A Web-Assisted Tobacco Quality Improvement Intervention
In this report of patients from 143 community-based dental practices across multiple states, we found that a successful real-world, web-assisted tobacco control quality improvement program resulted in improvement in dental practice provision of brief tobacco use cessation advice. However, this practice improvement intervention was not strong enough to result in differentially higher rates of six-month tobacco use cessation among tobacco users seen at practices participating in the http://www.dtc.cme.uab.edu intervention, compared with usual-care control practices. Our evaluation adds considerably to the literature, extending the possibility to change provider behavior and the challenges of changing patient behavior.
Prior Dental tobacco interventions have had varied affects. A Cochrane review published in 2012 reported of 14 interventions in dental settings, pooled results suggested these interventions can increase tobacco abstinence rates (odds ratio [OR] 1.71, 95% confidence interval [CI] 1.44 to 2.03) at six months or longer, but with notable heterogeneity of effect. Gordon and colleagues provided in-person 5As training in one public health, and compared with one non-randomized control, and found that Patients in the intervention group were more likely to quit than those receiving usual care (15.5 versus 4.3 percent) and after 12 months (18.8 versus 4.6 percent). Attempting to integrate a tobacco Quitline into dental clinics increased tobacco cessation in some, but few dental patients were engaged with the Quitline. The majority of interventions reported in the above Cochrane collaboration review provided more intensive training for dental practices, and thus include a small number of practices (with an mean patients per study of 750). In contrast, our study recruited 143 dental practices and used a pragmatic, technology-assisted approach to training (light touch, high tech). Thus, our study had a greater reach, achieved a significant difference in provider behavior, but was not able to demonstrate an impact on dental patient behavior.
Despite the evidence of tobacco risk and the efficacy of tobacco control interventions, 30% of current smokers report that they have never been advised to stop smoking by a healthcare provider. As discussed, the dental practice provides an opportune time to speak with tobacco users about the dangers associated with tobacco (both smoking and smokeless tobacco), and also provide advice on ways to quit. In their care, dentists and hygienist already provide advice regarding flossing, teeth brushing and oral hygiene, so it is a natural setting for patients to receive advice about another health issue that influences oral hygiene and care. Increasing delivery of tobacco use cessation counseling and related tobacco control practices at point of care is necessary to increasing the impact of dental providers on the behavior of their patients. Although we successfully increased provision of tobacco use cessation advice in intervention dental practices, this improvement was not sufficient to actually change the behavior of the patients exposed to the improvement practices.
The goal of a randomized trial is to demonstrate the causal relationship between an intervention exposure and an outcome of interest. The benefit of randomization is to create balance in measured and unmeasured characteristics in the intervention and control groups, thereby isolating the intervention effect. Cluster-randomized trials are imperfect in their ability to produce balance of measured and unmeasured characteristics at levels below the level of randomization. Despite the large number of dental practices randomized in this study, we found a clear imbalance in patient characteristics between the two groups. Without adjustment, we might have concluded that increasing rates of tobacco use cessation advice at the point of care in dental practices resulted in lower cessation rates, approaching statistical significance. However, this was clearly confounded by the higher self-reported illness in the patients seen in http://www.dtc.cme.uab.edu intervention practices and who agreed to participate in follow-up. This may suggest that our practice-level quality improvement intervention resulted in providers reaching out to sicker tobacco users who may be more difficult to persuade to quit. After adjustment, we appropriately found no difference at all in the intervention or control. Future cluster-randomized trials need to again carefully measure patient characteristics that may be imbalanced and confound results in the trials.
Our http://www.dtc.cme.uab.edu results are important, despite the patient-level p-value, as they highlight the difficulty associated with changing tobacco behavior. While many interventions do exist, we still need innovations. Thus, the results from this study provide useful and cautionary results. Interventions targeted at the dental practice level, may be more effective if paired with patient-level interventions designed to provide continued support after the dental visit. We had a limited number of tobacco users per practice available for follow-up, limiting the precision of our estimates. As noted, our cluster-randomized trial did not achieve balance in patient-level characteristics, lending to somewhat limited generalizability. In addition to our main conclusion that quality improvement interventions like http://www.dtc.cme.uab.edu may be necessary but not sufficient to impact dental patient behavior; our analysis provides an interesting specific example of the potential problems of cluster-randomization.
Discussion
In this report of patients from 143 community-based dental practices across multiple states, we found that a successful real-world, web-assisted tobacco control quality improvement program resulted in improvement in dental practice provision of brief tobacco use cessation advice. However, this practice improvement intervention was not strong enough to result in differentially higher rates of six-month tobacco use cessation among tobacco users seen at practices participating in the http://www.dtc.cme.uab.edu intervention, compared with usual-care control practices. Our evaluation adds considerably to the literature, extending the possibility to change provider behavior and the challenges of changing patient behavior.
Prior Dental tobacco interventions have had varied affects. A Cochrane review published in 2012 reported of 14 interventions in dental settings, pooled results suggested these interventions can increase tobacco abstinence rates (odds ratio [OR] 1.71, 95% confidence interval [CI] 1.44 to 2.03) at six months or longer, but with notable heterogeneity of effect. Gordon and colleagues provided in-person 5As training in one public health, and compared with one non-randomized control, and found that Patients in the intervention group were more likely to quit than those receiving usual care (15.5 versus 4.3 percent) and after 12 months (18.8 versus 4.6 percent). Attempting to integrate a tobacco Quitline into dental clinics increased tobacco cessation in some, but few dental patients were engaged with the Quitline. The majority of interventions reported in the above Cochrane collaboration review provided more intensive training for dental practices, and thus include a small number of practices (with an mean patients per study of 750). In contrast, our study recruited 143 dental practices and used a pragmatic, technology-assisted approach to training (light touch, high tech). Thus, our study had a greater reach, achieved a significant difference in provider behavior, but was not able to demonstrate an impact on dental patient behavior.
Despite the evidence of tobacco risk and the efficacy of tobacco control interventions, 30% of current smokers report that they have never been advised to stop smoking by a healthcare provider. As discussed, the dental practice provides an opportune time to speak with tobacco users about the dangers associated with tobacco (both smoking and smokeless tobacco), and also provide advice on ways to quit. In their care, dentists and hygienist already provide advice regarding flossing, teeth brushing and oral hygiene, so it is a natural setting for patients to receive advice about another health issue that influences oral hygiene and care. Increasing delivery of tobacco use cessation counseling and related tobacco control practices at point of care is necessary to increasing the impact of dental providers on the behavior of their patients. Although we successfully increased provision of tobacco use cessation advice in intervention dental practices, this improvement was not sufficient to actually change the behavior of the patients exposed to the improvement practices.
The goal of a randomized trial is to demonstrate the causal relationship between an intervention exposure and an outcome of interest. The benefit of randomization is to create balance in measured and unmeasured characteristics in the intervention and control groups, thereby isolating the intervention effect. Cluster-randomized trials are imperfect in their ability to produce balance of measured and unmeasured characteristics at levels below the level of randomization. Despite the large number of dental practices randomized in this study, we found a clear imbalance in patient characteristics between the two groups. Without adjustment, we might have concluded that increasing rates of tobacco use cessation advice at the point of care in dental practices resulted in lower cessation rates, approaching statistical significance. However, this was clearly confounded by the higher self-reported illness in the patients seen in http://www.dtc.cme.uab.edu intervention practices and who agreed to participate in follow-up. This may suggest that our practice-level quality improvement intervention resulted in providers reaching out to sicker tobacco users who may be more difficult to persuade to quit. After adjustment, we appropriately found no difference at all in the intervention or control. Future cluster-randomized trials need to again carefully measure patient characteristics that may be imbalanced and confound results in the trials.
Limitations
Our http://www.dtc.cme.uab.edu results are important, despite the patient-level p-value, as they highlight the difficulty associated with changing tobacco behavior. While many interventions do exist, we still need innovations. Thus, the results from this study provide useful and cautionary results. Interventions targeted at the dental practice level, may be more effective if paired with patient-level interventions designed to provide continued support after the dental visit. We had a limited number of tobacco users per practice available for follow-up, limiting the precision of our estimates. As noted, our cluster-randomized trial did not achieve balance in patient-level characteristics, lending to somewhat limited generalizability. In addition to our main conclusion that quality improvement interventions like http://www.dtc.cme.uab.edu may be necessary but not sufficient to impact dental patient behavior; our analysis provides an interesting specific example of the potential problems of cluster-randomization.
SHARE
