Low-Burden Strategies to Promote Smoking Cessation Treatment in Patients with Serious Mental Illness

Abstract

Patients with serious mental illness have high smoking prevalence and early mortality. Inadequate implementation of evidence-based smoking cessation treatment in community mental health centers (CMHCs) contributes to this disparity. This study evaluates the effects of quality improvement strategies on treatment and cessation outcomes. Low-burden strategies of decision support and academic detailing with data-driven feedback were implemented in 4 clinics from 2014 to 2016. Pre- and post-implementation data from pharmacy and medical records were analyzed. With these quality improvement strategies, we found an increase in patient receipt of cessation medication from 4.6% to 18.0% (p<0.0001), and a decrease in smoking prevalence from 57.4% to 54.3% (p=0.0088). This study provides a successful example of a quality improvement approach that holds great potential for increasing the level of smoking cessation care for patients treated in CMHC settings. Decision support and academic detailing with feedback may be effective strategies to promote best practices.

Introduction

Despite the profound health costs of cigarette smoking in serious mental illness patients, and evidence that cessation treatment works in this population (1), mental health providers consistently report that they infrequently (<20%) provide smoking cessation treatments to patients with serious mental illness in Community Mental Health Centers (CMHCs) (2). Contrary to the common perception among health providers (3), patients with serious mental illness are frequently willing (4) and able to quit smoking with appropriate support (1). While evidence exists on comprehensive organizational change models in mental health settings, some strategies are so complex that implementation may be neither scalable nor sustainable (5). Given the limited time and resources in CMHCs, it is vital to identify low burden strategies to overcome barriers and increase the delivery of evidence-based treatment (EBT) for smoking (5).

Major barriers to delivering smoking cessation treatment in CMHCs include provider ‘lack of time’, ‘lack of training’, and ‘assumption of low patient interest’ (2, 3). Using a Consolidated Framework for implementation (6), we hypothesized that decision support tools, shown to save time in primary care, would overcome barriers of time constraint and provider misassumption. To address clinicians’ perceived training deficits, low burden academic detailing (provider training with feedback) has been used in primary care to increase provider implementation of EBT for smoking cessation (7). Using the decision support tool combined with academic detailing and data-driven feedback, we can improve care by gathering patient-reported information to prompt the clinician to deliver treatment based on patient needs.

Given that these low-burden strategies hold promise in primary care settings, we present evidence to assess the effect of these strategies on treatment delivery in the CMHC settings. We evaluated two quality improvement strategies, decision support and academic detailing with data-driven feedback, in four CMHC clinics. We examined whether these strategies increased delivery of EBT for smoking cessation, and whether they were temporally associated with decreased patient smoking prevalence in CMHCs.

Low-Burden Implementation Strategies

A quality improvement initiative was conducted in four CMHC clinics in Missouri to implement evidence-based smoking cessation treatment from mid-2014 to mid-2016. This multi-disciplinary quality improvement team held monthly phone conferences. This protocol was approved by the institutional review board of Washington University/BJC Healthcare. All patients received an annual assessment that included tobacco use. All clinics offered group smoking cessation counseling.

1. Decision support:

Data were collected from patients via a paper/pencil survey in the waiting room in two months (6/2014 and 6/2015) regarding (a) current smoking status, (b) interest in quitting or reducing smoking, and (c) interest in receiving cessation medication and cessation counseling. These data were immediately available for providers to facilitate care during the patients’ appointments.

2. Academic detailing with data-driven feedback:

This included 1-hour annual training and feedback for all providers (psychiatrists and caseworkers). We developed the 5As (Ask, Advise, Assess, Assist, Arrange) training cards from the 2008 Guideline which included talking points and resources tailored for patients in 4 smoking cessation phases (i.e., motivation, cessation, maintenance, relapse recovery). These cards were available in all offices and the clinic intranet site. Training also included medication use (nicotine replacement therapy, varenicline, and bupropion). We provided feedback to all providers regarding patient interest in cessation treatment and treatment rates.

Pharmacy and medical records were used to evaluate the following: 1) monthly smoking cessation pharmacotherapy prescriptions (% patients receiving prescriptions / all patients served by the pharmacy), and 2) annual patient smoking prevalence (% patient smoking / all clinic patients) before, during, and after the quality improvement initiative. We used data routinely collected in these clinics. Current smoking is assessed annually and defined as active use of any tobacco products in the past 90 days (missing data <0.9% in 2014–2016). A homogeneity z test was used to compare proportions of patients receiving treatment and the overall clinic smoking prevalence before and after the intervention.

Results

CMHC patients have high smoking prevalence. The demographics of adult patients: 55.1% female, mean age 45.1 (SD=14.0), 55.1% White, 40.9% Black, and 4.0% other races. The diagnosis distribution are schizophrenia/schizoaffective disorders (33%), mood disorders (63%), post-traumatic stress disorder (PTSD; 11%), and borderline personality disorder (8%) based on the patient profile at these clinics. Amongst adult patients served by the clinics in 2014 (N=3,379), 57.4% (N=1940) of patients were current smokers, 42.6% (N=1439) were not current smokers (either former smokers or never smokers).

These low burden strategies were well received. For the decision support patient survey, 300 patients (25% of all patients who showed for appointments in two implementation months) were offered the survey by the front desk staff and when offered, 100% of patients completed the survey. For academic detailing with feedback, 108 providers (90% of all 120 providers) received academic detailing with feedback.

Low-burden strategies increased provider treatment and decreased smoking prevalence. With these strategies, we found an increase in provider adoption of EBT as indicated by the 4-fold increase in prescriptions for cessation medication (including nicotine patch, nicotine gum, nicotine lozenge, varenicline, but not bupropion due to its multiple indications) issued from 2014 to 2016 (Supplementary Figure). The proportion of patients receiving cessation medication increased from 4.6% to 18.0% (69 out of 1,492 smokers received medication in 2014; 264 out of 1,465 smokers received medication in 2016, z=12.3, p<0.0001) based on data from the in-clinic pharmacies. This increase was observed in both pharmacies (6.0% to 16.0% in pharmacy 1; 0.3% to 24.3% in pharmacy 2). In addition, we found a decrease in smoking prevalence from 57.4% to 54.3% among clinic patients from 2014 to 2016 (N=3,379 in 2014, N=3,088 in 2016, z=2.63, p=0.0088, Supplementary Figure).

Discussion and Conclusions

We implemented two low-burden strategies, decision support and academic detailing with feedback, in four CMHC clinics, and found an increase in patient receipt of evidence-based cessation pharmacotherapy. We found a modest, but significant, decrease in patient smoking prevalence. Similar strategies have been used successfully in primary care settings (7). We find that these low burden strategies could be effective in the CMHC setting. Our findings also suggest the feasibility of tablet-assisted decision support as a strategy in CMHC settings for implementing evidence-based smoking cessation treatment and a potentially useful tool in the future.

These results should be interpreted in the context of several limitations. First, this was a quality improvement project without an experimental design that would enhance internal validity (e.g., using a control condition). This necessarily reduced the strength of inference regarding causality. For instance, a secular trend could have explained these findings. When compared to the 57% smoking prevalence in all Missouri CMHCs based on the state 2014–2016 data, the post-intervention 2016 smoking prevalence of 54.4% in these clinics is relatively lower. This suggests that the modest decrease in smoking prevalence is unlikely driven by the secular trend. Future research should use an experimental design that permits greater strength of inference, controlling for personal factors and temporal influences. Second, smoking in this study was patient-reported or provider-documented without biochemical confirmation. Past data on smoking was unavailable (e.g., data before 2014 had more than 1/3 missing values, consistent with evidence that smoking status was not consistently assessed by mental health specialists (2, 3). Third, we used data from in-house pharmacies which captured the majority of patients (77% in 2014 and 88% in 2016) in these clinics. We had no access to past data or outside pharmacy data. Fourth, this is one of multiple ongoing quality improvement initiatives with limited existing clinic resources and variable level of implementation across staff and clinics. We present only suggestive evidence and suggest a future randomized controlled trial to determine the efficacy of these intervention strategies. For example, the effect and optimal frequency of academic detailing needs to be further examined, given existing evidence on lack of smoking cessation education among health professionals (8, 9). Finally, the smoking prevalence in this population is much higher than the general population. This disparity identifies a great need for implementation research in this vulnerable population.

In summary, these results suggest that two relatively low-burden strategies, decision support and academic detailing with feedback, hold great potential for increasing the provision of smoking cessation treatment to patients with serious mental illness, and increasing their likelihood of quitting as well. Given the consistent high smoking prevalence in the population with serious mental illness in existing evidence (10), future research should explore the effects of such strategies to decrease the prevalence of smoking and its profound health consequences.