This is a cross-sectional study, and the sample was adults over the age of 30 years participating in the National Health and Nutrition Examination Survey (NHANES) 2015–16. NHANES is designed to assess the health and nutritional status, including interviews and physical examinations [21], among adults and children in the United States. Variables were selected from the demographic, oral health and acculturation modules. The de-identified data was obtained from open access sources with no requirement for institutional board ethics approval.
The outcome of interest, use of oral health care services, was measured in terms of dental visiting and was derived from the question ‘When did you last visit a dentist?’ The variable originally included 7 categories from less than 6 months to never; in this analysis, we used visiting within the past 12 months and dichotomised the response options (‘Yes’ vs ‘No’).
Disparity variables (for inequality and inequity) were education and income.
Education was based on the highest level of school completed, according to the ‘Education Level-Adults 20+’ classification [21], and categorised as ‘less than high school’, ‘high school’, ‘some college’ and ‘college’.
The income variable was total annual household income (summed across all household members) and categorised as ‘< $20,000’, ‘≥ $20,000 to < $45,000’, ‘≥ $45,000 to < $100,000’ and ‘≥ $100,000’. There were 7.5% missing data for ‘Income’ which could not be assumed to be missing at random, therefore these missing data were not included in the analyses.
Sociodemographic characteristics included gender, race/ethnicity, age, main language spoken at home, country of birth and citizenship.
The original race/ethnicity variable included the following categories: Mexican–American (MA), other Hispanic, non-Hispanic White (NHW), non-Hispanic Black, non-Hispanic Asian (NHA), and another race including multiracial. For the present analysis, MA and other Hispanic were combined into one group and NHA and other or multiracial in another group.
Age was categorised into ‘≥ 30 to < 40 years’, ‘≥ 40 to < 50 years’, ‘≥ 50 to < 60 years’, ‘≥ 60 to < 70 years’ and ‘≥ 70 years’ groups. The target population was adults over the age of 30 because at this age there is a greater likelihood of socio-economic independence and stability (e.g., not dependent on parental income and having completed schooling).
Main language merged two variables: (1) Language at interview which was either Spanish or English and (2) whether an interpreter was required during the interview; it was categorised as ‘English’ or ‘Other’.
Country of birth defined whether the person was born in the US or born in other countries.
Citizenship status indicated whether the person was a US citizen or not.
Need for oral health care was represented by self-rated oral health and impaired oral health.
The NHANES question for self-rated oral health asked participants for an overall rating of the health of their teeth and gums. Self-rated oral health was analysed including the following three categories (1) excellent or very good, (2) good, and (3) fair or poor.
Impaired oral health combined NHANES questions for pain, difficulty at work or school and embarrassment due to problems in teeth, mouth or dentures. The original NHANES questions were related to the frequency of the events (very often, fairly often, occasionally, hardly ever or never). This analysis used a binary variable, with impairment defined as one or more responses of ‘very often’, ‘fairly often’ or ‘occasionally’.
Data analysis
For all analyses, the outcome was defined as NOT visiting the dentist in the last 12 months.
Inequality in dental visits
To characterize existing inequality in the use of oral health care services, we examined bivariate relationships, used indices of inequality and depicted inequalities through concentration curves (CC). The measures of inequality included absolute and relative estimates using the health disparity calculator from SEER—The Surveillance Epidemiology and End Results Program of the US National Cancer Institute [22]. The range difference and ratio were measured between extreme groups in the social gradient. The absolute concentration index (ACI) and relative concentration index (RCI) were used to measure the extent to which the outcome (i.e., no dental visiting in the past 12 months) is concentrated in a particular group. The ACI is calculated by multiplying the RCI by the mean level of the outcome in the population (μ): ACI = μRCI. Thus, if there is no inequality (i.e., if RCI = 0), the ACI is 0 [23]. The slope index of inequality (SII) and relative index of inequality (RII) are regression-based and assume the relationship between ‘no dental visiting in the past 12 months’ and the disparity variables (i.e., education and income) is in a linear manner. The SII represents the absolute effect on ‘no dental visiting in the past 12 months’ of moving from the lowest socio-economic level to the highest. The SII can be calculated by Weighted Least Squares to allow heteroscedasticity of the error terms [24]. The RII is obtained by dividing the SII by the mean level of ‘no dental visiting in the past 12 months’ in the population [23]. The CC provides a graphical view of how dental visiting varied across education and income [13]. The CC plots the cumulative percentage of ‘no dental visits in the last 12 months’ against the cumulative percentage of the population ranked from the lowest/poorest to the highest/richest education and income. If everyone, irrespective of their educational level and income, had the same value of dental visiting, the concentration curve would overlap with the line of equality; that is, it would be a 45-degree line. If the curve is above the line of equality, it means a higher concentration of ‘‘no dental visiting in the past 12 months’ among the lower educational attainment and income groups [13].
Inequity in dental visits
Assessing inequity in dental vising requires comparing utilization at various education and income levels for the same level of need; we measured ‘actual’, ‘need-expected (need-predicted)’ and ‘need-standardized’ dental visiting. Actual use was a factual depiction of the extent of inequality in the distribution of dental visiting. Need-expected dental visiting adjusted actual use by self-rated oral health and impaired oral health (i.e., need). The gap between ‘‘actual’ and ‘‘need-expected’ dental visiting reflects different use of oral health care services. Need-standardized use quantified the extent of inequity through the difference between actual and expected use, plus the mean of expected use in the study population [13]. As dental visiting is a binary response, a non-linear model (i.e., Probit regression model) was used with the probability of no dental visiting as the dependent variables to indirectly standardize the dental service utilization [13].
Education and income were strongly associated; therefore, separate models were run for each and the prediction values did not control for each other when presented across levels of education or income. Both models were adjusted for age, sex, race/ethnicity and main language (‘Country of birth’, ‘US Citizen’ and ‘Main language’ were strongly associated, so only the latter was included in the models). ‘Self-rated oral health’ and ‘Impaired oral health’ were strongly associated and therefore, only impaired oral health was used in the models to measure need for oral health care. The assumption in this method is that vertical equity is satisfied—we assume differences in the use of health care among individuals with different needs are appropriate. Any residual inequality after standardization by need for care is interpreted as horizontal inequity.
Weights were used to account for the sampling methodology of the survey; SAS version 9.4 (SAS Institute Inc., Cary, NC, USA) and STATA version 16 (Corporation, College Station, TX, USA) were used for the analyses.