Diagnosis of Carious Lesions

The team's review revealed two principal shortcomings of the existing literature describing histologically validated assessments of diagnostic methods. First, the coverage was spotty in terms of combinations of lesion types, tooth surfaces, tooth types, patient types, and diagnostic methods for which assessments were available. Second, the literature was characterized by designs that are open to threats to internal validity and are problematic in terms of external validity. Efforts must be made to increase the "coverage" of the literature and at the same time address the design characteristics that limit the applicability of existing studies. Merely acquiring additional studies similar to those currently available is at best an inefficient approach to advance the understanding of the performance of methods for dental caries diagnosis.

Perhaps one of most limiting aspects of the literature is that a majority of the studies were performed in vitro. This characteristic of current research is eminently understandable for practical reasons, but in vitro studies have serious limitations. They are more difficult to generalize to the environment of dental practice for several reasons, not the least of which is that they permit careful selection of individual teeth or surfaces for assessment, rather than forcing the inclusion of a more representative set of teeth or surfaces. In vitro studies also minimize many limitations imposed by working within the oral cavity, arguably leading to improved performance, and they tend to emphasize certain tooth types. Because most in vivo studies also have had this latter limitation, rethinking the source of research material for in vivo studies of dental caries diagnostic may be necessary. One possibility might be to conduct such studies post mortem, although this approach is not strictly speaking in vivo, and would tend to overrepresent elderly subjects. Another possibility might be to expend more effort recruiting subjects from among patients in dental care systems where extractions are part of planned treatment. The former approach would facilitate the use of multiple examiners, thereby addressing another substantial limitation of the current literature.

Attention to the method used for histologic validation of the sample is also necessary. Work is needed to determine an acceptable standard technique for determination of the presence of a carious lesion from among techniques based on microscopy, stereomicroscopy, and microradiography. Standards for sectioning method and thickness, number of sections surveyed, magnification, dyes, and criteria for identification of a lesion all must be ascertained. Minimum expectations for number of examiners, reliability, and reporting methods should also be specified as a part of the standard. A conference or workshop of invited experts would represent a possible mechanism for standard setting.

The majority of studies were found to be deficient in terms of complete descriptions of important study characteristics, including the criteria for positive diagnoses of carious lesions, the criteria for selection of the sample of teeth or surfaces to be diagnosed, the background and training of the examiners, and examiner reliability. All reports should include a minimum set of descriptions in a standard format to facilitate comparisons among studies. The development of such a standard could be undertaken by one or more dental organizations sponsoring journals in which caries diagnosis reports appear. The standards might be developed along the lines of the CONSORT statement for reports of clinical trials. ³³

The issues of outcome measures and disease prevalence in diagnostic studies should be addressed in the standards document. This review included only studies reporting outcomes in terms of sensitivity and specificity. This inclusion criterion was limiting in that some studies reporting results in terms of areas under ROC curves were excluded. ROC results permit comparison of studies on the basis of a single number reflecting the "tradeoff" between sensitivity and specificity across a range of examiner confidence levels, but require collection of additional information from examiners regarding their certainty for each diagnostic decision. Although the utility of providing this type of outcome data for dental caries diagnosis studies has yet to be demonstrated, the standards should address the circumstances where one or both outcomes might be reported. The prevalence of carious lesions in the samples of current studies often represent barriers to generalization of the results of these studies, if not threats to internal validity. Prevalence of lesions in any in vivo or in vitro sample should be reasonably representative of the population prevalence for the same type of lesion on the same surfaces.

Once a set of standards is in place to guide investigators in designing studies and preparing reports that will facilitate the assessment of the validities of diagnostic methods for dental caries, some attention to the coverage of those assessments will be beneficial. Clearly, more assessments of newer methods are necessary. FOTI and digital radiographic methods are two obvious candidates. EC methods also would benefit from stronger assessments, and laser fluorescence methods will also require more assessment in the near future. Equally important, studies must include assessments on primary teeth and on root surfaces of permanent teeth.

Because identification of a lesion at one examination may not furnish sufficient information to provide an accurate assessment of the lesion's activity status and prognosis, consideration must be given to how various diagnostic methods facilitate longitudinal assessment of changes in lesion volume. Finally, diagnostic studies must begin to evaluate more than just the immediate outcomes of the use of the particular method or methods being assessed. Although the validity of diagnosis must be the principal concern in such assessments, some attention should be paid to the outcome in terms of the appropriateness of the treatment provided in response to the diagnosis. These longer term considerations represent the ultimate outcomes of diagnostic procedures. To this end, the procedures must be evaluated in terms of their benefit to patients, an outcome mediated by dentists' application of the information provided by the diagnostic method.

Caries Management Studies

At the most general level, additional clinical studies examining outcomes of management strategies for noncavitated lesions and for caries-active patients are clearly needed. In part, the number of available studies may be small because of the expense involved in mounting such studies and the understandable substitution of model systems, particularly for remineralization studies. Nevertheless, for all professionally applied remineralization methods, as well as for almost all professionally applied preventive interventions in caries-active/high-risk individuals, the evidence for efficacy is incomplete. The dental profession is just beginning to consider the issues surrounding evaluating carious lesions longitudinally and delaying surgical intervention until lesions are well advanced. The delay permits a period to evaluate whether a lesion is progressing or is inactive, necessitates a second evaluation that may lead to correction of initial false positive diagnoses, and offers an opportunity for nonsurgical treatment to arrest or reverse progression if it is occurring. But without nonsurgical treatments with proven efficacy, an important rationale for minimizing immediate surgical intervention is weakened. The same situation exists for management of caries-active patients. Dentists are just beginning to appreciate the role that risk assessment can play in the management of their patients, but again, in the absence of demonstrably efficacious treatments for those at heightened risk, much of the attractiveness of risk assessment will be lost.

The simplistic goal of acquiring more studies may be an inefficient solution to the problem of determining efficacy of current methods for management of noncavitated lesions and caries-active individuals. For example, although there were four times as many studies reviewed for the caries-active question as the noncavitated lesion question, the additional studies contributed to the resolution of only one efficacy question. Investigators must be encouraged to contribute studies that fill identified gaps, that build on existing findings, and that use methods that facilitate comparison across studies. The crazy quilt of intervention protocols and study designs found in this review of studies involving caries-active individuals suggests that without such a scheme, much of the research effort expended on a topic may not be very useful in basic determinations of efficacy. Gap-filling research and comparability could be encouraged by funding sources that place more emphasis on acknowledging these gaps and explain how the proposed findings will complement existing knowledge. Comparability could also be improved through methods that encourage rather more complete reporting of study methods and results than has been the norm in many dental trials. At a minimum, adherence to the CONSORT criteria ³³ should be expected by editors as a guide for providing a minimum level of information about study procedures. In addition, to facilitate comparison of caries studies in particular, more complete descriptive information about community and individual preventive dentistry exposures is needed.

Comparison group regimens for future research on preventive interventions for noncavitated lesions should be designed so that the evaluation of the experimental intervention will describe its efficacy compared with the most commonly used alternative nonsurgical treatment. Similarly, studies of preventive interventions for caries-active individuals should compare efficacy with the most common alternative preventive intervention. In both instances, the most common alternative is doing nothing unusual for the lesion or the individual, which is translated into "usual care." Because "usual care" is seldom the same between studies, or even for all group members within a study, investigators should, at a minimum, document the professional preventive care received by each member of the comparison group. Controlling the care received by all subjects is an alternative approach, but one that could add to the cost of doing the research.

In addition to specific treatment received by members of the comparison group, some of the reviewed studies indicated that experimental and comparison groups were exposed to other professional and/or community preventive dentistry regimens. Again, such exposure should be documented at the individual level. These exposures together with those accruing through participation in the comparison group should be included as covariates in the analyses. What is important is that the efficacy of the experimental intervention be evaluated under conditions that are duplicable by other researchers, easily generalizable to dental practice, and, where all possible threats to internal validity are known, reported and, if possible, controlled.

These recommendations are nothing more than an appeal to execute well-designed studies. Attention to the preceding issues, as well as to sample sizes and needed power, regimen compliance, attrition, and examiner reliability, would all represent needed strengthening of this literature. Not only should efficacy studies seek to maximize regimen compliance, but also the level of compliance should be accurately assessed and reported. Compliance can directly affect efficacy and might well explain substantial proportions of the variation among reports, but only if it were known.

Finally, there is an opportunity to increase the amount of information available about management of both noncavitated lesions and caries-active individuals through secondary analyses of existing data. Since almost all trials of preventive agents include baseline caries assessments, secondary analyses of outcomes stratified by baseline caries prevalence are theoretically possible. Some studies may also have collected information describing other risk factors for caries, which could be used in such stratified analyses. Similarly, an unknown number of trials include initial, or D₁, lesions in the examination criteria. Analyses of the fate of such lesions identified at baseline could be readily accomplished and would add to the very limited store of knowledge.