However, dentists may be less likely than other health care providers to address domestic violence within their role as health professionals.⁴ While universal screening for domestic violence may not be warranted, dentists must be able to recognize and respond appropriately to signs of abuse, a goal of the American Dental Association.⁵ Unfortunately, research shows that few dentists are knowledgeable about the relationship between head and neck injuries and domestic violence.⁶ A national survey of dentists found that only 39 percent of dentists responded that they often or always screened for domestic violence when a patient had visible signs of trauma on the head or neck.⁷

Many health care providers perceive the process of asking questions and intervening with victims of domestic violence to be difficult and complex, with the potential of opening Pandora’s box.⁸ Based on these concerns and on our own research about how health care professionals have been able to help survivors of domestic violence,^9,10 we developed an approach that simplifies health care professionals’ role in addressing domestic violence.¹¹

– asking patients about abuse;

– providing validating messages that acknowledge that battering is wrong while confirming the patient’s worth;

– documenting signs, symptoms and disclosures in writing and with photographs;

– referring victims to domestic violence specialists in the community.

Highly suitable for encounters with dental patients, the AVDR approach offers a simplified method of helping patients without imposing a requirement that dentists solve the problem of abuse, an unrealistic goal.¹⁰

Advances in technology have made the dissemination of multimedia educational programs highly efficient, and computer-assisted instruction has been shown to be an effective method of medical education. A recent study found that a two-hour, case-based educational program for physicians delivered over the Internet improved physicians’ confidence and beliefs in treating domestic violence patients as effectively as a more intensive, classroom-based approach.¹² To date, there have been no reports of assessments of any domestic violence training efforts designed specifically for the dental profession.⁴

Advances in technology have made the dissemination of multimedia educational programs highly efficient.

	METHODS

TOP ABSTRACT METHODS RESULTS DISCUSSION CONCLUSION REFERENCES

 
Development of AVDR tutorial. To help dentists learn the AVDR approach, we created an interactive, multimedia tutorial tailored to dental professionals that, if shown to be effective, would be easily reproducible as a compact disc or widely accessible from an Internet server. The tutorial depicts an interaction between a dentist and a patient who displays signs of battering (tooth mobility and laceration of the labial mucosa) at her dental visit.

A recent review of medical and nonmedical educational programs found such person-to-person interactions to be a promising use of virtual reality computer programs.¹³ Programs that have both video and audio components are useful because they require little text reading. Further advantages of this approach include interactivity, visualization and feedback through the learning experience.¹⁴ To develop the script, we created a model of battered women’s positive experiences with the health care system and health care providers’ reports of their successful strategies for helping battered women, which were gathered during previous research.^9,15

Two skilled actors appear in the tutorial; one is a real-life dentist who portrays a sage, or expert, and a second actor portrays a dental patient. The subject is asked to interact with this virtual patient by choosing questions and statements to deliver. These inputs elicit a variety of reactions from the patient. The sage then follows up with commentary and guidance about the simulated dentist-patient interaction. Additional teaching tools include a graphic display of principal messages and changes of color backgrounds, as the tutorial proceeds through the four stages of the AVDR lesson. The tutorial takes about 15 minutes to complete.

Study design. We conducted a controlled, randomized trial using three study groups in a modified Solomon four-group design to examine the impact of the tutorial, as well as the pretest.¹⁶ A pretest and posttest experimental group took a pretest, followed by the tutorial and then a posttest; a posttest-only group first took the tutorial and then the posttest; and a two-test control group took a pretest, followed immediately by the posttest. This control group received the tutorial after the posttest. (A posttest-only group with no pretest and no intervention was excluded because its addition would serve no apparent utility.)

Hypothesis. We hypothesized that, at the posttest, subjects in the two experimental groups would report having improved knowledge, attitudes and intentions to behave appropriately regarding assessment for domestic violence and intervention with victims than would subjects in the control group. We also hypothesized that any differences between the pretest and posttest experimental group and the posttest-only experimental group (which might occur because of a testing or learning effect) would be small.

Subject recruitment. We recruited dental students with clinical experience and faculty members from the University of California San Francisco, or UCSF, and the University of the Pacific, or UOP, dental schools to participate in the study. At UCSF, a four-year dental school, third- and fourth-year students were recruited, while at UOP, a three-year dental school, second-and third-year students were recruited.

The tutorial was available to students and faculty members in cafeterias and common rooms during lunchtime and breaks; it also was available in a family dental clinic, a site for rotating fourth-year students. We posted signs and distributed flyers inviting students and faculty members to participate when they had an available half-hour. Participation required between 15 and 25 minutes, and students received a $6 gift certificate for participating. Subjects were randomized to one of the three study groups by the computer. We obtained signed informed consent according to a study protocol approved by the UCSF Committee on Human Research.

Domestic Violence Assessment Instrument. We developed a test that included a series of 24 questions delivered via the computer. All subjects received the questions in the same order. The posttest repeated the 24 questions, with some differences in the order of the questions and, for some questions, a reversal of the direction from positive to negative. The box shows the the exact wording of the 16 questions about which we report, as well as the rating scales. (We did not report the results for eight of the questions, because they examined a range of attitudes and beliefs about the subject area that were outside our main area of focus.) In addition to the 24 questions, the computerized assessment included several questions that enabled us to determine a demographic profile of the three study groups.

Statistical analysis. We used polytomous logistic (generalized logit) regression to assess balance (distribution) of characteristics at baseline among all three groups globally (that is, simultaneously rather than each characteristic assessed individually) with a multiple degrees of freedom, or df, score statistic (race/ethnicity, sex, generalist/specialist, number of hours of prior domestic violence training and domestic violence training venue). We used logistic regression to assess baseline balance between the two experimental groups and the control group globally with a multiple df score statistic (the factors listed above plus the number of annual domestic violence consultations and 24 baseline questions about knowledge, beliefs and attitudes).

To account for multiple endpoints and multiple groups, we assessed the efficacy of the training intervention with resampling-based stepdown bootstrap multiple testing¹⁷ (1 million resamples) of means of the 24 knowledge, belief and attitude questions, both among all three groups (two experimental groups versus one control group) for posttest values, as well as between the pretest and posttest experimental group and the two-test control group. Furthermore, we tested interactions between the intervention and other factors (racial/ethnic group, sex, student status and prior domestic violence training) with interaction contrasts (to compare group x factor means) (for example, to compare if the pretest-to-posttest change between the intervention groups and the control group differed between men and women for the six treatment x sex strata).

	RESULTS

TOP ABSTRACT METHODS RESULTS DISCUSSION CONCLUSION REFERENCES

 
Of 174 subjects, 161 (93 percent) were students, 119 (68 percent) were affiliated with UCSF, 92 (53 percent) were male, 90 (52 percent) were Asian/Pacific Islanders and 51 (29 percent) were white. The computerized randomization assigned 56 subjects to the pretest and posttest experimental group 59 subjects to the posttest-only experimental group and 59 subjects to the two-test control group. There were no statistical differences between the groups in regard to the demographic variables (including sex, race and age) or baseline measures (three-group score statistic = 29.9, 34 df, P = .668; two-group score statistic = 45.9, 48 df, P = .315).

After the tutorial, both experimental groups demonstrated significantly improved attitudes and knowledge regarding the need to ask patients about domestic violence.

The table shows all study outcomes. For each item, mean scores and confidence intervals are shown for the pretest and posttest for the two-test control group, the posttest only for the first experimental group, and the pretest and posttest for the second experimental group. After the tutorial, both experimental groups demonstrated significantly improved attitudes and knowledge regarding the need to ask patients about domestic violence and validate patients’ worth (two of the four AVDR stages) compared with the control group. Similarly, the experimental groups experienced significant improvements in attitudes and knowledge about the need to document and refer patients, the other two stages of the AVDR model. The two experimental groups did not differ significantly from each other, suggesting that although testing may have familiarized subjects with some of the intervention concepts, it did not significantly improve subjects’ overall responses.

Four additional items assessed subjects’ attitudes about other related aspects of domestic violence. For two of these items (that is, within dentist’s role to ask and helping in terms of difficulty), the pretest and posttest experimental group scored significantly higher than did the two-test control group. For the item that asked about the difficulty of helping patients who are victims of domestic violence (1 = difficult to 7 = easy), the mean score increased by 1.6 points (from 2.6 to 4.2) for the experimental group versus only 0.1 points (from 2.7 to 2.8) for the control group. However, only two (25 percent) of these eight attitude posttest scores for both of the experimental groups were significantly different from scores for the control group.

Overall, at least one of the experimental groups had a mean posttest score that was significantly higher than the control group’s mean posttest score on 14 (88 percent) of the 16 assessment items (adjusting for multiple comparisons).

We found no significant differences when testing for intervention effects by sex, faculty members versus students, school (UOP versus UCSF) and having had previous domestic violence training (any versus none).

	DISCUSSION

TOP ABSTRACT METHODS RESULTS DISCUSSION CONCLUSION REFERENCES

 
We set out to develop a program that would help dentists recognize and respond to signs of domestic violence. Our study results strongly suggest that the program achieved this goal. Results also confirmed both hypotheses: the tutorial would improve subjects’ scores, and the observed improvements could not be attributed to a testing effect.

The first step in the process of addressing domestic violence—asking patients about possible abuse—is often thought to be the most difficult step for health care providers. After completing the tutorial, most subjects stated they intended to frequently inquire about patients’ safety after recognizing injuries to the head or neck—a large improvement from the typical preintervention response (which was "seldom"). This change alone could greatly increase the number of victims of domestic violence who are identified at their dental visits.

We also noted that improvements in the asking and validating domains were greater than improvements in the documenting and referring domains. In a recent survey of primary care physicians, Gerbert and colleagues¹⁸ found that once abuse is identified, providers are quite likely to follow up with counseling or referrals (70 to 80 percent would often or always do so). The present study also found high mean pretest scores (3.9 and 4.1) for intentions to provide referrals when one suspects a patient is being abused. Perhaps asking and validating are viewed as more optional and, therefore, are more malleable to change than are documenting and referring once a situation of domestic violence has been identified.

The results of our study also showed significant improvements in attitudes and beliefs about domestic violence screening and intervention. Although this learning experience was quite brief (15 minutes) and although domestic violence is a sensitive, complex issue, the results demonstrate that significant improvements may be achieved by using an engaging, interactive, computer program to present a model that simplifies the provider’s role.

An important strength of the study was our use of a modified version of the Solomon four-group design to examine whether any interaction between pretesting and the tutorial occurred. Also, randomization achieved balance within the study groups and would tend to balance even unmeasured baseline factors (such as actual clinical experiences with domestic violence).

One limitation of the study is that the posttest immediately followed the tutorial, so we cannot infer a lasting effect of the tutorial. Another limitation is that we did not measure actual behaviors or health outcomes. A third limitation is that we used a convenience sample of dental students and educators who were willing to take the tutorial for a small reimbursement. Although the analytic method presented assumes equal spacing between the ordinal categories (that is, change between any two consecutive categories is the same), ² trend analyses (maintaining the ordinal responses) yielded nearly identical results.

	CONCLUSION

TOP ABSTRACT METHODS RESULTS DISCUSSION CONCLUSION REFERENCES

 
As more dentists and other health care providers receive education about domestic violence, it will be important to conduct research to determine whether education leads to actual changes in screening, intervention and other behaviors. We plan to administer the tutorial to a sample of practicing dentists, and to compare their responses with the responses of dental students reported here. Following additional evidence of effectiveness, we hope to make the program available to all dental students and practicing dentists.