Sociodemographic Variation of Caries Risk Factors in Toddlers and Caregivers

Objectives. Dental caries is the most common chronic childhood disease, with numerous identified risk factors. Risk factor differences could indicate the need to target caregiver/patient education/preventive care intervention strategies based on population and/or individual characteristics. The purpose of this study was to evaluate caries risk factors differences by race/ethnicity, income, and education. Methods. We enrolled 396 caregiver-toddler pairs and administered a 105-item questionnaire addressing demographics, access to care, oral bacteria transmission, caregiver's/toddler's dental and medical health practices, caregiver's dental beliefs, and caregiver's/toddler's snacking/drinking habits. Logistic regressions and ANOVAs were used to evaluate the associations of questionnaire responses with caregiver's race/ethnicity, income, and education. Results. Caregivers self-identified as Non-Hispanic African-American (44%), Non-Hispanic White (36%), Hispanic (19%), and “other” (1%). Differences related to race/ethnicity, income, and education were found in all risk factor categories. Conclusions. Planning of caregiver/patient education/preventive care intervention strategies should be undertaken with these caries risk factor differences kept in mind.


Introduction
According to the first-ever U.S. Surgeon General's report on oral health published in May 2000 [1], dental caries is the most common chronic childhood disease. In addition, most (75%-80%) of the caries in children in this country occur in a small segment of the population (20%-25%) [2], and this problem is particularly prevalent in minorities and immigrants and lower-income children [3]. Unfortunately, the funds to provide preventive care to all children who are either from a minority population or the lower SES levels are simply not available. Therefore, it is reasonable to attempt to identify those at highest risk in these populations and concentrate what limited financial and manpower resources are available in these "highest of the high" [4]. Additionally, the cultural, and behavioral determinants of disease and the barriers to access dental health services in these populations may be dissimilar to those in other social groups [5].
Understanding these differences may eventually influence different preventive strategies and alternative ways to help health providers to communicate with these groups in order to enhance health related behaviors and conditions. Therefore, identifying a child's level of risk for development of dental caries and the reasons behind it are necessary first steps in managing dental caries.
Risk-based prevention and disease management have been recognized as the cornerstones of modern caries management [6][7][8], especially in young children [8][9][10]. The fact that the existence of past restorations is one of the greatest indicators of risk for the development of new caries lesions [9,11] only proves that the act of surgically treating the caries lesion does little to reduce the risk of developing the next lesion, generally makes no significant difference to bacterial loading, nor on the enactment of selfpromoting health behaviors such as brushing one's teeth [12][13][14]. The etiology of the dental caries process is multifactorial in nature and involves a combination of factors including diet, a susceptible host, and microflora, which interplay with a variety of social, cultural and behavioral factors. Additionally, most young children appear to acquire some cariogenic microorganisms (i.e., mutans streptococci-MS) from their mothers or primary caregivers [15]. Transmission happens through saliva and can be affected among other variables by the frequency of the contact (e.g., sharing of food and utensils, kissing, etc.), which could have cultural and behavioral determinants and, therefore, may vary among different ethnic and cultural population groups. Because of the multifactorial nature of the dental caries disease process, and the fact that the disease is very dynamic, but not continuous (e.g., lesions can progress and/or regress), studies on risk assessment tend to be complex, with a multitude of variables challenging the prediction at different times during the life of an individual [10]. For a clinician, the concepts of assessment of risk and prognosis are an important part of clinical decision making, individualized counseling, and anticipatory guidance.
In addition, risk factors may vary based on race, culture, and ethnicity [16][17][18][19][20][21]. Unfortunately, there are very few high-quality, longitudinal caries risk studies focusing on infants and toddlers [10,22]. Furthermore, existing studies have been conducted primarily in selective populations in Northern Europe [23][24][25][26][27][28][29], diminishing the generalizability of these results to the US population. One recent study has been conducted in a low SES, African-American U.S. community [30]. In addition, Gao et al. [31] have recently suggested that practical biopsychosocial caries risk models without biological markers, such as the one tested here, are effective (sensitivity/specificity was 82%/73%) and promise to be cost-effective to reach children in a variety of settings.
Others have studied dental habits, attitudes, and beliefs in a range of settings. However, these studies have drawbacks in application to caregivers of toddlers due to the populations studied (age, race/ethnicity, and/or geographic location) and due to the range of topics covered. Dental habits, beliefs, and attitudes have been studied in adults [32][33][34][35][36][37][38][39][40], finding variability in beliefs and attitudes which may affect their own dental outcomes, but was not necessarily examined in the context of adults who were caring for toddlers. In parents of young children in Great Britain, knowledge and attitudes were found to vary due to education, ethnicity, and area of residence [41]. Dental knowledge, attitudes, and practices may also be impacted by the overall health of the child. Research has shown beliefs were found to differ between parents of children with and without congenital heart disease [42]. Relationships of caries in 3-year olds in Japan with child-rearing behaviors and mother's health behaviors were examined [43], finding a stronger association with the childrelated behaviors than the mother's behaviors.
The purpose of this study was to evaluate how known caries risk factors evaluated longitudinally in young U.S. children differ by the ethnicity, income, and education of the caregiver. These factors had been identified through previous research as possible risk factors and were included as part of a one-year longitudinal risk study. If differences were found in the risk factors, as expected, this could indicate the need to target caregiver/patient education and preventive care intervention strategies based on the characteristics of the population or individual.

Methods
The study population included caregiver-toddler pairs in Indianapolis and Connersville, Indiana, USA. Subjects were recruited through four sites: (1) a primary-care-based studyrecruitment system affiliated with a large metropolitan hospital serving a generally underserved and lower-income population, (2) the Oral Health Research Institute of the Indiana University School of Dentistry, (3) the Hispanic Center of Indianapolis, and (4) the rural town of Connersville. At sites 2-4 above, recruitment was done by radio and newspaper advertisements, as well as contacting an IRB-approved database of people who had participated in previous studies with us at those locations. The adult accompanying the child was required to self-identify as being the primary caregiver for the child. We defined "primary caregiver" (PCG) as the individual consistently responsible for the housing, health, and safety of the child. Toddlers ranged in age from 16 to 36 months at the time of recruitment, and were generally healthy based on the caregivers' responses to a medical history questionnaire. The study protocol, letter of informational consent, and other supporting documents were approved by the Indiana University Medical Center Institutional Review Board prior to their use. Written informed consent was obtained from all PCGs (and parent/legal guardian if different from the child's PCG) prior to their enrollment.
A caries risk questionnaire was developed to include questions related both to the PCG and the child regarding social, cultural, functional, psychological, sociodemographic, dietary, and biological factors that may affect transmission, development of caries, and access to care in these populations. Many of the questions were taken or modified from other risk assessment questionnaires and tools. An external review panel, which ranged from practitioners (pediatric dentists and pediatric physicians) to experts in the area of cariology, predictive modeling, and behavioral science, were provided a copy of the questionnaire and asked to review/edit the questionnaire to ensure that the initial draft of the questionnaire was reasonable in scope and that no established risk indicator had been omitted. After receiving separate IRB approval, the draft questionnaire was tested in a panel of 25 caregivers (nearly equal numbers of English and Spanish speaking), similar to the target population (had to consent to participate and have a child between 18 and 36 months of age), to ensure that the questions that were asked were worded appropriately for nonprofessionals, to eliminate jargon, to define or eliminate confusing terminology (e.g., words such as frequent, often, etc.), to ensure use of culturally-sensitive language, to finalize the organization of the items, and to verify the consistency of the structure of similar items. In most cases, it was believed that the majority of persons to be interviewed as PCG would be the mothers, but others (e.g., grandmothers, fathers) were to be included if it was found that they were responsible for International Journal of Dentistry  : Relationships of race/ethnicity with oral bacteria transmission questionnaire responses (% responding "Yes"). Analyses were performed for each questionnaire item using multivariable logistic regression models with race/ethnicity, income, and education as predictors. Differences among race/ethnicity groups were found for "ever breast fed" (P = .0004, highest for Hispanics), "kiss child on lips" (P = .0001, lowest for Hispanics), "share plate/bowl/glass" (P = .0001, lowest for Hispanics), and "pacifier in your mouth" (P = .0156, lowest for Hispanics).
providing the largest percentage of the child's care. Based on the results of the focus group data, some changes were made in the wording of questions, some questions were eliminated and some were reordered, and the questionnaire was finalized.
The final version of the questionnaire, which included 105 items (see Appendix), was administered by study personnel to the PCG (n = 396) using a multiple choice format, with responses recorded directly into a web-based database system. The caregiver chose whether to use the English or Spanish version of the questionnaire. Topics included in the questionnaire were categorized into: demographics, access to care, possible routes for oral bacteria transmission, usual dental and medical health practices of the caregiver and the toddler, dental beliefs of the caregiver, and snacking and drinking habits of the caregiver and the toddler. In addition, a subset of the caregivers (n = 250) was invited to participate in an additional investigation of health literacy. After additional informed consent, caregivers were administered the Short Test of Functional Health Literacy in Adults (S-TOFHLA) [44], with the caregiver given the option of using either the English or Spanish version.
The associations of PCG education and household income with race/ethnicity were tested using ANOVA, and Spearman correlation coefficients were calculated to measure the association between PCG education and household income. For the analyses, education levels were collapsed into 8th grade or lower, some high school, completed high school, some college, 4-year college, and postgraduate. We analyzed each survey item individually to assess the need to modify caregiver/patient education and preventive care intervention strategies based on demographic factors. To examine the associations of individual survey items (dependent variables in separate models) with the caregiver's race/ethnicity, the caregiver's education, and the household income simultaneously (three independent variables), multivariable logistic and linear regression analyses were used for survey items with qualitative responses and quantitative responses, respectively; thus race/ethnicity comparisons are adjusted for income and education, income comparisons are adjusted for race/ethnicity and education, and education comparisons are adjusted for race/ethnicity and income. P-values presented for the race/ethnicity comparisons are for the overall tests for any effect among the three groups; individual pairwise results are presented when significant but the P-values are not provided. Odds ratios presented for education and income are for a one-level change in the response categories. A 5% significance level was used for all analyses; although a large number of tests were performed, we did not adjust for multiple testing. A less restrictive cutoff without a multiple-testing adjustment provides a larger pool of possible differences that can be targeted when revising caregiver/patient education and preventive care intervention strategies.

Results
The study enrolled 396 caregiver-toddler pairs at baseline (two additional pairs were screened but did not qualify due to the child's medical condition), which is estimated to be approximately 70% of those invited to participate. Nearly all of the primary caregivers (378) were the child's mother, Table 1: Relationships of race/ethnicity, education, and income with questionnaire responses for demographics and access to care. Association of PCG education with race/ethnicity was tested using ANOVA. Remaining analyses were performed for each questionnaire item using multivariable logistic regression models with race/ethnicity, income, and education as predictors. Race/ethnicity P-values are for the overall test of any difference among the three groups. Questionnaire item numbers are listed in the left-most column (see the appendix for questionnaire). N(%) for race/ethnicity, odds ratio (OR) for Education and Income. NH-AA = Non-Hispanic African-American, NH-W = Non-Hispanic White. with the remaining caregivers consisting of 14 fathers, 2 grandmothers, 1 aunt, and 1 other. The caregivers' ages ranged from 18 to 64 years, with an average age of 28 (SD = 6) years. The children ranged in age from 16 to 36 months, with a mean of 26 (SD = 6) months, and ages did not differ significantly by race/ethnicity, income, or education of the caregiver. 51% of the toddlers were female. One hundred seventy-five (44%) of the caregivers self-identified themselves as Non-Hispanic African-American, 141 (36%) were Non-Hispanic White, 75 (19%) were Hispanic (all races), and 5 (1%) did not fall into one of the previous three categories.
Nearly one-third of Hispanic caregivers reported difficulty understanding information they receive from physicians and dentists, while the rate was less than ten percent in Non-Hispanic African-Americans and Non-Hispanic Whites (Table 1). Concurrently, caregivers with less education also were more likely to report these difficulties. Furthermore, health literacy, collected on a subset of 250 caregivers, was not different among race/ethnicity groups but was weakly associated with education (r = 0.18, P = .02). Non-Hispanic Whites were more likely to use city water as their primary drinking water source as opposed to bottled or well water. Interestingly, drinking water source was not related to income or education in this cohort. There was a moderately high correlation between education and income (r = 0.56, P = .0001) and moderate correlations for caregiver age with education (r = 0.42, P = .0001) and income (r = 0.38, P = .0001).
Habits of the caregivers that might lead to transmission of bacteria to the toddler differed by race/ethnicity ( Figure 1), education, and income. Hispanic caregivers were less likely than Non-Hispanic African-American and Non-Hispanic White caregivers to put the toddler's pacifier in their own mouth (12% versus 37% and 31%, P = .0156), which was also associated with higher education (odds ratio 1.3, 95% CI 1.0-1.7, P = .0212) but not with income (odds ratio 1.0, 95% CI 0.8-1.1, P = .44). Tasting the child's food or drink using the same fork/spoon or glass was common in all race/ethnicity groups (approximately 70%, P = .87), but was more common with those reporting a higher income (odds ratio 1.3, 95% CI 1.1-1.4, P = .0013). Sharing food with the child using the same bowl/plate/glass and kissing the child on the lips occurred with nearly all Non-Hispanic African-American and Non-Hispanic White caregivers but was less frequent among Hispanics (P = .0001) and was more common with higher income (odds ratio 1.3, 95% CI 1.1-1.6, P = .0028). However, 87% of Hispanics ever breast-fed compared to 50% of Non-Hispanic African-Americans and 62% of Non-Hispanic Whites (P = .0004); breast-feeding was also more common with higher education (odds ratio 1.6, 95% CI 1.2-2.1, P = .0004) and higher income (odds ratio 1.1, 95% CI 1.0-1.3, P = .0458). Table 2: Relationships of race/ethnicity, education, and income with questionnaire responses for dental and medical health practices of the primary caregiver (PCG) and the toddler. Analyses were performed for each questionnaire item using multivariable logistic or linear regression models with race/ethnicity, income, and education as predictors. Race/ethnicity P-values are for the overall test of any difference among the three groups. Questionnaire item numbers are listed in the left-most column (see the appendix for questionnaire). N(%) or Mean(SD) for race/ethnicity, odds ratio (OR) or correlation (R) for education and income. NH-AA = Non-Hispanic African-American, NH-W = Non-Hispanic White. a Frequency rated on a 1-6 scale: 1 = more than once per day, 2 = once per day, 3 = several times a week, 4 = several times a month, 5 = a few times a year, and 6 = never. b Ratings on a 1-5 scale: 1 = excellent, 2 = very good, 3 = good, 4 = fair, and 5 = poor.
Although caregivers with more education more often reported that their child had a dentist (Table 2), there were no differences in whether the child had ever been to the dentist. Because the toddlers may have similar access to care as their caregivers, the questionnaire also asked about dentist and physician visits made by the caregiver. Seventy-one percent of Non-Hispanic White caregivers, 53% of Non-Hispanic African-American caregivers, and 29% of Hispanic caregivers had a dentist (Table 1), and having a dentist was also associated with higher education attainment and higher income. Approximately half of Non-Hispanic African-Americans caregivers reported going to the dentist for regular checkups, while nearly 40% of Hispanic caregivers reported never going to the dentist. Interestingly, higher income was associated with caregivers going to the dentist for checkups, while lower education but not income was associated with never going to the dentist. In addition, patterns of caregiver visits to the physician differed by race/ethnicity (Table 1) but were not as affected by income or education, where only regular visits to the physician were associated with higher income.
Hispanic caregivers reported their children's teeth were brushed less frequently than teeth of Non-Hispanic African-Americans and Non-Hispanic Whites ( Table 2). Caregivers with lower income were more likely to have problems with dry mouth when eating. Hispanic caregivers were more likely to be bothered by the appearance of their own teeth, which was not associated with education or income. Flossing was associated with more education but not with income or race/ethnicity. While there were differences among the race/ethnicity groups in how the caregivers felt about their child's and their own dental and medical health, education and income were generally not related to these ratings. Beliefs and knowledge (Figure 2) differed by race/ethnicity-adults eventually losing all their teeth (P = .0001, higher response of "false" for Non-Hispanic Whites), most children getting cavities (P = .0304, lower response of "false" for Hispanics), bad teeth being mostly inherited from parents (P = .0119,  Figure 2: Relationships of race/ethnicity with dental beliefs of caregivers. Analyses were performed for each questionnaire item using multivariable logistic regression models with race/ethnicity, income, and education as predictors. Differences among race/ethnicity groups were found for "cleaning should start" (P = .0001, earlier for Non-Hispanic Whites), "bad teeth are mostly inherited from parent" (P = .0119, lower response of "false" for Hispanics), "most children will get cavities" (P = .0304, lower response of "false" for Hispanics), and "adults lose all teeth as they get older" (P = .0001, higher response of "false" for Non-Hispanic Whites). lower response of "false" for Hispanics), and when tooth cleaning should start (P = .0001, earlier for Non-Hispanic-Whites), with also a trend for when the child's first dental visit should be (P = .06, earliest for Non-Hispanic African-Americans and latest for Non-Hispanic Whites). Belief that adults will eventually lose all their teeth was associated with less education (odds ratio 1.4, 95% CI 1.1-1.8, P = .0072) and lower income (odds ratio 1.1, 95% CI 1.0-1.3, P = .0376), and belief that most children will eventually get cavities was associated with less education (odds ratio 1.3, 95% CI 1.0-1.6, P = .0258), while none of the other beliefs/knowledge assessed were significantly associated with education or income. Hispanic toddlers were more likely drink from a bottle (29%) compared to Non-Hispanic Whites toddlers (11%) and Non-Hispanic African-American toddlers (4%), while Non-Hispanic White toddlers and Non-Hispanic African-American toddlers were not significantly different. Non-Hispanic African-American toddlers were also less likely to drink from a sippy cup (67%) compared to Non-Hispanic Whites (84%) and Hispanic (87%) toddlers, who were not significantly different from each other (Table 3). Hispanic children were most likely to receive a bottle or sippy cup at bedtime or naptime. Although Hispanic caregivers cleaned their child's teeth after removing the drink more frequently than Non-Hispanic African-Americans or Non-Hispanic Whites, cleaning the child's teeth after removing the drink was rare for all races. Less than half of Hispanic children regularly sipped on drinks between meals, while nearly all Non-Hispanic African-American and Non-Hispanic White children did. Types of snacks and drinks usually eaten/drank between meals varied considerably among race/ethnicity groups for toddlers (Table 3) and for PCGs (Table 4), while snacking and between-meals drinks were typically not associated with education or income, with a specific exception of nondiet soda being associated with less education.

Discussion
Despite a decrease in dental caries prevalence in permanent teeth for most Americans since the early 1970s, oral health disparities remain across some population groups, and dental caries is still the most prevalent chronic disease of childhood [1]. Furthermore, between 1988Furthermore, between -1994Furthermore, between and 1999Furthermore, between -2004, caries experience in primary teeth of children aged 2-5 years has significantly increased from 24% to 28%, primarily due to an increase in the percent with fillings [45]. Unfortunately, as mentioned earlier, our current understanding of caries risk and etiological factors derived from longitudinal studies in young children in the United States is limited. Available caries risk questionnaire tools are, for the most part, expert-based tools. Examples include the Caries Risk Tool of the American Academy of Pediatric Dentistry [46], the ADA's Caries Risk Tool for children Table 3: Relationships of race/ethnicity, education, and income with questionnaire responses for snacking and drinking habits of the toddler. Analyses were performed for each questionnaire item using multivariable logistic or linear regression models with race/ethnicity, income, and education as predictors. Race/ethnicity P-values are for the overall test of any difference among the three groups. Questionnaire item numbers are listed in the left-most column (see the appendix for questionnaire). N(%) or Mean(SD) for race/ethnicity, odds ratio (OR) or correlation (R) for education and income. NH-AA = Non-Hispanic African-American, NH-W = Non-Hispanic White. a Frequency rated on a 1-6 scale: 1 = more than once per day, 2 = once per day, 3 = several times a week, 4 = several times a month, 5 = a few times a year, and 6 = never. younger than 6 [47], and the Caries Management by Risk Assessment (CAMBRA) tool for children younger than 6 [48,49]. While other studies have identified caries risk factors in low-SES rural [50] and low-SES African-American [30] communities, the prevalence of the risk factors may affect both the disease prevalence and the types of interventions that may be effective in preventing and/or treating caries. Age, socioeconomic status, and race/ethnicity differences as well as in non-US populations studied previously provided individual risk factor prevalence estimates, but only indirect evaluations of the effects of the sociodemographic factors on the risk factors could be made. In the present study, multiple factors from the caries risk questionnaire within the access to care, oral bacterial transmission, dental and medical health practices of the caregiver and the toddler, and snacking and drinking habits of the caregiver and the toddler areas were directly compared and differed by race/ethnicity, income, and/or education. Having general and pediatric dentists understand that these differences exist is only a first step. The information must be incorporated in improved strategies to treat and/or prevent caries in toddlers.
With the limited sample size and single location sampled in this study, it is difficult to differentiate the effects of cultural influences, health knowledge gained through educational background, and income-based health utilization disparities on the risk factors; in other words, we were unable to look at the influence of interactions among the three factors or stratify the analyses. And while the study included three race/ethnicity groups, the single location of the study (Indiana) may not fully represent responses nationwide. A larger multisite study would be needed for increased generalizability as well as provide the sample size needed to differentiate among the cultural, income, and education influences on the risk factors. A large number of risk factors were examined, based on the extensive list of factors proposed or identified previously. Some of the risk factors differing by sociodemographic factors are likely to be false positives. Nevertheless the information from our study can provide useful risk factor prevalence data when revising caregiver/patient education and preventive care intervention strategies. Table 4: Relationships of race/ethnicity, education, and income with questionnaire responses for snacking and drinking habits of the primary caregiver (PCG). Analyses were performed for each questionnaire item (items 80 through 82-see the appendix for questionnaire) using multivariable logistic regression models with race/ethnicity, income, and education as predictors. Race/ethnicity P-values are for the overall test of any difference among the three groups. N(%) for Race/Ethnicity, Odds Ratio (OR) for education and income. NH-AA = Non-Hispanic African-American, NH-W = Non-Hispanic White. As mentioned above, our sample size was not large enough to justify a detailed examination of the 3-way interaction among race/ethnicity, income, and education to differentiate the effects of cultural influences, health knowledge gained through educational background, and income-based health utilization disparities on the risk factors. Regardless of the underlying "cause", as others have suggested based on observations in various populations [32,36], education and intervention strategies can be targeted generally to the population seen in the practice and specifically to individual patients. It is noteworthy to mention efforts in this country by medical (e.g., American Academy of Pediatrics and American Medical Association [51,52]) and dental (e.g., American Dental Association [47], American Academy of Pediatric Dentistry [46,53]) associations, among others, to stress not only the importance of a dental home early in life, but also the importance of riskbased preventive interventions and anticipatory guidance provided in a variety of settings to reach young children. In fact, a variety of programs have evolved in different places around the country. The "Into the Mouth of Babes" (IBM) program in North Carolina is one of the best examples of the effort resulting from the partnership between dentists and pediatricians to improve the oral health of children. The IMB program was initiated in 2000 and has led to a substantial increase in access to preventive dental services by enabling Medicaid children younger than 3 years of age to receive dental screening, counseling, and fluoride varnish in physicians' offices [54]. More work will certainly be needed to evaluate the acceptability and effectiveness of education and intervention strategies in targeted populations.
One problem hindering treatment and prevention of caries in high-risk children is that they may not seek care from dentists regularly, if at all. Despite the importance of establishing a "dental home" in the first year of life, most children do not receive a dental examination, nor do the parents receive needed education on oral health [55]. This is especially true for those at the highest risk. While 89% of infants and one-year-olds have been examined by a physician, only 1.5% has had a dental appointment [53]. Some of the factors identified above could be included in discussions of "healthy behaviors" with the caregivers at wellchild checkups. Patient education materials could also be developed to be made available through pediatrician and family practice offices. The results from our study may be useful to future investigators to focus the materials on factors prevalent in specific offices, such transmission of bacteria through sharing drinks or foods in higher income practices and providing drinks at bedtime or naptime in offices that have a high proportion of Hispanics.
In conclusion, significant differences were found in all areas of the questionnaire related to race/ethnicity, income, and/or education. A larger followup study may be able to explore more detailed differentiation of the effects of cultural influences, health knowledge gained through educational background, and income-based health utilization disparities on the risk factors. Patient education and preventive care intervention studies may need to be targeted based on the characteristics of the population to achieve increase effectiveness. I only take the child to the dentist if he/she has pain or a problem with his/her teeth. I take the child to the dentist regularly because he/she has problems with the teeth or gums. I take the child to the dentist for regular checkups. I do not take the child to the dentist as often as the dentist wants me to. I never take the child to the dentist.