Intended for healthcare professionals

Practice Quality Improvement Report

A framework for tobacco control: lessons learnt from Veterans Health Administration

BMJ 2008; 336 doi: (Published 01 May 2008) Cite this as: BMJ 2008;336:1016
  1. Scott E Sherman, associate professor of medicine1
  1. 1VA New York Harbor Healthcare System/New York University School of Medicine, New York, NY 10010, USA
  1. Correspondence to: S E Sherman, VA New York Harbor Healthcare System (111), 423 East 23rd Street, New York, NY 10010scott.sherman{at}
  • Accepted 7 February 2008

Despite considerable progress, smoking remains the leading preventable cause of death in the United States, responsible each year for 435 000 deaths1 and $157bn (£79bn; €103bn) in health related losses.2 Each pack of 20 cigarettes leads to $3.45 in medical expenditures and $3.73 in lost productivity.2 When Maciosek et al recently prioritised 25 preventive interventions, factoring in burden of disease and cost effectiveness, tobacco control was tied for the top priority, and better screening followed by brief intervention yielded a greater benefit in quality adjusted life years than the next 10 interventions combined.3 Similarly, the Institute of Medicine recently identified improving treatment for tobacco use as one of the top 20 healthcare priorities.4 5

No clear guidance exists for improving the performance of a healthcare system to the goals set by Maciosek.3 The Public Health Service guidelines suggest that providers focus on five steps, outlined in figure 1,6 and these are also relevant for healthcare systems. Systems that more consistently deliver these “5 A’s” to patients are providing better quality care for smoking cessation. The guidelines also include recommendations for healthcare systems (every clinic should implement a system to identify tobacco users, for example), but these are individual items rather than an approach to improve organisational performance. The task for healthcare systems is how to increase delivery of the five steps.


Fig 1 “Five A” model guiding providers’ behaviour with smokers

In this article I use the VHA’s experience with large scale quality improvement and system change in tobacco control to create a framework for how organisations approach tobacco control, moving through four increasingly complex stages, from no system for tobacco control to a system that provides tailored smoking cessation care to the entire population. By anticipating these stages and approaching the challenges of each in sequence, systems may be able to set goals for quality improvement appropriate to their levels of readiness.


  • Problem Existing clinical guidelines provide recommendations for treating smokers but little guidance on how to implement changes that will improve the delivery of smoking cessation care.

  • Design Extrapolation from experience in improving the quality of smoking cessation care in the Veterans Health Administration (VHA) to propose a conceptual framework describing the stages for improving the reach and effectiveness of organisations’ tobacco control efforts.

  • Setting United States Department of Veterans Affairs healthcare system.

  • Key measures for improvement Improved rates of screening patients for smoking status, advising smokers to quit, and helping smokers with a quit attempt.

  • Strategies for change The framework identifies four stages of organisational change during quality improvement for smoking cessation. In stage 1, no system exists to structure tobacco cessation care. Stage 2 is characterised by a systematic approach to asking about smoking and advising tobacco users to quit. In stage 3 there is an organised approach to helping tobacco users quit at their healthcare visits. The final stage focuses on providing tailored help for smoking cessation to the entire population within a healthcare system.

  • Effects of change By moving from stage 1 to stage 3, the VHA greatly improved tobacco use screening rates and tobacco counselling rates. Although prescription rates for smoking cessation drugs have shown a promising rise over the past four years, it is too early to evaluate the effect of system changes on the VHA’s rate of helping smokers to quit.

  • Lessons learnt Six key lessons can be drawn from the VHA’s experience with improving tobacco control: tobacco control is a system issue; practice guidelines are necessary; measuring performance is essential; performance measures should be tied to incentives; achieving guideline adherence requires additional system support; and best practices should be systematically identified and disseminated. Healthcare organisations as well as small practices can adapt these lessons to improve treatment for smokers, and the lessons can be modified to improve care delivery for other conditions, such as alcohol misuse and depression.

The four stages of the framework described below use data from the Veterans Health Administration (VHA) as a case study. The VHA, one of the largest managed care organisations in the United States, has undergone a series of key structural changes in the past decade leading to major quality improvements in care delivery, including tobacco control. For each stage I will discus what characterises the stage; the experiences of VHA with this stage; what data are typically available; and what is necessary to move to the next stage. I will then distil the lessons learnt and discuss how these steps apply to other health conditions, such as depression and alcohol misuse.

A four stage framework for tobacco control

Stage 1—no system for tobacco control

The chief feature of the first stage (in the VHA, before 1996) is the lack of a system. Providers act independently, as professionals have done for hundreds of years. Care is haphazard, with major variations between sites and between providers. Opinion leaders—influential peers—have a major effect on the behaviour of those around them,7 8 9 so the likelihood that smokers will be counselled depends on where they go for care.

Before the 1990s, smoking cessation was not a high priority for the VHA. Although it was one of the first healthcare systems to adopt a “smoke free” policy,10 this had little direct impact on patient care: health care was largely episodic, and VHA doctors were generally prohibited from providing primary care. Though many medical centres had smoking cessation programmes, few offered drugs for cessation. All facilities were required to have a smoking cessation coordinator, but this role was not clearly or consistently defined. As a result, facilities with a local smoking cessation champion often had an excellent intensive on-site smoking cessation programme, while other facilities had few local options. The likelihood that a smoker would be advised to quit varied dramatically between medical centres and between providers.

Nor was there a system for collecting data on clinical performance. The only data available were from researchers. Within the VHA, it was difficult to determine the answers to basic questions, such as the percentage of patients who use tobacco, how many were screened or advised to quit, and whether smokers were offered counselling or treatment. Published reports provided limited evidence, typically for selected subpopulations.11 12 13

Efforts to improve smoking cessation care at this time relied largely on motivated individual providers. Nearly all medical centres used a specialty clinic approach, referring interested smokers to a smoking cessation programme. Given this structure, one could view stage 1 efforts as driven by believers and zealots. What led to the transition to the next stage was the mandated use of evidence based guidelines and measurement of the level of adherence.

Stage 2—systematic screening

Stage 2 (in the VHA, 1996-2002) is characterised by an organised approach to asking about smoking and advising smokers to quit. Practice guidelines advocate systematically asking about smoking at every visit, and making smoking status a vital sign is an effective way to accomplish this.14 15 The health plan employer data and information set (HEDIS) used by the National Committee for Quality Assurance assesses the rate at which smokers are advised to quit, thereby systematising not just asking about smoking but also advising smokers to quit.

Within the VHA, the mid-1990s was a time of radical change. Under the leadership of Kenneth Kizer and others, the VHA changed from a specialty based system of episodic care to a managed care organisation based in primary care.16 The thinking and vision behind these changes, which came at a time of serious discussion about eliminating the VHA and of creating a national health insurance system, have been discussed in two books.17 18

An important aspect of this transition was the nationally mandated adoption of evidence based practice guidelines. Performance was measured through annual review of 60 000-80 000 charts by an external peer review organisation. For tobacco control, the guidelines measured rates of screening and advising smokers to quit in the previous year. The salary of each facility’s director was tied to the facility’s level of adherence to the guidelines,19 20 and little or no guidance was given on how the improved performance might be achieved. Directors received financial bonuses for meeting performance measurement targets and risked losing their jobs for consistently failing to meet performance targets. Facilities were usually not focused on being the best performers, but rather were extremely concerned about not falling below the median. Although other factors might explain the performance increase shown in figure 2, those involved with quality measurement in the VHA often point to this holding the top leadership accountable as the main cause, resulting in an unequivocal message to managers and providers that performance matters.


Fig 2 Rates of asking about smoking and advising smokers to quit in VHA and non-VHA facilities

These mandated (and measured) guidelines had a large effect on care and also dramatically changed what data were available. For the first time, systematic data were available for every VHA facility on the prevalence of smoking and the rate of screening and advice. Perhaps more important was that each facility knew it would be held accountable. The nationwide rate of screening (as measured by external chart review) increased from 49% to 86% the year that adherence to guidelines became mandatory, and the rate of advising smokers to quit increased from 35% to 79%. The rate of asking and advising rapidly approached 95% nationwide, where it has remained. The high rate of screening and advice was confirmed by the 1999 survey sent to 1.4 million enrollees (41% of all VHA users).21 Among the nearly 900 000 respondents, 81% reported being asked about smoking in the previous year and 72% reported being advised to quit.

How did facilities improve their performance at asking and advising? There was no consistent or recommended approach. The primary intervention was to mandate the outcomes, leaving facilities almost complete freedom to achieve these goals. This approach fosters innovation but tends to be inefficient, as effort is duplicated and many sites implement changes that are ineffective. From my own experience, some facilities tried using doctors to deliver the intervention and had a difficult time. The most effective intervention seemed to be including the screening questions as part of the nurse’s routine job.

What catalysed the transition to the next stage was the realisation that although the system was uniformly asking and advising, the rates of assisting smokers remained quite low. Although 60% of smokers had quit for at least one day in the previous year, only 28% reported being referred or treated for smoking within the past year.21 Furthermore, although 80% of smokers indicated they wanted services to help them quit, only 21% reported usually or always receiving these services. These results were supported by the quality improvement trial for smoking cessation, which showed that though 66% of smokers (at 18 sites) reported that a VHA provider talked about cessation within the past year, only 28% were referred to a programme and only 9% attended.22 About a quarter of patients reported receiving a prescription for nicotine patches within the past year. Thus, patients were routinely being advised to quit, but most were not being offered or receiving the treatments that would increase their chances of success.

Stage 3—systematic treatment

While stage 2 was a systematic approach to asking and advising, stage 3 (in the VHA, from 2002) is characterised by a systematic approach to assisting smokers in quitting. Stage 2 and stage 3 may seem similar, but they occur sequentially (smokers must be identified before they can be offered treatment) and they require different organisational processes. Most facilities rely on nursing staff to systematically ask about smoking, but the approach to offering counselling has been much more variable.23 Offering smoking cessation drugs requires a systematic approach from providers, something that is generally difficult to achieve.24 25 26 The VHA’s experience, where the prevalence of smoking changed very little even though nearly all smokers received advice, suggests that advising patients to quit may be necessary but not sufficient. Stage 3 therefore represents the next step—ensuring that all smokers are offered help in quitting, through both counselling and drugs.

The VHA’s approach to increasing treatment rates represented a shift to a public health approach to tobacco control, orchestrated at the national level by VHA public health and tobacco cessation experts. After reviewing the data mentioned above and additional data showing that only 7-10% of smokers received drugs to help them quit,27 we considered where to focus our effort. Rather than try to increase interest in quitting among the 40% of smokers not trying to quit in a typical year, it was more effective to focus on increasing treatment rates among the 60% who do try to quit. Our efforts since 2002 have included strategies based on our knowledge of the literature and our experience at changing clinical practice and health policy (box).28 29 The concerns raised ranged from providers’ competing demands on their time to smoking cessation experts feeling that we were endorsing substandard treatment.

Smoking cessation strategies used at VHA

  • Mandate the availability of drugs to help smoking cessation—Before 2003, most facilities restricted smoking cessation drugs to smoking cessation clinics. This new policy enabled doctors to prescribe drugs without restrictions,30 an important concern given the low rates of attendance at clinics.22

  • Update guidelines to promote treatment—The 1999 Veterans Health Administration/Department of Defense clinical practice guidelines focused on encouraging patients to attend a smoking cessation programme, which were considered the optimal treatment. The 2004 guidelines instead recommended offering drugs and counselling to all smokers in the most intensive setting they are willing to attend.31

  • Decreasing financial barriers to treatment—The copayment was removed from smoking cessation treatment (although not from smoking cessation drugs) in October 2005.

  • Revised performance measures—In 2007, the VHA adopted new smoking cessation performance measures (similar to those used by HEDIS), focusing on whether smokers were offered counselling, smoking cessation drugs, and referral to an intensive programme. Since performance measures apparently had a large effect on advising smokers to quit, these newer measures will hopefully have a comparable effect on increasing treatment rates.

The rate of prescribing smoking cessation drugs was remarkably constant from 1999 to 2002 (fig 3), but the prescribing rate has increased steadily over the past four years. Though it is not possible to separate out the individual effect of each change to the system, the clear upward trend is encouraging. Since these drugs are effective even with brief counselling in a primary care setting, this increased prescribing of smoking cessation drugs should lead to a corresponding decrease in the prevalence of smoking.


Fig 3 Percentage of smokers receiving smoking cessation drugs

Considerably more data are available to guide current efforts to improve performance. The survey of healthcare experiences of patients, administered to about 400 000 VHA users annually since 2002, has included questions on smoking habits and services offered. These surveys are cross sectional rather than longitudinal, so provide no data on changes in the prevalence of smoking. They show a downward trend, but this may be due partly to an increase in affluent patients (with a lower prevalence of smoking)32 using the system for prescription benefits.

The feature essential for transition to stage 4 is a comprehensive database of smokers. Nearly all VHA facilities use electronic clinical reminders for smoking cessation. Unfortunately, these reminders (and the accompanying data structure) have differed between facilities, making it difficult to systematically identify all smokers nationwide. This will change when new smoking cessation performance measures are assessed with a national clinical reminder, resulting in consistent storage of smoking status and a national database of smokers.

Stage 4—population health

Unlike stages 2 and 3, which are reactive approaches focused on advising and assisting all smokers during a clinic visit, stage 4 (a future stage for the VHA) focuses on assisting all smokers covered by the healthcare system. Because it does not depend on patients visiting the facility, this proactive, population based approach requires being able to identify all smokers. It incorporates outreach to all smokers and opens the possibility of tailoring the approach on the basis of smokers’ need, interest in quitting, and other factors.

As an example of how this might work, a facility could generate a list of all current patients who smoke (regardless of whether they had an upcoming visit) and send a letter to each smoker describing a new telephone quit line, along with a voucher for free smoking cessation drugs. The letter might be tailored to medical diagnoses and include an endorsement from the patient’s primary care provider. A health educator could then call to follow up. Additional interventions might be triggered each time a smoker visited the facility.

Our goal is to move the VHA to this fourth stage. The availability of a national database of current smokers should help this transition. It should also allow the use and testing of evidence based and promising strategies, such as direct marketing,33 34 disease management for smoking cessation, and “recycling” relapsed smokers.35 The VHA’s electronic medical record has allowed the creation of a virtual cohort of patients with human immunodeficiency virus to monitor trends in utilisation of VHA services, toxicity of drugs taken, and outcomes,36 and a similar approach could be used for tobacco use. The VHA is currently deciding what additional strategies to use at this point.

Lessons learnt

Tobacco control is a systemissue

At the Veterans Health Administration, performance did not improve meaningfully until tobacco control became a system issue (a mandated guideline with monitored performance). It requires a more concerted effort than simply having a smoking cessation clinic and waiting for referrals. Rather than using the approach advocated in the movie Field of Dreams—“If you build it, they will come”—a facility or system needs to think about how to meet the needs and preferences of its smokers.

Practice guidelines are necessary

To improve the quality of care, evidence based standards of care are needed. Without a standard, there is no consensus on how to measure quality, and standards that are not evidence based are more likely to be resisted by providers. Practice guidelines provide a convenient method to group the standards of care for a topic and describe the extent to which each is evidence based.

Measuring performance is essential

It is difficult or impossible to manage care without accurate, meaningful data. Without data, providers and administrators overestimate the quality of care.37 Measuring performance regularly helps raise awareness of the level of adherence to guidelines, particularly when quality is less than desired. Accuracy of measurement is essential, or few will believe the assessment of quality.

Performance measures must be tied to incentives

Performance measurement has little value if the data are not used or viewed as important. The perceived importance depends on many factors, such as whether clinicians agree with the underlying guideline and measure, the level of endorsement (both explicit and implicit) by middle and top level administrators, and any consequences associated with poor adherence. The approach taken by the VHA to increase guideline adherence—holding top leadership personally accountable for performance—seemed to be effective at changing providers’ behaviour, even though facilities received little or no guidance on how to make the change.

Achieving adherence to guidelines requires additional system support

The initial VHA plan to adopt 10-20 additional guidelines each year (with a goal of 70-80 guidelines) was modified for several reasons. Firstly, as the number of guidelines increased, so too did anxiety and dissatisfaction among providers, who were often expected to improve their level of adherence with little support from the healthcare system. Secondly, a VHA cooperative study showed that computer reminders for doctors led to better adherence to 13 standards of care, but this benefit diminished over time.38 Additional strategies seemed to be needed to sustain the improvement, but it was not clear what they should be. The final reason the initial plans were scaled back was the consensus that it was more important to have large improvements in a few measures than small improvements in many measures.

The trend across VHA facilities in more recent years is to change the system to help providers adhere to guidelines. Most facilities now rely on nursing staff to systematically ask patients about tobacco use.23 Computer reminders for smoking cessation are in use throughout the VHA. Many facilities use audit and feedback to keep providers informed about their level of performance. Most recently, doctors’ pay within the VHA was restructured to provide a financial incentive for meeting locally and regionally determined performance goals, which will likely improve performance in these areas.39

Systematically identify and disseminate best practices

The VHA mandated outcomes by measuring specific processes, such as asking about smoking. Since each facility largely developed its own approach to achieve these outcomes, this was a fertile ground for innovation, albeit an inefficient one. Some sites made smoking cessation a quality improvement project, while others took a more top-down approach of specifying the approach to care. A major constraint is that smoking was only one of many areas that were simultaneous targets of improvement, thereby limiting the effort that could be expended on smoking or any other individual problem. In retrospect, it would have been worthwhile to systematically identify best practices and disseminate them to foster effective and efficient change. (The VHA did have a website for innovative programmes, but these were not systematically identified and were used haphazardly.) The most effective solutions have in general been ones that keep the provider’s role as brief as possible, relying on other health professionals (nursing staff, pharmacists, etc), and on information systems to monitor and remind providers at all levels.

Relevance for other healthcare systems

This framework seems consistent with the experience of other healthcare systems. Group Health Cooperative of Puget Sound has long been recognised as a leader in tobacco control.40 41 It progressed through the stages described above, increasing rates of screening from about 25% to 95% (stage 2);working on increasing the rate of intervention by doctors (stage 3); and intending to create a registry of smokers to allow a tailored, population based approach (stage 4). This framework could easily be superimposed on a recent description of tobacco control efforts in non-profit health maintenance organisations.42

This approach has worked for the Veterans Health Administration, but is it relevant for smaller healthcare systems? Some would argue that the VHA’s large size, electronic medical record, and stable patient population make it a special case, but I feel this approach can be generalised. Improving preventive care is widely recognised now as a systems issue, regardless of whether the system is a health maintenance organisation, a multispecialty group practice, or a doctor in solo practice. The six lessons summarised above apply to any system, large or small.

Next steps

The characteristics of these stages I have proposed, along with what facilitates transition to the next stage, are listed in the table. Data from the VHA illustrate one system’s past, current, and future progress through these stages.

Characteristics of the four stages for smoking cessation and elements needed for transition

View this table:

This approach applies to conditions other than smoking cessation—managing major depression, for example. Increasing the screening rate for depression has been straightforward.43 Many healthcare systems are now at stage 3, using a collaborative care model to ensure that all patients identified as depressed receive effective treatment.44 45 46 Similarly, management of alcohol misuse within the VHA has progressed from stage 1 (no system) to stage 2 (near universal screening with the AUDIT-C alcohol screening tool),47 and current efforts focus on increasing rates of treatment for alcohol misuse (stage 3).

This framework suggests that these steps and transitions were logical, well thought out, and well orchestrated—but it was derived retrospectively. In a large, complex healthcare system such as the VHA, change is typically messy and chaotic. It is difficult to change large healthcare systems, so the steady improvements in tobacco control over the past decade are indeed an accomplishment. The VHA approach was to mandate the outcome (almost entirely a “pull” approach),48 with little guidance as to how the outcome might be achieved. At that time there was scant literature on guideline implementation to guide this system change. Considerable thought and planning went into the approach to performance measurement underlying the transition from stage 1 to stage 2 and into the shift in focus from stage 2 to stage 3. Facilities did not receive any additional resources to make these changes and in fact were reducing staff at the time, so the improved performance at smoking cessation may have come at the expense of diminished performance on other, unmeasured problems.

Whether these future directions will be effective in reducing tobacco use within the VHA remains to be seen. Ongoing programmes by the Department of Defense should reduce the numbers of people who start smoking and increase stopping among military personnel on active duty, leading to a lower prevalence of tobacco use among patients entering the VHA. My hope is that this framework will help make future efforts at large scale system change go more smoothly.


  • I thank Laura York for invaluable editorial assistance, and Melissa Farmer, Marc Gourevitch, Adina Kalet, Lisa Rubenstein, Mark Schwartz, and Nirav Shah for reviewing an earlier draft of this article. The views expressed in this article are solely those of the author and do not necessarily represent those of the Department of Veterans Affairs.

  • Contributors: SES is the sole contributor.

  • Competing interests: None declared.

  • Provenance and peer review: Not commissioned; externally peer reviewed.