| Sign In to gain access to subscriptions and/or personal tools. |
Estimating Rates of New Root Caries in Older Adults
1 Centers for Disease Control and Prevention/Division of Oral Health/Surveillance, Investigations, and Research Branch, 4770 Buford Highway, MSF10, Chamblee, GA 30341, USA; and Correspondence: * corresponding author, sig1{at}cdc.gov
Although older adults are keeping their teeth longer, no national data are available on new caries in this age group. To characterize the extent of caries among older adults, we systematically reviewed studies on root caries incidence, increment, attack rate, and annual total (root + coronal) caries increment. We used a random-effects model to estimate annual summary measures and their 95% confidence intervals (CI) and tested for heterogeneity. For the 9 studies reporting root caries incidence, the summary measure equaled 23.7% (CI = 17.1–30.2%). For the 9 studies reporting root caries increment, the summary measure was 0.47 surfaces (CI = 0.34–0.61). For the 7 studies reporting total caries increment, the summary measure equaled 1.31 surfaces (95% CI = 1.01–1.61 surfaces). Because of heterogeneity, summary measures should be interpreted with caution. This research suggests, however, that older adults experience high rates of new caries and could benefit from caries-prevention programs.
Key Words: root caries • incidence • increment
US adults are keeping their teeth longer (US Department of Health and Human Services, 2000); however, increased retention of teeth means that more adults are at risk for root caries. Gingival recession caused by normal aging puts root surfaces especially at risk (US Department of Health and Human Services, 2000). Among dentate adults, the prevalence of root caries experience is about 8 times higher among those aged 75+ yrs (55.9%) than among those aged 18 to 24 yrs (6.9%) (Winn et al., 1996). Because the overall US population is growing, with the proportion of the US population aged 65 yrs or older projected to increase from 12.5% in year 2000 to 20.1% in 2030 (Administration on Aging, 2003), root caries experience can be expected to increase in future years. Very few public dollars are allocated to adult dental disease. Dental care is one of the few health areas not covered by Medicare (US Government program that covers medical costs for adults over age 65 years), and only a limited safety net exists to provide dental care to low-income adults. Medicaid frequently covers only emergency dental care for adults; approximately 15% of state Medicaid programs (US program funded by Federal and state governments that covers medical costs for the poor) provide no adult dental benefits, and approximately 45% cover only tooth extraction and emergency services (Oral Health America, 2003). With the exception of water fluoridation, virtually all community prevention programs target children and youth (Association of State and Territorial Dental Directors and Centers for Disease Control and Prevention, 2002). As a result, decay among adults is more likely to remain untreated; the mean number of untreated decayed surfaces among American adults (1.8) (Winn et al., 1996) is about 4 times that among US schoolchildren (0.4) (Kaste et al., 1996). Documenting the burden of root caries is important, because this condition can be prevented and/or arrested in its early stages with fluorides, and such early preventive treatments have been shown to lower overall treatment costs (US Department of Health and Human Services, 2000; Leake, 2001). In this study, we systematically reviewed published studies of root caries incidence (the proportion of population developing caries in at least 1 root surface), root caries increment (the mean number of root surfaces becoming decayed or filled per person), and root caries attack rate (proportion of root surfaces at risk for caries [i.e., sound, exposed], which developed caries). We also estimated the total caries increment (root plus coronal) by analyzing data from studies that presented both root and coronal caries increments. One previous study has summarized the findings of 5 North American studies on new root caries (Leake, 2001). Our study adds the findings of 6 studies conducted among populations living in advanced market economies and uses different assumptions on how to adjust for negative reversal and to derive annual estimates of incidence. Our study also differed from the earlier one in its use of standard meta-analysis methods to derive summary statistics, its use of a random-effects model, and its use of a test for heterogeneity to determine if study outcomes should be pooled.
Search for Studies A librarian at the Centers for Disease Control and Prevention searched MEDLINE and EMBASE (a major biomedical and pharmaceutical database) for studies on dental caries incidence, increment, and attack rate conducted after 1980 among adult populations and published in English. We reviewed the search abstracts and secured articles with study populations 45 yrs and older living in countries with established market economies. In addition, we searched the references of each ordered article for additional studies.
Data Abstraction We sorted articles by location and assigned all articles with the same location to the same team to make sure that we were not double-counting some populations that may have appeared in more than one article. After each team member independently abstracted the assigned article, the team completed a consensus abstraction form.
Inclusion Criteria
Data Adjustment To estimate annual incidence, we calculated the probability that no disease was incurred over the study interval and then took the nth root of this value (where n represents number of yrs in the study) to calculate the probability that no disease was incurred in a given yr. We subtracted this value from 1 to obtain the annual incidence. Some studies reported separate results for specific groups. For these studies, we estimated the outcome measure for the entire study population by taking the weighted average for specific groups and calculating the associated standard error according to the following formula:
We excluded studies that reported increment or attack rate but did not report the variance, standard deviation, or standard error for that rate. For studies reporting incidence rates without the standard errors for these rates, we calculated the standard error according to the following formula:
To make the findings of studies that adjusted for negative reversals more comparable with the results of studies that did not, we also, when possible, reported estimates of new disease that did not subtract out negative reversals.
Analysis We tested for heterogeneity among outcome measures with a chi-square test (QW) (Normand, 1999). To identify potential sources of heterogeneity, we examined the Pearson correlation coefficients between each root caries measure and the population’s baseline risk (as measured by mean number of decayed and filled root surfaces at baseline, mean number of exposed root surfaces, and mean number of teeth; Appendix Table 4, the population’s baseline age, and study design characteristics (as measured by study length, and whether the study subtracted out negative examiner reversals; Appendix Tables 4 and 5).
The DSL summary annual root caries incidence for the 9 studies reporting this measure was 23.7% (95% CI, 17.1–30.2%), and rates in individual studies ranged from 10.1% in Canada to 40.6% in Washington (Fig. 1  ). For the 6 studies meeting the validity criteria, the DSL summary annual estimate was 20.7% (95% CI, 13.4–28.0%), and for the 6 US studies, it was 25.8% (95% CI, 17.9–33.7%). Among US studies, the lowest incidence rate was 14.1% in North Carolina. Because heterogeneity was present, the summary measures should be interpreted with caution. Incidence increased with age; thus, age may have been one source of heterogeneity (Appendix Table 4).
The DSL summary annual root caries increment for the 9 studies reporting this measure was 0.47 root surfaces (CI, 0.34–0.61) and ranged from 0.20 surfaces in Canada to 0.98 surfaces in Massachusetts (Fig. 2 ). For the 7 studies meeting the validity criteria, the DSL summary estimate was 0.47 surfaces (95% CI, 0.32–0.62), and for the 6 US studies it was 0.49 surfaces (95% CI, 0.38–0.60). Again, we could not reject the null hypothesis of homogenous means when we included non-US studies in the summary measure calculations. Among US studies, however, we could reject the null at the 5% level of significance but not at the 10% level.
Only 5 of the 11 studies had estimated annual root caries attack rate, which ranged from 0.9% in Sweden to 3.9% in Washington (Appendix Fig.). The DSL summary measure equaled 1.9% (95% CI, 0.9–2.8%). Of the 3 US studies that estimated this attack rate, the lowest reported estimate was 1.3% in North Carolina. Because there were so few studies, we did not report separate summary measures for US studies or for studies meeting our validity criteria. Again, summary findings should be interpreted with caution, because the value of QW suggested that we could reject the null hypothesis of homogenous means with 95% confidence. Attack rate increased as age increased and as number of teeth at baseline decreased. The positive correlation between this outcome measure and baseline root surfaces also approached statistical significance (Appendix Table 4).
Seven studies reported total annual caries increment (root plus coronal caries increment) (Fig. 3
At baseline, the populations in the studies we analyzed (which were conducted from 1983 to 1999) had similar numbers of teeth and slightly higher levels of root caries experience than did the population for NHANES III, phase 1 (which was conducted from 1988–1991; Winn et al., 1996) (Table ).
With the exception of annual root caries increment among US studies, we could not reject the null hypothesis of homogenous outcome measures. Sources of heterogeneity included age and number of teeth at baseline. In addition, for 3 of the 5 caries-prevention studies, we did not confine the outcome measure to the control group—2 studies reported outcome measures for all participants because the treatment effect was not statistically significant, and 1 examined the effect of fluoride toothpaste. We did not use any of these clinical trials to estimate summary measures for studies meeting the validity criteria, however, because they did not use probability sampling. The results of this systematic review were, in some instances, markedly different from those reported at the Consensus Development Conference (Leake, 2001) because of differences in underlying assumptions. For example, using our methodology to estimate summary incidence for the NIH study population, we estimated that incidence to be almost twice as high (15.3% vs. 8.2%). The main reasons for the discrepancy were that we included only adults aged 45 and above, used the shortest study interval rather than the longest, and, when possible, reported unadjusted or crude findings. We believe that using the shortest study duration reduces sample bias, because study subjects at highest risk are more likely to drop out (an occurrence especially likely among the elderly). For example, in the North Carolina study, subjects who dropped out of the study between the third- and fifth-year examinations had significantly higher rates of caries incidence over the first 3 yrs than those who did not (Lawrence et al., 1996). In the Iowa study, most of the subjects who dropped out between 18 mos and 3 yrs did so because of severe illness or death, and those who dropped out had significantly higher caries increments during the first 18 mos of the study (Hand et al., 1988a). We also suspect that annualizing incidence may be problematic, because distortions attributable to annualization may become more pronounced as the study duration increases if the probability of developing caries is positively correlated with the probability of having developed caries in previous periods (i.e., recurrent caries within the person). For example, consider a two-year study consisting of two subjects, A and B, in which A developed new caries in both years and B developed caries in neither year. If the incidence of caries was measured separately for each year, the incidence for each year would equal 50%, whereas if the incidence was measured from baseline to the end of the second year and was annualized by being divided by the number of study years, it would equal only 25%. Although studies with longer intervals are less likely to capture recurrent caries at both the person and tooth levels, the problem is more pronounced for incidence measures, because recurrent caries on the same tooth will underestimate both incidence and increment, whereas new caries on a different tooth within the same individual will underestimate only incidence. We also believe that crude estimates may provide better estimates of new caries than do estimates that net out negative reversals. Beck et al.(1995) argue that as baseline caries prevalence increases, the probability of examiner reversals increases and the probability of examiner increments decreases. Thus, they propose a prevalence-based adjustment for examiner reversals—multiply the crude increment by the complement of the number of reversals divided by baseline frequency. With Beck’s adjustment as the gold standard, the percentage deviation from the true value is higher when netting out negative reversals as compared with the crude method. Beck et al.(1995) found that, relative to the adjusted increment, the crude method overestimated the root caries increment by 10%, while netting out negative reversals underestimated it by 38%. Slade and Caplan (2000) found that, for root caries increment and attack rate, estimates from the crude method overestimated those obtained with the adjusted method by approximately 21%, while net estimates underestimated adjusted values by approximately 45%. In conclusion, we found that the annual root caries increment and root + coronal increment were 0.47 and 1.31 surfaces, respectively. A review of studies conducted in the late 1970s and 1980s among children aged 5 to 15 yrs found that, among children living in fluoride-deficient communities, the mean annual coronal caries increment was 1.4 surfaces, and among children living in fluoridated communities, it was 0.8 surfaces (Garcia, 1988). Thus, older adults may experience similar or higher levels of new caries than do schoolchildren, the primary recipients of caries-prevention programs.
We thank Brenda Mazzocchi, CDC Reference Librarian, who conducted all literature searches, and the National Science Foundation, which partially funded this research (Award #0223364). The remainder of the research funding was from the authors’ institutions.
A supplemental appendix to this article is published electronically only at http://www.dentalresearch.org. Received for publication November 24, 2003. Revision received April 30, 2004. Accepted for publication May 28, 2004.
Journal of Dental Research, Vol. 83, No. 8,
634-638 (2004)
|
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
|
|
 |
|
Sign In to gain access to subscriptions and/or personal tools.
TOP
ABSTRACT
INTRODUCTION
