Predicting Caries in Medical Settings: Risk Factors in Diverse Infant Groups

Abstract

Expanded partnership with the medical community is a promising strategy for reducing disparities in dental caries among young children. However, no validated caries risk instrument exists for use in primary health care settings. To help resolve this gap, a 52-item caries risk questionnaire was developed and targeted to primary caregivers (PCGs) to test in a 3-y prospective study. To begin to understand the validity of the questionnaire items, the purpose of this study was to compare responses to the questionnaire based on key demographic characteristics known to be associated with disparities in caries experience (e.g., race/ethnicity and insurance status). A total of 1,323 one-year-old children were recruited primarily through 3 medical research networks. Baseline questionnaire responses were analyzed via logistic regression. The sample was 49% female. Its racial/ethnic makeup was as follows: 13% Hispanic, 37% White, 37% Black, and 13% other or multiracial. Sixty-one percent were enrolled in Medicaid, and 95% resided in urban communities. Mothers represented 94% of PCGs. There were significant differences (P < 0.05) in baseline responses based on Medicaid status and race/ethnicity. As compared with those not enrolled in Medicaid, children in the Medicaid group were significantly more likely (after adjusting for race/ethnicity) to 1) go to sleep while nursing or drinking something other than water, 2) eat sugary snacks between meals, 3) consume sugary drinks between meals, 4) receive topical fluoride from a health professional, 5) visit the dentist, and 6) not have an employed adult in the household. PCGs of children enrolled in Medicaid were significantly more likely to be the mother, have bleeding gums, eat sugary snacks between meals, consume sugary drinks between meals, eat or drink something other than water before going to bed, and not get regular dental checkups. In conclusion, there are significant differences in caries risk questionnaire responses based on Medicaid status and race/ethnicity that provide construct and criterion validity to the developed caries risk tool (ClinicalTrials.gov NCT01707797).

Introduction

Targeted health care delivery to those at greatest risk is an important strategy to address health care costs and resource constraints and can improve overall health. The management of dental caries is a prime target for this approach, especially given the marked disparity in distribution of dental caries among U.S. children (Capurro et al. 2015; Dye et al. 2015) and the fact that dental caries is one of the most common chronic diseases of children (Kassebaum et al. 2015). In addition, young children of minority groups and/or low socioeconomic status often have limited or no access to regular dental services (U.S. Department of Health and Human Services 2016). Primary care physicians have been called on to assess and prevent oral diseases (American Academy of Pediatrics Section of Pediatric Dentistry and Oral Health 2008; Moyer 2014). However, currently there is no validated and easily scored tool to accurately assess risk and triage those infants or toddlers at high caries risk.

The available caries risk assessment tools are primarily expert opinion–based tools (Tellez et al. 2013; Fontana 2015), are not validated on the basis of prospective longitudinal studies of U.S. infants or toddlers, and are too long and cumbersome to be useful in the quick-paced medical setting—for example, the Caries Risk Tool (American Academy of Pediatric Dentistry 2014), the Caries Risk Assessment Form (American Dental Association 2008), the Caries Management by Risk Assessment (Ramos-Gomez et al. 2007), and the Oral Health Risk Assessment Tool (American Academy of Pediatrics 2011). An interdisciplinary project was funded to develop a practical and easily scored short caries risk tool to be constructed according to responses to a caries risk questionnaire for use in primary medical settings to identify U.S. infants/toddlers at high risk for developing caries.

The objective of this study is to describe the distribution of baseline responses to the developed caries risk questionnaire by Medicaid status and race/ethnicity to evaluate the construct and criterion validity of a caries risk instrument. Construct validity refers to the degree to which an instrument measures the construct that it was designed to measure (Key 1997). In the case of the risk questionnaire, it refers to the measurement of theoretical concepts associated with dental caries. Criterion-related validity refers to detecting the presence of ≥1 criteria considered to represent traits or constructs of interest (e.g., sugar consumption is associated with caries experience). To test this, we compared baseline responses to the questionnaire among groups known to experience significant disparities in caries. It was hypothesized that positive responses to questions associated with known risk factors for caries development would be more common in child population groups that experience higher rates of caries in the United States (e.g., Medicaid-enrolled racial and ethnic minorities).

Materials and Methods

The overall approach for the parent caries risk assessment project included a 3-y multisite prospective study managed and coordinated at the University of Michigan (Ann Arbor, MI). Three well-established primary care medical research networks—Pediatric Research Network (Indianapolis, IN), Iowa Research Network (Iowa City, IA), and Duke University’s Primary Care Practice-Based Research Network (Durham, NC)—enrolled 1,323 children (mean ± SD age at baseline visit: 12 ± 3 mo old). The study population was stratified by Medicaid status and represented diverse racial/ethnic groups and urban/rural residence. Children were identified for possible recruitment through well-child doctor appointments, neighborhood centers, day cares, and advertising (Daly, Levy, Xu, Levy, and Fontana 2016). The 3 research teams additionally facilitated 2 follow-up visits (child’s age, 2.5 and 4 y). Institutional Review Board approval was obtained at all 4 institutions. The sample size of children initially recruited for this study was based on the final number of subjects needed at the age 4 follow-up examination for estimation of sensitivity and specificity of the caries risk tool.

Each child was paired with his or her primary caregiver (PCG). The PCG was defined as the individual primarily responsible for the child’s housing, health, and safety and, for recruitment purposes, was limited to PCGs who were also the parent or legal guardian. Inclusion criteria included the following: the PCG 1) was at least 18 y old or an emancipated minor, 2) was able to read and speak English and/or Spanish, 3) provided written informed consent for the PCG and the child, 4) completed the risk questionnaire, 5) allowed a dental examination of the child, and 6) was available for 2 follow-up study visits and responded to multiple intermediate contacts between examination visits. The child had to be 12 ± 3 mo old and generally healthy. Exclusion criteria for the child included being in foster care, requiring antibiotic and/or sedative premedication prior to a dental examination, having a history of uncontrolled epilepsy, undergoing current cancer therapy, or having an unrepaired congenital heart defect.

During the baseline clinical visit, the PCG completed a 52-item paper version of the caries risk questionnaire, which was designed to be self-administered with no need for special instructions or training. PCGs were asked to circle the answer to each question. The study team was trained on how to 1) respond if the PCG had questions and 2) check the returned questionnaire for completeness and ensure that unanswered questions were not accidental.

The caries risk questionnaire initially included 143 items abstracted from existing caries risk instruments and relevant sociodemographic and clinical risk variables. These items were pilot tested in a 12-mo trial with a panel of 399 toddlers from an underserved urban U.S. population to identify risk factors to predict caries progression in primary health care settings (Eckert et al. 2010; Fontana et al. 2011). The pilot study defined “high risk” to include only children who developed and/or had lesions progress to cavitation (International Caries Detection and Assessment System score ≥3) or had fillings placed during the test period. The questionnaire was then refined for use in the current study by the following steps:

• 1) After pilot testing (Fontana et al. 2011), the data were analyzed with classical test theory (Crocker and Algina 1986) and Rasch modeling (Masters and Wright 1984), followed by predictive modeling to assess how well the items worked together to produce a valid caries risk assessment. Redundant, invariant, and independent items (item total correlations <0.2) were eliminated, resulting in a 52-item pool.

• 2) To refine the questionnaire and narrow the relevant underlying constructs for each question, a conceptual framework of dental caries was formulated, including social and biological disease determinants and specific relevant indicators of the oral environment (e.g., diet), oral flora, personal hygiene, and treatment factors. Analyses of these initial items yielded modest sensitivity (0.8) and specificity (0.6) values for 2 administrations of the questionnaire (baseline and follow-up; Fontana et al. 2011).

• 3) Because some items were initially worded in a way that may interfere with the ability to correctly assess the risk factor sufficiently, items were reworded, and a “think-aloud” session (Collins 2003) was conducted with 5 respondents matched according to their relevance to the target population (i.e., PCGs of children 9 to 24 mo). During the think-aloud, item content and response options were assessed regarding whether they produced the cognitive process and responses consistent with expectations.

• 4) Based on the feedback obtained, the item content and response options of all psychometrically validated items were edited to ensure that all respondents, regardless of background, could answer the questions in accord with expectations. The final instrument contained 52 nonredundant and clear items (i.e., face valid for experts) to be assessed in the predictive risk assessment study (Appendix).

• 5) Finally, after the study started and the questionnaire was administered to 125 PCGs, 8 items were reworded for clarity based on requests by the PCGs or staff personnel.

• 6) English and Spanish versions of the questionnaire were developed.

This article presents analyses of baseline questionnaire responses by Medicaid status and race/ethnicity (per STROBE guidelines). The self-reported categories of race (White, Black, Asian, Native Hawaiian or Pacific Islander, American Indian or Alaskan Native, other) and ethnicity (Spanish, Hispanic, or Latino: yes or no) for child and adult were collapsed into 4 categories for the baseline analyses: Hispanic (if of Hispanic ethnicity, regardless of race), White (if non-Hispanic and White), Black (if non-Hispanic and Black), and other (if non-Hispanic and a race other than Black or White). Analyses were performed with multivariable logistic regression, including Medicaid status and race/ethnicity in the models simultaneously. A 5% significance level was used for all tests; all questionnaire items were evaluated individually with no adjustment of significance levels for multiple testing.

Results

A total of 2,173 individuals were approached to achieve the enrollment goal of 1,323 children and their PCGs who completed the baseline visit and signed the informed consent during a 16-mo period. Among the PCGs, 1,246 (94%) were the mothers of the children. Mothers were significantly (P < 0.05) more likely to be the PCGs in the Medicaid group (96%) than non-Medicaid group (92%), with no difference by race/ethnicity. The infants (51% males, 49% females) were 11.4 ± 2.0 mo old at baseline, while the PCGs were 28.7 ± 6.0 y old. PCGs in the non-Medicaid group were significantly older (31.9 ± 4.8 y) than in the Medicaid group (26.6 ± 5.6 y). PCGs who were White reported being significantly older (30.9 ± 5.3 y) than the group who was Black (26.9 ± 5.8 y). Sixty-one percent of the children were enrolled in Medicaid, 37% were non-Medicaid; and 2% did not know their Medicaid status. The ethnic/racial distribution of children was as follows (rounded percentages): Hispanic, 13%; Black, 37%; White, 37%; and multiracial or other race, 13%. The mean completion time for the questionnaire was 8.3 ± 6.4 min and was significantly less for the non-Medicaid group (7.4 ± 7.7 min) than the Medicaid-enrolled group (8.8 ± 5.8 min).

There were differences in the questionnaire responses by Medicaid status and race/ethnicity (Table). After adjustment for race/ethnicity, children in the Medicaid group were significantly (P < 0.05) more likely than those in the non-Medicaid group to 1) go to sleep while nursing or drinking something other than water, 2) eat sugary snacks and consume sugary drinks between meals, 3) receive topical fluoride from a health professional, 4) go to visit the dentist, and 5) not have an employed adult in the household. PCGs of children enrolled in Medicaid were significantly more likely to 1) be the mother; 2) have bleeding gums; 3) have had cavities, fillings, or teeth pulled in the past 2 y; 4) eat sugary snacks and consume sugary drinks between meals; 5) eat or drink something other than water before going to bed; 6) more frequently kiss their child on the mouth; 7) not get regular dental checkups; and 8) less frequently state that they do an excellent or very good job at taking care of their child’s medical health.

Table.

Responses to the Caries Risk Questionnaire by Medicaid Status and Race and Ethnicity.

Questionnaire Item: Response	All	Hispanic	Black	White	Medicaid	No Medicaid
Does your child have any teeth?
Yes	93	92	90	95	91	95
No	7	8	10	5	9	5
Does your child have any cavities or fillings?
Yes	1	1	2	1	2	1
No	82	73	82	86	80	86
Don’t know	16	26	16	13	18	13
Did your child’s doctor or dentist prescribe fluoride drops or tablets?
Yes	1	1 a,b	2 a	<1 b	2	1
No	95	96	94	97	95	95
Don’t know	4	4	4	3	3	4
Does your child wear any oral appliances, such as space maintainers?
Yes	<1	0	<1	0	<1	0
No	99	99	99	100	99	100
Don’t know	<1	1	1	<1	1	<1
Does your child receive topical fluoride from a health professional (doctor, dentist, nurse, hygienist, etc.)?
Yes	11	10	16	7	17 a	2 b
No	88	89	83	91	82	97
Don’t know	1	1	1	1	1	1
How often does an adult brush your child’s teeth?
Daily	56	54	56	56	55	57
Weekly	16	15	14	18	14	18
Monthly	2	1	1	2	1	2
Never	27	29	30	23	30	23
How often are your child’s teeth brushed with toothpaste?
Daily	31	37	30	30	31	32
Weekly	10	10	10	10	10	10
Monthly	1	0	1	1	1	1
Never	58	52	59	59	59	56
How often are your child’s teeth brushed with nonfluoride toothpaste?
Daily	22	28	21	22	22	23
Weekly	8	10	8	7	8	8
Monthly	1	0	1	2	1	1
Never	69	61	71	70	70	68
How often does your child share a toothbrush with another person?
Daily	1	1	1	<1	1	<1
Weekly	2	1	<1	3	1	3
Monthly	<1	0	<1	1	0	1
Never	97	98	99	95	98	96
How often do you check your child’s teeth for anything unusual?
Daily	44	48 a	55 a	31 b	51 a	33 b
Weekly	33	27	27	41	30	39
Monthly	7	8	4	10	5	12
Never	16	17	14	18	15	17
When brushing, how often do your child’s gums bleed?
Daily	1	2	1	1	1	<1
Weekly	<1	2	<1	<1	1	0
Monthly	<1	0	<1	1	<1	1
Never	98	96	98	98	98	99
How often do you clean inside your child’s mouth and/or gums?
Daily	56	51	61	54	59	53
Weekly	18	18	15	20	16	20
Monthly	3	4	2	4	3	4
Never	23	26	22	22	22	23
Does your child usually (throughout the day) drink from a bottle or sippy cup?
Yes	95	92 b	98 a	92 b	96	93
No	5	8	2	8	4	7
How often does your child go to sleep while nursing or while drinking something other than water from a bottle/sippy cup?
Daily	47	69 a	54 b	31 c	57 a	30 b
Weekly	11	7	13	11	12	10
Monthly	2	3	1	3	1	3
Never	40	22	32	55	30	57
How often does your child eat or drink anything other than plain water before going to bed (and after you have brushed his or her teeth, if teeth are brushed)?
Daily	64	71	65	59	66	60
Weekly	12	9	14	12	13	11
Monthly	2	1	2	3	2	3
Never	22	20	19	26	19	26
How often does your child typically drink tap water—including filtered water from the refrigerator?
Daily	57	48 b	52 b	66 a	52	66
Weekly	13	13	14	13	15	11
Monthly	3	1	4	3	3	2
Never	27	39	30	19	30	21
How often do you give your child sugary snacks such as raisins, candy, cookies, cakes, or cereal between meals?
≥3/d	4	6 b	6 a	2 c	6 a	2 b
1 to 2/d	41	47	45	35	47	33
Weekly	24	18	24	26	23	26
Monthly	7	2	6	9	5	9
Never	23	25	18	28	19	30
How often do you give your child sugary drinks such as regular soda, sweet tea, chocolate milk, strawberry milk, or fruit juice between meals?
≥3/d	4	6 a	6 a	1 b	5 a	1 b
1 to 2/d	26	35	38	10	35	10
Weekly	18	16	21	16	21	14
Monthly	5	7	4	5	4	5
Never	48	36	32	68	34	69
How often do you clean your child’s pacifier with juice, soda, honey, or sweet drink?
Daily	2	2	3	0	3	0
Weekly	2	3	4	<1	3	<1
Monthly	1	2	2	<1	2	<1
Never	54	46	53	57	51	59
Don’t use pacifier	41	47	38	42	41	41
How often do you clean your child’s pacifier by putting it in your mouth?
Daily	15	18 b	20 a	8 c	19	8
Weekly	6	1	6	7	5	7
Monthly	2	2	2	2	2	3
Never	34	29	30	40	30	39
Don’t use pacifier	44	50	42	43	45	43
How often do you share/taste food with your child using the same spoon, fork, glass, or other utensil?
Daily	46	41 b	59 a	36 b	53	36
Weekly	21	14	13	33	15	31
Monthly	4	3	2	7	3	7
Never	28	42	26	25	29	26
How often do you kiss your child on the mouth?
Daily	61	44 c	71 a	57 b	65 a	55 b
Weekly	12	9	10	15	9	16
Monthly	3	2	2	4	2	5
Never	24	45	17	24	24	24
How often do you take your child to the dentist?
Twice yearly	8	8	8	8	9 a	7 b
Yearly	5	4	6	4	6	3
Only when in pain	1	0	1	<1	1	<1
Never	86	88	85	87	84	90
Is it very difficult to get your child to the doctor or the dentist?
Yes	3	5	4	1	4	1
No	97	95	96	99	96	99
Is your child’s care covered by Medicaid or state insurance? (Medicaid is the U.S. health program for certain people and families with low incomes and resources)
Yes	61	80	91	25	100	0
No	39	20	9	75	0	100
Is your child covered by additional health insurance?
Yes	45	23 b	23 a,b	76 a	15 b	94 a
No	54	77	76	24	85	6
Don’t know	1	0	1	<1	1	<1
Is your child covered by additional dental insurance?
Yes	33	18 b	20 a	52 a,b	13 b	64 a
No	64	81	76	45	83	32
Don’t know	3	1	4	3	3	4
Does your child participate in public assistance programs?
Yes	59	77 a	87 a	23 b	90 a	10 b
No	41	23	13	76	10	90
Don’t know	<1	1	<1	<1	<1	<1
Was your child born >3 wk (premature) before the expected delivery date?
Yes	15	13	17	13	16	12
No	85	87	82	86	83	87
Don’t know	1	0	1	1	1	<1
Was your child delivered by C-section?
Yes	33	29	34	33	32	33
No	67	71	66	67	68	67
Don’t know	<1	0	<1	0	0	<1
Do you have any natural teeth?
Yes	99	96 b	99 a	99 a,b	98	99
No	1	4	1	1	2	1
Have you had cavities, fillings, and/or teeth pulled in the last 2 y?
Yes	53	51 b	63 a	44 b	59 a	44 b
No	47	49	37	56	41	56
How often do your gums bleed when you brush?
Daily	10	13	13	6	14a	4b
Weekly	13	13	15	11	15	10
Monthly	22	24	18	25	20	24
Never	55	49	55	58	51	63
How often do you brush your teeth?
Daily	98	96	97	99	96	100
Weekly	2	2	2	1	2	<1
Monthly	<1	1	1	0	1	0
Never	1	2	<1	<1	1	0
How often do you use toothpaste when you brush?
Daily	98	97	98	99	97	100
Weekly	1	1	1	1	1	<1
Monthly	<1	0	1	<1	1	0
Never	1	2	<1	<1	1	0
How often do you eat sugary snacks such as raisins, candy, cookies, cakes, or cereal bars between meals?
≥3/d	14	13 b	24 a	4 c	20 a	4 b
1 to 2/d	46	53	45	46	47	45
Weekly	31	25	22	43	24	43
Monthly	6	6	7	6	6	6
Never	2	2	2	2	2	2
How often do you drink sugary drinks such as regular soda, sweet tea, chocolate milk, strawberry milk, sports drinks, or fruit juice between meals?
≥3/d	22	21 b	37 a	7 c	33 a	5 b
1 to 2/d	37	46	39	32	40	32
Weekly	22	22	17	27	18	28
Monthly	9	6	4	15	5	15
Never	10	5	3	19	4	20
How often do you eat or drink anything other than plain water before going to bed (and after brushing your teeth, if teeth are brushed)?
Daily	40	42 b	64 a	14 c	56 a	14 b
Weekly	17	15	18	16	18	15
Monthly	6	5	5	9	4	10
Never	37	37	13	62	22	62
How often do you see your health care provider for regular check-ups?
Twice per year	34	36 b	49 a	18 c	44 a	17 b
Yearly	50	42	39	65	38	70
Every other year	8	12	6	10	9	8
Never	8	11	7	8	10	5
How often do you get dental check-ups?
Twice per year	44	36 b	35 a	55 a,b	31 b	64 a
Yearly	27	23	32	22	32	18
Every other year	17	17	19	15	20	13
Never	12	23	14	7	17	5
Do you have health insurance?
Yes	84	63 b	82 a	93 a	75 b	97 a
No	16	37	18	7	25	3
Do you have dental insurance?
Yes	70	49 b	69 a	79 a	62 b	83 a
No	30	51	31	21	38	17
Do you primarily speak a language other than English at home?
Yes	18	65 a	14 b	5 c	23	10
No	82	35	86	95	77	90
Is an adult in the child’s household employed?
Yes	82	88 b	65 c	96 a	71 b	99 a
No	18	12	35	4	29	1
I do a/an —— job taking care of the child’s teeth and/or gums (past behavior)
Excellent	22	17 b	34 a	11 b	28	13
Very good	31	27	31	32	30	32
Good	34	39	25	41	30	39
Fair	11	11	9	13	10	13
Poor	3	6	1	3	2	4
I do a/an —— job taking care of the child’s medical health (past behavior)
Excellent	69	63 b	75 a	64 a,b	71 b	64 a
Very good	26	26	21	32	22	33
Good	5	10	4	4	6	3
Fair	<1	1	1	0	1	0
Caregiver relation to child
Father	5	6	4	6	3a	8b
Grandmother	<1	0	<1	<1	<1	<1
Legal guardian	<1	0	<1	<1	<1	0
Mother	94	94	96	93	96	92
Urban/rural zip code
Rural	5	5 b	2 a	9 b	4	8
Urban	95	95	98	91	96	92
Household income
<$5,000	16	15	29	3	26	<1
$5,000 to $9,999	7	6	13	2	12	<1
$10,000 to $19,999	9	14	12	4	15	<1
$20,000 to $29,999	10	18	11	7	15	4
$30,000 to $39,999	6	6	5	6	7	4
$40,000 to $49,999	6	6	4	8	4	10
$50,000 to $79,999	13	9	5	23	4	28
$80,000 to $99,999	7	2	1	16	<1	19
≥$100,000	12	3	2	26	<1	31
Don’t know	12	20	17	4	17	3

Values are presented as percentages. Significant differences between race and ethnicity groups (P < 0.05) and between Medicaid-enrolled groups are indicated by superscript letters (bolded), with those having different letters being significantly different from each other (e.g., a vs. b).

Regarding differences based on race/ethnicity (Table), children who were Black were significantly (P < 0.05) more likely than children who were White to receive prescription fluoride drops or tablets. They were also significantly more likely than children who were Hispanic to be enrolled in Medicaid, and their PCGs were significantly more likely to have their own teeth, get dental checkups, and state that they do an excellent job at taking care of their children’s medical health. Children who were Black were significantly more likely than children who were White or Hispanic to drink from a bottle or sippy cup throughout the day, have their PCGs share or taste food with them using the same utensil, and live in an urban community, and their PCGs were significantly more likely to have had cavities, fillings, or teeth pulled in the last 2 y and state that they do an excellent job at taking care of their child’s teeth and/or gums.

PCGs who were Black were significantly more likely than PCGs who were Hispanic (and those who were Hispanic significantly, more likely than those who were White) to frequently clean their child’s pacifier by putting it in their mouth, frequently eat sugary snacks and consume sugary drinks between meals, eat or drink something other than water before going to bed, and go to their health care provider for regular checkups. PCGs who were White were significantly less likely than PCGs who were Black or Hispanic to check their child’s mouth for anything unusual, frequently give their child a sugary drink between meals, and have their children participate in public assistance programs. PCGs who were White were significantly less likely than PCGs who were Hispanic (and those who were Hispanic, less likely than those who were Black) to frequently kiss their child on the mouth and give their children sugary snacks in between meals. PCGs who were Hispanic were significantly more likely than those who were Black (and those who were Black, significantly more likely than those who were White) to speak a language other than English at home and frequently put the child to sleep after drinking something other than water. PCGs who were Hispanic were significantly less likely than those who were Black or White to have health and dental insurance and have children covered by any insurance other than Medicaid. Finally, children who were Hispanic or Black were significantly less likely to typically drink tap water than children who were White.

Discussion

The purpose of this article was to describe the distribution of baseline responses to the developed caries risk questionnaire by Medicaid status and race/ethnicity to evaluate the construct and criterion validity of a caries risk instrument. To test this, we compared baseline responses to the questionnaire among groups known to experience significant disparities in caries. From birth to age 19 y, treatment of caries is ranked as the fifth-highest condition of health care spending in the United States, surpassed only by expenses associated with well newborn, attention-deficit/hyperactivity disorder, other dental (e.g., orthodontics), and asthma (Bui et al. 2017). In fact, dental caries is one of the most prevalent chronic diseases among children (Kassebaum et al. 2015) and one of the most common unmet health care needs of poor children. Untreated cavities result in pain, loss of tooth structure, and infection of peridental tissues, with lasting effects on function, growth, development, and quality of life. According to the most recent dental caries prevalence data from the National Health and Nutrition Examination Survey, disparities among children continue to persist, and caries experience more than doubles from 23% among children aged 2 to 5 y to 56% among those aged 6 to 8 y (Dye et al. 2015). For children aged 2 to 8 y, Hispanic (46%) and non-Hispanic Black (44%) children are substantially more likely to experience dental caries as compared with non-Hispanic White children (31%). Our findings demonstrate significant differences in baseline caries risk questionnaire responses, for both the child and the PCG, based on Medicaid status and race/ethnicity. Significantly higher caries risk-associated responses were noted in population groups that have higher prevalence of caries (Dye et al. 2015), supporting the construct validity of the risk tool. Significantly more Medicaid than non-Medicaid enrollees reported cariogenic dietary behaviors, such as putting the child to sleep with anything other than water, the child consuming frequent sugary snacks in between meals, and the PCG consuming sugary snacks or drinks in between meals and prior to going to bed. Significantly more frequent responses to similar risk-associated dietary behaviors (e.g., providing sugary snacks or drinks in between meals, putting the child to bed with anything but water, not drinking tap water) were seen among children who were Black or Hispanic as compared with those who were White. In addition, Medicaid PCGs rated significantly less frequently that they did an excellent/very good job of taking care of their child’s medical health. The importance of this remains unclear; however, further analyses of the data concluded that 1) two-thirds of the PCGs perceived that they provided better care of the infants’ medical health than their oral health and 2) PCGs who thought that they provided good oral care were more likely to have better personal dental health behaviors (Daly, Levy, Xu, Jackson, et al. 2016). Significant differences in responses to other questions support expected outcomes for different population groups (e.g., it was expected that the Medicaid group would respond more frequently that the child participated in public assistance programs and that a language other than English would be spoken more frequently at home for the Hispanic group).

The questionnaire developed was based on a conceptual framework of dental caries, including social and biological determinants proximal and distal to the biofilm-tooth interface, and it specified relevant risk indicators associated with the oral environment, diet, oral flora acquisition, personal hygiene, and access-to-care/population-level factors (e.g., education, socioeconomic status; Fisher-Owens et al. 2007; Kim Seow 2012; Lee and Divaris 2014). In fact, the list of variables that may directly or indirectly influence caries risk is long, especially for young children (Fontana 2015) and includes the following: clinical and biological factors (e.g., caries experience of child and caregiver, plaque/microbiology, saliva, tooth developmental defects, medical factors, genetics); environmental factors (e.g., exposure to fluoride); and behavioral, psychosocial, and sociodemographic factors (e.g., diet, oral hygiene habits, age, parenting styles, child temperament, caregiver’s education level, socioeconomic status, insurance status, access to dental care). In addition, some of these risk factors not only influence dental caries but have much broader impacts on general health. For example, diet is one of the common risk factors, with a role in dental caries, obesity, diabetes, heart disease, stroke, and cancer (Scottish Intercollegiate Guidelines Network 2014). The role of diet is so prominent in caries experience (Pitts et al. 2017) that current guidelines call for limiting free sugars intake to <10% of total energy intake to minimize the risk of dental caries throughout life (Moynihan and Kelly 2014), among other individual and public health approaches (World Health Organization 2017).

In addition, because of the multiple influences at the individual, family, and community levels in early caries development and risk (Fisher-Owens et al. 2007), parental factors were the focus of extensive research and review (Fontana 2015) and are part of the current tested risk tool. For example, a review by the Scottish Intercollegiate Guidelines Network (2014) concluded that parental deprivation is a risk indicator for caries development in children. Maternal oral health status (Dye et al. 2011) and maternal weight and intake of sugar and fat during pregnancy (Wigen and Wang 2011) were also associated with and/or found to be strong predictors of caries among children. Compared with children delivered by C-section, vaginally born Thai children experienced increased caries prevalence (Pattanaporn et al. 2013).

The current risk assessment tool offers several advantages over previous tools. First, the item pool was developed on the basis of the functional and developmental factors that lead to dental caries. The logic underlying this risk assessment model provides both a structure and a causal framework for future intervention studies. Second, these factors provide differential risk assessment. Oral flora acquisition and prevalence, for example, represent a necessary condition, whereas other factors—oral environment, personal hygiene, and treatment—support or inhibit the metabolism and proportions of flora that lead to dental caries. Future work may identify risk profiles that lead to differential prevention practices. Third, by using the developmental and etiologic pathways (e.g., diet, oral hygiene), it was possible to identify representative questions that most likely characterize those risk factors. A few behaviors for each pathway were chosen to ensure sufficient coverage, but they also favored brevity for a less burdensome tool. Future developers may alter the existing items to be more representative of behaviors as customary child-rearing practices change. These 3 advancements enable future prospective risk assessment studies to more readily identify the behaviors most likely to contribute to dental caries development and provide clinicians and patients with more effective prevention strategies.

In conclusion, there are significant differences in baseline caries risk questionnaire responses based on Medicaid status and race/ethnicity that match higher frequency of cariogenic behaviors in population subgroups that traditionally have higher caries prevalence rates in the United States, thus providing construct and criterion-related validity to the developed caries risk tool. The population is being followed over time to examine the ability of items in the instrument to predict development of dental caries in primary health care settings so that selected items can be identified and weighted to develop a shorter caries risk tool for implementation in practice. We will then also compare predictive caries models for different population groups (e.g., Medicaid and non-Medicaid, racial and ethnic) to evaluate whether the groups can be served by a single model, which would facilitate implementation, or whether they should have different predictive models, achieved by targeting a variety of population subgroups. Whether a single risk tool for different groups is justified or not, knowledge of the prevalence of risk factors among subgroups may lead to different intervention strategies.

Author Contributions

M. Fontana, G.J. Eckert, contributed to conception, design, data acquisition, analysis, and interpretation, drafted and critically revised the manuscript; M.A. Keels, R. Jackson, B.P. Katz, A.R. Kemper, B.T. Levy, S.M. Levy, contributed to conception, design, data acquisition, analysis, and interpretation, critically revised the manuscript; E. Yanca, contributed to design, data acquisition, analysis, and interpretation, critically revised the manuscript; S. Kelly, J.M. Daly, B. Patterson, contributed to design, data acquisition, and analysis, critically revised the manuscript; P. McKnight, contributed to design, data analysis, and interpretation, drafted and critically revised the manuscript. All authors gave final approval and agree to be accountable for all aspects of the work.