Medicaid is a needs-based program in the United States that subsidizes medical and dental care for minors. The purpose of this study was to test for compliance differences between self-pay and Medicaid-supported patients.
Medicaid patient records (n = 88) were perused retrospectively for characteristics that distract from an orthodontist’s workflow (missed appointments, broken brackets, treatment duration, and so on) and compared with a sample (n = 145) of self-pay patients from the same teaching clinic. Differences in treatment difficulty were adjusted by subject selection and statistically (analysis of covariance).
Medicaid patients were younger (mean age, 14.1 vs 14.9 years) and significantly more likely to be dismissed from treatment (19% vs 4%), generally for noncompliance. Broken brackets and missed appointments were more common in the Medicaid sample. There was no difference in the number of appointments in those completing treatment, but treatment duration was significantly longer for the Medicaid patients who completed treatment (29 vs 25 months). Commute distance and estimated driving time were significantly shorter for the Medicaid-assisted group.
Greater difficulty in managing Medicaid patients may partly explain why they are underserved. Prospective studies are needed to clarify the causes of the differences.
We studied compliance differences between self-pay and Medicaid-supported patients.
Differences in treatment difficulty were adjusted by case selection and statistically.
Medicaid patients missed significantly more appointments on average.
Treatment duration was 4 months longer for Medicaid patients to complete treatment.
Medicaid patients did not continue treatment as frequently as self-pay patients.
Medicaid is a needs-based insurance program funded at both state and federal levels. Medicaid and the smaller Children’s Health Insurance Program cover 1 in every 3 children in the United States: more than 31 million children in total.
There has been a considerable increase in the number of people enrolled in Medicaid, mostly children. However, only a small proportion of orthodontists provide treatment to Medicaid patients, leaving a large underserved pool of children without access to orthodontic care despite their eligibility through Medicaid. In Washington state, approximately 25% of orthodontists participated in Medicaid in 1999, but most treated only a few patients. Only 10 orthodontists provided 81% of the Medicaid-funded orthodontic treatment statewide. In North Carolina, just 10 orthodontists provided 80% of the statewide Medicaid orthodontic treatment.
In 1966, a task force formed through the American Dental Association recommended that treatment of malocclusions should be included as part of covered treatment services for certain people. The next year, 1967, dental care, including orthodontic treatment for handicapping malocclusions, was made available to Medicaid-eligible citizens under 21 years of age. Treatment of these qualified minors is considered “medically necessary.”
In the United States, orthodontic eligibility for Medicaid coverage varies from state to state. A child may be eligible to receive Medicaid-covered orthodontic treatment in one state but not in another. There is no federal standard determining orthodontic eligibility for Medicaid patients. Each state determines the occlusal index to be used and the score required for eligibility.
In Tennessee, the Salzmann index currently is used to quantify the severity of the malocclusion, and a score of at least 28 points typically qualifies a person for Medicaid-covered orthodontic treatment (in addition to verification of financial need). Each prospective Medicaid patient is reviewed individually. The process involves submitting a Salzmann index form, facial and intraoral photographs, and panoramic and cephalometric radiographs. Certain preexisting conditions, such as a craniofacial anomaly, cleft lip or palate, or deep impinging overbite with visible soft tissue laceration, currently are automatically eligible for Tennessee Medicaid-covered orthodontic treatment.
Several studies have aimed at answering the question, “Why are providers reluctant to accept Medicaid patients?” The most common answers are (1) low fee reimbursement, (2) broken appointments, and (3) excessive or complex paperwork. Additional reasons include (1) difficulty in collecting fees from Medicaid, (2) delays in receiving payments, (3) prior authorization requirement, (4) potential for loss of coverage during treatment, (5) low patient compliance, (6) patients missing appointments, (7) patients arriving late to appointments, and (8) patients canceling appointments at the last minute.
Orthodontists are not alone in their reluctance to treat Medicaid patients. Pediatric dentists choose not to participate or limit their participation in Medicaid because of broken or canceled appointments and poor patient compliance. Pediatric Medicaid patients miss significantly more appointments than do non-Medicaid patients.
Arguably, the predominant aspect of orthodontic compliance is meeting appointments. With orthodontic treatment lasting 2 years or more, maintaining appointments is important. In a study at Virginia Commonwealth University, Medicaid patients failed to meet 15% of their appointments, and self-pay patients failed 8% of their appointments, a highly significant statistical (and financial) difference.
However, in a study in North Carolina, researchers surveyed orthodontists and reported that Medicaid and non-Medicaid patients did not differ substantially with respect to effectiveness of treatment or compliance with treatment. Due to limited and conflicting claims, in this study, we evaluated the compliance between self-pay and Medicaid patients receiving comprehensive orthodontic treatment. The null hypothesis was that there is no difference in compliance between self-pay and Medicaid-covered patients receiving comprehensive orthodontic treatment.
Material and methods
Institutional review board approval was granted by the University of Tennessee for this retrospective study of patient records. Medicaid patients (n = 88) were from the postgraduate orthodontic clinic at the University of Tennessee, Health Science Center, in Memphis. The records were hand-searched; we selected recently completed, full-treatment patients (2007-2012), including finished patients in retention. A sample of 145 self-pay control records of adolescents was also analyzed; the one inclusion criterion was a high discrepancy index (DI) (>14) assessed by the American Board of Orthodontics DI method. The DI—like all summary numbers—is an imperfect measure of case difficulty, but it is commonly used and familiar to most orthodontists. The DI was only used to relate samples in terms of treatment difficulty. It played no role in study outcomes. Each patient’s record was perused to collect pertinent information ( Table ). Because the study was retrospective, no standardized measure of oral hygiene could be formulated, although a potential difference here (and in demineralized spots) would be informative. Most patients completed treatment, but those who had to be dismissed from the clinic were eliminated from the analysis as indicated. Start of treatment was defined as the initial placement of appliances (mean age, 14.6 years), not initial records. End of treatment was the final treatment appointment when the appliances were removed (mean age, 18.8 years), not the final records appointment.
|Variable||Self-pay||Medicaid subsidized||P value||Statistic|
|n||% or mean||n||% or mean|
|DI (score) ∗||139||19.6||71||23.2||0.0002||ANOVA|
|Age at start||145||14.9 y||88||14.1 y||0.0831||ANOVA|
|Age at end||145||19.3 y||88||18.0 y||0.0047||ANOVA|
|Treatment duration ∗||139||25.1 mo||71||28.8 mo||0.0012||ANOVA|
|Lost brackets ∗||139||3.2%||71||3.6%||0.4955||ANOVA|
|Broken appliances ∗||139||0.13||71||0.13||0.9900||ANOVA|
|Missed appointments ∗||139||1.0||71||3.7||<0.0001||ANOVA|
|Distance to clinic||145||56.3 km||88||21.6 km||<0.0001||ANOVA|
|Travel time||145||40.4 min||88||17.9 min||<0.0001||ANOVA|
The data were either nominal (eg, boy, girl) or ratio-scale (continuous). Statistically significant differences for the nominal data were tested with Fisher exact tests, and continuous data were assessed with analysis of variance (ANOVA). Corrections were not made for multiple comparisons ; an alpha of 0.05 was preset as the level for significance, and all tests were 2-tailed.
Statistical results are given in the Table . There was no significant difference for starting age between payment methods (self-pay vs Medicaid), but both the durations of treatment and the patients’ chronologic ages at the end of treatment were statistically different. The sex distribution did not differ between the groups because both groups had a preponderance of girls (60%; 140 of 233).
The Medicaid group had a significantly higher DI score because this (and patient age) is the prime measure of eligibility. Patients with low DI scores tended not to be eligible. DI scores of the Medicaid sample had a lower limit of 11 (mean, 23.2; SD, 7.7; range, 11-55). This resulted in a highly significant difference in DI scores because it was impractical to find a self-pay sample with equivalent scores (means, 19.6 for the DI of the self-pay patients vs 23.2 for the Medicaid patients; P <0.0001). This disparity was treated statistically using the patient’s DI as a covariate, but the adjustment had no important effect on the results, so those results are not shown. Also, the DI score was not predictive of treatment duration in either payment group.
Age at the start of treatment did not differ statistically ( P = 0.0831) for patients completing treatment, nor did the number of appointments ( P = 0.2223). The frequencies of patients treated with permanent tooth extractions (ranging from1-5 teeth) did not differ between groups ( P = 0.2201), but the mean numbers of extracted teeth (excluding third molars) were marginally higher ( P = 0.0464) in the Medicaid group (means, 1.8 and 2.3 teeth, respectively), perhaps because of higher DI scores.
Significantly more Medicaid patients were dismissed from treatment before completion, mostly for noncompliance (19% vs 4%). By odds ratio, the Medicaid patients were 5.5 times more likely to be dismissed as the self-pay patients (95% confidence limits, 2.09 and 14.69).
Regarding orthodontic visits, the grand average was 25.0 appointments (SD, 6.6), distributed across an average of 26.5 months in this teaching environment. There was no statistically significant difference in the number of appointments of those completing treatment ( P = 0.2223).
The average number of detached brackets did not differ statistically between the groups ( P = 0.4955). Neither the number of broken brackets (grand mean brackets, 3.5; SD, 4.06) nor the number of broken appliances (grand mean, 0.13 appliances; SD, 0.51) differed significantly by payment method; both were uncommon. The number of broken appliances ranged from none to 5. Treating these data as nominal (no, yes), the frequencies still did not differ significantly, although the incidence was higher in the Medicaid sample (13% vs 8%).
Missed appointments were significantly more common in the Medicaid sample ( P <0.0001). The average number of missed appointments was 1.0; patients missing more than 1 appointment were flagged in the data set. The Medicaid sample exceeded this average 65% of the time vs 24% in the self-pay sample. By odds ratio, the Medicaid patients were 6.0 times more likely to miss an appointment as were the self-pay subjects (confidence limits, 3.14 and 11.11).
Google Maps (Google, Mountain View, Calif) was used to calculate the distance from each patient’s home address to the clinic (in kilometers). This software also estimated commute time (in minutes). The supposition tested was that if few specialists accept Medicaid, then approved patients would have to drive farther to be treated in an accepting practice. There was a highly significant difference for both measures (commute distance and time), but their nature was opposite of our expectations. Self-pay patients (mean, 56.3 km; SD, 3.6 km) drove much farther than did the Medicaid-assisted patients (mean, 21.6 km; SD, 4.6 km), with proportionate differences in commute times (40 vs 18 minutes).