root canal treatment is, the procedure, success rates and alternatives — see canonical card KM-DENTAL-18 (in production) and domain article P03, root canal treatment. This article handles only the verification question of “cancer causation / focal infection”.

2-7 The NICO (neuralgia-inducing cavitational osteonecrosis) cavitation thesis

What the claim is: that “cavitation” lesions inside the jawbone, hard to detect on imaging, are the cause of chronic facial pain and need to be treated by surgical curettage.

What the evidence says: this section needs particularly careful wording, because the patient's pain is real — what is being questioned is this diagnostic label and the surgical decisions attached to it, not the patient's symptoms.

NICO has been described as a debated diagnostic entity proposed as a cause of chronic orofacial pain [Fn76]. A scoping review searching 1979 to 2024 yielded 603 records and included 41 studies after screening [Fn87a], and what it collates is this: those 41 studies revealed recurrent methodological limitations [Fn87] — absence of control groups, non-standardised diagnostic criteria, limited histological evidence, and interpretive bias [Fn3]; the proposed mechanisms (such as RANTES/CCL5 overexpression) lack independent validation [Fn10]; diagnostic tools remain unvalidated, while surgical interventions continue on the basis of unclear criteria [Fn78]; and the evidence consists predominantly of low- to moderate-quality retrospective case series, with no randomised controlled trials [Fn88]. The authors' core conclusion is that current evidence is insufficient to support NICO as a distinct, validated clinical entity [Fn77]. The same paper also notes that many patients diagnosed with NICO may be more appropriately classified, under the international classification of orofacial pain, as having persistent idiopathic facial pain or painful trigeminal neuropathy [Fn79].

An earlier systematic review (searched to June 2020; 29 studies included out of 4,051) [Fn86] is consistent in its findings: with the exception of hereditary coagulopathies, which were identified as potential risk factors in five of those studies, suggestions concerning the aetiology varied widely [Fn85a] (that exception has to be written out too, or “no consensus at all on aetiology” would be more absolute than the source itself); no gold standard diagnostic means could be identified [Fn82]; treatment was most often performed by surgical curettage of the affected bone [Fn84a]; and the surgical results were likewise divergent — significant facial pain remission was reported at rates as low as 66% and as high as all cases, with follow-up periods spanning 2 months to 18 years [Fn85b], while no or little relief, and recurrences, were reported in up to a third of cases [Fn85]; all studies were observational in their design [Fn4], and all were rated as poor quality because of high risk of bias and non-transparent reporting [Fn84]. Its summary is that the evidence concerning the aetiology, diagnosis and treatment of NICO is poor [Fn83].

The key to reading this passage: figures of the “as low as sixty-six per cent, as high as complete remission” kind, in observational studies with no control group [Fn4], are almost impossible to separate from the natural course of the condition, from placebo effects or from regression to the mean; and the follow-up periods run from 2 months to 18 years [Fn85b] — follow-up duration is precisely the key variable for reading an uncontrolled remission rate, and without it that range is not even internally comparable. This is the framework of §1-4 in practical application.

Where the risk lies: the scoping review authors put it very directly — the substantial physical and psychological risks of surgical intervention underscore the ethical imperative to base treatment decisions on validated evidence [Fn80]; clinicians should exercise caution, prioritising evidence-based non-surgical management and avoiding irreversible surgical interventions without clear justification [Fn81].

Box: F (the evidence is insufficient to support it as a distinct disease entity). This is not the same as “your pain is imaginary” — the alternative route the literature offers is a return to the validated classification of orofacial pain [Fn79]. The full framework for chronic orofacial pain belongs to domain article P09, the temporomandibular joint and the maxillofacial region, and is not expanded here.

2-8 “Tooth detox” and oral detoxification claims

What the claim is: that certain practices or products can draw “toxins” out of the teeth or the mouth.

What the evidence says: this article's search obtained no clinical study at all with “oral detoxification” as a research outcome — and that in itself is part of the answer. What can be retrieved is this: the term is recorded as one of the unsubstantiated therapeutic claims in internet advertisements, listed alongside claims of antibacterial, antifungal and antiviral action [Fn24]; and for the product category that carries those claims, neither the safety nor the efficacy is substantiated by the available clinical and laboratory data [Fn25].

This section deliberately draws no further inference: this article does not claim that “detoxification is entirely ineffective”, because disproving an outcome that has never been defined is methodologically impossible; it states only the verifiable fact — that this is a term classified in the literature as an “unsubstantiated claim” [Fn24].

The portable verification move: when you meet a word like “detox”, ask first “what is the toxin, how is it measured, and what measured change is seen after it has been drawn out”. The three-level check of §1-6 applies here in full.

Where the risk lies: the same as §2-2 — the main cost of putting resources and time into unsubstantiated claims is delay.

3. “Is the treatment my dentist described being sold to me?”

This is the section of the article that needs the most careful wording. Everything below is a research phenomenon at population level; none of it is an accusation against any individual clinician, institution or country. This article also provides no checklist for “how to spot a dishonest clinician” — the literature does not support such a checklist.

3-1 The phenomenon has indeed been recorded within study designs

A pseudo-patient cross-sectional study conducted in South Korea took place between August and December 2018, in 196 private dental clinics with 58 pseudo-patients [Fn89]; these trained pseudo-patients had been confirmed beforehand by two experienced dentists to have no previous oral disease, including dental caries [Fn89a]. The results: in 33.2% (65/196) of the interactions, the pseudo-patients were diagnosed as having no dental caries [Fn90]; while in 20.9% (41/196) of the sample, five or more carious teeth were diagnosed [Fn91]. The authors conclude that the study findings indicate that Korean dentists tend to over-diagnose dental caries, which could pose a threat to public health both in Korea and worldwide [Fn92].

The boundary of that figure has to be nailed down: it comes from a single country, a single period and a single study design, and used visual and tactile examination methods only. The claim it supports is “overdiagnosis is a phenomenon that has been recorded by a systematic study design”; it cannot be read as “the average overdiagnosis rate of dentists worldwide”, still less as a judgement on any individual dentist [Fn89][Fn92]. This article's search obtained no comparable pseudo-patient study from another country for cross-national comparison.

Another questionnaire study, conducted in Spain, asked 42 final-year dental students (52 questionnaires were sent out) about their diagnostic and treatment decisions across 10 clinical scenarios [Fn98a]: in the scenario of “repairing a defective existing restoration”, an overtreatment was proposed by 87.8% of the students [Fn98]; and a tendency to perform unnecessary complementary tests for caries diagnosis was observed [Fn99]. The figures pointing the other way in the same abstract have to be read alongside, or the sampling is selective: that study also records that the caries treatment proposals were in accordance with available evidence in the majority of the cases [Fn99a], that the treatment for carious lesions at different stages was correct in between 51.2% and 92.7% of the cases [Fn99b], and that 58.8% of the students correctly identified an early carious lesion on the occlusal surface of a molar [Fn99c]. The subjects of that study are students, not practising clinicians, the sample is small, and it cannot represent clinical reality in any country; the reason this article cites it is that it points to a different cause — criteria and training, rather than motive.

3-2 A more constructive explanation: part of overdiagnosis comes from reading the images

Radiographic diagnosis of caries is often complicated by optical effects on the image, such as the Mach band effect and triangular-shaped radiolucencies (TSR) [Fn93a]; these phenomena can give rise to false-positive diagnoses, especially on bite-wing radiographs, thus influencing clinical decision-making and possibly leading to overdiagnosis and overtreatment [Fn93]. A systematic review with meta-analysis (registered with PROSPERO; only 5 studies included out of 640 reports) [Fn95] quantified this: the overall prevalence of non-carious TSR on maxillary molars was 26.44% (270/1021) [Fn94a]; and the meta-analysis showed that these effects led to false-positive diagnoses of caries or fractures in approximately 13% of observations (60/464) [Fn94].

The limits: only 5 studies were included, heterogeneity was high, and the certainty of the evidence was classified as low to very low [Fn96]. That review's clinical recommendation is that clinicians should interpret radiographic findings with caution and always correlate them with a thorough clinical examination [Fn97].

Why this section matters: it turns the question from “is this clinician honest” into “what is this diagnosis based on”. The latter is a question a patient can ask directly and a clinician can answer directly; the former is not.

3-3 Upstream structural factors

The Lancet oral health Series points out that in high-income countries the current treatment-dominated, increasingly high-technology, interventionist and specialised approach is not tackling the underlying causes of disease [Fn100]. That is a systemic commentary on an entire model of care, not an assessment of individual practitioners; it belongs to the same set of structural observations as the conflict-of-interest issue in §1-5 [Fn9][Fn10a].

3-4 The portable verification moves (taking the framework back into the surgery)

Below, the five questions of §1 are translated into forms that can be asked out loud in the surgery, each corresponding to the evidence in this section:

• Ask what the diagnosis is based on: what is this judgement based on? Does what is seen on the image match the clinical examination? — corresponding to the fact that imaging artefacts may produce false positives [Fn93][Fn97].
• Ask about the time line: what happens if this is not dealt with now, and how long before it is different? — corresponding to the evidence that “divergence in diagnosis itself exists” [Fn90][Fn91].
• Ask about alternatives: besides this approach, what other options are there, and what are the trade-offs of each? — corresponding to the thinking behind modern standardised assessment pathways [Fn70a].
• Ask how the cost is composed: what items does this cost include, which of them are variables, and what might be added later? (This article gives no amounts.) — corresponding to the part most easily overlooked when cost is the motive for a decision [Fn37][Fn35b].
• A second opinion is not distrust: in situations where the certainty of the evidence is rated low [Fn96][Fn75], divergence between different clinical judgements is to be expected.

The risk in the other direction has to be stated just as plainly: nothing in this section amounts to advice “not to have treatment”. As already written in the previous section, it is untreated apical lesions that the literature is concerned about [Fn68a]; and “delay” is the principal risk this article points to from beginning to end.

Adjacent specific questions (there is a dedicated domain article; this article does not expand them)

- Symptom triage, when to seek care and red-flag criteria — see domain article P13, symptom triage and seeking care; the universal logic of costs and insurance and the differences between localities are in domain article P12, dental costs and insurance systems. Local systems and costs are in the corresponding canonical cards (TW).

4. Folding the framework back up: a reusable self-check table

5. When not to stop and verify, but to seek care straight away

Verification exists to make better decisions, not to postpone them. The greatest risk this article points to throughout is delay, and in the following situations the evidence all points to “time itself is a variable”:

• A tooth knocked out (avulsed): this is one of the most serious dental injuries [Fn44], prompt and correct emergency management is essential to obtaining the optimal outcome [Fn45], and the time window for replantation and the storage medium are key variables listed explicitly in the literature [Fn48]. In this situation, do not go online to check and do not look for an adhesive; follow first-aid principles and seek dental help immediately [Fn47]. The specific content of the first-aid steps and the time window is in canonical card KM-DENTAL-30 (in production) and is not restated here.
• Self-treatment that has already caused symptoms: self-fabricated appliances and super glue may cause severe and irreversible hard and soft tissue reactions [Fn53]; in the elastic-band case, severe acute periodontitis had already developed by the time of intervention [Fn41].
• Undergoing unsupervised orthodontic treatment and experiencing discomfort or abnormal progress: reliance on self-monitoring and peer advice may delay recognition of complications [Fn34].
• Belonging to a group more susceptible to mercury exposure and having concerns about a filling: the instruction on the official page is that if you have a relevant health condition you may discuss the need for removal and replacement with your dentist and/or physician [Fn58a] — rather than deciding to have it taken out on your own [Fn55]; the susceptible groups and the allergy contraindication listed on that page are in [Fn57][Fn58].
• Chronic orofacial pain: prioritise evidence-based non-surgical management and avoid irreversible surgical intervention without clear justification [Fn81]; and return to the validated classification framework for orofacial pain for assessment [Fn79].

This article does not compile a list of symptoms. The complete triage and red-flag criteria belong to domain article P13; this section marks only “in these situations, verification should not take priority over seeking care”.

6. Risk factors (indications / adverse effects / contraindications and limits)

Indications (the scope within which the verification framework described here applies)

• This framework applies to assessing a piece of information. It does not apply to assessing the state of your own teeth — that requires clinical examination cross-checked against imaging [Fn97].
• Assessment involving oral foci and systemic health is, in modern practice, brought into a standardised scoring and referral pathway [Fn70a], and falls within the scope of clinical professional judgement.
• The official patient-information page states explicitly that deciding what filling material to use to treat dental decay is a choice that should be made by you and your dentist [Fn55a].

Possible adverse effects and outcomes (those already recorded in the literature)

• Charcoal-based products: deleterious outcomes have been reported in research, including increased caries and enamel abrasion [Fn23]; neither their safety nor their efficacy has been substantiated [Fn25].
• Differences in abrasivity (in-vitro data, not a clinical adverse outcome): an in-vitro study assigned specimens to six groups for testing [Fn31a], and the relative dentine abrasion values for all six groups combined (regular and whitening toothpastes together) ranged from 26 to 166 [Fn31]; the same paper records that the abrasion value of the whitening toothpaste containing hydrogen peroxide was not high compared with the regular toothpaste [Fn31b], and that abrasion values vary with ingredients and mechanism [Fn30] — so the class as a whole cannot be judged, and this range cannot be hung on the single class of “whitening toothpaste”. These are in-vitro measurements on bovine specimens; they are listed here to mark “abrasivity varies” as a variable, not to treat them as an adverse outcome recorded in humans.
• Unsupervised orthodontics: user experiences already recorded include physical discomfort requiring self-management, uncertainty in treatment progress, and complications leading to separate dental consultations [Fn35a][Fn38]; reliance on self-monitoring may delay recognition of complications [Fn34].
• Self-adhesive treatment: severe and irreversible hard and soft tissue adverse reactions may occur [Fn53]; in the elastic-band case, two permanent teeth were ultimately extracted [Fn42].
• Actively removing an intact amalgam filling: this causes unnecessary loss of healthy tooth structure and exposes the patient to a temporary increase in mercury vapour [Fn56]; the removal process itself volatilises significant amounts of mercury vapour [Fn64].
• Irreversible surgical intervention on the jawbone: this carries substantial physical and psychological risks [Fn80] and should be avoided in the absence of clear justification [Fn81].
• Overtreatment and unnecessary tests: these have been recorded within research settings [Fn98][Fn99]; optical effects on images may lead to false-positive diagnoses and consequent overtreatment [Fn93].

Contraindications and limits of application

• The explicit contraindication for amalgam: people who are allergic or otherwise sensitive to mercury or to the other components of dental amalgam (silver, copper, tin) should not have amalgam fillings [Fn58]; the association's position statement likewise excludes people with allergies to amalgam components and those with severe renal impairment from its statement of safety [Fn59].
• Groups needing a separate discussion: pregnant women and their developing fetuses, women planning to become pregnant, nursing women and their newborns and infants, children (especially those younger than six years of age), people with pre-existing neurological disease, and people with impaired kidney function [Fn57].
• The ceiling of the evidence: several of this article's core sources state their own certainty of evidence as low or very low — the overall quality of the body of evidence in the oil-pulling meta-analysis was very low [Fn17]; the GRADE rating of the meta-analysis on root canal treatment and inflammatory markers was low [Fn75]; the certainty of the meta-analysis on imaging artefacts was low to very low [Fn96]; and the NICO systematic review rated all included studies as poor quality [Fn84].
• Geographical and population limits: the quantitative data on overdiagnosis come from a single study in a single country [Fn89][Fn92]; the questionnaire study on DIY orthodontics was a single country with 101 people [Fn39]; the data on information behaviour on social media come from a self-selected sample in a single country [Fn12][Fn13] — none of these may be extrapolated into global proportions.
• Evidence gaps (left blank honestly by this article): no high-quality epidemiological research directly investigating the association between root canal treatment and cancer incidence was obtained; no systematic review designed around “treating dental problems oneself with super glue” was obtained; no clinical study on “using an over-the-counter temporary filling material oneself” was obtained; no systematic review of the human safety of activated charcoal published after 2017 was obtained; and no comparable study from another country was obtained against which the South Korean pseudo-patient study could be set. This article fills in none of those five gaps and speculates about none of them.
• This article contains no medication, dose, procedural instruction or product recommendation; every statement at the level of ingredients is a compilation of the literature, not a claim about product performance.

⚠ This section is a disclosure of medical risk and does not constitute treatment advice for any individual. The actual method of treatment and its effect vary from person to person and must be assessed by a dentist.

Every clinical statement in this article is mapped line by line to its cited source (see the sources and evidence chain below). It has not been clinically reviewed by a licensed practitioner. This is health information, not individual advice; assessment by a clinician is required.