There has been a debate in and outside the dental profession about the appropriate time for a child’s first dental visit.¹⁴ Regardless, there are relatively few general dentists who have children enter their practices for regular care before the age of three years.¹⁵ General dentists frequently do not treat young children if they "have an available reference source."¹⁵ In 1999, 39.3 percent of children aged 2 to 4 years had a dental visit¹⁶; thus, it is imperative that young children be able to find dental treatment.

Dentists’ referral patterns can have a direct impact on dental work force issues and access to care. Thus, it is necessary to identify which variables are associated with referral patterns. The purpose of our study was to determine which dentist and practice variables predict general dentists’ referrals to pediatric dentists of children aged 3 to 5 years—an age range that corresponds with an increased utilization rate—and to describe which patient characteristics are associated with the referral of these preschool-aged children.

	SUBJECTS AND METHODS

TOP ABSTRACT SUBJECTS AND METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

 
We developed a 25-item survey to determine which variables were associated with the likelihood of making referrals. We pilot tested the survey for content and organization. The survey emphasized two domains—practice characteristics and dentists’ characteristics. Practice characteristics included items such as the total percentage of children in the practice, while dentists’ characteristics included items such as perceived adequate exposure to children in dental school. The survey also included queries about patient characteristics.

We obtained a list of all Iowa licensed private practice general dentists (N = 1,089) from the Iowa Dentist Tracking System (IDTS), a comprehensive statewide database that monitors dental practices on an ongoing basis. IDTS staff members work with the state dental board to track information pertaining to dentists’ work environments and basic demographic information. One of the primary foci of our research was to detect differences between male and female general dentists’ referral patterns. Therefore, on the basis of a 35 percent expected response rate, we calculated the sample size needed to detect a 10 percent difference with P < .05 and an 80 percent power. Owing to the small number of female general dentists (117), we sent the survey to all 1,089 Iowa general dentists after obtaining approval from The University of Iowa’s institutional review board. We sent the initial mailing in November 2003, and we sent a second mailing one month later. We tested nonresponder bias using the following variables from the IDTS: number of hours worked per week, practice type, alma mater, year of graduation, age of the dentist and sex of the dentist.

We collected referral pattern information for patients in three age groups: younger than 3 years, 3 to 5 years of age and 6 to 14 years of age. Based on anecdotal information about when general dentists may initially treat children, we chose ages 3 and 6 years as the break points among groups.

The main survey question was "In the past 12 months, when children came to your practice requesting care, how likely were you to refer those children (to a pediatric dentist) for care?" Respondents who had any children in their practices were asked to choose their responses from a Likert-type scale that included the choices "never," "sometimes," "often" and "always." To determine which dentists usually referred children, we dichotomized the responses into "never/sometimes" versus "often/always" for bivariate and logistic regression analyses. We conducted separate analyses for children younger than 3 years and children 3 to 5 years of age because we hypothesized that the predictor variables associated with the referral of children would differ by age group. Additionally, we collapsed the total percentage of patients with public insurance (Medicaid) into 0 to 5 percent versus greater than 5 percent groups. We chose this split to determine if the referral patterns of general dentists who never or rarely accepted patients with public insurance differed from those of general dentists who accepted more patients with public insurance.

We double-entered data into a database in which we analyzed them using statistical software (SAS 9.0, SAS Institute, Cary, N.C.). Descriptive statistics were reported as means and frequencies. We used nonparametric tests (² and Mann-Whitney) for bivariate analyses because the data were not distributed normally. We used a generalized logistic regression analysis to explain which variables were most likely to predict the likelihood of general dentists’ referring children aged 3 to 5 years. Before entry into the logistic regression models, we examined associations among predictor variables to test for collinearity. We tested interactions in the final logistic model.

Dentists who often/always referred children aged 3 to 5 years had a smaller mean percentage of children in their practices compared with dentists who never/sometimes referred children aged 3 to 5 years.

We analyzed the logistic models using stepwise selection. We set inclusion and exclusion criteria for the stepwise procedure at P .1 for those variables that were significant at the P .2 level from the bivariate analyses. Although we found that distance was not significant in the bivariate analysis, we included distance in the logistic model owing to its significance in the literature.^2,10–12 We calculated adjusted odds ratios for variables that we found to be statistically significant (P < .05) in the final model.

	RESULTS

TOP ABSTRACT SUBJECTS AND METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

 
Of the 1,089 surveys we mailed, 104 women and 608 men returned useable surveys, resulting in an adjusted response rate of 65.4 percent. Table 1 shows the responders’ practice and demographic characteristics by sex. Nonresponder bias testing showed that there were no significant differences between responders and nonresponders in ownership status, mean number of hours worked, age or years since graduation. Responders, however, were more likely to be graduates of The University of Iowa College of Dentistry (P < .01).

To better understand why dentists refer children, we asked the dentists to indicate the frequency with which they would refer children of any age (0–14 years) given a list of patient characteristics (Table 3, page 658). We found that 74.3 percent of the dentists who reported that they never/sometimes refer children aged 3 to 5 years stated that they would refer uncooperative patients, whereas 93.1 percent of dentists who reported that they often/always refer children aged 3 to 5 years stated the same. Severe caries/extensive treatment needs was the next most frequent reason for referral, followed by the child’s having a special health care need. Dentists were more likely to refer children with public insurance than children with private or no insurance. Dentists who often/always referred children aged 3 to 5 years referred 33.3 percent of emergency patients, whereas dentists who never/sometimes referred children aged 3 to 5 years were more likely to retain emergency patients in their practices. In general, dentists who often/always referred children aged 3 to 5 years were more likely to refer for each patient type or situation than were dentists who never/sometimes referred children aged 3 to 5 years.

	DISCUSSION

TOP ABSTRACT SUBJECTS AND METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

Klooz and Lewis² found that Canadian dentists who had higher percentages of children in their practices were less likely to refer children aged 0 to 4 years and 5 to 9 years. They speculated that dentists who voluntarily had high percentages of children in their practices presumably wanted to treat children in their practices.² Our analysis also suggests that as the percentage of children increased in a practice, dentists were less likely to often/always refer children aged 3 to 5 years. It is necessary to consider the percentage of children in the practice to understand the potential effect. Our findings corroborate those of others who found that the mean percentage of children in U.S. dental practices was nearly 20 percent.¹⁵ However, more than 60 percent of dentists in the United States and in Iowa had pediatric patient percentages of less than 20 percent.¹⁵ Thus, the possibility exists for large discrepancies to occur in dentists’ likelihood of making referrals. The percentage of children in a practice may be acting as a proxy for other variables such as whether dentists like children and feel comfortable treating them. Additional studies may better delineate the specific variables that influence the total percentage of children in a practice and how those variables influence referral patterns.

A U.S. survey found that fewer than 50 percent of general dentists see Medicaid-insured children aged 4 to 15 years in their practices.¹⁵ Similarly, approximately 40 percent of Iowa dentists treat publicly insured patients (adults and children) in their practices. In our study, these dentists were more likely to often/always refer children aged 3 to 5 years despite having high mean percentages of children in their practices. While dentists with higher percentages of children in their practices were less likely to refer the children to a pediatric dentist, dentists with more children receiving public assistance were more likely to refer children.

The literature is contradictory regarding whether advanced training beyond dental school influences the likelihood of referring patients. Atchison and colleagues⁴ reported that referrals differed by procedure. For example, AEGD-trained dentists were more likely to refer patients who needed stainless steel crowns than were general dentists and GPR-trained dentists.⁴ In contrast, Seale and Casamassimo¹⁵ found that the percentage of children in practices did not differ among dentists who had received additional training compared with those who had not. Our study demonstrated that dentists who received additional training were more likely to often/always refer children aged 3 to 5 years. Perhaps advanced training programs are not providing adequate exposure to young children. Another possibility is that AEGD-/GPR-trained dentists prefer to provide more complex treatment to adult patients. Studies should be conducted to provide a better understanding of advanced training programs’ impact on the treatment and referral of children.

We found numerous differences between male and female general dentists. Bivariate analysis indicated that female dentists had higher mean percentages of children in their practices than did male dentists. Furthermore, men were more likely to often/always refer children aged 3 to 5 years than were women. This suggests that female general dentists may have a higher capacity for accepting children in their practices. Despite the statistically significant difference in the likelihood of referring children aged 3 to 5 years found through the bivariate analysis, no sex differences emerged in the multivariable analyses. As more women enter the work force, referral patterns among general dentists for children aged 3 to 5 years may stay the same. Consequently, the number of children who receive care from general dentists may increase as more women enter the profession, thus minimizing the effect of the pediatric dentistry work force shortage.

Nearly 30 percent of general dentists reported that they would refer children aged 3 to 5 years more often if a pediatric dentist were closer to their practice. Distance, however, was not a significant predictor for the likelihood of referring children aged 3 to 5 years. This contrasts with other studies that have demonstrated that the further a dentist is located from a specialist, the less likely the dentist is to refer patients to the specialist.^2,10–12 Since our study included variables that were not considered in the other studies, the effect of distance on the likelihood of referring children aged 3 to 5 years may have been masked by the strong associations of the other variables. The finding that distance was not a significant predictor, even when forced into the model, suggests that dentists will refer children regardless of the potential barrier associated with distance. Thus, policy-makers may be able to place less emphasis on the geographic maldistribution of general dentists when estimating the proper general dentist-to-pediatric dentist ratio.

We found that various patient characteristics were associated with the likelihood of referring children aged 3 to 5 years. The finding that dentists who never/sometimes refer children aged 3 to 5 years were more likely to often/always refer children with public insurance than children with private or no insurance is consistent with the literature. Seale and Casamassimo¹⁵ found that more than 50 percent of dentists reported that they never treat children with Medicaid coverage. This is troubling since a national study of dental schools found that 88 percent of pediatric patients treated in dental schools were from low-income families.¹⁷ This suggests that variables beyond predoctoral experiences need to be considered when determining how to increase general dentists’ likelihood of treating children receiving public insurance.

Unlike other dental specialties in which general dentists refer patients based on the complexity of the case, referrals to pediatric dentists also relate to patient management issues. A national study found that dental students are not being trained properly to treat challenging children. In general, predoctoral students mainly treat well-behaved children with minimal restorative needs; thus, general dentists may not feel competent treating more difficult patients in their practices.¹⁷ Our study showed that dentists who perceived that they had received adequate predoctoral exposure to children aged 3 to 5 years were less likely to often/always refer children. In contrast, dentists who received additional training beyond dental school were more likely to refer children. To increase the number of general dentists who are comfortable treating more challenging patients, efforts should be made at the predoctoral level to increase the types and numbers of experiences that students receive with preschool-aged children.

There are limitations to our study. The definition of "referral" was not specified on the survey. As a result, "referral" was dependent on the dentists’ interpretation. Thus, the timing and extent of the dentists’ involvement with the referral process is unknown. It appears as though other studies also have left the definition of referral up to the study participants’ and the readers’ interpretation, since "referral" has not been defined explicitly in the literature.^{2,4,5,10–12} While the lack of a definition for "referral" is a limitation among all of the studies, our study provides information pertaining to the referral of children aged 3 to 5 that is lacking in the literature (Tables 3 and 4).

There also were limitations associated with the measurements used in our study. To increase survey response, we used the qualitative measurements of "never," "sometimes," "often" or "always" to measure the dependent variable "likelihood of referring" rather than more precise quantitative measurements. Since we left these categories open to each dentist’s interpretation, the perceived definitions of these categories may have differed among dentists. Because we dichotomized the dependent variable for the analyses, we may have lost specific details associated with each level of referral. Our study, however, had a high level of return, which lends it credibility.

While our study revealed associations among the statistically significant variables and the likelihood of making referrals, it could not show causality because it was a cross-sectional study. The paucity of research in this area allows for the exploration of other reasons for dentists’ referrals. Future studies should include both subjective and objective variables, as well as specific clinical situations. If additional studies confirm these findings, then more confidence can be placed in developing policies pertaining to pre- and postdoctoral educational experiences and work force needs.

	CONCLUSIONS

TOP ABSTRACT SUBJECTS AND METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

 
The American Academy of Pediatric Dentistry suggests that general dentists have an ethical obligation to ensure that children can obtain dental care.¹⁸ Our study suggests that efforts should be made to ensure that general dentists have access to pediatric dentists to facilitate children’s acquisition of care and to help guarantee that the care they receive is competent. Dialogue should take place between general and pediatric dentists to ensure that expectations are uniform regarding the appropriateness of referrals. Policy-makers and educators should use this information to determine whether there is a proper ratio of pediatric to general dentists and whether these ratios should change based on the changing demographic profiles of general dentists. Efforts also should be made to increase general dentists’ willingness to treat preschool-aged children and thus reduce the strain placed on the pediatric dentistry work force.