In this cluster-randomized controlled trial, introduction of an EHR-based intervention improved several processes of tobacco treatment in primary care practices. The intervention increased documentation of smoking status, referral of smokers for tobacco treatment, and documented contact of smokers with cessation counseling, an assist
that other studies show roughly doubles the odds of quitting.5–7
The intervention did not increase the prescription of cessation medications. In addition to improving several process measures, there was a suggestion that outcomes of care may have also improved. Among patients who were documented smokers at the start of the intervention period, more were documented nonsmokers by the end of the intervention period in the intervention practices compared with control practices, although this outcome may be simply attributable to improved documentation.
Effective models for implementing the assist
step in tobacco treatment by clinicians is critical because clinicians provide assistance at much lower rates than they document smoking status or provide brief advice.8–12
Increasing rates of assistance have been resistant to intervention, in part owing to multiple competing demands for physicians' time and smokers' resistance to acceptance of counseling.13,25–27
The negligible rate of performance of these actions in our control group reflects that challenge. Our intervention had a strong effect in increasing referral rates, but the absolute rates of our outcomes, even in the intervention group, were modest. Even so, small differences in counseling and quit rates could have a large influence on premature death.2,4
One explanation for the modest absolute effect of the intervention was the low intensity of the introduction and promotion of the intervention, which consisted of a single site visit and periodic e-mails. A more intensive introduction and promotion and the provision of ongoing feedback about the intervention could have resulted in more clinicians using the intervention and an increase in tobacco treatment rates.28,29
The success of our intervention in improving documentation of smoking status and facilitating referral to a counselor suggests that similar interventions could be widely disseminated successfully among practices using an EHR.
Documentation improved in the control practices, consistent with secular trends,12
but documentation increased to a greater extent in intervention practices. In particular, documentation in the intervention practices increased for former smokers and never smokers but not current smokers. Documentation is an important step in tobacco treatment for individual patients and populations. In the United Kingdom, proactively identifying smokers increased use of cessation services and quit attempts.30
Clear, coded documentation, as opposed to free-text entries in visit notes, could facilitate outreach to smokers outside scheduled visits and enable population-based tobacco management within a health care system.31
The failure of our intervention to increase rates of prescription of smoking cessation medications has several potential explanations. First, having referred patients to counseling, clinicians in intervention practices may have thought a medication prescription was unnecessary. Second, our study period coincided with the introduction of a new pharmacotherapeutic agent, varenicline, whose marketing included direct-to-consumer advertising. This may have increased rates of its prescription in the control group. Third, nicotine replacement products can be purchased without a prescription. Clinicians are less likely to enter over-the-counter medications in the EHR than prescription medications.32
Because the 3-part intervention was introduced as a package, we cannot determine the relative value of individual intervention components. For example, the smoking status icon, the reminder, or both might have increased smoking status documentation. Access to the Tobacco Smart Form likely facilitated e-mail referrals to the tobacco counselor and fax referrals to the state Quitline. Clinicians may have preferred accessing the tobacco counselor by e-mail because it was logistically easier than generating and sending a fax. Alternatively, clinicians may have preferred a within-system counselor over the state Quitline.
Other similar interventions have varied in their effectiveness. In an uncontrolled intervention at 10 California Veterans Administration practices, Sherman and colleagues33
used EHR-based referrals to care coordinators, who subsequently attempted to contact and connect smokers with the state Quitline. The number of referrals to the state Quitline increased markedly after implementation, but, as in our study, no change was found in prescribing of medications. In a pre-post study, Szpunar and colleagues34
introduced an EHR-based tobacco counseling system that involved both nurses and physicians. It increased the rate at which patients were asked about tobacco use but not the rate of assistance of patients. Bentz and colleagues29
used EHR data to provide monthly feedback to clinicians in a cluster-randomized controlled trial. The feedback resulted in increases in assessment, advising, and assistance of patients to quit using tobacco. The combination of such retrospective feedback with prospective decision support tools, such as those we implemented, has the potential to further improve tobacco treatment.35,36
This study has limitations that should be considered. First, the study took place in a network of academically affiliated primary care practices using a locally developed EHR. However, the basic form of the intervention—a smoking status icon and e-mail referral capability—is easily generalizable to other practice types and EHRs. Indeed, portions of our intervention have been built into other EHRs, although not as a package and not as rigorously tested.29,33,37,38
Our results could also differ in settings with varied access to different types of tobacco treatment programs. Second, our main outcomes are only proxies for actual smoking cessation, but they have repeatedly been shown to be associated with increased quit rates.5–7
Third, patients could have participated in tobacco treatment programs other than those we assessed. However, there should have been balanced outcomes ascertainment between the intervention and control practices. Fourth, we did not assess the acceptability of the intervention to clinicians or the opportunity costs of the intervention. Despite these limitations, a major strength of our study is that the intervention was made available to all clinicians in randomly selected practices. Participation was not selective by patients, clinicians, or practices with a particular interest in tobacco treatment.
Tobacco use is the leading preventable cause of death in the United States, and tobacco treatment is highlyeffective.1,5
Despite the availability of national tobacco treatment guidelines since 1996, primary care clinicians fail to identify the smoking status of all their patients and provide treatment to few patients because of time constraints, competing demands, lack of expertise, and pessimism about the probability of success.13
Interventions that allow primary care physicians to quickly and efficiently identify and refer patients to tobacco treatment might increase rates of tobacco counseling and increase smoking cessation rates. In a cluster-randomized controlled trial, we found that the use of an EHR-based enhancement increased rates of documentation, referrals, and contact with tobacco treatment. Health care systems and clinicians may be able to provide more efficient, effective tobacco treatment by using health information technology to centralize their tobacco treatment efforts.