Both the theoretical aspects and practical consequences of dental fear have been investigated over the past 50 years [
Studies of dental fear play an important role in predicting behavioural problems in dentally fearful individuals that can result in avoidance or irregular visits [
Generally, dental fear is measured according to cut-off points on validated self-reported scales [
There is considerable variability in the instruments used to determine dental fear and there is no recognised gold standard [
The data for this study were derived from a preexisting field study database [
Potential participants were approached during lecture classes and asked if they were prepared to participate in the study.
The Brazilian version of the DFS was used to collect data [
The two questionnaires were administered in a pilot study of 80 students from the three courses on two occasions, separated by an interval of two weeks. These students did not participate in the main study. The intraclass correlation coefficient (ICC) for test-retest reliability of the DFS for mathematics, dentistry, and psychology undergraduate students was 0.969 (95% CI: 0.945–0.986), 0.968 (95% CI: 0.953–0.980), and 0.949 (95% CI: 0.911–0.977), respectively. Additional information on dental fear was collected using a single question from the second questionnaire “Are you fearful of going to the dentist?” with four response options, “not fearful”; “a little fearful”; “very fearful”; and “extremely fearful.” Cohen’s kappa statistic for test-retest to the question “Are you fearful of going to the dentist?” showed a good degree of agreement (
Dental fear was the main outcome variable; it was indexed by total DFS score and treated as a continuous variable. The distribution of DFS scores was analysed to define two cut-off points using the receiver operating characteristic (ROC) curve, thus creating three dental fear categories: a “not fearful” group, a “low fearfulness” group, and a “highly fearful” group. The independent variables were gender, negative dental experiences in childhood, and the response to the question “Are you fearful of going to the dentist?”
Statistical analysis was conducted using descriptive statistics, the ROC curve, the Kruskal-Wallis test, and multinomial logistic regression; the significance level was set at 5% for all tests. The Statistical Package for the Social Sciences, version 22.0 for Windows (SPSS Inc., Chicago, IL), was used to conduct statistical analysis.
To calculate the ROC curve responses to the question “Are you fearful of going to the dentist?” it was coded as a binary variable (“not fearful” and “fearful”), to distinguish between individuals without fear and those who were fearful. A dichotomisation was performed to define a cut-off point to identify individuals who were highly fearful. Participants who had responded “not fearful” were assigned to the “not fearful” category; all other participants were assigned to the “fearful” category (“a little fearful”; “very fearful”; and “extremely fearful”). Then a different binary categorisation was used: “highly fearful” and “not fearful/low fearfulness.” The “highly fearful” category was made up of participants who indicated that they were “very fearful” or “extremely fearful,” whilst participants who had responded that they were “not fearful” or “a little fearful” were assigned to the “not fearful/low fearfulness” category. This dichotomisation was important for determining the values of DFS scores that identify individuals with dental fear.
The ROC curve was used to determine DFS cut-off points based on self-reported fear of going to the dentist. The ROC curve is a graphical plot of sensitivity against 1 − specificity at various discrimination cut-off points. The best cut-off point is the one that represents the best compromise between sensitivity and specificity.
The normality of the data was assessed using the Kolmogorov-Smirnov test; this test indicated that the data were not normally distributed
Multinomial logistic regression was used to assess whether the independent variables (gender and negative dental experience in childhood) were related to fear of going to the dentist in terms of the three categories defined by DFS scores. Observed values were the numbers in the fear categories (“not fearful,” “low fearfulness,” or “highly fearful”) defined in terms of DFS cut-off points. Predicted values were the numbers in these respective groups of fear predicted by the logistic regression model.
The two self-report questionnaires were completed by 1,256 dentistry, psychology, and mathematics undergraduates, representing a response rate of 80.25%. The age of participants ranged from 18 to 65 years, with a mean of 22.3 years (SD = 5.1); 37.1% were men and 62.9% were women. In response to the question “Are you fearful of going to the dentist?” most participants (71.5%) reported that they had no fear of going to the dentist and only 3.6% indicated that they had a high level of fear (“very fearful” or “extremely fearful”). A statistically significant relationship was found between total DFS score and directly self-reported fear; that is, participants with a lower DFS score were more likely to belong to the “not fearful” group whereas participants with a higher DFS score were more likely to belong to the “highly fearful” group (
Total DFS scores for participants grouped according to response to the direct question “Are you fearful of going to the dentist?” (
Groups of fear | DFS total scores | ||||
---|---|---|---|---|---|
|
Minimum | Maximum | Median | Mean (SD) | |
Not fearful | 895 | 20.0 | 63.0 | 27.0 | 29.4 (8.0) |
A little fearful | 312 | 24.0 | 76.0 | 45.0 | 45.2 (10.4) |
Very fearful | 27 | 45.0 | 100.0 | 76.0 | 75.0 (12.3) |
Extremely fearful | 18 | 48.0 | 89.0 | 62.0 | 63.5 (12.2) |
Kruskal-Wallis test (
SD = Standard Deviation.
Conclusion = not fearful < a little fearful < highly fearful (very fearful and extremely fearful).
DFS cut-off points were based on responses to the question “Are you fearful of going to the dentist?” Two cut-off points were identified from the ROC curve and used to classify DFS respondents into three dental fear groups: “not fearful” (DFS score ≤ 35), “a little fearful” (36 ≤ DFS score ≤ 52) and “highly fearful” (DFS score ≥ 53) (Figure
Determination of DFS cut-off points on the basis of the receiver operating characteristic (ROC) curve for responses to the question “Are you fearful of going to the dentist?” Not fearful (DFS score ≤ 35), a little fearful (36 ≤ DFS score ≤52), and highly fearful (DFS score ≥ 53).
The ROC curve was used to estimate cut-off point scores that reflected the binary “not fearful” and “fearful” classification of responses to the single direct question about fear of going to the dentist. The area under the curve (AUC) was 0.903 (
Determination of DFS cut-off points sensitivity and specificity based on responses to the question “Are you fearful of going to the dentist?”
Cut-off point (total DFS score) |
|
| ||
---|---|---|---|---|
Sensitivity | Specificity | Sensitivity | Specificity | |
20.5 | 1.000 | 0.102 | 1.000 | 0.075 |
25.5 | 0.997 | 0.375 | 1.000 | 0.279 |
30.5 | 0.952 | 0.638 | 1.000 | 0.487 |
|
|
|
1.000 | 0.657 |
40.5 | 0.678 | 0.902 | 1.000 | 0.764 |
45.5 | 0.527 | 0.954 | 0.956 | 0.846 |
50.5 | 0.359 | 0.981 | 0.889 | 0.913 |
|
0.328 | 0.984 |
|
|
55.5 | 0.255 | 0.991 | 0.822 | 0.949 |
60.5 | 0.182 | 0.997 | 0.800 | 0.973 |
65.5 | 0.106 | 1.000 | 0.622 | 0.992 |
70.5 | 0.073 | 1.000 | 0.556 | 0.999 |
75.5 | 0.050 | 1.000 | 0.378 | 0.999 |
A DFS score of 35.5, representing the best compromise between sensitivity (0.81) and specificity (0.81), was chosen as the cut-off point for
Cut-off points are given in bold.
Receiver operating characteristic (ROC) curve for the Dental Fear Survey (DFS). Responses to the question “Are you fearful of going to the dentist?” were assigned to binary categories, “not fearful” or “fearful.” A DFS score of 35.5, representing the best compromise between sensitivity (0.81) and specificity (0.81), was chosen as the cut-off point for “fearful.” Area under the curve (AUC) = 0.903;
Receiver operating characteristic (ROC) curve using the Dental Fear Survey (DFS). Responses to the question “Are you fearful of going to the dentist?” were assigned to binary categories, “not fearful/low fearfulness” (“not fearful” and “a little fearful”) and “highly fearful” (“very fearful” and “extremely fearful”). A DFS score of 52.5, representing the best compromise between sensitivity (0.89) and specificity (0.92), was chosen as the cut-off point for “highly fearful.” Area under the curve (AUC) = 0.977;
A multinomial logistic regression controlling for undergraduate course showed that the two independent variables, gender and negative dental experiences in childhood, were associated with dental fear. High dental fear was more prevalent among females (OR = 1.63; 95% CI = 1.09–2.44). Individuals who had had negative dental experiences in childhood were more likely to report high dental fear in adulthood (OR = 5.09; 95% CI: 3.47–7.44) (Table
Multinomial logistic regression model of independent variables for fear categories based on DFS cut-off points.
Variables | Not fearful/highly fearful | Low fearfulness/highly fearful | ||
---|---|---|---|---|
OR/95% CI |
|
OR/95% CI |
|
|
Intercept |
|
|
||
Gender | 1.63 (1.09–2.44) |
|
1.15 (0.74–1.79) | 0.520 |
Negative dental experience in childhood | 5.09 (3.47–7.44) |
|
2.09 (1.41–3.11) |
|
Model adjusted for undergraduate course.
Reference = high fearful category.
Statistically significant results are given in bold.
OR = Odds Ratio.
CI = Confidence Interval.
The multinomial logistic regression model was used to assign individuals to one of the three DFS groups (“not fearful,” “low fearfulness,” or “highly fearful”) to get an estimated or predicted fear categorisation. The logistic regression model was evaluated by comparing the “observed” and “predicted” DFS categories (Table
Multinomial logistic regression model adjusted for predicting the fearfulness of going to the dentist comparing observed values (responses to direct question) with predicted values (based on DFS scores and cut-off points) (
Observed value | Predicted value | Correct classification (%) | |||
---|---|---|---|---|---|
Not fearful | Low fearfulness | Highly fearful | Total | ||
Not fearful |
|
14 | 24 | 794 | 95.2 |
Low fearfulness | 230 |
|
53 | 293 | 3.4 |
Highly fearful | 108 | 11 |
|
168 | 29.2 |
Total |
|
|
|
1,255 | 64.9 |
69.1% | 28.6% | 38.9% |
Different cut-off points have been used to produce categorical variables from continuous scales and this makes it difficult to compare studies [
Although the range of possible DFS scores is 20 to 100, this study showed that the cut-off point for “highly fearful” was only 53. The high sensitivity (88.9%) and specificity (92.5%) at this cut-off point indicated good agreement between categorisation on the basis of DFS score using this cut-off point and responses to the direct question “Are you fearful of going to the dentist?” In this present study using a cut-off point of 53 provided a reliable classification; a similar cut-off point (55) was defined by Firat et al. [
The multinomial logistic regression model assigned 1,094 individuals to the “not fearful” group; however, the observed responses of 230 (21.0%) and 108 (9.9%) of these individuals placed them in the “low fearfulness” and “highly fearful” categories, respectively (Table
Gender contributed to variance in dental fear. Women had more dental fear than men when the three categories of fear were compared; however when dental fear was coded as a binary variable (“highly fearful”; “low fearfulness”) there was no effect of gender (
The limitations of the study should be taken into account when interpreting the results. Self-report instruments which assess past dental experiences may be subject to recall bias, respondents may differ in how they interpret the various statements about fear, and evaluations may be influenced by the salience of particular dental experiences [
These results are directly relevant to dental research and practice. Previous study with Brazilian undergraduates demonstrated a mean score for DFS of
Dental fear has been recognised as an important problem and is often a barrier to successful dental treatment [
The results suggest that ROC curves can be used to identify cut-off point for dental fear categories based on optimising sensitivity and specificity. A DFS score of 53 represented the best compromise between sensitivity and specificity and was selected as the cut-off point for high dental fear. There was a significant association between responses to a direct question about dental fear “Are you fearful of going to the dentist?” and categories based on DFS score using this cut-off point. Individuals with a DFS score ≥ 53 must be treated as highly fearful and precautions must be taken to avoid behavioural problems during dental treatment.
The authors declare there is no conflict of interests.
This study was supported by the Coordination for the Improvement of Higher Level Education Personnel (CAPES), the National Council for Scientific and Technological Development (CNPq), the State of Minas Gerais Research Foundation (FAPEMIG), and Pro-Reitoria de Pesquisa da Universidade Federal de Minas Gerais (PRPqUFMG), Brazil.