Checking Smoking Status as a Routine Vital Sign: A Cluster-Randomized Trial of Its Effect on Cessation Counseling

Two national clinical practice guidelines (CPGs) have advised primary care physicians to implement an office-wide system for screening and recording tobacco-use status for every patient at every office visit. These guidelines specifically recommended checking smoking status as a vital sign. This tracking system is elegantly simple: As a prompt for the physician, the nursing staff records the patient's response to a brief inquiry of their smoking status (current, former, never) with the other vital sign data.

Scientific evidence that this measure increases physician counseling or smoking cessation is limited. The proposed study evaluates the effectiveness of the vital sign intervention through a practice-level randomized controlled trial.

18 practices will be recruited for the 9 month study. Eligible practices will include primary care practices (family practice or internal medicine) with an average of two or more full-time equivalent providers (including physician assistants and nurse practitioners), not already using (or planning to institute during the study period) a practice-wide tobacco identification and reminder system, and willing to participate in the study.

Data collection will occur in two phases; baseline and post intervention. The purpose of baseline data collection is to provide a baseline counseling rate for each office prior to randomization.

Over the course of both data collection periods, research assistants (RAs) will visit individual practices on a rotating basis (about 16 visits per site on average). On average, practices will be visited once a week during the baseline month and twice a month during the intervention phase. The frequency of visits will be adjusted in an ongoing fashion by the PI, as practices with a smaller number of providers, a lower prevalence of smokers, or a lower survey participation rate will necessitate proportionally more visits to keep data collection balanced at each site.

The RA will recruit patients following their interaction with the provider to participate in a one page in-office survey. Upon returning the in-office survey, self-identified adult current smokers who visited a physician that day will be recruited to participate in a 4 page mail-back survey.

Both surveys are anonymous. A matching key code will be recorded on the returned in-office survey for those patients given a mail-back survey to allow linked analysis, while maintaining complete patient anonymity.

Following the baseline data collection, practices will be matched in clusters (by baseline cessation counseling rate) and randomized within each cluster to intervention and control.

In the intervention practices, the research coordinator will work with the office manager to implement the most appropriate method of recording the vital sign for that practice. To facilitate timely implementation, the research coordinator will begin these discussions with all the practices early in the study so that it can be implemented smoothly once the randomization is determined. Depending on how the practice currently records routine vital signs and documents visits in the chart, smoking as vital sign could be implemented with a stamp on the visit note, modification to forms and/or progress note paper, or modification to a vital sign flow sheet.

Immediately prior to implementation of the smoking vital sign in the 9 intervention practices, the research coordinator will conduct an in-service training session with the practices' nursing staff. For the six months intervention period in the intervention practices, the nursing staff will record the patient's response to a brief inquiry of their smoking status (current, former, never) with the other vital sign data. This information is to be assessed routinely at each adult patient visit to the practice.

With the exception of nursing staff assessing and recording the smoking vital sign, all other work of data collection will be done by VCU research staff. No intervention with providers will be done.

During the six months intervention period and only in the intervention sites, the office or nursing manager will be provided with periodic reports on how often patients' report having been asked whether they smoke (a measure of implementation of the intervention) and asked to share this with the nursing staff through the practice's usual QA/QI mechanisms to reinforce continued performance of the intervention.

Taking into account the clustered design, statistical significance for the primary and secondary hypotheses will be tested by hierarchical logistic regression using the SAS and Sudaan statistical software. The mean rates for intervention practices will be compared to the mean rates for control practices. Temporal trends for patient report of intervention performance (having been asked their smoking status) and of the rates of cessation counseling for intervention and control practices will be portrayed in time-series graphs. A repeated-measures design will be incorporated into the categorical models to test the statistical significance of temporal trends. Secondary data analysis will examine possible influences of visit and smoking-related factors, provider, patient demographics, and health status by adding these covariables to the categorical models.

Resolving scientific uncertainty about the vital sign's effectiveness is important. Tobacco remains the leading cause of death and disability in this country. Few practices have adopted the vital sign or any other type of tobacco tracking or reminder system. Evidence that this intervention increases physician counseling with smokers would therefore argue for a change in current practice patterns. If the vital sign intervention is not effective, then expectations that practices adopt it (as recommended in the CPG) may be misguided. Increasingly, primary care physicians are being encouraged to modify their practices to incorporate quality-of-care interventions that lack scientific evidence of effectiveness. Using randomized controlled trials to test the effectiveness of guidelines, including those that are evidence-based, is necessary to insure that they are relevant and useful when applied in typical primary care practices.