Assessing inequities in unmet oral care needs among adults in Sweden: An intersectional approach

Objectives: The goal of the Swedish oral healthcare system is to achieve good oral health and equitable access to care for the entire population. However, considerable inequities in oral health and care are evident and occur across a range of social dimensions. This study uses an intersectional approach to examine complex inequities in unmet oral care needs among adults in Sweden over the period 2004– 2021. Methods: Data were obtained from 14 Health on Equal Terms surveys conducted during 2004– 2021

[5][6][7] In Sweden, oral care is organized by 21 administrative regions with patients freely choosing among public or private care provider. 6e oral care system has historically undergone several reforms to make care affordable for all.However, the ambition for an equitable system remains ambiguous.
On the one hand, the governing Dental Care Act explicitly emphasizes its goal to achieve good oral health and care provision on equitable terms for the entire population. 8This is mainly implemented through tax-financed subsidies; specifically, while oral care is free of charge for children and youth up to 23 years, adults between 30 and 64 years receive an annual allowance of 28 EUR (in 2022), with double the amount (56 EUR) for younger (24-29 years)   or older (> 64 years) adults.An additional special allowance of 112 EUR/year is given to adults with certain chronic conditions, and a high-cost protection applies for all with great care needs (50% of patient costs >279EUR/year). 6 the other hand, oral care is not considered by the regulations of general health care. 9Hence, it is neither included in the regular public healthcare insurance nor regulated by the principle that the ability to pay should not affect the possibilities of receiving care.As a result, private spending covers as much as 65% of total oral care expenditures. 6Swedish residents without means to pay can seek municipal subsidies only for basic care, while asylum seekers and undocumented people are solely entitled to emergency care.In addition, the public oral sector has during the last decade faced a huge shortage of human resources, affecting accessibility and financing.This together with a widespread presence of private actors and deregulated pricing of services has contributed to oral care being the most profitable healthcare sector for private companies. 6ere are worries that these arrangements of the Swedish oral care system could accentuate the effects of existing social inequities in oral health and care, and practitioners and policy makers have called for evidence to support development of equity-promoting strategies. 6,103][14] Despite this pattern of ubiquitous inequities, little is known about how such social dimensions interact and simultaneously shape complex population patterns in oral health and care outcomes. 16e recognition of such complex inequities has recently started to gain attention in the field of oral health, guided by the concept of intersectionality. 16,17Intersectionality challenges how social categories traditionally have been examined as separate entities, and instead emphasizes the largely neglected feature of how they influence and intersect in individuals' lives in complex ways. 17Accordingly, social categories are not considered individual characteristics or risk factors, but contexts shaped by social determinants and processes conditioning the distribution of power and resources and consequently health. 18,19e intersectionality framework originates in social sciences and has traditionally been approached by qualitative methods. 16,17wever, the application of intersectionality has more recently also been advocated within social epidemiology.This has been motivated by its potential to inform effective policy and intervention for equity in health and care, by providing a more precise mapping of social inequities and uncovering complex patterns of advantage/ disadvantage. 19 this vein, new methodological approaches have been developed in recent years, including the (multilevel) analysis of individual heterogeneity and discriminatory accuracy (AIHDA/MAIHDA). 18,19ese methods estimate not only the different average risk of experiencing an outcome across a range of categories defined by complex social positions, but also the ability of such categorization to discriminate between those with and without the outcome.In practice, this provides a comprehensive picture of social heterogeneity in health and guide the choice of universal versus targeted interventions. 20fortunately, these new methods have mostly been applied to general health, [20][21][22][23] with only one recent MAIHDA study examining periodontal disease in England, 23 and no study on the oral care field.
In response to an international call for intersectional methodologies in oral health research, and in a national and global context of political debate about how to tackle oral health and care inequities, this study uses an intersectional approach to examine inequities in unmet oral care needs in the adult population in Sweden over the period 2004-2021, to contribute to the current knowledge and strategies for a more equitable oral care system.This study was approved by the Swedish Ethical Review Authority (approval no.2021-02398).

| Variables
Unmet care needs refers to the inability to obtain needed care and is a concept commonly used in health services research to indicate barriers to care. 24In this study, the outcome variable unmet oral care needs was assessed through the survey question: 'Have you during the past three months, considered yourself in need of dental care, but refrained from seeking care?', with response options 'Yes' (coding = 1) or 'No' (=0).
The exposure variables consisted of five sociodemographic inequity dimensions which were retrieved and linked from population registers of Statistics Sweden.They were categorized and coded as follows: gender, defined by the proxy variable of biological sex Survey year was identified as a potential confounder and therefore included as a covariate.To avoid overadjustment for potential intervening variables, no other covariates were included.

| Data analysis
An intersectionality-informed analysis of individual heterogeneity and discriminatory accuracy (AIHDA) was conducted based on the procedure described elsewhere. 20The individual heterogeneity component consists of modelling a specific health outcome using a regression analysis of individuals nested within a matrix defined by the intersection of several inequity dimensions (intersectional strata) for example gender and income.As a complement, the discriminatory accuracy provides information on the accuracy that the inequity dimensions in the model discriminates individuals who have unmet needs from those who do not. 19rst, prevalence ratios (PR) with 95% confidence intervals of unmet oral care needs were estimated by generalized linear models using log family and identity link, in three consecutive models: base model including the crude models for each of the inequity dimensions and survey year (Model 1); the five single dimensions of inequity adjusted for survey year (Model 2) and the intersectional strata variable adjusted for survey year (Model 3).
Second, the discriminatory accuracy of the single inequity dimensions and the intersectional strata for identifying individuals with unmet oral care needs was estimated by calculating the area under the receiver operating characteristic curve (AU-ROC, or AUC for short) for the corresponding three models.A known classification was used to interpret the discriminatory accuracy as follows: (i) 'absent or very small' (AUC = 0.5-0.6),(ii) 'moderate' (AUC >0.6-≤0.7),(iii) 'large' (AUC >0.7-≤0.8)and (iv) 'very large' (AUC >0.8). 22nally, the incremental change in the AUC value (ΔAUC) was calculated to quantify the improvement in the discriminatory accuracy compared to the reference model (Model 1).All analyses were performed in STATA 17.0.

| Inequities in unmet oral care needs by single dimensions
Overall, the prevalence of unmet oral care needs was 15%.It displayed a decreasing trend between 2004 and 2014 and then a gradual increase until 2021.The distribution of unmet needs among single inequity dimensions showed the highest prevalence among immigrants (24%) followed by those aged 26-44 years (20%) and those with low income (18.6%)(Table 1).
Table 2 shows results from the crude and adjusted analysis (Models 1 and 2, respectively).Overall, the descriptive patterns were confirmed in both the crude and adjusted models, with independently higher risk (prevalence ratios) for unmet needs among 3)], and with a lower risk for women [0.9 (0.8-0.9)] in the adjusted model.

| Intersectional inequities
The analysis of the 48 intersectional strata generally followed the patterns suggested in the single-dimension analysis, but also showed great heterogeneity in the prevalence of unmet oral care needs which was not discernable in the preceding analyses.Overall, women displayed a slightly lower prevalence of unmet needs.The lowest prevalence of 5.6% was found among the reference stratum consisting of older non-immigrant men with high education and income, and the highest prevalence in the group of men aged 26-44 years, immigrant with both low education and income (42.1%).A high prevalence was also found among the correspondingly disadvantaged women (34.9%) (Table 3).
As shown in Table 3 and Figure 1, the six groups with highest risk for unmet oral care needs (PR >6) were all immigrants, and five of them also had low income.In contrast, most groups with lowest risk (PR <2) were natives.However, several exceptions to these general patterns were also seen.In both women and men, the combination of being an immigrant and having low income did not universally entail a notoriously high risk for unmet needs as long as they did not carry other disadvantages (PR 2.7 and 3.7, respectively).However, the risk increased markedly for those who were younger (PR = 5.4-6.1) and also had low education (PR = 6.5-7.8).Old women and men with an immigrant background, high education and high income did not show a notoriously higher risk than the most privileged stratum (PR = 1.7-1.8).In addition to this, low-educated old women and men experienced a low risk (PR = 1.2-1.5),as long as they had a high income and were natives.

| DISCUSS ION
13]23 Our findings confirmed that there is a higher risk of unmet needs among immigrants, 12,24 groups with both low income 3,12,24 and education, 12 and that gender does not play a major role in explaining differences in unmet needs. 3,4vergent findings were, however, observed regarding age, as our study reported highest disadvantage among young adults in contrast to previous research. 4This is probably explained by the differences in the public oral health coverage among countries and the extent to which elderly's oral care is subsidized.Moreover, it is known that younger adults can commonly experience high lifestyle stress related to several factors including the formation of a family and transition to parenting, high demands for financial stability and achievement of career aspirations. 25,26e results expand previous knowledge on inequities in unmet oral care needs in several ways.We identified a particularly high risk of unmet needs among strata mostly consisting of individuals who were both immigrants and had low income.Relatedly, a considerable portion of immigration to Sweden comprises asylum seekers from conflict countries, for example Iraq, Syria and Somalia, and for whom integration to the labour market is challenging.However, being an immigrant and/or having a low income did not universally entail a high risk but varied by the social position along other axes, particularly age and education.These observations illustrate two intersectionality hypotheses.
The first one is the 'multiple jeopardy hypothesis' that has been integral to intersectionality research since its conception; that is a synergistic effect where the more socioeconomic disadvantages an individual has the higher is her/his risk of experiencing unmet needs. 21,27is is illustrated by the high risk of the multiply disadvantaged group of immigrant young women with both low education and income compared to the almost five-fold lower risk of the more advantaged group of Sweden-born old women with both high education and income.
The second hypothesis is the contingency of the inequities, suggesting that the risk or protection from a particular social disadvantage depends on other social positions.For instance, immigrant young women who had both high income and education still  reported a risk of 2.5 times higher risk than the most privileged group, despite their favourable socioeconomic positions.A similar phenomenon has been described regarding inequities in periodontal disease in England where the stratum with highest risk displayed a mix of both disadvantage and privilege. 23These findings illustrate that considering socially disadvantaged groups as particularly vulnerable to unmet oral care is only contingently accurate and that an intersectional approach can uncover patterns of advantage/disadvantage and identify the most vulnerable groups who remain undiscernible when considering inequities as singular.
Regarding the discriminatory accuracy analysis, the moderate AUC (0.66) of the intersectional model and the marginal increase in AUC in the intersectional model constituted other important findings.They indicate that the intersection of inequity dimensions did not improve the accuracy for identifying individuals with unmet oral care needs compared to the simple combination of all those dimensions.Measures of discriminatory accuracy, such as the AUC, are commonly interpreted as informative for policy decision based on the principle of proportionate universalism 20 that is that action towards equity in health need to be universal, but with a scale and intensity proportionate to the level of disadvantage. 28The AUC estimates found in this study were of moderate size but nevertheless of similar or larger size than those found for established cardiovascular risk factors. 29 To our knowledge, this is one of few studies using a quantitative intersectional approach to the oral health field. 23The results provide an improved mapping of the socioeconomic disparities in unmet oral care needs in Sweden and can therefore contribute to ongoing political discussions for future reforms.However, one limitation was the low to moderate participation rate (from 60.8% in 2004 to 44.1% in 2021).This and the fact that men, immigrants and individuals with both low income and education are less likely to participate are indications of selection bias, which could lead to underestimation of the inequities, and would limit the generalizability of the results to the target population.
The outcome measure 'unmet oral care needs' has been increasingly used in recent national and international reports examining oral care disparities, 6,15 despite that it can be considered unspecific and It might capture several access barriers and can alternatively be assessed more comprehensively through multiple questionnaire items. 24,30Nevertheless, the item used in this study is the same or similar to measures used in previous studies on various population groups and countries, including Sweden.The simple categorization of each inequity dimension was another weakness because of the risk of underestimating heterogeneity in unmet needs; for example only binary sex was available to measure the participants´ gender.These categorizations were necessary to maintain statistical power while simultaneously considering multiple inequity indicators and remain a challenge for intersectionality-informed quantitative research.
Furthermore, our approach to 'operationalize' intersectionality has been challenged from a theoretical perspective which considers that the intersectional categories represent complex social contexts that exist prior to any research question and cannot be captured by using statistical analyses. 19Indeed, our approach confirms that the interaction of effects goes beyond the statistical concept of additive or multiplicative interaction, implying that the absence of any interaction effect across several intersectional strata does not refute the existence of intersectional patterns. 20Nevertheless, this study aligns with efforts to advance the intersectionality-informed epidemiological research and suggests that it could be complemented by qualitative methods to better understand the complex mechanisms by how intersectional disadvantages can affect health. 16ken together, these findings suggest some considerations for policy towards equity in oral health and care in Sweden.Strategies directly targeting disadvantaged groups, that is immigrants, may not necessarily be feasible nor effective.Moreover, they could lead to stigmatization of large and heterogenous population groups while failing to reach the most vulnerable strata.In contrast, policy and interventions may be the most effective if approaching the whole population, combined with selected interventions directed at the most disadvantaged intersectional strata, according to the principle of proportionate universalism.One example would be to regulate the price of dental services while strengthening the high-cost protection and special subsidies for the groups with greatest care needs.
Our findings could reflect a similar situation in other Nordic or European countries with similar welfare and oral care systems.
However, generalization should be done with caution since our outcome is associated with complex phenomena shaped by preconditions of the particular national context.
Furthermore, this study contributes to recent international initiatives advocating the application of an intersectionality perspective in oral health research, but which also notices challenges for quantitative methods. 16We have attempted to overcome these challenges by using an empirical testable method which could serve to investigate oral health and care in other contexts and population groups.
Finally, intersectionality-informed quantitative research in oral health and care is scarce, and further evidence is needed to develop the available methods and to guide policy process.Future research should therefore explore inequities in oral health and care focusing on multiple intertwined inequity dimensions across different contexts and oral care systems.Moreover, it could benefit from interdisciplinary methodologies to deeply understand underlying mechanisms that can serve as the targets for action towards an equitable oral healthcare system. 16

CO N FLI C T O F I NTE R E S T
There are no conflicts of interest.
Data were obtained from 14 Health on Equal Terms (HET) surveys conducted by the Public Health Agency of Sweden during the period 2004-2021, excluding 2015 and 2016 as unmet oral care needs was not included in the survey, and 2017 and 2019 as no survey was conducted these years.The HET is a self-administered questionnaire from a yearly random sample of the Swedish population aged 16-84 years.It gathers information about general and oral health, lifestyle and healthcare utilization.These data are linked to information on sociodemographic conditions from population registers.The survey response rate had a decreasing trend across the study period (from 60.8% in 2004 to 44.1% in 2021) resulting in a total of 151 674 responses with an overall response rate of 49%.Respondents aged 16-25 years [N = 15 471 (10.2%)] were excluded as the majority in this age group is studying and has therefore not yet entered the labour market, making it difficult to accurately assess educational level and income.Moreover, oral care services in Sweden are free of charge for residents younger than 24 years (up to 16000528, 2023, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/cdoe.12836 by Umea University, Wiley Online Library on [19/06/2023].See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions)on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License people compared to the rest of the population.Thereafter, respondents with missing responses on the outcome and exposure variables were excluded [N = 6730 (4.9%)].The final study population consisted of N = 129 473 individuals.

TA B L E 1
Distribution of the sample and prevalence of unmet oral care needs across categories of inequity dimensions and survey year in 129 473 participants in the Swedish Health on Equal Terms surveys (2004-2021).

FU
The study was financially supported by the Strategic research grant from the Faculty of Medicine, Umeå University, to PEG (FS 2.1.6-339-20).

crude) PR (95% CI) Model 2 (adjusted) PR (95% CI)
Prevalence ratios (PR) and 95% confidence interval (CI) of unmet oral care needs by each inequity dimension and survey year (Model 1) and by the inequity dimensions adjusted for survey year (Model 2) among 129 473 participants in the Swedish Health on Equal Terms surveys (2004-2021).
28,29suggest that predominantly universal interventions combined with interventions targeting the most disadvantaged intersectional strata would be the most effective strategy for addressing unmet oral care needs inequities among adults in Sweden.28,29Model 1 ( Distribution, prevalence and prevalence ratios (PR) with 95% CI of unmet oral care needs across the 48 intersectional strata.