A financial case to enable state health jurisdictions to invest in tobacco control

There is universal consensus about the risks for ill health and premature death attributable to tobacco. In Australia, in 1996, cigarette smoking was responsible for the loss of about 227 000 disability-adjusted life-years (DALYs).1 Worldwide, tobacco causes about 8.8% of deaths and 4.1% of DALYs.2 By 2020, the disease burden attributable to tobacco will have doubled.2 Yet a 25% reduction in smoking would result in 17 million fewer DALYs in 2010 and 22 million fewer in 2020.2 Nonetheless, criticism from the World Health Organization continues — that most services, resources and research in healthcare focus on the treatment of tobacco-related disease, rather than its prevention.2

Unlike preventive approaches for other public health issues, for which the evidence may be less compelling or the dividend from health promotion interventions less scientifically defensible, tobacco control represents an uncontroversial investment in population health. There is compelling evidence of both short-term and long-term health benefits from tobacco control.3-7 Furthermore, there is national consensus about health promotion programs that are effective in reducing tobacco use,8-11 as well as a priority-driven research agenda to furnish interventional evidence in gaps most immediately relevant to policy and practice.12

In Australia, bureaucratic responsibility for evidence-based tobacco control is unclear in the existing legislative arrangements between Commonwealth and state (including territory) health jurisdictions.8 State health jurisdictions carry primary responsibility for inpatient hospital services and directly bear the financial burden of hospitalisations,13 including “avoidable” admissions.14 By contrast, the Commonwealth directly funds general practice and other fee-for-service professionals, and specific public health programs.13 Within these financial arrangements, there is no clear delegation of accountability for tobacco control. Hence, the necessary coalition of government interests to reduce tobacco use appears split counterproductively between jurisdictions. Commentators outside the healthcare system have criticised the paucity of government expenditure on tobacco control in any jurisdiction.15,16

Smokers use inpatient hospital services more than people who have never smoked.17-19 Smokers are also more likely than non-smokers to use services such as emergency and outpatient departments,19 to heal less quickly20 and to be admitted after surgical procedures to intensive care.21 As these costs are met by state jurisdictions, the states are likely to benefit most from tobacco control through reduction of hospitalisations. We hypothesised that hospital managers in state jurisdictions have yet to be presented with a coherent “business case” for tobacco control.

We first referred to available data sources to determine a list of conditions for which there is clear and quantified evidence of causation by tobacco. Specifically, aetiological fractions were originally published by English in 199522 and recently updated by Ridolfo and Stevenson.23 The aetiological fraction, also known as attributable risk or attributable fraction, is a recognised calculation quantifying morbidity and mortality due to a specified risk factor.23 It is the proportion of cases to have been caused directly by exposure to this specific risk factor.22 Furthermore, these cases occur only because of this exposure.22 Box 1 shows aetiological fractions by age and sex for 30 tobacco-attributed diseases in four broad categories: cancer, ischaemic heart disease, chronic obstructive pulmonary disease, and others. Almost all cases (90%) of lung cancer in men and nearly two-thirds (65%) in women are directly attributable to personal tobacco use (Box 1). Because of unresolved epidemiological uncertainty,22,23 we did not include passive smoking and fire injuries. Hence, Box 1 is a conservative list of tobacco-related conditions.

By using these aetiological fractions, the proportion of hospital admissions for these 30 conditions attributable to tobacco can be calculated. Through the Health Outcomes Information and Statistical Toolkit,24 we obtained the most recent hospital separation data and average length of stay (hospital bed-days) data for a 12-month period (1999–2000 financial year) from the New South Wales Inpatient Statistics Collection (ISC). ISC is a census for admissions to all services provided by public hospitals, public psychiatric hospitals, public multipurpose services, private hospitals and private day-procedure centres.24 Hospital separations are coded according to one principal diagnosis. We used ISC data to calculate the total number of hospital separations and the average length of stay by sex for each of the diseases listed in Box 1. Hospitalisations directly attributable to smoking were calculated by multiplying the sex and age-specific hospital separations for each disease by the relevant attributable fraction. Smoking-attributable hospital bed-days for each disease were calculated by multiplying the number of hospital separations attributable to tobacco for this disease by the relevant average length of stay. Hospital costs of smoking-attributable hospitalisations for patients admitted to hospital in the 12-month period were estimated by multiplying published amounts for average bed-day cost25,26 by totals of smoking-attributable bed-days. Our statistical program is available on request.

Armed with these estimates, a persuasive argument can be mounted to invest in tobacco control to reduce hospital costs. More particularly, the amounts to be dedicated to tobacco control could be calculated on the basis of an objective determination of tobacco-related costs. Indeed, health service managers could be asked to nominate a priori the proportion of hospital costs they would consider appropriate to levy in their setting to reduce demand. Our initial consultations with senior executives in the Division of Population Health obtained responses ranging from 5% to 10%. A value of 3% is consistent with productivity savings typically required of public-sector agencies.

As proposed here, our method can be criticised. For example, taxes already are levied on the tobacco industry, although these are inconsistently allocated to health promotion by state jurisdictions.27 Furthermore, aetiological fractions are a function of the relative risks associated with the disease in question and the population prevalence. Hence, significant variation in smoking prevalence should be factored into subsequent applications of our method. Time lags between tobacco exposure and disease will determine the realisation of benefit from an investment in tobacco control. Given the lead-time between the act of smoking and the onset of disease, policy makers who move funds to tobacco control will not see all returns from this investment in terms of fewer hospitalisations, and hence lower hospital costs, in the short term. However, epidemiological sources used by us23 addressed prevalence estimates and time lag.

Could our method be applied to other public health issues? Theoretically, this is possible, but could transgress the diligence required of an evidence-based approach. Specifically, tobacco control meets two fundamental criteria characteristic of evidence-based healthcare. First, the causal pathway between tobacco and subsequent disease has been convincingly established. There is no speculation about the validity of aetiological fractions due to direct tobacco use. Second, there is a clear body of evidence demonstrating the effect of health promotion interventions encompassing prevention, legislation and cessation, although dimensions of equity and, particularly, differential impact of such interventions by socioeconomic class remain poorly understood. Where effectiveness is not yet proven, a responsive research agenda for tobacco control has been proposed.8,9 In other aspects of population health, either or both of these criteria are not yet met. In obesity, for example, the portfolio of effective interventions is yet to be defined. With respect to emerging techniques for population-based screening, governments do not advocate screening in the absence of compelling evidence. In such situations, determination of a budget for promotional activities would be premature. By contrast, many government policies contain positive political commitment to tobacco control.9,10,28

Our approach has revealed how much is spent on tobacco-related hospitalisations. Healthcare costs will increase unabated unless persuasive arguments can be mounted to secure sustainable budgets for evidence-based prevention. As pressures to reconsider health priorities confront all governments, senior executives, clinicians and consumers,29 our novel yet methodologically defensible argument appears accessible to a wide audience.