Structured Abstract

Objectives

Dental caries is a widespread chronic infectious disease, experienced by almost 80

percent of children by the age of 18 and over 90 percent of adults. Substantial

variation exists in dentists' diagnoses of carious lesions as well as in the methods

dentists use to prevent and manage carious lesions. In addition, new methods for

identifying carious lesions are beginning to appear, and new approaches for the

management of individual carious lesions and for the management of individuals

deemed to be at elevated risk for experiencing carious lesions are emerging. A

systematic review of the literature was conducted to address three related questions

concerning the diagnosis and management of dental caries: (1) the performance

(sensitivity, specificity) of currently available diagnostic methods for carious lesions,

(2) the efficacy of approaches to the management of noncavitated, or initial carious

lesions, and (3) the efficacy of preventive methods in individuals who have

experienced or are expected to experience elevated incidence of carious lesions.

Search Strategy

We conducted two detailed searches of the relevant English language literature from

1966 to October 1999 using MEDLINE, EMBASE, and the Cochrane Controlled Trials

Register. We did not pursue reports in the gray literature, i.e., information not

appearing in the periodic scientific literature. We did hand-search current journals up

to the end of 1999. One search focused on six diagnostic methods (visual and

visual/tactile inspection, radiography, fiberoptic transillumination, electrical

conductance, laser fluorescence) and combinations of these methods. A second

search focused on preventive or management methods for carious lesions, including

fluorides, pit and fissure sealants, health education, dental prophylaxis, oral hygiene,

dental plaque, chlorhexidine, dental sealants, and cariostatic agents.

Selection Criteria
We included studies in the diagnostic review that used histologic validation of caries

status and either reported results as sensitivity and specificity of the diagnosis or

reported data from which these measures could be calculated. We excluded reports of

diagnostic methods not commercially available. For the review of the dental caries

management literature, we included only reports concerning methods applied or

prescribed in a professional setting. Also, we included only studies performed in vivo

and having a comparison group. In the literature describing the management of

noncavitated carious lesions, we included only studies where the lesion was the unit

of analysis. In the literature describing the management of subjects at elevated risk

for dental caries, we included only studies where such determinations had been made

on an individual subject level based on carious lesion experience and/or bacteriologic

testing.

Data Collection and Analysis

We selected studies for inclusion from among 1,407 diagnostic and 1,478

management reports through independent duplicate reviews of titles, abstracts, and,

where necessary, full papers. We abstracted data (single abstraction, subsequent

independent review) on 39 diagnostic studies and 27 management studies using

different forms for the diagnostic and management studies. Similarly, a separate

quality rating form was completed by the scientific director for the each study.

Different rating forms were employed for the two types of studies.

Main Results

We judged the strength of the evidence describing the validity of all diagnostic

methods evaluated to be poor. There were almost no reports of diagnostic

performance of any method applied to primary teeth, anterior teeth, and root

surfaces. For posterior occlusal and proximal surfaces of permanent teeth, the

number of available studies was sufficient for some but not all methods. However,

where numbers of studies were sufficient, their quality and/or the variation among

studies precluded establishing unambiguous assessments of sensitivity and specificity.

The variation in sensitivity among methods was generally similar to the variation
reported within methods. With the exception of electrical conductance, dental caries

diagnostic methods featured criteria that maximized specificity at the expense of

sensitivity: false positive diagnoses were proportionally infrequent compared with

false negative diagnoses. In addition to the limited numbers of studies for certain

teeth and methods, the literature on diagnosis displayed a variety of serious

limitations, including the predominance of in vitro studies, small numbers of

examiners, high prevalences of lesions, and inadequate descriptions of subject

selection, examiner training and reliability, and criteria for diagnoses.

The literature on the management of noncavitated carious lesions consisted of five

studies describing seven experimental interventions. Because these interventions

varied extensively in terms of management methods tested as well as other study

characteristics, no conclusions about the efficacy of these methods were possible. We

rated the evidence for efficacy of methods for the management of noncavitated

lesions as incomplete. Standardization for the determination of noncavitated status is

needed for future studies.

The literature on the management of individuals at elevated risk of carious lesions

consisted of 22 studies describing 29 experimental interventions. We rated the

evidence for the efficacy of fluoride varnish for prevention of dental caries in high-risk

subjects as fair and the evidence for all other methods as incomplete. Because the

evidence for efficacy for some methods, including chlorhexidine, sucrose-free and

xylitol-containing gum, and combined chlorhexidine-fluoride methods, is suggestive

but not conclusive, these interventions represent fruitful areas for further research.

Conclusions

The strength of the evidence available to estimate the validity of diagnostic methods

for carious lesions dental caries is insufficient to the task. For many applications,

there are few studies, and when sufficient numbers of studies are available,

substantial variation among studies and/or the quality of the studies is problematic.

The literature describing the management of two specific dental caries-related

conditions, nonsurgical interventions for noncavitated lesions and prevention of

lesions in persons at elevated risk for new lesions, is inadequate to permit conclusions

about the efficacy of most methods. Only for two specific applications, fluoride

varnishes in caries-active, high-risk individuals and fluoride-based interventions for

individuals receiving radiotherapy was the evidence rated as fair. For all other

management methods, the evidence was judged to be incomplete. The need for

efficacy determinations is acute as much of modern preventive dental practice is

predicated on the efficacy of management methods for these conditions.

This document is in the public domain and may be used and reprinted without

permission except those copyrighted materials noted for which further reproduction is

prohibited without the specific permission of copyright holders.

Suggested Citation

Bader JD, Shugars DA, Rozier G, et al. Diagnosis and Management of Dental Caries.

Evidence Report/Technology Assessment No. 36 (prepared by Research Triangle

Institute and University of North Carolina at Chapel Hill Evidence-based Practice

Center under Contract No. 290-97-0011). AHRQ Publication No. 01-E056. Rockville,

MD: Agency for Healthcare Research and Quality. June 2001.

Summary

Overview

Dental caries, or cavities, is a chronic infectious disease experienced by more than 90

percent of all adults in the United States. Recent changes in the epidemiology of

dental caries have altered the presentation of the disease so that among children age

5 to 17 years, about 75 percent of the disease is now experienced in 25 percent of the

population. Also, as understanding of the disease process has matured, the range of

management strategies for dental caries has broadened.

Interventions to arrest or reverse the demineralization process that characterizes the

development of a carious lesion are available, and several strategies for identifying
those persons representing the quarter of the population who will experience an

elevated incidence of dental caries have been reported.

The growing sophistication in available interventions for prevention and nonsurgical

treatment of dental caries is matched by a similar increase in the available methods

for diagnosis of carious lesions. The diagnosis of carious lesions has been primarily a

visual process, based principally on clinical inspection and review of radiographs.

Tactile information obtained through use of the dental explorer or "probe" has also

been used in the diagnostic process. The development of some alternative diagnostic

methods, such as fiberoptic transillumination (FOTI) and direct digital imaging

continue to rely on the dentist's interpretation of visual cues, while other emerging

methods, such as electrical conductance (EC) and computer analysis of digitized

radiographic images, offer the first "objective" assessments, where visual and tactile

cues are either supplemented or supplanted by quantitative measurements.

This relatively recent growth in alternatives available for both diagnosis and

management of dental caries has yet to be fully assimilated by dental practice.

Thorough reviews of methods for diagnosis and management of dental caries should

assist in that assimilation process.

Reporting the Evidence

The clinical questions in this report were developed in conjunction with the planning

committee for the Dental Caries Consensus Development Conference on the Diagnosis

and Management of Dental Caries Through Life (to be held in 2001). The questions

reflect three aspects of the diagnosis and management of dental caries where the

committee perceived either that current clinical practice might not reflect current

knowledge regarding efficacy and effectiveness, or that a review of current evidence

might help stimulate new research.

The first question addresses methods used in caries diagnosis asking what the validity

of each diagnostic technique is. Diagnoses of carious lesions must be made in a

variety of sites -- primary and permanent teeth, occlusal and smooth surfaces, and

coronal and root surfaces.

Several diagnostic techniques are available, and the ability of these different

techniques to detect carious lesions on specific sites is not widely understood.

The second question concerns the efficacy of nonsurgical strategies to arrest or

reverse the progress of carious lesions before tooth tissue is irreversibly lost. The

relative effectiveness of these conservative treatments is not well identified.

The third question addresses the efficacy of preventive methods among those

individuals who have experienced, or are expected to experience, an elevated

incidence of carious lesions. Dentists are now being urged to identify individuals with

elevated caries activity, but this risk assessment strategy has not been complemented

by the identification of the most effective interventions to mitigate the expected caries

attack.

Methodology

Search Process and Inclusion Criteria

The Evidence-based Practice Center (EPC) review and investigative team conducted

two detailed searches of the relevant English language literature from 1966 to

October 1999 using MEDLINE, EMBASE, and the Cochrane Controlled Trials Register.

The team did not pursue reports in the gray literature (i.e., information not reported

in the periodic scientific literature). The team hand-searched current journals up to

the end of 1999.

One search focused on the following diagnostic methods -- visual and visual tactile

inspection, radiography, fiberoptic transillumination, electrical conductance, laser

fluorescence, and combinations of these methods -- using keywords for the disease

(dental caries, tooth demineralization), diagnostic concepts (oral diagnosis, oral

pathology, dental radiography), and study characteristics and design.

A second search focused on dental caries preventive or management methods, using

keywords for methods (fluorides, pit and fissure sealants, health education, dental
prophylaxis, oral hygiene, dental plaque, chlorhexidine dental sealants, cariostatic

agents) and study characteristics and design in addition to the disease keywords.

The EPC team applied several inclusion and exclusion criteria to the reports identified

in our literature search. The team included studies in the diagnostic review that used

histological validation of caries status, and either reported results as sensitivity and

specificity of the diagnosis or reported data from which these measures could be

calculated. The team excluded reports of diagnostic methods not commercially

available. For the review of the dental caries management literature, the team

included only reports concerning methods applied or prescribed in a professional

setting, and only studies performed in vivo and having a comparison group.

The two disease management questions that were addressed by the team used the

results of the management review and featured additional inclusion criteria. For the

management of non-cavitated carious lesions, the team included only studies where

the lesion was the unit of analysis. The team accepted several different descriptions of

noncavitated lesions (including the terms "incipient" and "initial)." From the literature

describing the management of subjects at elevated risk for dental caries, the team

included only studies where the classification of elevated risk had been made for

individual subjects and was based on carious lesion experience and/or bacteriological

testing. The team accepted the elevated risk classification described in the paper.

The EPC team selected studies for inclusion from among 1,407 diagnostic and 1,478

management reports through independent duplicate reviews of titles, abstracts, and,

where necessary, full papers, with discussion leading to consensus where

disagreement occurred. Two team reviewers agreed on inclusion status for 97 percent

of the reports at this stage. In addition, the reviewers separately identified six studies

evaluating preventive methods in patients who had received radiotherapy for head

and neck neoplasms (a special high-risk group) and seven studies evaluating

preventive methods in patients with orthodontic bands or brackets (another special

high-risk group). The team believed that these studies should be included in the

review, but not combined with the main group of studies due to substantial

differences in lesions and study methods.

The team abstracted data (single abstraction, subsequent independent review) on 39

diagnostic studies and 27 management studies, using different forms for the

diagnostic and management studies. Four reviewers were involved in the abstraction

process, with reviewer agreement rates of 100 percent for results and 88 percent for

other study descriptors. Separate quality rating forms were completed by the EPC

team's scientific director for the two types of studies. The quality rating scales

assessed several elements of internal validity, including study design, duration,

sample size, blinding, baseline assessments of differences among groups, loss to

followup, and examiner reliability. Two items also requested the reviewer's subjective

assessment of both the internal and external validity of the study.

The team compiled the abstracted data in a series of six evidence tables, one each for

in vivo and in vitro radiographic studies, studies of management of noncavitated

carious lesions and individuals at elevated risk for carious lesions, and studies of

special populations of orthodontic patients and patients who received head and neck

radiotherapy. The team then graded the evidence summarized in the tables.

For the diagnostic question, the strength of the evidence was judged in terms of the

extent to which it offered a clear, unambiguous assessment of the validity of a

particular method for identifying a specific type of lesion on a specific type of surface.

The three possible ratings were:

• Good (A): The number of studies is large, the quality of

the studies is generally high, and the results of the studies

represent narrow ranges of observed sensitivity and specificity.

• Fair (B): There are at least three studies, the quality of

the studies is at least average, and the results represent

moderate ranges of observed sensitivity and specificity.

• Poor (C): There are fewer than three studies, or the

quality of the available studies is generally lower than average,

and/or the results represent wide ranges of observed

sensitivities and/or specificities.

For purposes of this question, a narrow range is defined as no more than 0.15 on a

scale of 0.0 to 1.00, a moderate range is no more than 0.35, and a wide range is

more than 0.35. High quality is defined as most study scores at or above 60, and

average quality is defined as most study scores at or above 45.

For the management studies, the team used a scheme based on several

considerations, including the magnitude of the results reported, the quality rating

scores of the studies, the number of studies, and the consistency of the results across

studies. The EPC team's scientific and clinical directors independently rated the

interventions and developed an adjudicated final rating. The four possible ratings

were:

• Good (A): Data are sufficient for evaluating efficacy.

The sample size is substantial, the data are consistent, and the

findings indicate that the intervention is clearly superior to the

placebo/usual care alternative.

• Fair (B): Data are sufficient for evaluating efficacy. The

sample size is substantial, but the data show some

inconsistencies in outcomes between intervention and

placebo/usual care groups such that efficacy is not clearly

established.

• Poor (C): Data are sufficient for evaluating efficacy. The

sample size is sufficient, but the data show that the

intervention is no more efficacious than placebo or usual care.

• Incomplete Evidence (I) Data are insufficient for

assessing the efficacy of the intervention, based on limited

sample size and/or poor methodology.

Findings

Diagnostic Methods

The EPC team evaluated the strength of the evidence describing the performance of

diagnostic methods separately for cavitated lesions, lesions involving dentin, enamel
lesions, and any lesions. The team also separated the evaluations by the surface and

tooth type involved. The team found 39 studies reporting 126 histologically validated

assessments of diagnostic methods.

• There are few assessments of the performance of any

diagnostic methods for primary or anterior teeth and no

assessments of performance on root surfaces. The strength of

the evidence describing the performance of any method for

these teeth and surfaces is poor.

• Among studies assessing diagnostic performance for

proximal and occlusal surfaces in posterior teeth, the team

rated the strength of the evidence describing the performance

of visual/tactile, FOTI, and laser fluorescence methods as poor

due to the small numbers of studies available.

• The team also rated the strength of the evidence for

radiographic, visual, and EC methods as poor for all types of

lesions on posterior proximal and occlusal surfaces. However,

these ratings were due less to inadequate numbers of

assessments than to variation among reported results. In one

instance, the quality of the available studies was the principal

reason for the rating.

• For all but EC assessments, specificity of a diagnostic

method was generally higher than sensitivity. Thus, false

negative diagnoses are proportionally more apt to occur in the

presence of disease than are false positive diagnoses in the

absence of disease.

• The evidence did not support the superiority of either

visual or visual/tactile methods. The number of available

assessments was small and there was substantial variation

among reports for each method.

• The evidence suggests, but is not conclusive, that some

digital radiographic methods offer small gains in sensitivity

 compared with conventional film radiography on both proximal

and occlusal surfaces.

• The evidence also suggests, but is not conclusive, that

EC methods may offer heightened sensitivity on occlusal

surfaces, but at the expense of specificity.

• The diagnostic performance literature is limited in terms

of numbers of available assessments for most diagnostic

techniques overall, and especially for primary teeth, anterior

teeth, and root surfaces and for visual/tactile and FOTI

methods. The literature is further limited by threats to both

internal and external validity represented by incomplete

descriptions of selection and diagnostic criteria and examiner

reliability, the use of small numbers of examiners,

nonrepresentative teeth, samples with high lesion prevalence,

and a variety of reference standards of unknown reliability.

Management of Noncavitated Carious Lesions

We found only five studies addressing this topic. The evidence was rated as

incomplete.

This literature is limited by:

• Differences in treatment provided to comparison groups

and in how noncavitated lesions are defined.

• Problems in the identification and control of patient

exposure to community-based and individual preventive dental

procedures.

• High loss to followup due in part to limiting analyses

only to full participants.

All of these limitations make drawing conclusions difficult.

Management of Caries-Active Individuals

The EPC team evaluated the evidence for nine management methods: fluoride

varnishes, fluoride topical solutions, fluoride rinses, chlorhexidine varnishes,

chlorhexidine topicals, chlorhexidine rinses, combined chlorhexidine-fluoride

applications, sealants, and other approaches. The team based its review on 22 studies

that described 29 experimental interventions evaluating these. The team also

examined 13 studies of special at-risk populations (orthodontic and head and neck

radiotherapy patients).

• The team rated the evidence for the efficacy of fluoride

varnishes as fair, and the evidence for all other methods as

incomplete.

• The evidence for efficacy was suggestive for

chlorhexidine varnishes and gels, for combination treatments

including chlorhexidine, and for sucrose-free gum, but in each

instance the number of studies was too small or the results

were too variable to be conclusive.

• Among subjects undergoing orthodontic treatment with

attached bands or brackets, the team found the evidence for

efficacy of fluoride interventions to be suggestive but

incomplete. Evidence was also incomplete for all other

prevention methods for these subjects.

• Among patients receiving head and neck radiotherapy,

the literature offers fair evidence of the efficacy of fluoride-

based interventions. The evidence was incomplete for any other

types of preventive interventions among these patients.

• The team found no reports of substantive harms

associated with any interventions.

• The team found the number of available studies for any

specific method to be a serious limitation. Among studies

addressing a method, the variety of experimental protocols,

comparison groups, and other community and individual

preventive dentistry exposures further restricted the

 opportunity to draw conclusions about the efficacy of the

method. Finally, generalization from the studies to the broader

U.S. population is problematic, as nearly all studies included

only children and evaluated changes only in the permanent

dentition.

Future Research

Research is needed to evaluate the performance of all diagnostic methods currently

available to dental practitioners. Such research should focus on in vivo settings to the

extent possible, despite difficulties imposed by the requirement for histological

validation in that environment. Methods for histological validation should be

standardized, and a standard reporting format for evaluation of diagnostic

performance should be formulated. Several aspects of study designs in this literature

should be strengthened, including using samples with representative lesion

prevalences and presentations, increasing the numbers of examiners whose

performance is assessed, and ensuring examiner blinding for determinations of both

experimental diagnoses and reference standards. Finally, research is needed to

evaluate the "downstream" performance of diagnostic methods; i.e., the

appropriateness of treatment provided in response to the diagnosis and diagnostic

performance in detection of changes in lesion volume.

Additional clinical studies examining outcomes of management strategies for

noncavitated lesions and for caries-active patients are clearly needed. Here

investigators must be encouraged to contribute studies that fill identified gaps, that

build upon existing findings, and that use methods that facilitate comparison across

studies. Funders and editors are important gatekeepers in this respect.

Whenever possible, studies should use comparison groups representing the most

common alternative treatment, and they should document all professional,

community, and individual preventive dentistry exposures for all subjects. Intention to

treat analyses, where all outcomes of all subjects enrolled at baseline are included in

the analyses, are to be encouraged as well.

Secondary analyses of existing studies of preventive agents might be exploited in the

short-term to augment the meager store of knowledge for both noncavitated lesions

and caries-active individuals. However, some additional efforts need to be extended

for the development of valid standard criteria for these classifications.

Chapter 1. Introduction

Primary Objectives and Scope of this Evidence Report

The National Institute of Dental and Craniofacial Research (NIDCR) is collaborating

with the Agency for Healthcare Research and Quality (AHRQ) in supporting a series of

systematic analyses of oral health topics at the beginning of the new century. NIDCR

selected dental caries as the inaugural topic in the series partly because current

approaches to diagnosis, treatment, and prevention of this most widespread of

chronic diseases have become subjects of increased interest as the expression of the

disease in the population has changed.

NIDCR has scheduled a National Institutes of Health (NIH) NIDCR Consensus

Development Conference (CDC) on Diagnosis and Management of Dental Caries

Through Life. The conference will address most aspects of the diagnosis and

prevention of dental caries. The Evidence-based Practice Center (EPC) was asked to

"anchor" the conference through the preparation and presentation of evidence-based

reviews for selected aspects of conference topics. An objective of the CDC, and the

principal objective of the evidence report, is to identify valid diagnostic methods for

various lesions and effective professional preventive strategies for specific types of

lesions and patients.

The treatment of dental caries has long claimed the majority of dentists' efforts, and

until the last three decades, much of that effort was devoted to repairing teeth that

had suffered irreversible loss of tissue and removing teeth deemed unsalvageable.1

With the advent of water fluoridation, fluoridated dentifrices, and both community-

based and professional fluoride treatments, the nature of the disease has gradually

changed for the majority of the population, who now experience what can be
characterized as a more gradual and limited caries onset, with fewer lesions

manifesting and progression of these lesions seemingly occurring more slowly.2 For a

minority of individuals, however, caries incidence continues unabated, with the result

that although caries is still ubiquitous, a relatively small proportion of the population

now bears a large majority of the disease in terms of the number of lesions

experienced.3 Thus, dentists now routinely encounter a distribution of disease among

their patients that was uncommon two decades ago.

Concomitantly, as knowledge of the carious process is progressively refined, dentists

are increasingly urged to view dental caries as a chronic infection, and more attention

is being paid to the elimination of the infection as a key step in treatment.4 Also,

nonsurgical treatment interventions are gaining in popularity as alternatives to

mechanical replacement of damaged tooth tissue with artificial materials. Thus, at the

same time that differences among patients in caries activity and perceived caries risk

are raising new questions about the appropriate preventive and treatment strategies

for individual patients, the range of possible strategies that can be applied to these

patients is increasing.

This growing complexity in methods for caries management is matched by a similar

increase in the complexity in methods for caries diagnosis. The diagnosis of carious

lesions has been primarily a visual process, based principally on clinical inspection and

review of radiographs. Tactile information obtained through use of the dental explorer

or probe has also been used in the diagnostic process. Chiefly because these methods

depend on subjective interpretation of subtle visual and tactile cues, variation among

dentists' diagnoses had tended to be extensive.5 Some developing alternative

diagnostic methods, such as fiberoptic transillumination (FOTI) and direct digital

imaging, continue to rely on dentists' interpretation of visual cues, whereas other

emerging methods, such as electrical conductance (EC) and computer analysis of

digitized radiographic images, offer the first "objective" assessments, where visual

and tactile cues are either supplemented or supplanted by quantitative

measurements.

Key Clinical Questions

The clinical questions in this report were developed in conjunction with the planning

committee for the CDC. They reflect three aspects of the diagnosis and management

of dental caries where the committee perceived either that current clinical practice

might not reflect current knowledge regarding efficacy and effectiveness or that a

review of current evidence might help stimulate new research.

The first question addresses methods used for identifying carious lesions. At issue is

the validity of each diagnostic technique. Lesions must be identified in a variety of

sites -- primary and permanent teeth, occlusal and smooth surfaces, and coronal and

root surfaces. Several diagnostic techniques are available, and the ability of these

different techniques to detect carious lesions on specific sites may not be completely

appreciated.

The second question concerns the effectiveness of strategies to arrest or reverse the

progress of carious lesions before tooth tissue is irreversibly lost. Early stages of

dental decay involve demineralization of tooth tissues with minimal loss of the organic

matrix. In some instances, dentists can promote remineralization of the matrix, thus

effectively reversing the caries process.6 In other instances, the affected area can be

covered with a protective material without any surgical removal of tooth tissue. The

efficacy of these conservative, nonsurgical caries treatments is not well identified.

The third question addresses the effectiveness of preventive methods in those

individuals who have experienced, are experiencing, or are expected to experience an

elevated incidence of carious lesions. Dentists are now being urged to identify

individuals with elevated caries activity,7 but this "risk assessment" strategy has not

been complemented by the identification of the most effective interventions to

mitigate the caries attack in these high-risk individuals.

Technical Expert Advisory Group Involvement

Guidelines from AHRQ require identification of technical experts in diagnosis and

management of dental caries. The Technical Expert Advisory Group (TEAG) (see

Appendix B for its composition) was expected to contribute to (a) advancing AHRQ's

broader goals of creating and maintaining "science partnerships" and "public-private

partnerships" and (b) meeting the needs of a broad array of potential users of its

products. Thus, it was both a resource and a sounding board throughout the project.

The TEAG included seven members, three technical experts, two individuals

representing the public health perspective of the population at large, and two

potential users of the final evidence report or other materials.

To ensure scientifically robust work, the TEAG was called upon to provide reactions to

work in progress and advice on substantive issues or possibly overlooked areas of

research. TEAG members participated in conference calls and e-mail solicitations:

• At the beginning of the project to discuss the key

clinical questions, initial drafts of causal pathways, and

proposed inclusion and exclusion criteria for research articles.

• During the development of abstracting forms to provide

comments concerning the forms, the content proposed for

inclusion in the evidence tables, and the final versions of the

key clinical questions and causal pathways.

• When the draft evidence tables were produced to

discuss the content of the tables and the completeness of the

search.

Because of their extensive knowledge of the caries literature and their active

involvement in professional societies, TEAG members were also asked to participate in

the peer review process by commenting on the draft report. In addition to the

contribution of the TEAG, the preparation of the evidence report also benefited from

the contributions of three consultants whose advice was sought informally during all

phases of the project. Subject to their availability, the consultants also participated in

the conference calls.

Dental Caries: Background and Significance

Dental caries is a chronic infectious disease that results in the destruction of tooth

tissue. It is caused by a complex interaction of oral microorganisms in dental plaque,

diet, and a broad array of host factors ranging from societal and environmental
factors to genetic and biochemical/immunologic host responses.8 Dental caries is also

site specific as each tooth and each site have different susceptibilities because of their

unique anatomical, physiologic, and environmental characteristics. The crown or

coronal portion of a tooth is covered by a layer of enamel. The occlusal surfaces of

posterior crowns have invaginations termed pits and fissures, whereas the facial,

lingual, and proximal aspects of tooth crowns typically are smooth. In contrast, the

tooth root consists of dentin covered by only a thin layer of cementum. These

anatomical variations provide different environmental niches that permit very

different forms of plaque to flourish.

Dental tissues are in a constant state of mineralization and demineralization because

the acidogenic plaque adjacent to enamel surfaces. When this dynamic balance is

disrupted, the caries process can proceed and can result in the destruction of tooth

tissue. Initially there is a diffusion of acids into the enamel and subsurface

demineralization begins to occur. Loss of subsurface enamel can result in a

noncavitated lesion, or white spot lesion. If the balance of the equilibrium shifts to

remineralization, the subsurface layer of enamel can be reformed by deposition of

calcium and phosphate. However, when demineralization dominates, the subsurface

lesion becomes so large that the surface layer of enamel collapses causing cavitation.

Cavitated and noncavitated lesions can progress through the enamel to the

dentoenamel junction (DEJ). Once in dentin, the lesion progresses by following the

dentinal tubules and spreads laterally in a saucer-shape fashion. Root surfaces, which

are composed of a thin layer of cementum over dentin, are much rougher than

coronal (enamel) surfaces, facilitating plaque formation. Compared with coronal

lesions, root lesions have less well-defined margins and exhibit a broad pattern of

progression through the dentin.

Prevalence of Carious Lesions

The Third National Health and Nutrition Examination Survey-Phase I (NHANES III),

conducted from 1988 to 1991, provides the most recent estimates of the prevalence

of carious lesions in the United States.9,10 This survey, which produced nationally

representative estimates for the civilian noninstitutionalized U.S. population, found

that the mean decayed and filled surfaces (dfs) of primary teeth score in children age

2 to 9 was 3.1. The score varied among racial-ethnic categories: non-Hispanic whites

(2.5), non-Hispanic blacks (2.7), and Mexican-Americans (4.8). Overall, 83 percent of

children age 2 to 4 years had experienced no lesions in the primary dentition, with

this percentage dropping to 50 percent in children 5 to 9 years of age.

Table 1. Mean DS, DMFS, and % DS/DMFS per person by race-ethnicity (more...)

Table 1. Mean DS, DMFS, and % DS/DMFS per person by race-ethnicity in

U.S. children and adolescents age 5 to 17, 1988-91

1
Race/Ethnicity DS (SE) DMFS (SE) % DS/DMFS (SE)

Total 0.4 (0.1) 2.5 (0.2) 19.7 (1.5)

Non-Hispanic Whites 0.3 (0.0) 2.4 (0.3) 14.6 (2.2)

Non-Hispanic Blacks 0.8 (0.1) 2.5 (0.2) 37.9 (3.1)

Mexican-Americans 0.7 (0.1) 2.7 (0.1) 36.4 (2.8)

The overall mean decayed, missing, and filled surfaces (DMFS) of permanent teeth

scores for children age 5 to 17 in various racial-ethnic categories are shown in Table

1, together with the D component scores and the D/DMF proportion, which represents

the relative proportion of an individual's disease experience that has not received

treatment. In this age group, overall DMFS was similar across race-ethnicity

categories, but the proportion of DS in the DMFS varied substantially by race-ethnicity

categories, which may be more of a reflection of access or utilization issues rather

than disease patterns.

In children age 5 to 11, 74 percent had no carious lesions in the permanent dentition,

whereas for children age 12 to 17, the proportion falls to 33 percent. Overall, carious

lesions in children are not evenly distributed; 75 percent of overall caries experience

in permanent teeth occurred in approximately 25 percent of the population.3 Thus,

although the majority of children have moderate decay or less, carious lesions are a

recurring problem for a substantial minority. Carious lesions are also not equally

distributed across tooth surfaces in this population, as occlusal surfaces experience

lesions five times more frequently than the mesial and distal (smooth) surfaces.

Table 2. Mean DS, DFS, and %DS/DFS per person by race-ethnicity in (more...)

Table 2. Mean DS, DFS, and %DS/DFS per person by race-ethnicity in U.S.

dentate adults, ages 18 Years and older, 1988-91

1
Race/Ethnicity DS (SE) DFS (SE) %DS/DFS (SE)

Total 1.8 (0.0) 22.2 (0.9) 14.2 (0.8)

Non-Hispanic Whites 1.5 (0.1) 24.3 (1.5) 10.6 (0.8)

Non-Hispanic Blacks 3.4 (0.3) 11.9 (3.4) 35.4 (2.8)

Mexican-Americans 2.8 (0.3) 14.1 (2.8) 31.0 (2.6)

In adults 18 and older, evidence of past or present coronal carious lesions was found

in 94 percent of the population. The mean DFS score for dentate adults (those with

one or more teeth) was 22.2. Females exhibited a higher mean number of treated and

untreated surfaces per person. Coronal DFS scores, shown in Table 2, varied by race-

ethnicity, and the proportion of decayed surfaces in the DFS score varied as well.
Carious root lesions were found in 23 percent of the dentate population overall and in

more than 47 percent of individuals 65 or older. The average number of treated and

untreated root surfaces ranged from a low in non-Hispanic whites of 1.1 to 1.4 in

Mexican-Americans and 1.6 in non-Hispanic blacks.

Burden of Illness from Dental Caries

Because treatment and/or prevention of carious lesions is one of several reasons for

visiting a dentist, and because accurate information describing reasons for dental

visits is not available, the extent to which carious lesions necessitate dental visits is

not known. However, from the preceding section on the prevalence of carious lesions

and filled tooth surfaces, it would seem that racial and ethnic minorities receive

proportionally less treatment for carious lesions than do white non-Hispanics. This

observation is supported from existing data on dental visits. The Surgeon General's

report on oral health11 has assembled data from a number of Federal agencies that

paint a clear picture of differences in the receipt of oral health care by race/ethnicity,

income, and insurance status. In white, non-Hispanic adults, 64 percent reported a

dental visit in 1993 compared with 47 percent of black, non-Hispanic adults and 46

percent of Hispanic adults. For individuals with incomes at or above poverty level, 64

percent reported a dental visit in the previous year compared with 36 percent for

those with incomes below poverty level. In 1989, 70 percent of individuals 2 years

and older with private dental insurance reported a dental visit within the preceding

year compared with 51 percent for those without private dental insurance. Finally,

those individuals who rate their oral health as very good or excellent are more likely

to have visited a dentist in the preceding year (61 percent) than were those assessing

their health as fair or poor (45 percent).

The economic cost of dental caries is also difficult to assess precisely. In 1998,

Americans spent more than $53 billion on dental services.11 From an analysis of

insurance claims, approximately 40 percent of charges are related to restorative

dental services, which are usually, but not always, performed to repair teeth damaged

by carious lesions.12 Thus, even without adding in the cost of more complex services
necessary to restore function lost as a result of the sequalae of the dental caries

process, expenditures related to caries were more than $20 billion.

Beyond the direct economic costs of dental treatment, there are the less directly

calculable costs associated with the loss of working time, missed school, and reduced

levels of social functioning. The National Health Interview Survey (NHIS) indicates

that 2.9 million acute dental conditions occurred in the U.S. population during 1994.

These dental conditions accounted for an estimated 3.9 million days of missed work in

persons 18 years of age and over, 1.2 million days of missed school in youth 5 to 17

years of age, and 12.2 million days of restricted activity across all ages (e.g.,

nonperformance of usual family role activities).13 The NHIS methods may

underestimate the actual amount of missed time from school and work and restricted

activity days for dental conditions.14,15

Studies of how dental caries affects quality of life are much less empirically

compelling, but experts agree on what the potential effects of dental caries are likely

to be in the short and long term.16 In the short term, physical discomfort and pain are

the most likely consequences of untreated lesions. Physical impacts can be felt

directly as through the pain of toothaches, infections, and temporomandibular joint

disorder resulting in part from a loss of posterior teeth and the failure to replace them

when necessary. The possible eventual inability to eat -- both bite and chew --

because of tooth loss can lead to unnecessary dietary restrictions and nutritional

deficiencies as well as complicate the dietary management of other chronic health

conditions.

The psychological pain of self-consciousness and social isolation may also accompany

the embarrassment of the unsightly deterioration of anterior teeth caused by dental

caries. The same psychological distress can result from the embarrassment of missing

anterior teeth, the communication dysfunction associated with not being able to be

easily understood by others, and the isolation or withdrawal from social intercourse

because of missing teeth. In the long term, left untreated, carious lesions may lead to

the loss of such teeth, the replacement of which may be needed for functional, social,

cosmetic, and physical and mental health reasons.

Caries Diagnosis

The local result of the dental caries infection is a process of demineralization of tooth

tissue (enamel, dentin, cementum). Acid produced by bacteria as a product of

carbohydrate fermentation causes the demineralization. The diagnosis of dental caries

at a particular site on a tooth is based on either direct or indirect detection of the

demineralized tooth structure.

The principal methods dentists use to diagnose carious lesions -- visual and

visual/tactile examinations and radiographic assessment -- have been employed with

little change for decades. Refinement in techniques, rather than development of new

technology, has characterized these methods over the years. Illumination has

improved and magnification is more easily employed for visual examinations, whereas

radiation doses have decreased for radiographic assessment as both equipment and

film have been improved.

Visual inspection is based on a search for signs of demineralization, which include

changes in color and in surface consistency and contour. Tactile inspection is usually

accomplished with a fine-tipped dental explorer or probe that is passed over smooth

surfaces of teeth as well as pits and fissures. On smooth surfaces, the surface texture

is assessed for roughness as well as breaks in contour. In pits and fissures, the probe

is usually pressed with differing levels of force into depressed areas to assess whether

any penetration is possible and whether there is any resistance to withdrawal of the

probe. Radiographic assessment is based on identification of demineralization of tooth

tissue through differential exposure of film. Demineralized tooth tissue is less

resistant to the passage of ionizing radiation and thus appears darker on film images.

More recently, new technologies have begun to appear that further refine radiographic

diagnosis of carious lesions and offer alternatives to this technology. Digital

radiographic techniques eliminate film by capturing radiographic images on phosphor

storage plates or charge-coupled devices. The images can then be manipulated to

enhance diagnostic features. Fiberoptic transillumination, passing a narrow beam of

light through tooth tissue, has become an adjunctive diagnostic method now used for
both anterior and posterior teeth, principally on proximal surfaces. Demineralized

tooth tissue appears dark when transilluminated because of its decreased

transmission of light. This method represents refinement of the traditional technique

of transilluminating the anterior proximal surfaces using a mouth mirror and the

operatory light. Measuring the resistance of tooth tissue to an electrical current

passed through it is another approach to caries diagnosis, especially of occlusal

fissure caries. First demonstrated in the 1950s, the technique has been progressively

refined, with devices available commercially since the 1980s. The technique depends

on the fact that when enamel becomes demineralized, it loses much of its resistance

to electrical charges, hence its conductance increases.

The extent of variation in the diagnosis of dental caries is substantial among dental

practitioners using the traditional techniques. Typically agreement among several

dentists is poor to moderate, with kappa values ranging from 0.30 to 0.60 in several

studies.5 The range of positive diagnoses (proportion of teeth diagnosed as carious) is

typically wide for any given sample, often spanning 30 to 40 percentage points.5 The

problem of calibrating dental practitioners to an objective standard for caries

diagnosis results to a large extent from the absence of objective criteria for the

diagnosis;17,18 consequently, dentists tend to develop widely different subjective

patterns or "scripts" that they then use for identification of carious lesions.19 This

variation in the diagnosis of carious lesions is a principal contributor to the still

greater variation in the decision to restore teeth through irreversible surgical

intervention20,21 and the concomitant variation in associated costs of those decisions.22

Professionally Administered Methods of Caries Prevention

Caries prevention as accomplished in dental practice has traditionally been viewed as

a combination of several procedures, including oral prophylaxis, topical application of

fluoride, oral self-care instruction, sealants for fissured surfaces, and restoration of

existing carious lesions. Although oral self-care instruction and oral prophylaxis

methods have not changed appreciably over the years, application of topical fluoride

has seen continuing modifications, both in delivery vehicles and in solutions and

concentrations used. Dental sealant technology has similarly become refined, with
changes in materials and in etching and polymerization techniques. In recent years,

an additional intervention has become available: prescription antimicrobial

mouthrinses. Also, the number of "over-the-counter" (OTC) products that dentists can

specifically recommend for home use has increased, such as remineralization rinses,

salivary substitutes for persons with decreased salivary flow, and candies and gums

with nonfermentable sugars. Finally, simplified testing for mutans streptococci (mS),

the putative pathogen for dental caries, has become commercially available.

As the incidence of carious lesions experienced by most children has decreased in the

past three decades, available approaches to prevention in both children and adults

have become more specific to individual clinical circumstances. The two circumstances

on which this review is focused involve the management of noncavitated carious

lesions and the prevention of carious lesions in caries-active individuals. Noncavitated

carious lesions are areas where demineralization has started, but is not extensive. In

theselesions, no tissue has been lost and no loss of contour or break in continuity of

the enamel surface is detectable. Strategies for preventing these lesions from

progressing to irreversible tissue loss, or cavitation, can include all of the traditional

and more recently developed preventive techniques. The prevention of new carious

lesions in caries-active individuals also can involve the full gamut of professionally

applied preventive procedures.

Variation in Methods to Control Noncavitated Lesions

Little is known about dentists' strategies to reduce or eliminate progression of

noncavitated carious lesions and hence the necessity for surgical intervention. The

previously cited literature on variation in dentists' decisions to initiate treatment

includes some studies of dentists' treatment thresholds. These studies suggest there

is variation in the extent of progression of a carious lesion that individual dentists are

willing to tolerate before they intervene surgically. Unfortunately, these types of

studies must be done using patient vignettes, and there is some suggestion that what

dentists say they do with respect to intervention is often different than what they

actually do in practice.23-26 These studies show that a sizable proportion of dentists

routinely intervene when radiographic evidence of dental caries manifests itself in the
enamel prior to cavitation. No recent studies are available to document circumstances

surrounding application of nonsurgical means of control, although the continuing

controversy about "sealing over caries"27 suggests that dentists vary in their

willingness to use sealants as a method for the control of unidentified occlusal lesions.

Variation in Methods to Control Caries in Caries-Active

Individuals

Knowledge of dentists' practices in addressing caries control in caries-active

individuals is exceedingly limited. Only recently have the concepts of "caries risk" and

"medical management" emerged in the clinical dental literature.4,7,28,29 These

discussions suggest that practitioners' preventive approaches may not be routinely

based on a careful assessment of the magnitude of the caries challenge. Information

from insurance claims suggests that topical fluoride applications tend to vary by

practitioner, but not by patients within a practice, who all receive the same preventive

care even though they have different rates for restoration receipt.30 Also, a survey of

practitioners shows that commonly used clinical protocols are not congruent with

current recommendations for low-risk individuals.31 Clearly, there is a potential for

both over use and under use of prevention and control methods in a caries-active

population; but studies that examine dentists' preventive treatment behaviors are

rare, and none differentiate treatment by an individual's caries activity.

Organization of this Report

The remainder of this report is organized in the following sections. Chapter 2 provides

details about the literature search and review methods describes the causal pathway

for key questions and approaches to establishing inclusion and exclusion criteria,

conducting the systematic review, abstracting data from articles, maintaining quality

control, assigning quality scores to individual articles, and similar details. Chapter 3

presents the results for the three key clinical questions -- diagnostic methods,

management of noncavitated lesions, and management of caries-active individuals.

Chapter 4 provides conclusions, and Chapter 5 offers recommendations concerning

research on diagnosis and management of dental caries. References cited in the body
of the report, the six evidence tables, and a list of all literature reviewed for the

preparation of the tables follow. The appendixes contain acknowledgments (Appendix

A), information on the TEAG (Appendix B) and the peer reviewers (Appendix C), data

extraction forms (Appendix D), and acronyms and abbreviations used in this report

(Appendix E).

Chapter 2. Methodology

Overview

This chapter of the report documents the procedures that the Research Triangle

Institute-University of North Carolina at Chapel Hill Evidence-based Practice Center

(RTI-UNC EPC) used to develop a comprehensive evidence report that describes and

contrasts the approaches currently used in the diagnosis of dental caries and in

management of two specific clinical presentations of dental caries. To set the

framework for review, the key questions and their underlying causal pathway are

presented first. This is followed by a detailed description of the literature search,

which includes descriptions of the Medical Subject Headings (MeSH terms) used in the

principal search, other search sources, the inclusion and exclusion criteria, and the

application of these criteria to the results of the searches. Once the RTI-UNC EPC

team determined that studies met the inclusion/exclusion criteria and were eligible for

inclusion, the team abstracted data onto Data Extraction Forms and then transferred

critical information to evidence tables; these forms are also described in this chapter.

The chapter also discusses quality issues, i.e., the RTI-UNC EPC's quality control

procedures with regard to determining the eligibility for inclusion, carrying out the

data abstraction, and developing a quality rating scheme for individual studies. An

evidence report requires an extensive search of all types of literature. Because the

criteria for quality ratings will vary by type of study design, the RTI-UNC EPC

developed quality rating forms specific to the two types of studies included in the

diagnosis and management reviews. This section describes the development of the

rating system and its use in the analysis.

Key Questions and Causal Pathways

This report addresses three questions. The first concerns diagnosing carious lesions,

the second examines strategies for treatment of early carious lesions, and the third

focuses on management of patients who have multiple carious lesions or are

perceived to be at high risk for developing lesions. All the questions were put in final

form with input from the TEAG and the consultants after an original set of questions

was identified in initial discussions with the planning committee for the CDC on the

Diagnosis and Management of Dental Caries Throughout Life.

Final Key Questions

The key questions address issues of caries diagnosis and management that arise in

the professional treatment of dental caries, i.e., those procedures that are provided

by dentists and allied dental personnel in dental practices and clinics. Thus, the

procedures are limited to those commercially available at the time of this review.

Further, the caries management questions focus on issues that accompany the

"modern" view of dental caries as an oral infection that, at specific sites, initially leads

to demineralization and ultimately destruction of tooth tissue. The key questions,

stated in final form, are as follows:

• Question 1.

• What are the validities of the available diagnostic

methods for detecting carious lesions in primary and

permanent teeth?

• Question 2.

• What are the efficacies of the nonsurgical methods

available for stopping or reversing the progression of a

noncavitated coronal carious lesion in a primary or a permanent

tooth?

• Question 3.

• What are the efficacies of the methods available for

reducing the incidence of new coronal carious lesions in primary

 and permanent teeth in individuals who are deemed to be

"caries active" or at "high caries risk"?

The first question addresses only the diagnosis of primary caries, i.e., the first carious

lesion on a tooth surface. Both coronal and root surfaces are included in the review,

and for coronal surfaces, both primary and permanent teeth are included. Following

discussion with the TEAG, assessment of test validity was operationalized as the

sensitivity and specificity of a diagnostic test. The methods to be assessed included all

those diagnostic methods that are commercially available, including visual and visual-

tactile inspection, radiography, FOTI, EC, laser fluorescence, and combinations of

those methods.

The second question focuses on individual early carious lesions, where

demineralization has occurred but cavitation has not yet occurred. In the past, this

type of lesion was either removed surgically and replaced with a restoration or

monitored or "watched." Dentists generally assumed that many noncavitated lesions

would progress to cavitation, and based treatment decisions on this assumption. More

recently, the possibility of remineralizing or at least arresting the demineralization of

these noncavitated lesions has been considered as an alternative to surgical removal

and restoration. Another nonsurgical technique, placing dental sealants, is also

available for noncavitated lesions on fissured surfaces. The question includes

consideration of a still wider range of potentially useful methods, including

professional fluoride applications and prescribed supplements, other remineralization

agents, professional oral hygiene and plaque control programs, and combinations of

these methods.

The third question focuses on patients rather than individual carious lesions. It

reflects the need for information about how to manage patients who have active

carious lesions or who are at risk of developing such lesions. Recommendations for

the "medical management" of such patients have appeared; yet the methods to be

included in such an approach are not well defined. This question includes

consideration of professional fluoride applications and prescribed supplements,

sealants, antimicrobial therapy, salivary enhancements, nutritional/diet counseling,

professional oral hygiene/plaque control programs, and combinations of these

methods.

Causal Pathways

Figure 1. Causal pathways for the diagnosis, nonsurgical management, (more...)

Figure 1. Causal pathways for the diagnosis, nonsurgical management, and

prevention of carious lesions

Because the questions are closely linked in the typical examination and treatment

sequence that occurs in dental practice, the RTI-UNC EPC team chose to construct a

single causal pathway that defines the relationship of the three questions (Figure 1
). The diagnosis of carious lesions is, in reality, an exhaustive search for signs of

disease on all surfaces of all teeth, using a variety of search techniques. The results of

the search will drive subsequent treatment decisions. Information from the search will

include the presence or absence of carious lesions and their pattern of occurrence, the

degree of penetration of each identified lesion, and whether a lesion is cavitated, i.e.,

has lost organic material to the extent that the enamel surface has lost its contour.

The first question examines the accuracy with which the presence or absence (i.e.,

"any caries") and the depth of penetration (caries affecting the dentin or inner

structure of the tooth) are identified, as well as the accuracy with which cavitation can

be detected.

The degree of penetration of the lesion is thought to be the principal criterion that

most dentists use in making treatment decisions, with penetration to the dentin

seemingly the threshold for restoration reported most often. In view of caries

progression, whether a lesion is cavitated or not may represent a more logical

criterion for differentiating between opportunities to arrest or reverse caries

progression nonsurgically and the necessity for removal of the lesion and replacement

of the lost tissue. The use of dentin penetration as the surgical intervention criterion

may result in the treatment of noncavitated, potentially reversible lesions. The causal

pathway reflects the lack of a cavitation criterion for nonsurgical intervention.

For those patients found to have one or more carious lesions, in addition to surgical or

nonsurgical treatment directed specifically at the lesion(s), there is an opportunity to

provide treatment for the purpose of reducing the likelihood for the development of
further lesions. As noted, although dentists have long provided professional

preventive procedures, linking the provision of these procedures to a patient's caries

activity status, when it has been done, usually has been done informally, with little

knowledge of the effectiveness of such preventive procedures in patients with high

rates of disease. Extending this type of targeted intensified prevention to patients

identified as being at risk for the development of carious lesions is less common.

Caries risk assessment is a relatively recent development in dentistry; and even

though a number of risk assessment instruments have been described, the approach

has not been validated when applied to individual patients.

Literature Search

This portion of Chapter 2 documents the literature search process, specifying the

terms used for each of the literature database searches conducted, as well as

describing other search strategies and listing the inclusion/exclusion criteria used for

the initial search and the review of identified studies. It also documents the steps

taken to identify the relevant studies from among those identified in the searches to

be included in the evidence report.

Search Terms

Table 3. Strategy and results of MEDLINE caries diagnosis search (more...)

Table 3. Strategy and results of MEDLINE caries diagnosis search

Wide search of early caries literature

1 exp dental caries/pa,di.ra 2,846

2 limit to human, English, 1966-75 219

Defining studies of caries

3 exp tooth demineralization/pa,di,ra 2,928

4 exp dental caries/ 21,830

5 3 or 4 21,904

Limiting 5 to diagnostic methods

6 exp diagnosis/, oral diagnosis/ 2,420

7 exp radiography/, dental radiography/, digital dental

816
radiology/

8 exp pathology/, oral pathology/ 4

9 1 or 6 or 7 or 8 2,539

10 limit to human, English 1,776

Limiting 10 to various study types

11 controlled clinical trial 21

12 meta analysis 4

13 randomized controlled trial 50

14 epidemiologic study characteristics 244

15 epidemiologic research design 333

16 comparative study 457

Combining results of 1966-75 "wide" search and searches for
specific study types

17 2 or 11 or 12 or 13 or 14 or 15 or 16 1,266

Adding all root caries studies

18 exp root caries/pa,di,ra 62

Total 1,328

Table 4. Strategy and results of MEDLINE caries management search (more...)

Table 4. Strategy and results of MEDLINE caries management search

Identifying management methods

1 exp fluorides, topical/tu 2,061

2 exp tooth remineralization/ 445

3 exp pit and fissure sealants/tu 667

4 exp health education, dental/ 4,287

5 exp dental prophylaxis/ 3,699

6 exp oral hygiene/ 8,624

8 exp dental plaque/pc,dh,dt,th 3,423

9 exp chlorhexidine/tu 1,126

10 exp xylitol/tu 162

11 exp tooth demineralization/pc,dt,th 10,162

12 exp cariostatic agents/tu 3,994

13 fluoride supplements 60

14 1 or 2 or 3 or 4 or 5 or 6 or 7 or 8 or 9 or 10 or 11 or
26,902
12 or 13

Identifying caries management methods

15 exp dental caries/pc,dh,dt,th 10,064

16 14 and 15 10,058

17 limit to human, English 5,057

Limiting 17 to various study types

18 controlled clinical trial 122

19 randomized clinical trial 177

20 epidemiologic study characteristics 762

21 epidemiologic research design 266

22 comparative study 758

Total of 18 or 19 or 20 or 21 or 22 1,435
Two separate literature searches were conducted for this evidence report -- one for

the caries diagnosis question and the other for the two caries management questions.

Tables 3 and 4 show the MeSH terms used for searching MEDLINE, the principal

database for each of these two searches, as well as the results of the searches. The

searches were run in October 1999. Although detailed sets of inclusion and exclusion

criteria had been developed prior to the searches (see following section), few of the

criteria are evident in the search strategies. Indexing for the dental literature is

sketchy and unreliable in the first 10 years covered by MEDLINE, and problems exist

well into the 1980s for some terms of interest in these searches (e.g.,

demineralization and remineralization). Thus, the search strategies tended to be

inclusive rather than exclusive. Only at the broadest level could either search be

limited to human studies, reports in English (because resource constraints), and a

rather wide variety of study types listed in the tables.

In the absence of effective exclusion criteria available in MEDLINE, it still might have

been possible to design relatively "tight" search strategies if certain critical keywords

were available to narrow the search focus. Unfortunately, this was not the case for

either search. The diagnosis search returned a large number of potentially eligible

studies (1,328) because preliminary searches had demonstrated that a key term,

"sensitivity and specificity," could not be assumed to identify accurately all eligible

studies.

In the management search, two critical features of eligible studies could not be

isolated through use of indexing terms. Neither noncavitated lesions nor caries-active

or "at-risk" patients are identifiable through the keyword structure. Thus, the

management search had to be designed to identify all possible evaluations of the

eligible preventive methods, with subsequent inspection of the abstract or full paper

required for a final determination of eligibility for either of the systematic reviews

based on patient sample or type of lesion included. The result was the identification of

1,435 citations.
Additional Searching

Table 5. Strategy and results for EMBASE caries diagnosis search (more...)

Table 5. Strategy and results for EMBASE caries diagnosis search

1. dental adjacent to caries 1,554

2. diagnosis 248,652

3. dental radiography 121

4. 2 and 3 248,677

5. 1 and 4 87

6. New citations added (not duplicates with MEDLINE) 79

Table 6. Strategy and results for EMBASE caries management search (more...)

Table 6. Strategy and results for EMBASE caries management search

1. dental adjacent to caries 1,554

2. topical fluorides 6

3. remineralizaion 79

4. dental sealants 13
5. sealants 96

6. chlorhexidine 0

7. cariostatic agents 0

8. 2 or 3 or 4 or 5 or 6 or 7 181

9. 1 and 8 48

10. New citations added (not duplicates with MEDLINE) 43

Subsequent to the principal literature searches in MEDLINE, the team completed

followup searches in EMBASE and the Cochrane Controlled Trials Register. The search

terms and results for the EMBASE searches are shown in Tables 5 and 6. The studies

not duplicated in the MEDLINE searches were added to the two groups of studies

included in the review. No new studies were found in the Cochrane Library.

A valuable supplemental search strategy was perusal of the reference sections of

papers identified in the searches. Again, the reason for the seeming inefficiency of the

MEDLINE searches is in large measure the imprecise indexing characteristics of dental

studies in the 1970s and 1980s. Not only are descriptors of study design

characteristics inexact or missing, but descriptors related to the condition or process

of interest are also often tangential in nature. This forces the search to be less

exclusive and at the same time increases the likelihood that some studies will be

missed, even with a fairly broad search strategy such as the one employed.

The team had elected at the outset not to complete a detailed search of the gray

literature. This is information not appearing in the periodic scientific literature, such as

dissertations, theses, industry reports, unpublished studies, abstracts, and other

nontraditional sources. The team made this decision because of both limited resources

and the prevailing experience opinion among RTI-UNC EPC staff that in the absence of

known sources for such literature, searches were unlikely to yield useful information.

The team did query NIDCR to identify any in-progress studies that might have
recently reported relevant data. The team did not identify other potentially fruitful

sources for gray literature for dental topics. Thus, no other sources were searched.

Finally, because the addition of dental articles to the MEDLINE database tends to lag

behind publication date by at least 6 months, the team hand-searched six of the more

fruitful journals for relevant studies for the period January 1998-December 1999

(Caries Research, Community Dentistry Oral Epidemiology, European Journal of Oral

Sciences, Journal of Dental Research, Journal of Dentistry, and Journal of Public

Health Dentistry).

Inclusion and Exclusion Criteria for the Literature Searches

Table 7. Inclusion (I) and exclusion (E) criteria for (more...)

Table 7. Inclusion (I) and exclusion (E) criteria for caries diagnosis studies

(1) Diagnostic question

I accuracy of the determination of the presence/absence of

natural caries at defined levels

E accuracy of measure of caries depth, volume, extent, etc.

(exclude if only question addressed)

E artificial caries achieved through demineralization or

drilling

(2) Diagnostic method

I visual or visual and tactile inspection

I film radiography with D- or E-speed film

 I digital radiography (charged coupled devices, storage
phosphor screens)

I fiberoptic transillumination (FOTI)

I electrical conductance techniques (Vanguard, Caries Meter

L, ECM, etc.)

I fluorescence/optical techniques if commercially available

E methods using equipment not available commercially

(3) Validation technique

I sectioning with visual inspection, microscopy,

stereomicroscopy, or macroradiography

I visual/tactile inspection of intact surface for cavitation only

E visual or visual/tactile inspection for caries level other than

cavitation

E radiography or other nonhistologic technique for caries

level other than cavitation

(4) Sensitivity/specificity determination

I reported or calculable from results presented

E not determinable
Table 8. Inclusion (I) and exclusion (E) criteria for noncavitated (more...)

Table 8. Inclusion (I) and exclusion (E) criteria for noncavitated lesions

studies

(1) Study design

I in vivo studies

I studies with concurrent comparison group (nil, placebo, or

active)

I studies with the lesion as the unit of analysis

E in vitro, in situ studies

E studies without concurrent comparison

E studies without baseline determination of individual lesion

status

(2) Intervention

I interventions requiring professional application and/or

prescription

I interventions likely to be undertaken only upon the

recommendation of a dentist

E fluoride dentifrice studies regardless of concentration, if

only intervention/control component

(3) Sample size

no exclusion criterion
 (4) Outcome

I outcome expressed or calculable as percent of lesions

identified at baseline that progressed

Table 9. Inclusion (I) and exclusion (E) criteria for (more...)

Table 9. Inclusion (I) and exclusion (E) criteria for studies of caries

prevention in caries-active individuals

(1) Study design

I in vivo studies

I studies with concurrent comparison group (nil, placebo, or

active)

E in vitro, in situ studies

E studies without concurrent comparisons

(2) "Caries active/high caries risk" (CA) designation

I studies where CA status is designated at the level of the

individual

I studies where DS, DFS, DMFS, ds, dfs, or defs score is

used for CA designation

I studies where microbiological testing is the basis for CA

designation
 E studies where CA is designated at a group, community, or
population level

E studies where CA is based solely on sociodemographic

characteristics

(3) Intervention

I interventions requiring professional application and/or

prescription

I interventions likely to be undertaken only upon the

recommendation of a dentist

E fluoride dentifrice studies regardless of concentration, if

only intervention/control component

(4) Sample size

no exclusion criterion

(5) Outcome

I outcome expressed as change in DS, DFS, DFT, DMFS,

DMFT, ds, dfs, dft, defs, or deft
Tables 7, 8, and 9 show the final inclusion and exclusion criteria applied to studies for

questions relating to diagnosis, noncavitated lesions, and caries-active individuals. As

noted earlier, the searches were electronically limited to human subjects, time periods

were 1966 or later, and publication language was limited to English. In addition, the

eligibility criteria restricted studies to settings that could realistically be generalized to

dental practices, although all geographic locations were potentially eligible.

A key criterion for the diagnostic question was the requirement for histologic

validation of caries status for each surface studied. By requiring this level of
validation, the team eliminated a large number of studies that compared two or more

diagnostic methods, with one method designated as the reference standard. No

currently available clinical diagnostic method is perfectly valid, i.e., has 100 percent

sensitivity and specificity compared with histologic evaluation. Thus, the team

excluded such studies because members were unwilling to include studies that would

automatically introduce error into the assessment process. The team made an

exception to the histologic reference standard where cavitation was the extent of

lesions to be detected. Here a reference standard of direct visual clinical examination

was deemed acceptable. Both in vivo and in vitro studies were accepted, although the

number of in vivo studies was understandably limited because of the requirement for

histologic validation.

The inclusion criteria for diagnostic studies also required that outcomes be expressed

in terms of sensitivity and specificity. This criterion resulted in the exclusion of several

studies where results were expressed only as receiver operating characteristic (ROC)

curves. Such outcomes are typically obtained when observers indicate their level of

certainty about a diagnosis on a five-point scale. The argument for using such an

analytic approach is that asking the observer to state a level of certainty helps

disassociate an observer's degree of leniency from the implications of any given

decision criterion, thus permitting diagnostic performance to be reflected independent

of an observer's perceived "cost" of an incorrect diagnosis. Many studies that employ

ROC analyses also report sensitivity and specificity outcomes for the combined levels

of "reasonably certain" and "certain" that a lesion is present. If these outcomes were

reported, the study was included in the evidence table. However, when it was

necessary to estimate values for sensitivity and specificity outcomes directly from

ROC curves because no data were reported in text or table, a study was excluded.

Finally, the team had originally set arbitrary limitations for caries prevalence and

sample size, but with no objective support for those specifications. Subsequently, the

team found that 15 percent of studies would be ineligible because they either did not

report caries prevalence in the sample or had a prevalence over the maximum of 80

percent. The team also found that the sample size exclusion criterion (less than 30)

would exclude 10 percent of identified studies. In light of the limited number of

studies available when other inclusion and exclusion criteria were applied, the team

decided to drop both of these exclusion criteria and include the studies.

Two sets of management inclusion and exclusion criteria were applied to the single

set of studies identified in the management literature. These two sets shared some

common criteria. Only in vivo studies were included; in vitro studies and in situ

studies, where exogenous tooth tissue was placed in the oral environment, were

excluded. Studies without concurrent comparison groups (either nil, active, or placebo

comparisons) were excluded. The team kept only studies with interventions requiring

professional provision (e.g., application, prescription, etc.) or with interventions

unlikely to be undertaken without the recommendation of the dentist (e.g., daily OTC

mouthrinses, gums, etc.).

The key inclusion criterion for noncavitated lesion studies involved the type of lesion

examined. The question addressed lesions for which there was some likelihood that

remineralization treatment would be successful. To identify studies that included only

this type of lesion, the team originally established "noncavitated" as a definition for

acceptable lesions in conjunction with the TEAG. However, the term "noncavitated"

was not in widespread use for much of the search period. Thus, the inclusion criterion

were broadened to include the terms "incipient" and "initial" lesions. This decision

permitted the inclusion of studies in which outcomes of interest had been reported in

analyses stratified by caries status at baseline.

One other criterion for studies of noncavitated lesions deserves mention. The unit of

analysis was the individual carious lesion. This is not the usual analytical unit in caries

studies, which are typically analyzed and reported in terms of total decayed, missing,

and filled (DMF) surfaces for a single subject. However, when the reversal of

individual lesions is at issue, an aggregated analytical unit that combines cavitated

and noncavitated lesions and cannot distinguish multiple lesions on a single surface is

unusable.

For the review of studies involving caries-active individuals, the definition of subjects

was the key inclusion criterion. As noted earlier, no consensus exists on

characteristics of either caries-active or at-risk individuals. For this reason, and

because the team realized that most of the extant studies would represent subgroup

analyses of controlled trials rather than studies recruiting at-risk or caries-active

subjects exclusively, the inclusion criteria were intentionally broad. The team

accepted designation of at-risk and caries-active individuals through any combination

of caries experience and/or mutans streptococci concentrations. The team did not

specify cut points or limit the relative size of the risk group with that of the total

sample. The team did insist on individual identification of subjects, thus excluding

studies where schools or communities were selected on the basis of mean caries

experience, socioeconomic status, or other group-level predictors of caries activity or

caries risk.

Criteria for outcomes of studies eligible for inclusion allowed a range of traditional

measures of caries experience in primary and permanent teeth. Again, studies were

required to have concurrent comparison groups; but because some findings were

expected to be subgroup analyses, minimum sample sizes were not established.

Title, Abstract, and Paper Review

The RTI-UNC EPC team performed an initial survey of the titles of the identified

papers from both searches and selected papers with titles that indicated some

possibility that the study was relevant and would be eligible, i.e., would satisfy the

inclusion criteria. The surveys were done independently by both the clinical and

research directors. Titles indicated by either one were placed on the potentially

eligible list.

The team then surveyed the abstract or, if no abstract was available, the full paper to

further refine the list of potentially eligible studies. Again, this process was relatively

inefficient; many full papers had to be photocopied because the searches had

identified substantial numbers of studies from the 1970s and early 1980s, when

abstracts were not routinely included in the MEDLINE database. Again, the research

and clinical directors worked independently, applying the full set of inclusion/exclusion

criteria, with all disagreements resolved through discussion.

When the abstract or full-paper survey indicated any likelihood that the paper would

be eligible, the full paper was obtained if necessary and the inclusion/exclusion

criteria reapplied to determine final eligibility. Again, all disagreements between the

research and clinical directors were discussed. In addition, the paper was examined

for citations to other, possibly eligible, studies that had not yet been identified.

Citations so identified were obtained and surveyed for eligibility, and if eligible, added

to the pool of papers to be abstracted. Finally, the team circulated a preliminary list of

included articles to TEAG members for their comments and possible additions.

Table 10. Search refinement results

Table 10. Search refinement results

Diagnosis Management
Step
Search Search

Database searches

Initial MEDLINE search 1,328 1,435

Initial EMBASE search

79 43
(nonduplicates of MEDLINE)

Total articles for review 1,407 1,478

Initial screening for inclusion

Surviving title review 285 487

Surviving title & abstract/paper 50 34

 Diagnosis Management
Step
Search Search

review

Final review for inclusion

Surviving final review

39 27
(included)

Separate review for special populations

Additional special population

13
papers
Table 10 shows the numbers of papers remaining after each major step in this review

process. The final numbers of papers included in the reviews were 39 diagnostic

studies and 27 prevention studies. The dramatic reductions in numbers from the

original searches were, as indicated, primarily because of a single inclusion criterion in

each search. For diagnostic methods, this criterion was histologic validation. For the

caries management questions, the key inclusion criterion was either the analysis of

individual noncavitated lesions or of caries-active individuals, depending on the

question being addressed. Of the prevention studies, 5 addressed the noncavitated

lesion question and 22 addressed the caries-active question. In addition, as a part of

the review process, the team separately identified six studies evaluating preventive

methods in patients who had received radiotherapy for head and neck neoplasms, a

special high-risk group, and seven studies evaluating preventive methods in patients

with orthodontic bands or brackets, another special high-risk group. Consultation with

the TEAG indicated that these studies should be included in the review but not

combined with the main group of studies because of substantial disparities in lesions

and study methods.

Data Abstraction
Data Extraction Forms and Reviewers

The scientific director and clinical director collaborated on the development of three

data extraction forms for the diagnostic, noncavitated lesion, and caries-active

individual questions. Draft versions of the forms, together with lists of evidence table

columns linked to the extraction forms, were circulated to the TEAG for review and

comment. Final versions of the forms incorporated TEAG comments and necessary

modifications that had been identified through pretesting. Copies of the forms appear

in Appendix D. Based on experience with previous abstraction forms developed for

other RTI-UNC EPC projects, the team included essential directions and criteria on the

forms and also endeavored to collect little information beyond that planned for

inclusion in the evidence tables. All three forms included discrete sections addressing

study design, subjects, examiners, caries criteria, intervention information, and

results.

Data extraction for the two management questions was performed by the scientific

director as the sole reviewer. In addition to extracting data, the reviewer also

calculated the number needed-to-treat (NNT) statistic from the data reported in the

study, where possible. The clinical director subsequently reviewed the evidence

tables, confirming entries directly with the published papers. All disagreements were

discussed and tables changed when indicated by the discussion.

Data extraction for each paper in the diagnosis review was completed by one of three

reviewers (a pediatric dentistry resident, a dental epidemiology resident [a dentist],

and a dental hygienist with a masters in education who specializes in dental

radiology). The reviewers were not blinded to journal, author(s), or institution. The

scientific director reviewed the completed extraction forms and subsequently verified

evidence table entries against the published papers. The three reviewers participated

in a training session that used six of the included studies, and all also completed an

extraction form for the same study toward the end of review period.

Quality Control, Adjudication, and Reliability

Quality control mechanisms for determining eligibility for abstraction have already

been described. All articles were reviewed by title by the scientific and clinical

directors and disagreements settled by consensus. Agreement on retention status for

this process was 96 percent for diagnosis articles and 97 percent for management

articles. At the next stage (abstract/full paper review), all articles were again

reviewed by both directors, with disagreement again resolved by consensus. An

agreement rate was not calculated for this stage. Finally, the pool of potentially

eligible articles was again subjected to a final review by both directors, with

disagreement arising on one diagnostic article (2 percent) and one management

article (3 percent).

Agreement among the three reviewers and the scientific director on descriptive data

for the study abstracted by all reviewers was 100 percent for results (sensitivity and

specificity for four different diagnostic methods) and 88 percent for study description

items. Agreement statistics for confirmation of evidence table entries by direct

comparison with articles were not recorded.

Quality Rating of Individual Articles

The team developed separate sets of quality rating items for the diagnostic and

management articles. The sets of items are unique, but most individual items are

either modified from or taken directly from existing rating scales used by the RTI-UNC

EPC. In developing the quality rating item sets, the team was guided by the

suggestions advanced by Lohr and Carey,32 both investigators in the RTI-UNC EPC.

For the prevention questions, CONSORT criteria33 figured prominently in the design;

and for the diagnostic question, several items were included in response to issues

examined in Lijmer, Mol, Heisterkamp, et al.34

The rating scales assess several elements of internal validity, including study design,

duration, sample size, blinding, baseline assessments of differences among groups,

loss to followup, and examiner reliability. Two items also request the reviewer's

subjective assessment of both internal and external validity of the study. The forms
were pretested on small groups of studies, which resulted in some changes in

wording.

Figure 2. Quality rating form -- caries diagnosis studies (more...)

Figure 2. Quality rating form -- caries diagnosis studies

Number of sites assessed:

3 150 or more

2 75-149

1 40-74

0 fewer than 40

Area assessed for any site:

1 Entire surface (occlusal, proximal, etc.)

0 Specific site on surface

Setting:

2 In vivo

0 In vitro

Tooth selection:

3 Both posterior and anterior teeth

2 Only posterior or only anterior teeth

1 Selected posterior or selected anterior teeth

0 Single tooth type (e.g., max. or mand. 3rd molar)

Validation method:

2 Light microscopy (stereo/mono) w/wo dye

1 Other visual or radiographic assessment of sectioned tooth

0 Assessment of unsectioned tooth

Validation criteria:

1 Criteria explicitly stated

0 Criteria not explicitly stated

Validation reliability:

1 Interevaluator or intraevaluator reliability reported

0 No validation reliability reported

Caries prevalence (calculate score for each lesion type

evaluated):

2 Less than 20%

1 20-49%

0 50% or more
Number of test evaluators:

2 4 or more

1 2-3

01

Test reliability reported:

2 Interevaluator and intraevaluator reliability reported

Interevaluator reliability reported or intraevaluator reliability

1
reported

0 No reliability reported

Criteria for caries call:

1 Specified prior to evaluation

0 Developed post hoc

Figure 3. Quality rating form -- caries management studies (more...)

Figure 3. Quality rating form -- caries management studies

Study type:

2 RCT
1 Other prospective design with control/comparison group

0 Uncontrolled or cross-sectional design

Duration:

3 5 years or more

2 2-4.9 years

1 1-1.9 years

0 Less than 1 year

Blinding:

2 Examiners and subjects

1 Examiners only

0 Subjects only or none

Baseline assessment of equality of treatment groups:

1 Reported and adjusted if necessary

0 Not reported or unadjusted when differences reported

Previous and concurrent caries prevention exposures (other

than intervention) described:

1 Yes

0 No
Sample size:

3 50 or more in smallest analysis group

2 20-49 in smallest analysis group

1 10-19 in smallest analysis group

0 <10 in smallest analysis group

Loss to follow-up per year:

1 Fewer than 15%

0 15% or greater or unreported

Analysis:

1 Intention to treat

0 Includes only those remaining at final exam

Reliability:

Intrarater and interrater reliability reported and interrater

2
reliability above 0.6 kappa, or 90%

Intrarater or interrater reliability only reported and above 0.6

1
kappa, or 90%

0 No rater reliability reported or reported level(s) below minimum

Active group (for Caries-Active Individual studies only):

1 Active group represents less than 50% of total sample/population

 from which it was selected

0 Active group proportion >50% or unknown

Probing (for Noncavitated lesion studies only):

1 Criteria for lesion not dependent on probing

0 Probing used in lesion criteria

Subjective assessment of external validity:

1 Reasonably wide generalization possible

0 Applicability limited to very specific populations

Subjective assessment of internal validity

1 Reasonably "tight" methods and design

0 One or more concerns re measurement, selection, etc.

The quality rating items and scoring are presented in Figures 2 and 3 for caries

diagnosis and caries management articles, respectively. A maximum of 20 points is

possible on either form, with all raw scores rescaled to a 0 to 100 scale. The same

items were used for scoring both noncavitated lesion and caries-active individual

studies, with one exception, as noted on the caries management scoring form. Only in

scoring the articles that appear in the accessory evidence tables, i.e., those

addressing prevention of special types of lesions in special subjects, were some of the

items not applicable to all studies. In those instances, the item value was removed

from both the numerator and denominator prior to calculating the quality score.

Quality scores for all articles included in the evidence tables were completed by the

scientific director. The scores were not used in inclusion/exclusion decisions. Rather,
they represented one of the considerations for grading the evidence available to

answer each key question.

Grading the Evidence

For the diagnostic question, the strength of the evidence was judged in terms of the

extent to which it offered a clear, unambiguous assessment of the validity of a

particular method for identifying a specific type of lesion on a specific type of surface.

The three possible ratings were:

• Good (A). The number of studies is large, the quality of

the studies is generally high, and the results of the studies

represent narrow ranges of observed sensitivity and specificity.

• Fair (B). There are at least three studies, the quality of

the studies is at least average, and the results represent

moderate ranges of observed sensitivity and specificity.

• Poor (C). There are fewer than three studies, or the

quality of the available studies is generally lower than average,

and/or the results represent wide ranges of observed

sensitivities and/or specificities.

For purposes of this question, a narrow range is defined as no more than 0.15 on a

scale of 0.00 to 1.00, a moderate range is no more than 0.35, and a wide range is

more than 0.35. High quality is defined as most study scores at or above 60; average

quality is defined as most study scores at or above 45.

As in previous RTI-UNC EPC systematic reviews, we used a four-level grading scheme

for judging the overall efficacy of each of the interventions reviewed in the two caries

management reviews. The scheme was based on four aspects of the situation as

depicted in the evidence tables, the number of studies, the magnitude of the effects

reported, the quality rating scores of the studies, and the consistency of the evidence

across studies. The scientific and clinical directors independently rated the evidence

and developed an adjudicated final rating. The four possible ratings were:
 • Good (A). Data are sufficient for evaluating efficacy.

The sample size is substantial, the data are consistent, and the

findings indicate that the intervention is clearly superior to the

placebo/usual care alternative.

• Fair (B). Data are sufficient for evaluating efficacy. The

sample size is adequate, but the data show some

inconsistencies in outcomes between intervention and

placebo/usual care groups such that efficacy is not clearly

established.

• Poor (C). Data are sufficient for evaluating efficacy.

The sample size is sufficient, but the data show that the

intervention is no more efficacious than placebo or usual care.

• Insufficient Evidence (I). Data are insufficient for

assessing the efficacy of the intervention, based on limited

number of studies and/or poor methodology.

Because the majority of comparative studies included in the systematic reviews for

the caries management questions are either randomized controlled trials (RCTs) or

nonrandomized controlled trials (21/27), the term "efficacy" is used throughout the

report.

Development of the Evidence Tables

The intent in developing the evidence tables has been to make them as "user friendly"

as possible. The team has tried to limit what is included to only the most essential

information for assessing the strength and the results of individual studies. In

particular, the team was concerned about the complexity inherent in the two tables

reporting results of diagnostic methods because of the multiple issues that are

assessed in the question. These diagnosis tables could each conceivably address 105

separate clinical questions -- the diagnosis of caries based on three different

diagnostic thresholds (enamel caries, dentin caries, cavitation) for five different

posterior sites (proximal and occlusal surfaces of primary and permanent teeth, root

surfaces of permanent teeth) and two different anterior sites (primary and permanent

proximal surfaces) using five basic methods (radiography, visual/tactile, visual, EC,
and FOTI) -- without even considering combinations of methods or sites. Because

diagnostic methods perform differently using different diagnostic thresholds on

different surfaces and tooth types, all these possible combinations must be considered

separately.

Because the study conditions are so different, it is essential that in vivo and in vitro

diagnostic studies be considered separately. The team constructed two diagnosis

tables: one reporting in vivo studies, which reports the results of studies under 10

subheadings for combinations of threshold, site, and method; the other reporting in

vitro studies, which has 24 such subheadings. The entries in these tables were refined

to ensure that the complexity was not increased unnecessarily by inclusion of

extraneous information. Both entries and studies were numbered in the diagnosis

tables to make it clear that the number of studies represented is smaller than the

number of entries in the table.

The evidence tables for the management questions are less complex, reporting results

for only a few specific types of interventions. Nevertheless, in keeping with the effort

to present only essential information, attempts were made to streamline these tables

as well, while still presenting information necessary for an informed evaluation of the

strength of individual studies. For the studies of management of caries-active

individuals, the team constructed (a) a main table presenting results from the studies

with samples representing the general population, and (b) separate tables for the

highly focused studies of patients who had received radiation therapy for head and

neck tumors and patients who had undergone orthodontic treatment.

The team attempted to make all of the evidence tables self-explanatory but found it

necessary to abbreviate certain oft-repeated words and phrases. A glossary of

acronyms and abbreviations appears at the beginning of the evidence table section.

Footnotes were limited to a very few occasions where more complete explanation was

necessary for unusual circumstances. For convenience, each footnote appears on the

page where it is cited.

Peer Reviewer Process

The draft report was reviewed by a group of 17 scientists, methodologists, clinicians,

and laypersons. These peer reviewers were asked to identify factual inaccuracies and

to comment on the team's interpretations and recommendations. All comments

received from the reviewers were recorded and any changes made in the reports in

response to the comments were documented. Modifications represented consensus

decisions of the scientific and clinical directors. The selection process and names of

the peer reviewers are shown in Appendix C.

Chapter 3. Results

Based on information presented in the evidence tables, this chapter describes the

principal findings of the reviews for the three key questions concerning the diagnosis

and management of dental caries. The chapter presents the results of the three

reviews separately starting with the review of diagnostic methods, continuing with the

review of the management of noncavitated lesions, and ending with the review of the

management of caries-active individuals.

Diagnosis of Carious Lesions

As noted, the charge to review methods for the diagnosis of carious lesions was a

broad one. The review was to include all commercially available methods for the

diagnosis of primary carious lesions (first occurrence on a surface) on coronal and

root surfaces of permanent teeth and coronal surfaces of primary teeth. The

systematic review of the literature for studies evaluating the validity of six methods

for diagnosing carious lesions yielded 39 studies -- 5 conducted in vivo, 32 conducted

in vitro, and 2 reporting both in vivo and in vitro results. Many of these studies

reported multiple assessments (usually either several diagnostic methods or several

variations of one diagnostic method, all using the same sample of teeth), and many

also reported results for multiple types of lesions (cavitated, dentinal, enamel, and

coronal) so that the total number of assessments reported in Evidence Tables 1 and 2

is 126.
Table 11. Number of diagnostic assessments by tooth (more...)

Table 11. Number of diagnostic assessments by tooth surface, tooth type,

lesion type, diagnostic method, and study setting

Proximal surfaces

Visual/Ta 1 2 3
Radio Visual FOTI ECM LS
Toot ctile
Lesion
h
Type
Type vi vitr viv vitr viv vitr viv vitr viv vitr
vivo vitro
vo o o o o o o o o o

Occlusal Surfaces

Poste Cavitat
5 1 2 1 1 1
rior ion

Dentin 6 1

Any
8
lesion

Ename
l only

Root surface
Proximal surfaces

Visual/Ta 1 2 3
Radio Visual FOTI ECM LS
Toot ctile
Lesion
h
Type
Type vi vitr viv vitr viv vitr viv vitr viv vitr
vivo vitro
vo o o o o o o o o o

Occlusal Surfaces

Anter
ior & Cavitat
poste ion
rior

Dentin 2

Any
3
lesion

Ename
2
l only

Root
surface

Prim Cavitat
1
ary ion
Proximal surfaces

Visual/Ta 1 2 3
Radio Visual FOTI ECM LS
Toot ctile
Lesion
h
Type
Type vi vitr viv vitr viv vitr viv vitr viv vitr
vivo vitro
vo o o o o o o o o o

Occlusal Surfaces

Dentin

Any
lesion

Ename
l only

Poste Cavitat
1 1 2
rior ion

Dentin 25 2 1 9 1 2 12 2

Any
7 2 1 3 1 7
lesion

Ename
4 2 1 1
l only

Root
surface
Proximal surfaces

Visual/Ta 1 2 3
Radio Visual FOTI ECM LS
Toot ctile
Lesion
h
Type
Type vi vitr viv vitr viv vitr viv vitr viv vitr
vivo vitro
vo o o o o o o o o o

Occlusal Surfaces

Anter
ior & Cavitat
poste ion
rior

Dentin

Any
lesion

Ename
l only

Root
surface

Prim Cavitat 1
Proximal surfaces

Visual/Ta 1 2 3
Radio Visual FOTI ECM LS
Toot ctile
Lesion
h
Type
Type vi vitr viv vitr viv vitr viv vitr viv vitr
vivo vitro
vo o o o o o o o o o

Occlusal Surfaces

ary ion

Dentin 1

Any
lesion

Ename
l only
Table 11 displays the distribution of the assessments across the categories of tooth

surfaces, tooth types, lesion types, diagnostic methods, and study settings (in vivo, in

vitro) considered in this review, with the exception of three assessments of visual and

radiographic methods used concurrently. The assessments predominately concerned

posterior teeth (n=113). There were 10 assessments of diagnostic methods applied to

permanent anterior teeth and primary teeth of any type, and there were no studies

evaluating diagnostic methods for root caries. Eleven assessments addressed the

validity of diagnostic methods for detecting carious lesions confined to enamel. As

noted, in vitro assessments were more frequent than in vivo studies. Finally, the

assessments disproportionately involved the validity of radiographic methods (51

percent of assessments), whereas fewer than 10 percent of assessments evaluated

the validity of arguably the most common diagnostic method in the United States, the

visual/tactile examination.

The studies included in the review represented a limited proportion of all assessments

of methods for diagnosis of carious lesions, i.e., those with a histologic reference

criterion, or "gold standard." An exception was made for cavitated lesions, where

direct visual inspection was an acceptable reference criterion. There was a larger

group of studies that compared two or more diagnostic methods without a histologic

reference standard. It is the team's impression that the distribution of tooth types and

lesion types assessed in this larger group of studies was not markedly different from

that displayed in Table 11. Also, a small group of studies with histologic reference

standards was excluded because sensitivity and specificity outcomes were not

reported and could not be calculated, and these studies also reflected similar

distributions of tooth and lesion types.

The study results are described and aggregated by tooth surface (proximal or

occlusal) and type of lesion being identified (cavitation, penetration into dentin, any

lesion, enamel only). The category of any lesion includes all lesions detected by the

examiner(s) applying the criteria employed in the study. Thus, the category of any

lesions typically includes all lesions penetrating into dentin and some, but not

necessarily all, lesions in enamel. Summary tables for posterior teeth that show

sensitivity and specificity values for assessments grouped by diagnostic method and

setting are presented. Details of the studies appear in Evidence Table 1 for in vivo

studies and in Evidence Table 2 for in vitro studies.

Cavitated Lesions on Proximal Surfaces

Table 12. Sensitivity and specificity values for the (more...)

Table 12. Sensitivity and specificity values for the detection of proximal

cavitated lesions
 1 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8 9 9 10
05
0 5 0 5 0 5 0 5 0 5 0 5 0 5 0 5 0 5 0

Radiographic -- in vivo

Rug
g-
Gunn
X O
,
1972
37

Dow
ner,
X O
1975
109

Meja
re,
X O
1985
38

Pitts,
1992 X O
63

Hint
ze,
X O
1998
36

Radiographic -- in vitro
 1 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8 9 9 10
05
0 5 0 5 0 5 0 5 0 5 0 5 0 5 0 5 0 5 0

Espe
lid,
X O
1986
110

Visual/Tactile -- in vivo

Meja
re,
X O
1985
38

Hint
ze,
X O
1998
36

Visual/Tactile -- in vitro

Dow
ner,
X O
1975
35

Visual -- in vitro

Dow
ner,
OX
1975
35
 1 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8 9 9 10
05
0 5 0 5 0 5 0 5 0 5 0 5 0 5 0 5 0 5 0

FOTI -- in vivo

Hint
ze,
X O
1998
36

Table 12 shows the 11 assessments of the diagnosis of cavitated lesions on proximal

surfaces of posterior teeth. Specificity (depicted as O) in these assessments tended to

be high, ranging from 0.89 to 1.00, indicating that false positive diagnoses were

uncommon. Sensitivities (depicted as X) ranged much more widely, from 0.04 to

0.94, with the highest values recorded in two assessments reported in the same in

vitro study, one for visual/tactile diagnosis and the other for visual diagnosis. This

study employed a single experienced examiner and used criteria for cavitation that

may have allowed lesions with enamel discontinuities to be considered sound.35

Sensitivity levels for radiographic methods were uniformly higher than for the

remaining two assessments of visual/tactile diagnosis, or for the single assessment of

FOTI which was highly insensitive.36

Among the assessments of radiographic techniques, two reported sensitivity levels

substantially lower (0.34, 0.35) than the remainder, (0.63 to 0.87).37,38 There were no

obvious characteristics of these two studies that could explain the difference.

Although one study37 required lesion penetration only to two-thirds of the thickness of

the enamel, the other employed a criterion essentially similar to the remaining

studies, with lesion penetration into the dentin. Both assessments used D-speed film

as did three of the four other assessments, and both were conducted in samples with

low caries prevalence (5 and 9 percent), similar to three of the four other radiographic

studies.
The variation in sensitivities displayed among studies of radiographic and

visual/tactile methods and the availability of only single studies of visual and FOTI

methods for the diagnosis of cavitated lesions on proximal surfaces of permanent

posterior teeth results in a poor rating for the strength of the evidence for assessing

the validity of these diagnostic methods. All of the assessments display high

specificity, but the range of sensitivities is wide within methods. However, one pattern

evident in these results is the tendency for sensitivity to be higher in in vitro than in

in vivo assessments.

Lesions Penetrating into Dentin on Proximal Surfaces

Table 13. Sensitivity and specificity values for the (more...)

Table 13. Sensitivity and specificity values for the detection of proximal

carious lesions penetrating into dentin

1 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8 9 9 10
05
0 5 0 5 0 5 0 5 0 5 0 5 0 5 0 5 0 5 0

Radiographic -- in vitro

Milema
n, X O
199039

Verdons
chot, X O
199141

Russell
X O
, 199342
 1 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8 9 9 10
05
0 5 0 5 0 5 0 5 0 5 0 5 0 5 0 5 0 5 0

Russell
X O
, 199342

Russell
X O
, 199342

Rickett
s, X O
1997d40

Visual/Tactile -- in vitro

Verdon
schot, X O
199141
Table 13 shows the seven assessments of the diagnosis of lesions penetrating into

dentin on proximal surfaces of posterior permanent teeth. These assessments were

reported in four studies. Six assessments from four studies involved radiographic

methods and one involved visual/tactile examination, all in vitro. Only two of the

studies reported the prevalence of lesions in the sample, 43 and 13 percent,

respectively.39,40 One study summarized the performance of 276 general practitioners,

who reviewed duplicated films in a mailed survey,39 two studies used three

examiners,41,42 and one used five examiners.40

The radiographic assessments all showed high specificity values, ranging from 0.92 to

0.99, whereas sensitivity values were lower and more variable, ranging from 0.16 to

0.54. One of the two studies with the lowest sensitivity employed a direct digital

radiographic technique.42 The single visual/tactile assessment41 showed lower

specificity (0.71) and sensitivity equivalent to the better radiographic assessments

(0.56). The diagnostic criteria for this visual/tactile study stressed the cavitation of

the lesion in addition to penetration to dentin, which may have caused more frequent

misidentification of enamel lesions as having penetrated dentin.

The strength of the evidence for assessing the validity of methods for detecting

proximal lesions penetrating into dentin is rated as poor for both methods. Although

the number of studies was too small for conclusions, the available studies suggested

that the radiographic method features high specificity, but a range of sensitivity

levels, from low values of 0.16 to 0.30 to moderate levels around 0.55. Thus, in any

given assessment, little more than one-half of all lesions extending into dentin will be

identified radiographically. The single visual/tactile study precludes any conclusions

about this method for diagnosing proximal lesions penetrating into dentin.

Any Carious Lesions on Proximal Surfaces

Table 14. Sensitivity and specificity values for the (more...)

Table 14. Sensitivity and specificity values for the detection of any proximal

carious lesions

1 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8 9 9 10
05
0 5 0 5 0 5 0 5 0 5 0 5 0 5 0 5 0 5 0

Radiographic -- in vitro

Heav
en, OX
199243

Russe X O
ll,
 1 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8 9 9 10
05
0 5 0 5 0 5 0 5 0 5 0 5 0 5 0 5 0 5 0

199342

Russe
ll, X O
199342

Russe
ll, X O
199342

Ricke
tts,
X O
1997d
40

Firest
one, O X
199844

Firest
one, X O
199844

Firest
one, X O
199844
Table 14 shows the eight assessments of the diagnosis of any carious lesion on

proximal surfaces of posterior permanent teeth. These assessments stemmed from

four studies, and all assessed the performance of radiographic methods in vitro.

Prevalence of carious lesions was reported in three studies and ranged from 37 to 75
percent. Two of the studies used a single "examiner," a computer-based image

analysis program;43,44 one used 3 examiners,42 one used 5,40 and two used 16

examiners.44

The specificities were generally high, although there was a greater range than was

seen for diagnosis of cavitated lesions and lesions penetrating the dentin (74 to 100

percent). The assessment with perfect specificity also reported perfect sensitivity. This

assessment, which employed image analysis software, evaluated only 16 surfaces, 75

percent of which were carious. Three other assessments reported moderately high

sensitivity levels, all stemming from a single study that evaluated automatic image

analysis software, D-speed film, and a digital image of the D-speed film read with the

results of the automatic image analysis displayed.44 Lesion prevalence in this study

was also high at 66 percent. The remaining four assessments returned low sensitivity

levels, 0.15 to 0.27. One of these assessments also employed digital imaging

techniques.42

Three of the studies reflected the same patterns seen in the performance of

radiographs in detecting lesions extending into dentin. Two studies reported

assessments with relatively low sensitivity and high specificity,40,42 whereas one study

reported assessments with higher sensitivity and reduced specificity.44 Again, the

variation among studies limited our ability to reach firm conclusions about the

performance of radiographs in detecting proximal carious lesions of any extent. The

strength of the evidence for assessing the validity of the method is rated as poor. The

specificity of this method appears to be slightly more variable than when it is used to

diagnose lesions into dentin or cavitated lesions. Sensitivity levels varied widely and

were highest when computer-based image analysis was applied to digital images.30

Cavitated Lesions on Occlusal Surfaces

Table 15. Sensitivity and specificity values for the (more...)

Table 15. Sensitivity and specificity values for the detection of occlusal

cavitated lesions

1 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8 9 9 10
05
0 5 0 5 0 5 0 5 0 5 0 5 0 5 0 5 0 5 0

Visual/Tactile -- in vivo

Dow
ner,
OX
1975
109

Visual/Tactile -- in vitro

Dow
ner,
O X
1975
35

Visual -- in vitro

Dow
ner,
O X
1975
35

Ketl
ey,
X O
1993
45

Table 15 summarizes the findings for four assessments from three studies of the

diagnosis of cavitated lesions on occlusal surfaces. One assessment evaluated the

visual/tactile method in vitro, one the visual method in vivo, and two the visual

method in vitro. Specificities were high, but again displayed some variability (0.78 to

0.98). Sensitivities were also high for three assessments, but quite low for the fourth

(0.31 to 0.94),45 which had the highest specificity. Criteria used in this study were

essentially similar to those used in the other visual in vitro study.35 Each study used a

single examiner, which may account for much of the difference between similar

studies. There was little difference in performance in an intrastudy comparison of

visual and visual/tactile methods.35 The strength of the evidence for assessing validity

of the methods is rated as poor because of the limited number of studies for any

given method.

Lesions Penetrating into Dentin on Occlusal Surfaces

Table 16. Sensitivity and specificity values for the (more...)

Table 16. Sensitivity and specificity values for the detection of occlusal

carious lesions penetrating into dentin

1 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8 9 9 10
05
0 5 0 5 0 5 0 5 0 5 0 5 0 5 O 5 0 5 0

Radiographic -- in vitro

Wenzel
X O
, 199049

Wenzel
X O
, 199049

Wenzel
X O
, 199150
 1 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8 9 9 10
05
0 5 0 5 0 5 0 5 0 5 0 5 0 5 O 5 0 5 0

Wenzel
X O
, 199150

Wenzel
X O
, 199150

Wenzel
X O
, 199150

Wenzel
X O
, 199150

Wenzel
X O
, 199251

Wenzel
X O
, 199251

Wenzel
X O
, 199251

Nytun,
O X
199248

Ketley,
X O
199345

Russell
X O
, 199342

Russell X O
 1 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8 9 9 10
05
0 5 0 5 0 5 0 5 0 5 0 5 0 5 O 5 0 5 0

, 199342

Russell
X O
, 199342

Lussi,
X O
199355

Verdon
schot, X O
199356

Lussi,
X O
199547

Rickett
s, X O
1994111

Huysm
ans, X O
1997112

Ekstra
nd, X O
199757

Rickett
s, X O
1997d40
 1 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8 9 9 10
05
0 5 0 5 0 5 0 5 0 5 0 5 0 5 O 5 0 5 0

Ashley,
X O
199852

Ashley,
X O
199852

Huysm
ans, X O
1997112

Visual/Tactile -- in vitro

Pennin
g, X O
199254

Lussi,
X O
199355

Visual -- in vivo

Rickett
s, X O
199546

Visual -- in vitro

Nytun,
O X
199248

Wenzel X O
 1 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8 9 9 10
05
0 5 0 5 0 5 0 5 0 5 0 5 0 5 O 5 0 5 0

, 199251

Lussi,
X O
199355

Lussi,
X O
199355

Verdon
schot, X O
199357

Deery,
X O
199563

Ekstra
nd, O X
199757

Ashley,
X O
199852

Huysm
ans, X O
1997112

EC -- in vivo

Lussi,
O X
199547
 1 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8 9 9 10
05
0 5 0 5 0 5 0 5 0 5 0 5 0 5 O 5 0 5 0

Rickett
s, O X
199546

Verdon
schot, X O
199356

Rickett
s, O X
199546

Rickett
s, O X
199546

Rickett
O
s,
X
1997a58

Rickett
s, O X
1997b59

Rickett
O
s,
X
1997c60

Ekstra O X
nd,
 1 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8 9 9 10
05
0 5 0 5 0 5 0 5 0 5 0 5 0 5 O 5 0 5 0

199757

Lussi,
X
199961

Huysm
ans, X O
199853

Huysm
ans, X O
199853

Huysm
X
ans,
O
199853

Ashley, X
199852 O

FOTI -- in vitro

Ashley,
X O
199852

Laser Fluorescence -- in vitro

Lussi,
X O
199961

Lussi, O X
 1 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8 9 9 10
05
0 5 0 5 0 5 0 5 0 5 0 5 0 5 O 5 0 5 0

199961
Table 16 shows the 54 assessments of methods for diagnosing carious lesions

extending into the dentin on occlusal surfaces. Six different methods were

represented: radiographic techniques, visual/tactile examination, visual technique,

FOTI, EC, and laser fluorescence. Three of the assessments, one visual and two EC,

were completed in vivo.46,47 The studies were quite varied within method in terms of

the details of the diagnostic methods employed as well as the specific diagnostic

criteria used. Caries prevalence in the samples tended to be high, with 57 percent of

assessments performed on samples with lesion prevalences equal to or greater than

40 percent.

The radiographic studies showed moderate variation in specificities, which ranged

generally from 0.75 to 1.00. Sensitivities appeared in two distinct ranges of values

with one value centered near 0.20 and another ranging from 0.45 to 0.70. Generally,

the assessments with the lowest values for sensitivity had specificity values near the

high end of the range of values. A single study reported a specificity of 0.50, with a

sensitivity within the upper range at 0.66. This assessment was based on teeth

selected because they showed signs of "fissure caries," and the prevalence of carious

lesions penetrating into the dentin was 77 percent.48 Ten of these assessments

involved digital techniques.42,49-53 Two assessments reported sensitivities in the lower

range of values (0.21, 0.24)42,52 and eight were in the higher range (0.54 to 0.72).

Five studies reporting assessments of digital methods also included assessments

based on direct examination of D- or E-speed films. In three instances, the

performance of the film image was not markedly different than any of the digital

comparisons reported in the same study.42,49,52 In one study, one of the digital

methods, edge enhancement of the digitized film image, yielded 0.23 and 0.04

improvement in sensitivity and specificity, respectively.51 This study featured a single

examiner, with 1-week intervals separating diagnostic sessions where different

methods were employed.

Two in vitro assessments of the visual/tactile method returned sensitivity values of

0.24, with specificities within the range established in the radiographic studies.54,55

One of these assessments, which reported perfect specificity, reflected standardized

(mechanical) probing of 1,140 sites on three teeth.54 The criterion for a positive

diagnosis was the force necessary to withdraw the probe from the surface of the

tooth, or "tugback." The visual component of the examination was limited to directing

the probe to all pits and fissures on the surface of the tooth.

Both in vivo and in vitro visual assessments were included in the review. The single in

vivo assessment reported a sensitivity of 0.03 and a specificity of 0.97 for what was

presumably a single examiner's diagnoses of specific sites on third molars.46 The in

vitro assessments generally reflected the same levels of performance as were seen in

the radiographic assessments. Two groups of sensitivities were reported, more

commonly between 0.10 and 0.25, with two studies between 0.45 and 0.55.51,56 Two

assessments reported higher sensitivity values. One assessment was based on the

sample of teeth selected for signs of "fissure caries"48 and the other used a histologic

criterion that required lesions to penetrate more than one-third of the width of the

dentin before dentin caries was declared.57

The two in vivo EC assessments both showed moderate-to-high specificities (0.56,

0.77) and extremely high sensitivities (0.93, 0.97).46,47 The in vitro EC assessments

showed much more variation in both sensitivities and specificities. In some studies,

the sensitivities were equal to or greater than the linked specificities. These studies

were characterized by relatively modest sample sizes, all between 76 and 107 sites,

and often a single examiner. Five of the 12 assessments were reported by the same

principal author,46,58-60 and four of these assessments were characterized by post hoc

determination of the optimal values for caries criteria.

The single FOTI assessment displayed a sensitivity value of 0.14, at the lower end of

the range of values in both radiographic and visual examinations, and a high
specificity value of 0.95.52 Finally, two assessments of a laser fluorescence method

utilizing newly available equipment also reported relatively high sensitivity levels

compared with radiographic and visual methods (0.76, 0.84), with specificities in the

lower half of the range of values for these other methods (0.87, 0.79). The

assessments represented one report,61 with the system tested with the teeth dry and

moist.

The general pattern of results suggested that radiographic and visual methods were

roughly comparable in their ability to accurately indicate the presence of carious

lesions penetrating into the dentin on occlusal surfaces. The two visual/tactile

assessments were not markedly different than the larger number of in vitro visual

assessments, and the single in vitro FOTI assessment returned a similar performance.

EC and laser fluorescence methods showed better sensitivity than these other

methods, with only a small penalty reflected in reduced specificities. However, the

strength of the evidence for assessing the validity of each of the methods in any

specific application is rated as poor because of the variation in sensitivity among

studies and the small number of studies reported. The evidence presented by studies

describing the EC method is the closest to meeting criteria for a fair rating and would

do so if either of two studies were to be discounted. The pattern of sensitivity and

specificity values for EC assessments presented a generally narrower range of

variation than any other method. Further as noted, specificity values were typically

lower than for other methods, and sensitivity values were typically higher, with the

result that for 8 of 14 assessments, sensitivity was higher than specificity.

Any Carious Lesions on Occlusal Surfaces

Table 17. Sensitivity and specificity values for the (more...)

Table 17. Sensitivity and specificity values for the detection of any occlusal

carious lesions
 1 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8 9 9 10
05
0 5 0 5 0 5 0 5 0 5 0 5 0 5 0 5 0 5 0

Radiographic -- in vitro

Russe
ll, X O
199342

Russe
ll, X O
199342

Russe
ll, X O
199342

Ricke
tts,
X O
1997d
40

Wenz
el, X O
199049

Wenz
el, X O
199049

Lazar X O
chik,
199511
 1 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8 9 9 10
05
0 5 0 5 0 5 0 5 0 5 0 5 0 5 0 5 0 5 0

Visual/Tactile -- in vitro

Penni
ng, X O
199254

Lussi,
199111 X O
4

Visual -- in vivo

Ricke
tts, X O
199546

Visual
-- in
vitro

Wenz
el, O X
199049

Lussi
,
X O
199111
4
 1 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8 9 9 10
05
0 5 0 5 0 5 0 5 0 5 0 5 0 5 0 5 0 5 0

Deery
, O X
199562

EC --
in
vivo

Ricke
O
tts,
X
199546

EC -- in vitro

Rock,
198811 X O
5

Ricke
tts, X O
199546

Ricke
O
tts,
X
199546

Ricke
tts,
X O
199611
6
 1 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8 9 9 10
05
0 5 0 5 0 5 0 5 0 5 0 5 0 5 0 5 0 5 0

Ricke
tts,
X O
1997a
58

Ricke
tts,
X O
1997b
59

Ricke
tts,
X O
1997c
60

Table 17 summarizes the 21 assessments of diagnostic methods for detecting any

carious lesions on occlusal surfaces. Seven assessments evaluated radiographic

methods, two evaluated visual/tactile methods, one and three studies evaluated

visual methods, respectively, in vivo and in vitro, and one and seven studies

evaluated EC in vivo and in vitro, respectively.

The radiographic assessments displayed the two ranges of sensitivity seen in

summaries of radiographic assessments on other surfaces and for other types of

lesions. The lower range of sensitivities was generally associated with higher

specificities. This pattern of two ranges of sensitivity levels in the aggregated results

of radiographic studies was most likely not associated with a single design feature of

these assessments. Principal determinants of sensitivity included the specific criteria

used for identifying lesions, the criteria for selection of sample teeth, the selection

and training of the examiner(s), and the extent to which the study design protected

against upward performance bias through familiarity with the sample. Assessments of
digital radiographic techniques reported sensitivities at the higher level for digitized

film49 and the lower level for a direct digital method.42

The visual/tactile assessments reflected the radiographic performance levels, with one

assessment each in the higher and lower sensitivity ranges. The in vivo visual

assessment fell in the same lower sensitivity range, whereas the in vitro visual

assessments reflected the higher range. In two of these three assessments,49,62 the

sensitivity values were greater than the specificity values. In these instances, the

extremely high prevalence of lesions in the samples, 89 and 97 percent, respectively,

may have influenced the examiners' diagnostic decisions.

The EC assessments reflected a pattern similar to such assessments of lesions

penetrating dentin, with the exception that the sensitivities tended to be slightly lower

as a group. Six of the seven studies were reported by the same principal author; all

featured one examiner; sample sizes were between 30 and 100 sites with the same

sample used in two separate studies;58,59 and relatively high lesion prevalence, 64 to

80 percent, was reported.

Overall, the performance of diagnostic methods for the detection of any carious

lesions on occlusal surfaces was similar to the performance of these methods in

detecting lesions penetrating dentin. Again, however, the strength of the evidence is

rated as poor for all radiographic visual and visual-tactile methods, as well as for in

vivo EC studies because of the variation in sensitivity and/or the number of available

studies. The strength of the evidence is rated as poor for in vitro EC studies because

of their low quality scores.

Enamel-Only Lesions on Occlusal Surfaces

Table 18. Sensitivity and specificity values for the (more...)

Table 18. Sensitivity and specificity values for the detection of occlusal

carious lesions confined to enamel

1 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8 9 9 10
05
0 5 0 5 0 5 0 5 0 5 0 5 0 5 0 5 0 5 0

Radiographic -- in vitro

Wen
zel,
X O
1990
49

Wen
zel,
X O
1990
49

Ashl
ey,
X O
1998
52

Ashl
ey,
X O
1998
52

Visual -- in vitro

Wen
zel,
O X
1990
49
 1 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8 9 9 10
05
0 5 0 5 0 5 0 5 0 5 0 5 0 5 0 5 0 5 0

Ash
ley,
X O
1998
52

EC -- in vitro

Ash
ley,
X O
1998
52

FOTI -- in vitro

Ashl
ey,
X O
1998
52

Eight in vitro assessments of enamel lesions on occlusal surfaces were reported: four

radiographic assessments, two visual, and one each EC and FOTI (Table 18). The

radiographic assessments reflected somewhat lower specificity levels for given

sensitivity levels than were reported for lesions into dentin. The four assessments,

reported in two studies, appeared to cluster into higher and lower sensitivity ranges,

but specificity levels were uniformly at or slightly below 0.80. The two assessments of

visual methods reflected higher sensitivity levels, with specificity dropping to near

0.75. The sole EC assessment reflected a similar result, and the only FOTI assessment

showed higher specificity with slightly lower sensitivity. Among the extremely small

number of assessments available, the visual, EC, and FOTI methods seemed to offer

higher sensitivities than the radiographic for the detection of enamel-only lesions on
occlusal surfaces. However, the strength of the evidence is rated as poor for all

methods because of the small numbers of studies.

Other Assessments

Thirteen other assessments of diagnostic methods were reported. Three assessments

reported the results of methods on primary teeth. Radiographs were extremely

sensitive (0.99) and highly specific (0.89) in detecting cavitated lesions on primary

tooth proximal surfaces in vivo,63 and slightly less so (sensitivity=0.93,

specificity=0.89) in detecting lesions into dentin on occlusal surfaces in vitro.45 Visual

methods for detecting cavitated surfaces on occlusal services in vitro were less

sensitive (0.45) but perfectly specific (1.00).45 The strength of the evidence for

assessing the validity of any diagnostic methods in primary teeth is rated as poor.

Seven assessments examined performance of radiographic methods in vitro on

combined anterior and posterior proximal surfaces. Six of these assessments were

reported in the same study64 and featured high specificity for dentin lesions (0.94 to

0.95) that drop somewhat for enamel lesions and any lesions (0.76 to 0.80), and

moderate sensitivities for dentin and any lesions (0.49 to 0.58) that drop somewhat

for enamel-only lesions (0.35 to 0.46).

The other assessment of any lesions reported high specificity (0.95) and low

sensitivity (0.33). The strength of the evidence is rated as poor.

Three assessments reported results for combined methods, all visual and radiographic

examinations of posterior occlusal surfaces to detect carious lesions penetrating into

dentin.48,55,65 All of the assessments involved at least 10 examiners, and when

reported, intra- and interexaminer reliability was low to moderate (kappa=0.30,

0.46). Performance of the combined methods varied, with moderately high

sensitivities (0.49 to 0.86) and specificities in the lower half of the range reported for

visual-only or radiographic-only methods (0.64 to 0.87). These results suggest that

the combination of visual and radiographic methods returns moderate to high

sensitivities, i.e., in the upper range of the dual range of sensitivity values described
earlier, but that specificity may be reduced from that of either method alone. The

strength of the evidence is rated as poor.

Limitations to the Evidence Base

The literature describing the histologically determined validity of methods for

diagnosing carious lesions had a variety of limitations, many of which represented

potentially serious threats to internal validity, and most of which represented barriers

to generalization of the reported results to dental practice. The most obvious

limitation has already been noted -- the virtual absence of any assessments of

diagnostic methods applied to primary teeth and to root surfaces of permanent teeth.

The breadth of reported studies also seriously restricted any conclusions about

differences in the validities of visual and visual/tactile examinations and possible

advantages of combining examination methods. A minority of method/surface/lesion

type combinations were represented by more than three studies; and for most of

these combinations, the variation among reported performances was extensive.

Although the team might have meta-analyzed these results for several combinations

of method, surface, and lesion type, two characteristics of the assessments

discouraged the use of meta-analysis. First, many were not independent assessments,

reflecting common examiners and sample teeth across studies. Second, the studies

did not all assess the same "outcomes," since criteria for diagnosis were different.

Thus, the available literature could not support specific conclusions about the

performance of various diagnostic methods and permitted only the occasional

generalization about relative differences in performance.

The quality scores for these studies tended to cluster in the low to mid range of the

scale of possible scores. The mean for in vivo studies was 61, and for in vitro studies

it was 45. The range was 5 to 70. Most studies had sample sizes of 75 or more sites

or surfaces. The choice of sites rather than surfaces may be problematic for external

validity, however, because most occlusal surfaces will present multiple sites for

assessment. The results of site assessment did not summarize the status of the entire

surface, as is routinely done in clinical practice. As noted, most of the studies were

performed in vitro, a practical necessity if histologic validation is to be easily

accomplished. There is some indication that in vitro assessments tended to return

higher sensitivity values than in vivo assessments, although the numbers of studies

were too small and the number of study variables too large to assume that the small

observed differences were the results of the setting per se. Most in vivo studies were

limited to premolars and third molars, which tend to be extracted more frequently in

good clinical condition. In vitro studies also often relied on these teeth for the same

reason -- that they are more frequently available with intact crowns. The problem is

that the teeth that most frequently experience occlusal and proximal surface decay --

the first and second molars -- differ from the premolars and third molars in ways that

may affect the performance of diagnostic methods. For example, occlusal surfaces of

third molars tend to have less well-coalesced fissures; and proximal enamel

thicknesses, both buccolingually and mesiodistally, are less in both premolars and

third molars.

The inclusion criteria for the review required a histologic validation, and a variety of

validation methods were represented in the included studies, with little assurance that

different methods are equivalent.66,67 Slightly fewer than one-half of the studies relied

on light microscopy, with an identical number using other methods for evaluating the

extent of caries on sectioned teeth. The remainder used visual criteria to confirm

cavitated lesions. Further, a majority of studies supplied no explicit criteria for the

validation, and a large majority did not report reliability information for the validation

despite known variability in this procedure.66 The result was that the reported

diagnostic performances may be biased, although not in any predictable direction, by

the validation procedures employed, making comparisons problematic.

The percent of sites to be evaluated that actually included a carious lesion was less

than 20 percent in only 5 percent of in vitro assessments compared with 53 percent of

in vivo assessments. Further, most of the in vivo sites with lesion prevalences above

20 percent were selected sites on third molars. Lesion prevalence proportions above

20 percent across all surfaces (a D value of 40 in a fully dentate individual) are rare in

clinical practice. The effect of elevated frequencies of occurrence in assessment

samples raised issues about examiner bias, since unusual presentations may alter

examiner alertness and behavior, albeit in an unknown manner. The criteria used for
selecting teeth for the samples also raised issues about both the comparability of

studies and the generalization of results to clinical practice. The criteria described for

selection did not always seem intended to reflect typical presentations in clinical

practice. When this limitation was coupled with the previously noted limitations

resulting from restricted tooth types and the use of sites rather than surfaces,

generalization to clinical practice was again problematic.

Only one-half of the studies reported the combined performance of four or more

evaluators. The studies relying on a smaller number of evaluators may present

difficulties in generalization because of the positive influence of particularly skilled

investigator/examiners. Single examiner studies, of which there were 17 (46 percent

of studies), were especially vulnerable to this phenomenon. Further, 17 studies

reported no reliability information for the examiner(s). When interexaminer reliability

was reported, the values often were low enough to underscore the threat to external

validity represented by the use of single examiners.

Finally, 14 of the assessments described post hoc determination of the optimal criteria

for lesion designation. Although the development of new diagnostic techniques often

requires such procedures, it is usually expected that the criteria will then be tested in

a second "validation" sample. Such a procedure was not reported in any of the 14

assessments.