The Relationship Between Preventive Dental Care and Overall Medical Expenditures

ABSTRACT

Objectives: To examine the relationship between preventive dental visits (PDVs) and medical expenditures while mitigating bias from unobserved confounding factors.

Study Design: Retrospective data analysis of Indiana Medicaid enrollment and claims data (2015-2018) and the Area Health Resources Files.

Methods: An instrumental variable (IV) approach was used to estimate the relationship between PDVs and medical and pharmacy expenditures among Medicaid enrollees. The instrument was defined as the number of adult enrollees with at least 1 nonpreventive dental claim per total Medicaid enrollees within a Census tract per year.

Results: In naive analyses, enrollees had on average greater medical expenditures if they had a prior-year PDV (β = $397.21; 95% CI, $184.23-$610.18) and a PDV in the same year as expenditures were measured (β = $344.81; 95% CI, $193.06-$496.56). No significant differences in pharmacy expenditures were observed in naive analyses. Using the IV approach, point estimates of overall medical expenditures for the marginal enrollee who had a prior-year PDV (β = $325.17; 95% CI, –$708.03 to $1358.37) or same-year PDV (β = $170.31; 95% CI, –$598.89 to $939.52) were similar to naive results, although not significant. Our IV approach indicated that PDV was not endogenous in some specifications.

Conclusions: This is the first study to present estimates with causal inference from a quasi-experimental study of the effect of PDVs on overall medical expenditures. We observed that prior- or same-year PDVs were not related to overall medical or pharmacy expenditures.

Am J Manag Care. 2024;30(2):e39-e45. https://doi.org/10.37765/ajmc.2024.89499

Takeaway Points

Results from an instrumental variable approach showed no significant relationship between preventive dental visits and total medical and pharmacy expenditures.

• Previous studies have shown that preventive dental care may lead to reduced overall medical expenditures, but these studies have been observational in design and subject to omitted variable bias.
• The instrument utilized in this study may feasibly generate valid findings using existing administrative claims data from states, insurance programs, and populations.
• Findings have implications for overall state public insurance policies because adult Medicaid dental benefits vary across states and may be targeted for reductions when state budgets are constrained.

Poor oral health is associated with pain,^1,2 decreased chewing function,³ negative social perceptions,⁴ and reduced quality of life.^5-8 Although the full causal pathways remain elusive, some evidence suggests an association between poor oral health and chronic diseases.^9-14 Bacterial infection and inflammation in the mouth (caused by dental plaque buildup and poor hygiene) may trigger or exacerbate the host’s overall inflammatory response, subsequently contributing to the progression of other systemic diseases. Thus, preventing or reversing poor oral health may positively affect overall health by improving physical, psychological, and social well-being.^15,16 For instance, some studies suggest that having a dental prophylaxis (commonly referred to as a dental cleaning) or a dental scaling reduces the risk of ischemic stroke,¹⁷ esophageal cancer,¹⁸ infective endocarditis,¹⁹ Parkinson disease,²⁰ and myocardial infarction.²¹

A hypothesized systemic-oral health link has led to investigations into whether provision of timely dental services can lead to improved health and reduced medical expenditures.^22-26 Private insurers have reported fewer hospital admissions (39% less) and fewer emergency department visits (36% less) among enrollees who receive preventive dental care at least once a year.²⁷ Among adults with gum disease, those who receive annual dental care have lower overall medical expenditures, fewer emergency department visits, and fewer annual inpatient admissions compared with adults without annual dental care.^27-29 Additionally, several studies have reported significantly lower medical expenditures among adults with periodontal disease who receive appropriate dental care treatments.^24-26,28,30 However, all these studies are weak in internal validity and subject to omitted variable bias. The effect of preventive interventions on health outcomes is often overestimated because, in general, individuals who seek preventive care are otherwise healthier than their counterparts.³¹ In addition, current evidence on whether dental care affects medical care expenditures is subject to bias from relevant unobserved factors (eg, an individual’s health literacy, hygiene habits, and level of health consciousness) that confound the relationship between dental care and medical outcomes.

This study mitigates bias from key unobserved confounding factors by using an instrumental variable (IV) approach to determine the extent to which preventive dental care may be causally associated with overall medical expenditures. This will be accomplished using an appropriate IV, which is not associated with the outcome except through its effect on the endogenous variable and therefore should be unrelated to unobserved characteristics that influence the outcome. We examine how preventive dental care is related to medical and pharmacy expenditures in a population of low-income, Medicaid-enrolled adults using a measure of provider preference as the instrument. Our findings have implications for overall state public insurance policy, as adult Medicaid dental benefits vary greatly across states and may be targeted for reductions when state budgets are constrained.^22,32,33 Further, our IV may be of interest to other researchers who examine the relationship between dental care and medical outcomes.

METHODS

Data and Study Population

We used administrative claims and enrollment data from the Indiana Family and Social Services Administration Office of Medicaid Policy and Planning. Additional economic and county-level population data were derived from the Health Resources and Services Administration’s Area Health Resources Files.³⁴ We included any adult who was continuously enrolled in the Healthy Indiana Plan (HIP) Plus program for at least 36 months between February 1, 2015, and December 31, 2018 (see eAppendix A for more details [eAppendices available at ajmc.com]). We used the first year of enrollment as a baseline year to define our explanatory variable and the second continuously enrolled year as the measurement year to calculate our dependent variables.

Dependent Variable

For the enrollee’s second 12-month period of enrollment, we summed annual expenditures from their medical and pharmacy claims. All expenditures from claims with International Classification of Diseases, Ninth Revision, Clinical Modification or Tenth Revision, Clinical Modification diagnosis codes related to oral conditions and diseases^35,36 were excluded because these expenditures are directly associated with receipt of dental care (eg, emergency department visits for nontraumatic dental care) (see eAppendices B and C). Finally, we excluded expenditures associated with trauma or injuries, as these expenditures were used for a separate falsification outcome analysis (see eAppendix D). Expenditures were adjusted for inflation using the 2019 Consumer Price Index.³⁷

Endogenous Main Explanatory Variable

The main explanatory variable of interest was a 1-year lagged binary variable indicating whether the adult had any preventive dental visit (PDV) within the first year of enrollment. Using the Healthcare Effectiveness Data and Information Set definition,³⁸ PDVs were identified by dental claims that included Current Dental Terminology (CDT) codes D0120 (periodic oral evaluation), D0150 (comprehensive oral evaluation), D1110 (adult prophylaxis), D1206 (topical application of fluoride varnish), D1208 (topical application of fluoride excluding varnish), D1351 (tooth sealant), and/or D1330 (oral hygiene instructions) but were absent of CDT codes D2000 through D9999 (nonpreventive dental procedures).³⁸ Thus, our explanatory variable identified preventive-only dental visits. As a secondary variable of interest, we use the binary indicator of whether the adult had a PDV within the same enrollment year that the dependent variable was measured.

Instrumental Variable

Because individuals who seek preventive dental care may also seek preventive medical care more often than other adults, our naive model specification is subject to omitted variable bias. Unobserved individual characteristics, such as oral hygiene behaviors and health consciousness, motivate adults to self-select dental and medical care services. To address this bias, we computed a preference-based IV that represented variation in dental providers’ preference to treat adult Medicaid enrollees or the level in which they engage with Medicaid enrollees (see eAppendix E). Specifically, our IV is the number of adult enrollees with at least 1 nonpreventive dental claim per total enrollees within a Census tract per year. A nonpreventive dental claim was defined as a claim with CDT codes representing nonpreventive procedures (D2000-D9999) but the absence of CDT codes representing preventive dental care (D1000-D1999). When calculating the IV ratio at the individual level, we removed the enrollee associated with the IV. For additional information on first-stage statistics, please see eAppendix F.

Analysis

Descriptive characteristics of the population are presented as frequencies and means. We investigated the validity of our instrument by calculating the standardized mean difference in covariates above and below the median value of the IV. Standardized mean difference values between –0.1 and 0.1 indicate balance in observable covariates across groups.³⁹ Given the nontrivial number of enrollees (n = 3159; 11.3%) who had no medical expenditures ($0) and the positively skewed distribution of health expenditure data, we used a 2-part model similar to that used by Biener et al for medical expenditure data.⁴⁰ In the first part, we specify a logit model to estimate the probability of having positive expenditures. In the second part, we specify a generalized linear model with γ variance structure and log-link function to estimate the amount of medical expenditures conditional on positive expenditures. After first estimating naive regression results where PDV status is treated as exogenous, we estimated all outcomes using a 2-stage residual inclusion method with our IV. We estimated separate models for lagged effects (PDVs measured in the first year of enrollment) and for concurrent effects (PDVs in the second year of enrollment). In models that estimated the lagged effect of PDVs, we restricted our sample to individuals who had no PDV in the same year as the measured outcome. All models were adjusted for sex, race/ethnicity, marital status, living arrangement, family size, year, annual chronic encounter, annual wellness visit, rural-urban commuting area designation, unemployment rate, total number of medical providers per Census tract, and health professional shortage area designation (see eAppendix G for information on the construction of these variables). We present the estimated marginal effects on annual medical expenditures (including pharmacy expenditures), and separately for pharmacy expenditures only, associated with having a PDV in the prior or same year. Estimates from the first and second part of the model are reported in eAppendix H.

We conducted the following sensitivity analyses (see eAppendix I): (1) examining outcomes in the enrollees’ third year of enrollment; (2) restricting our sample to individuals with diagnosed diabetes, a chronic disease associated with poor oral health, using the diabetes condition algorithm provided by the CMS Chronic Conditions Data Warehouse⁴¹; and (3) restricting our outcome to only inpatient expenditures. We also present 2-stage least squares and generalized linear regression models as robustness checks (see eAppendix J). Through falsification tests, we evaluated higher orders of the IV. All P values on overidentification tests were insignificant at an α of 0.05; thus, we failed to reject the null hypothesis that the instruments were uncorrelated with the residual error term. Finally, we present models from our falsification outcome test (expenditures attributed to trauma or injury) in eAppendix K. All estimates were reported at 95% CIs, and SEs were adjusted with bootstrapping methods. Data were managed with SAS 9.2 (SAS Institute) and analyzed with Stata/SE 17 (StataCorp LLC).

RESULTS

The study population consisted of 27,888 enrollees, of whom the majority were women (58.6%), White (75.8%), and lived in urban counties (77.6%) (Table 1 [part A and part B]). On average, an enrollee was approximately 45 years old. Individuals who had a PDV within a 12-month enrollment period were more likely to be female (67.0% vs 55.1%), married (32.5% vs 30.8%), and have an annual wellness visit (37.2% vs 23.0%) (all P < .001). Among those with any medical expenditures (88.7%), enrollees had a mean (SD) of $4059 ($6122) in annual medical expenditures. Among those with any pharmacy expenditures (77.7%), individuals had a mean (SD) of $1329 ($3043) in yearly pharmacy expenditures.

Table 2 presents observed covariates by IV status (ie, above and below the median IV value). Standardized mean differences between group covariates were between –0.1 and 0.1 (with the exception of unemployment rate), therefore differences between the 2 groups were not considered meaningful. Thus, our IV simulates a random “flip of a coin” in how it balances observable baseline covariates across groups. Although this falsification test does not definitively prove the exclusion restriction assumption, the balance of observable factors by IV status provides some validation that the IV would plausibly balance the distribution of unobservable factors between the 2 groups.

Naive and IV estimates of the effect of prior- and same-year PDV on medical and pharmacy expenditures during enrollees’ second year of enrollment are presented in Table 3. With our primary measure of PDV (measured in the prior year of enrollment), naive analyses found on average greater medical expenditures (β = $397.21; 95% CI, $184.23-$610.18) but no difference in pharmacy expenditures compared with enrollees with no PDV in the prior year. Point estimates from IV methods were similar to naive results but were not found to be significant for medical expenditures (β = $325.17; 95% CI, –$708.03 to $1358.37) or pharmacy expenditures (β = $265.35; 95% CI, –$127.27 to $657.97) for the marginal enrollee who had a prior-year PDV.

With our secondary measure of PDV (measured in the second year of enrollment), naive analyses found on average greater medical expenditures (β = $344.81; 95% CI, $193.06-$496.56) and no difference in pharmacy expenditures compared with enrollees with no PDV in the same year as expenditures were measured. Analyses from IV methods indicate that having a PDV in the same year was not significantly related to the marginal enrollee’s medical expenditures (β = $170.31; 95% CI, –$598.89 to $939.52) or pharmacy expenditures (β = $20.84; 95% CI, –$433.32 to $475.02).

Estimates from sensitivity analyses and robustness checks were consistent with our main results (see eAppendices H-J). Results from our falsification outcome test can be found in eAppendix K. Although the test did not find any association between PDVs and injury or trauma expenditures, point estimates were large and suggest uncertainty in these data.

DISCUSSION

This is the first study to present estimates with causal inference from a quasi-experimental study of the effect of PDVs on medical health, as measured by medical and pharmacy expenditures. We examined the effect of PDV on overall medical and pharmacy expenditures using a strong econometric technique to mitigate bias from unobserved confounding factors. Specifically, we used an IV regression approach, which has not previously been used within the context of PDV and medical expenditures. Our IV appears plausibly valid and may be a useful approach for future studies that evaluate the effect of adult preventive care on medical outcomes of interest. Randomized controlled trials are considered the gold standard for generating causal findings, but such designs face considerable ethical, cost, and timing challenges, especially when resources are limited. In contrast, the IV used in this study may feasibly generate valid findings using existing administrative claims data from other states, insurance programs, and populations.

Our naive regression results showed a positive relationship between PDV and medical expenditures, which contrasts with the results of previous studies.^27,29 Our findings may be related to differences in our study population (low-income adults) compared with previously studied populations (privately insured adults). One hypothesis is that preventive dental care can reduce medical expenditures, but a competing hypothesis is that people who receive preventive services spend more on health care and thus are more likely to have a PDV. Our naive results may also be affected by the timing of our study, which occurred during the first 4 years of HIP Plus program implementation following Medicaid expansion in Indiana. Given that many in our study population were previously uninsured, these adults may have had significant pent-up demand for both medical and dental care, thus positively skewing our naive results.

In the case of the marginal enrollee, our IV regression results showed no significant relationship between PDV and medical health, as measured by total medical and pharmacy expenditures. For some of the models, the point estimates generated by the naive and IV regressions similarly suggested higher medical costs among those who received PDVs. Overall, differences in our findings compared with those of previous studies are likely due to 4 reasons.

First, previous studies have had simple observational study designs that were not able to control for endogenous factors associated with PDVs and medical expenditures, including health behaviors, beliefs, and practices that influence how medical care is sought and engaged with by adults. We used a robust econometric approach to mitigate bias and present results with strong internal validity. Second, several previous studies have focused on a subset of the adult population with diagnosed periodontal disease, primarily because prevailing hypotheses claim that treatment of periodontal disease leads to reduced overall medical expenditures, particularly among adults who have other chronic diseases.^24,25,28,30 Our study did not narrow inclusion to adults who had periodontal disease. Thus, we recommend caution in extrapolating our observed effect of PDV to individuals with periodontal disease and the effect of periodontal treatments on medical expenditures. Third, our definition of preventive dental care varies from that of other studies. We evaluated the effect of preventive dental visits using only specific preventive procedures and included no treatment services whatsoever in our definition. In contrast, Lamster et al²⁶ found that preventive dental care was associated with reductions in inpatient admission and emergency department costs among Medicaid-enrolled adults in New York. However, their definition of preventive dental care included nonpreventive treatments for periodontal disease.²⁶ We recommend additional research that is strong in internal validity, to evaluate the effect of certain dental treatments on overall and disease-specific expenditures, particularly in low-income populations, who have higher rates of periodontal disease than other adults.⁴² Fourth, we examined the first 4 years of the HIP Plus program and, therefore, a study population that was previously uninsured. Thus, among a population with potential for significant pent-up demand, restoration of oral health may be needed before reductions in overall medical expenditures can occur through preventive care.

In addition, this current study cannot account for any delayed effects of preventive dental visits on downstream medical conditions. Given that certain medical conditions take years to develop, the effect of PDVs in the prior or same year may not capture potential changes in expenditures that could accumulate over time. A longer study period is needed to test these conjectures and to determine whether long-term, repetitive PDVs may have benefits not observed in our short study period.

In summary, this current study offers strong evidence on the short-term relationship between preventive dental care and medical expenditures in a limited population, but it is not exhaustive in answering all questions associated with the relationship of dental care and medical outcomes.

Limitations

Although this is the first study to present causal estimates of the effect of PDVs on overall medical expenditures, our study is not without limitations. Our IV approach has strong internal validity but limited external validity. Our results derived from the IV approach may not generalize to adults outside this study population, such as those with different socioeconomic status, insurance coverage, and disease status. More importantly, our IV estimates only inform policy with regard to the marginal enrollee, a person whose receipt of PDV is determined by the value of the IV. Further, our findings may not generalize to Medicaid-enrolled adults who disenroll over time or who reside in other states. Use of our proposed instrument in those populations should be considered. In addition, not all the assumptions associated with our IV approach are directly testable. However, as a strength, we conducted falsification tests by evaluating whether the instrument balanced observable covariates, testing higher-order IVs, and examining a falsification outcome to explore the validity of the proposed instrument. Notably, we excluded dental-related diagnoses from medical expenditures and thus cannot draw conclusions about any potential association with PDV in this way. For example, emergency department visits for nontraumatic dental care (eg, tooth pain) were excluded from our analysis and warrant further investigation with different methodologies. We did not track specific disease comorbidities but did attempt to control for an individual’s overall disease burden by adjusting for enrollee’s chronic disease encounters, and we explored the relationship between PDVs and medical expenditures among patients with diabetes and found similar nonsignificant results. Finally, given our short study time period, we presented results from a limited longitudinal data set. Future work should examine how the IV performs when predicting expenditures over time while accounting for different patterns of within-person PDV utilization.

CONCLUSIONS

This is the first study to present causal estimates of the effect of PDVs on overall medical expenditures. We found that prior- and same-year preventive dental visits did not have an effect on overall medical or pharmacy expenditures. Additional research is needed to explore the effect of PDV over a longer period of time, the effect of specific dental treatments on medical expenditures, and whether preventive dental care has an impact on emergency department visits and expenditures. 

Author Affiliations: Department of Health Policy and Management, Indiana University Richard M. Fairbanks School of Public Health (HLT, AMH, NM, JB), Indianapolis, IN; Center for Biomedical Informatics, Regenstrief Institute (NM, TS), Indianapolis, IN; Department of Health Care Organization and Policy, University of Alabama at Birmingham School of Public Health (BS), Birmingham, AL.

Source of Funding: Research reported in this publication was in part supported by the National Library of Medicine of the National Institutes of Health under award number T15LM012502. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health or the National Library of Medicine.

Author Disclosures: The authors report no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article.

Authorship Information: Concept and design (HLT, AMH, TS, BS, JB); acquisition of data (JB); analysis and interpretation of data (HLT, AMH, BS, JB); drafting of the manuscript (HLT); critical revision of the manuscript for important intellectual content (AMH, NM, TS, BS, JB); statistical analysis (HLT); obtaining funding (TS); administrative, technical, or logistic support (NM); and supervision (NM, TS, BS, JB).

Address Correspondence to: Heather L. Taylor, PhD, MPH, RDH, Department of Health Policy and Management, Indiana University Richard M. Fairbanks School of Public Health, 1050 Wishard Blvd, Ste 6185, Indianapolis, IN 46202. Email: hhavens@iu.edu.

REFERENCES

1. Kassebaum NJ, Smith AGC, Bernabé E, et al; GBD 2015 Oral Health Collaborators. Global, regional, and national prevalence, incidence, and disability-adjusted life years for oral conditions for 195 countries, 1990-2015: a systematic analysis for the global burden of diseases, injuries, and risk factors. J Dent Res. 2017;96(4):380-387. doi:10.1177/0022034517693566

2. Ardila CM, Agudelo-Suárez AA. Association between dental pain and caries: a multilevel analysis to evaluate the influence of contextual and individual factors in 34 843 adults. J Investig Clin Dent. 2016;7(4):410-416. doi:10.1111/jicd.12168

3. Ramsay SE, Whincup PH, Watt RG, et al. Burden of poor oral health in older age: findings from a population-based study of older British men. BMJ Open. 2015;5(12):e009476. doi:10.1136/bmjopen-2015-009476

4. Willis MS, Esqueda CW, Schacht RN. Social perceptions of individuals missing upper front teeth. Percept Mot Skills. 2008;106(2):423-435. doi:10.2466/pms.106.2.423-435

5. Naito M, Yuasa H, Nomura Y, Nakayama T, Hamajima N, Hanada N. Oral health status and health-related quality of life: a systematic review. J Oral Sci. 2006;48(1):1-7. doi:10.2334/josnusd.48.1

6. Ortíz-Barrios LB, Granados-García V, Cruz-Hervert P, Moreno-Tamayo K, Heredia-Ponce E, Sánchez-García S. The impact of poor oral health on the oral health-related quality of life (OHRQoL) in older adults:
the oral health status through a latent class analysis. BMC Oral Health. 2019;19(1):141. doi:10.1186/s12903-019-0840-3

7. Gil-Montoya JA, de Mello AL, Barrios R, Gonzalez-Moles MA, Bravo M. Oral health in the elderly patient and its impact on general well-being: a nonsystematic review. Clin Interv Aging. 2015;10:461-467. doi:10.2147/CIA.S54630

8. DeBaz C, Shamia H, Hahn J, Mithani S, Sadeghi G, Palomo L. Periodontitis impacts quality of life in postmenopausal women. Climacteric. 2015;18(4):637-642. doi:10.3109/13697137.2014.996124

9. Taylor HL, Rahurkar S, Treat TJ, Thyvalikakath TP, Schleyer TK. Does nonsurgical periodontal treatment improve systemic health? J Dent Res. 2021;100(3):253-260. doi:10.1177/0022034520965958

10. Bui FQ, Almeida-da-Silva CLC, Huynh B, et al. Association between periodontal pathogens and systemic disease. Biomed J. 2019;42(1):27-35. doi:10.1016/j.bj.2018.12.001

11. Kapellas K, Singh A, Bertotti M, Nascimento GG, Jamieson LM; Perio-CKD collaboration. Periodontal and chronic kidney disease association: a systematic review and meta-analysis. Nephrology (Carlton). 2019;24(2):202-212. doi:10.1111/nep.13225

12. Liccardo D, Cannavo A, Spagnuolo G, et al. Periodontal disease: a risk factor for diabetes and cardiovascular disease. Int J Mol Sci. 2019;20(6):1414. doi:10.3390/ijms20061414

13. Martinez-Herrera M, Silvestre-Rangil J, Silvestre FJ. Association between obesity and periodontal disease: a systematic review of epidemiological studies and controlled clinical trials. Med Oral Patol Oral Cir Bucal. 2017;22(6):e708-e715. doi:10.4317/medoral.21786

14. Sanz M, Ceriello A, Buysschaert M, et al. Scientific evidence on the links between periodontal diseases and diabetes: consensus report and guidelines of the joint workshop on periodontal diseases and diabetes by the International Diabetes Federation and the European Federation of Periodontology. J Clin Periodontol. 2018;45(2):138-149. doi:10.1111/jcpe.12808

15. Simpson TC, Weldon JC, Worthington HV, et al. Treatment of periodontal disease for glycaemic control in people with diabetes mellitus. Cochrane Database Syst Rev. 2015;2015(11):CD004714. doi:10.1002/14651858.CD004714.pub3

16. Outcomes and impact of oral disease. In: Report of the National Advisory Council on Dental Health. Australian Government Department of Health; September 3, 2012. Accessed July 14, 2020. https://web.archive.org/web/20200815024605/https://www1.health.gov.au/internet/publications/publishing.nsf/Content/report_nacdh~report_nacdh_ch1~report_nacdh_out

17. Lee YL, Hu HY, Huang N, Hwang DK, Chou P, Chu D. Dental prophylaxis and periodontal treatment are protective factors to ischemic stroke. Stroke. 2013;44(4):1026-1030. doi:10.1161/STROKEAHA.111.000076

18. Lee YL, Hu HY, Yang NP, Chou P, Chu D. Dental prophylaxis decreases the risk of esophageal cancer in males; a nationwide population-based study in Taiwan. PLoS One. 2014;9(10):e109444. doi:10.1371/journal.pone.0109444

19. Chen SJ, Liu CJ, Chao TF, et al. Dental scaling and risk reduction in infective endocarditis: a nationwide population-based case-control study. Can J Cardiol. 2013;29(4):429-433. doi:10.1016/j.cjca.2012.04.018

20. Chen CK, Huang JY, Wu YT, Chang YC. Dental scaling decreases the risk of Parkinson’s disease: a nationwide population-based nested case-control study. Int J Environ Res Public Health. 2018;15(8):1587. doi:10.3390/ijerph15081587

21. Lee YL, Hu HY, Chou P, Chu D. Dental prophylaxis decreases the risk of acute myocardial infarction: a nationwide population-based study in Taiwan. Clin Interv Aging. 2015;10:175-182. doi:10.2147/CIA.S67854

22. Vujicic M, Fosse C, Reusch C, Burroughs M. Making the case for dental coverage for adults in all state Medicaid programs. American Dental Association white paper. July 2021. Accessed July 9, 2021. https://web.archive.org/web/20210708183524/https://www.ada.org/~/media/ADA/Science%20and%20Research/HPI/Files/WhitePaper_0721.pdf

23. Yarbrough C, Vujicic M, Nasseh K. Estimating the cost of introducing a Medicaid adult dental benefit in 22 states. American Dental Association Health Policy Institute research brief. March 2016. Accessed August 9, 2020. https://web.archive.org/web/20170627125205/http://www.ada.org/~/media/ADA/Science%20and%20Research/HPI/Files/HPIBrief_0316_1.ashx

24. Nasseh K, Vujicic M, Glick M. The relationship between periodontal interventions and healthcare costs and utilization. evidence from an integrated dental, medical, and pharmacy commercial claims database. Health Econ. 2017;26(4):519-527. doi:10.1002/hec.3316

25. Smits KPJ, Listl S, Plachokova AS, Van der Galien O, Kalmus O. Effect of periodontal treatment on diabetes-related healthcare costs: a retrospective study. BMJ Open Diabetes Res Care. 2020;8(1):e001666. doi:10.1136/bmjdrc-2020-001666

26. Lamster IB, Malloy KP, DiMura PM, et al. Dental services and health outcomes in the New York state Medicaid program. J Dent Res. 2021;100(9):928-934. doi:10.1177/00220345211007448

27. Clinical insights drive better outcomes: Cigna dental report.Cigna white paper. May 2019. Accessed July 3, 2020. https://www.cigna.com/static/www-cigna-com/docs/employers-brokers/dental-white-paper.pdf

28. Albert DA, Sadowsky D, Papapanou P, Conicella ML, Ward A. An examination of periodontal treatment and per member per month (PMPM) medical costs in an insured population. BMC Health Serv Res. 2006;6:103. doi:10.1186/1472-6963-6-103

29. Better health outcomes for members, lower costs for employers: Aetna Dental/Medical Integration program.Aetna white paper. July 2018. Accessed July 3, 2020. https://www.neebc.org/assets/White-Papers/
Dental-%20Medical%20Integration-%20White%20Paper2018.pdf

30. Jeffcoat MK, Jeffcoat RL, Gladowski PA, Bramson JB, Blum JJ. Impact of periodontal therapy on general health: evidence from insurance data for five systemic conditions. Am J Prev Med. 2014;47(2):166-174. doi:10.1016/j.amepre.2014.04.001

31. Lousdal ML. An introduction to instrumental variable assumptions, validation and estimation. Emerg Themes Epidemiol. 2018;15:1. doi:10.1186/s12982-018-0069-7

32. Huang SS. Should Medicaid include adult coverage for preventive dental procedures? what evidence is needed? J Am Dent Assoc. 2020;151(8):607-613. doi:10.1016/j.adaj.2020.05.005

33. McGinn-Shapiro M; National Academy for State Health Policy. Medicaid coverage of adult dental services. State Health Policy Monitor. October 2008. Accessed September 2, 2021. https://web.archive.org/web/20220130164924/https://www.nashp.org/wp-content/uploads/2009/12/Adult-Dental-Monitor.pdf

34. Area Health Resources Files. Health Resources and Services Administration. Updated July 31, 2023. Accessed January 4, 2024. https://data.hrsa.gov/topics/health-workforce/ahrf

35. ICD-9-CM (ICD-9, ICD9, ICD9CM) diagnosis codes – group 52. Find-a-Code. Accessed May 29, 2021. https://www.findacode.com/icd-9/icd9-diagnosis-codes-52-group.html

36. Manz MC. Recommended Guidelines for Surveillance of Non-Traumatic Dental Care in Emergency Departments. Association of State and Territorial Dental Directors. April 2021. Accessed September 13, 2021. https://www.astdd.org/docs/recommended-guidlines-for-surveillance-of-ntdc-in-eds.pdf

37. Consumer Price Index. US Bureau of Labor Statistics. Accessed July 21, 2020. https://www.bls.gov/cpi/

38. HEDIS measures and technical resources. National Committee for Quality Assurance. Accessed August 2, 2020. https://www.ncqa.org/hedis/measures/

39. Joyce NR, McGuire TG, Bartels SJ, Mitchell SL, Grabowski DC. The impact of dementia special care units on quality of care: an instrumental variables analysis. Health Serv Res. 2018;53(5):3657-3679. doi:10.1111/1475-6773.12867

40. Biener AI, Cawley J, Meyerhoefer C. The medical care costs of obesity and severe obesity in youth: an instrumental variables approach. Health Econ. 2020;29(5):624-639. doi:10.1002/hec.4007

41. Condition categories. Chronic Conditions Data Warehouse. Accessed August 10, 2022. https://www2.ccwdata.org/web/guest/condition-categories

42. Eke PI, Thornton-Evans GO, Wei L, Borgnakke WS, Dye BA, Genco RJ. Periodontitis in US adults: National Health and Nutrition Examination Survey 2009-2014. J Am Dent Assoc. 2018;149(7):576-588.e6. doi:10.1016/j.adaj.2018.04.023