The Cost of Smoking in Canada, 1991

Volume 18, No.1 -1997

The Cost of Smoking in Canada, 1991
Murray J Kaiserman

Abstract

In 1991, smoking-attributable health care costs in Canada were $2.5 billion (CAN). Additional smoking-attributable costs included $1.5 billion for residential care, $2 billion due to workers' absenteeism, $80 million due to fires and $10.5 billion due to lost future income caused by premature death. Adjustments for future costs if smoking had not occurred and smokers had not died were estimated to be $1.5 billion. According to this analysis, smokers cost society about $15 billion while contributing roughly $7.8 billion in taxes. The results indicate that smoking-attributable costs in Canada have increased steadily since 1966 to the 1991 value of $15 billion. Nevertheless, while the determination of smoking-attributable costs is important, the issue continues to be public health. In addition, for the first time in Canada, the smoking-attributable cost for residential care has been estimated.

Key words: Canada; cost of illness; health care costs; smoking; smoking-attributable cost; tobacco

Introduction

As the twentieth century ends, governments at all levels are becoming more active in decreasing expenditures or, where possible, recovering costs.¹ One area of consideration has been smoking. Indeed, almost 20 American states, including Florida, Louisiana, Maryland, Massachusetts, Mississippi, West Virginia and Minnesota, have begun legal proceedings with Blue Cross/Blue Shield of Minnesota against the US tobacco industry to recover state-funded medical costs.²

While the notion of charging the tobacco industry for the death and disease caused by use of its products is an interesting one, the problem is how to determine those costs. Oftentimes the data necessary are either not available in a timely or useful fashion, or they are not available for the time period of interest. The result is that, while it is relatively easy to estimate smoking-attributable deaths, it is more difficult to estimate the cost of smoking to society and to provide this cost for the current year.

In Canada, since the publication of two reports estimating smoking-attributable mortality (SAM) at between 41,000 and 45,000 deaths in 1991, there has also been an interest in determining smoking-attributable costs.^3-5 Currently national-level data necessary to determine such smoking-attributable costs exist only for 1991⁶ and 1994, ⁷ when large national surveys were completed. Between those years, no national survey of sufficient size to provide adequate smoking and health data was undertaken. Since much of the 1994 data needed to estimate smoking-attributable costs is still not available, this study will estimate smoking-attributable costs in Canada for 1991.

Methods and Data Sources

The method used is primarily the one proposed and developed by Rice, Hodgson, et al.⁸ and later incorporated into the Smoking-Attributable Mortality, Morbidity and Economic Costs, Release II (SAMMEC) by the US Office on Smoking and Health.⁹ Since SAMMEC does not always conform to the Canadian model of health care and data sources, the method has been modified using methods suggested by MacMahon and Cole ¹⁰ and modified by Collishaw and Myers.¹¹ Notwithstanding these modifications, the analysis continues to be prevalence-based, providing an estimate of excess mortality and morbidity attributable to smoking that includes all health outcomes.

The following basic methods were used to determine costs.

Determination of Smoking-attributable Fractions⁸

Smoking-attributable fractions (SAFs) were calculated according to the following formula: (P(RR-1))/(1+P(RR-1)) where P is the prevalence of ever smokers and RR is the relative risk for that particular outcome, e.g. death, hospital visits, etc. Total outcomes are then multiplied by the SAF to determine the value that can be attributed to smoking. Table 1 provides an example of how this method can be applied.

Determination of Excess Utilization by Smokers

On average, ever smokers have higher levels of use of the health care system than never smokers.^8,12 The difference can be allocated to the effects of smoking and, when multiplied by the cost associated with that activity, the cost attributable to smoking can be determined.

Present Value AnalysisP> Present value analysis (PVA) was used to determine future costs and lost income attributable to those smokers who had died in 1991. For this analysis, it was assumed that the median age of death was halfway in the age group (so that, for example, calculations for smokers in the 15-24 age group start at age 20 and so on) and that, had these smokers not died, they would have lived to be 85 years old.¹³ Total costs, including changes between age groups, were calculated for deaths within each age group.

For the purposes of this report, a number of general assumptions were made as well as specific assumptions regarding certain calculations. Generally, it was assumed that the current levels of use would not change over time and that the discount rate was 4%. In addition, SAF calculations or differences led to results, in some cases, that seemed to indicate that smoking was protective. This occurred mostly in the determination of workers' absenteeism among certain age groups. Since it would be contrary to assume that smoking is protective of smokers, it was assumed that there were no smoking-attributable costs in those instances where smokers' absenteeism was less than that of non-smokers.8 All calculations are the result of determining costs for specific age groups: 15-24, 25-34, 35-44, 45-54, 55-64, 65-74 and 75+. Finally, current and former smokers were considered as one category, ever smokers.

The following is a detailed explanation of the sources, assumptions and calculations used.

Smoking-attributable Mortality

Smoking-attributable mortality (presented in Table 1) was calculated using the SAF method. Relative risks were obtained from the Cancer Prevention Study II sponsored by the American Cancer Society.¹⁴ The proportion of ever smokers was obtained from the 1991 General Social Survey (GSS '91).¹⁵ Age- and sex-specific deaths were obtained from reports published by Statistics Canada.¹⁶

Hospital Costs

Smoking-attributable costs of hospital visits were determined using the SAF method. Relative risks were calculated from self-reported hospital utilization data collected during the GSS '91. This survey also supplied ever smoker prevalence data.¹⁵ Data on hospital separation status (deceased or live) were obtained from 1991/92 hospital morbidity data.¹⁷ Average hospital costs and length of stay per visit were obtained from published reports.^18,19

In determining hospital costs, it was recognized that it was extremely costly to extract data for the 1991 calendar year. As a result it was assumed that, since 1991/92 fiscal year data accounted for nine months of 1991, they would provide a close approximation of the entire year. Indeed, overall hospital utilization in the years 1990/91, 1991/92 and 1992/93 differed by only about 1% per year.^20,21

As part of the analysis, future hospital costs were estimated for those ever smokers who died in 1991, using the PVA method. It was assumed that if they had not died from smoking, smokers would have used hospitals at some future date. Their hospital utilization was estimated using self-reported hospital use data of never smokers from the GSS '91 as the "normal rate of use."¹⁵

Physician Costs

Smoking-attributable costs were estimated using differences between self-reported data for never and ever smokers from the GSS '91.¹⁵ The average cost per visit was assumed to be $30 (Canadian Medical Association, personal communication).

The cost of future physician visits for those smokers who died in 1991 was calculated using the PVA method with never smokers' use as the "normal rate of use."¹⁵

Drug Costs

Smoking-attributable costs were estimated using differences between self-reported data for never smokers and ever smokers from the Health Promotion Survey (HPS) of 1990. ²² The HPS provided information regarding prescription drug use. Since there were no similar data available for 1991, the HPS results provided an approximation of drug use with the understanding that there could be a significant underestimation or overestimation due to differences in usage between the two years. The average cost per prescription was assumed to be $23.27, which is the average cost in Ontario (Canadian Pharmaceutical Association, personal communication).

The cost of future drug costs for those smokers who died in 1991 was calculated using the PVA method with never smokers' use as the "normal rate of use."¹⁵

Institutional Costs

While residents of long-term health care facilities are usually excluded from national surveys, the 1994 National Population Health Survey did include this population.²³ Data used from this survey provide smoking status, age and sex.²⁴ Using the never smokers as the baseline and both institutionalized and non-institutionalized prevalence data, it is possible to estimate the number of former and current smokers who should be found in these institutions.¹⁵ Based upon these calculations, it was estimated that there were approximately 38,000 excess female former smokers aged 65+ in such facilities. The calculations indicate that there are less male smokers (current and former) and less female current smokers than predicted.

While the data are for 1995, it was assumed that similar levels of use by ever smokers would apply for 1991. The smoking-attributable cost was determined by multiplying this value by the average per diem rate of $111.13.²⁵

Workers' Absenteeism

Workers absenteeism was calculated using differences between self-reported data from the GSS '91 for never smokers and ever smokers who were employed at the time of the survey.¹⁵ These values were then multiplied by the average annual salary divided by 261 working days, which represents an average five-day working week, including vacation and statutory holidays.

Future Earnings Lost by Reason of Death

Future earnings lost by reason of death were calculated using the PVA method and income data published by Statistics Canada.²⁶

Fires

The smoking-attributable cost of residential fires was obtained directly from data published by the Association of Canadian Fire Marshals and Fire Commissioners.²⁷ The smoking-attributable cost of forest fires was estimated to be about 10% of total costs as provided by the Canadian Forest Service (Department of Natural Resources Canada, personal communication).

Results and Discussion

It is estimated that roughly 6.5 million current smokers and 4.9 million former smokers accounted for approximately $2.5 billion in smoking-attributable excess health care costs in 1991, which is about 3.8% of the $66.7 billion 28 total health care bill in Canada (Table 2). Since ever smokers represent about 54% of the population, it must be remembered that they would also be responsible for about 54% of "normal" health care costs, or about $35 billion ¹⁵. An additional $1.5 billion was spent on former smokers in residential care facilities.

Despite the use of different methodologies, these costs are comparable to those of the state of California, which has about the same size of population as Canada. Using the Rice model, it was estimated that approximately 4.5 million Californian smokers aged 12 or more accounted for $2.2 billion in 1989.²⁹ In addition, the US Department of Health and Human Services, using a method similar to the one presented here, recently estimated that smoking-attributable health care costs in the US were $21.89 billion in 1993 for a population about 10 times the size of Canada's.³⁰

It is not surprising that all three studies would arrive at roughly the same estimates since they use the same basic SAF methodology and estimate the same health care categories: hospitals, physicians and drugs. Of the three studies, the one prepared by the US Department of Health and Human Services³⁰ is probably more complete because it incorporated other categories, including Medicare, Medicaid and nursing-home care, some of which do not exist in Canada. Nevertheless, whatever method used, the estimates are probably low because all of the costs cannot be estimated.

Excluding the costs for those smokers in residential homes, the largest amount of smoking-related health care costs in Canada was spent on hospital care (Figure 1). There was a difference in the use of medical services between male and female ever smokers. Indeed, smoking-attributable hospital costs for male ever smokers were almost twice as much as those for female ever smokers (Table 2)

The 1991 smoking-attributable costs by age and sex for ever smokers are presented in Figure 2. (To convert to daily costs, dollar values should be divided by the average daily rate of $516. 18 ) The graph shows age and sex differences in the smoking-attributable costs, with higher costs attributable to older smokers.

From self-reported data, the bulk of smoking-attributable costs of female ever smokers accumulates between the age groups of 15-34 and 65-74. While the peak in the early age group is probably due to the increase in the number of female smokers over the past 20 years, complications during childbirth could also be contributing. It is interesting to note that there is very little difference in hospital use by male and female ever smokers aged 25-34. For male ever smokers, who smoked at a much higher rate than females, the majority of hospital costs occur after age 45. These costs peak between ages 55 and 64 and start to decline after age 65. For females, the peak at age 65 is a reflection of the overall deterioration of general health and that of smokers in particular.²¹

Data from the GSS '91 indicated that ever smokers reported higher levels of absenteeism than never smokers.¹⁵ This difference in absenteeism, which is the smoking-attributable workers' absenteeism, is presented by age and sex in Figure 3 . Again, there are differences between males and females, with younger female smokers (under age 45) tending to be absent from work more often than older female and male smokers of the same age. Indeed, female ever smokers aged 15-24 were absent an average of almost two weeks longer than female never smokers. On the other hand, absenteeism among male ever smokers increased with age after 45. The differences between males and females at the older ages are probably due to the lower levels of females in the work force as well as their lower smoking rates. The exact reasons for absenteeism, such as illness, fatigue, etc., remain unknown.

Since 1968, there have been seven published studies (including this one)^11,31-35 and one unpublished study (Basavaraj S and Kaiserman MJ, unpublished observations) estimating smoking-attributable costs in Canada. The results of these studies, adjusted to 1991 dollars, are illustrated in Figure 4 . As the methodology became more refined and the data more available and as actual costs increased, the smoking-attributable cost rose to its present value. The major difference among these studies has been whether or not indirect costs were included, such as the expected value of future earnings lost due to the premature mortality of smokers and future health care costs attributable to those smokers not dying.

In 1992, Raynauld and Vidal published their estimates on smoking-attributable costs for 1986. ³⁵ As shown in Figure 4, their estimate was significantly lower than any estimate before or since. While some of their assumptions may be suspect, most notably the use of Saskatchewan data to estimate national levels of hospital use and the underestimation or dismissal of other costs such as workers' absenteeism, they did provide a good case for inclusion of future health care costs for those smokers who died in the year of interest. The assumption made was that had these people not smoked and not died as a result of their smoking, they would have lived and used the health care system at the same rate as never smokers. These future costs should be subtracted from current costs. In addition, Raynauld and Vidal estimated the future costs of residential home care in Canada for the first time. The net result was the value displayed in Figure 4 .

A similar analysis was conducted for this report. Smoking-attributable mortality in 1991 for those people aged 15 and over is presented in Table 1. It should be noted that these results are in agreement with previous SAM estimates for the same year.^3,4 The data in this table were then used to calculate the present value of future costs and lost income provided in Table 2.

Adjustments to the health care costs amount to approximately $1.5 billion, with residential care costs accounting for about 55% and hospital costs another 39%. As a result of these adjustments and their addition to estimated health care costs, smoking-attributable health care costs remain at just over $2.3 billion.

In addition, smoking-attributable disability accounts for another $2 billion, fires account for about $80 million and lost income, another $10.5 billion. The total smoking-attributable cost estimated for Canada in 1991 is therefore about $15 billion.

Conclusions

In 1991, smoking accounted for an estimated $2.5 billion in health care costs, $1.5 billion for residential care costs, $2 billion from workers' absenteeism and $10.5 billion in lost future earnings, totalling $16.5 billion. Nevertheless, had there been a tobacco-free society, the smokers who had died in 1991 would have lived longer, on average, and ultimately would have cost society for such services as pensions, medical care and residential care. Except for pensions, these "avoided" costs were estimated to be about $1.5 billion. Thus, overall, smokers cost Canadian society just under $15 billion in 1991.

These estimates are, of course, neither totally accurate nor complete. For example, future costs of pensions are not included because current federal government policy to make pensions self-financing makes their inclusion in this calculation uncertain.36 In addition, some of the calculations may either underestimate or overestimate costs. For the purposes of this report, it was assumed that these differences would tend to cancel each other out.

Finally, a number of smoking-attributable costs are not included because there are no methods, at present, to estimate their value. For example, there is very little information regarding the cost attributable to friends' and families' visits to sick and dying smokers, of long-term smoking-attributable disability or of smoking-attributable home health care. Furthermore, the psychological and emotional costs incurred by the families of dead or dying smokers are incalculable.

Yet, is the real cost of smoking economic? In 1987, Warner concluded that the biggest gain to a tobacco-free society would be "significantly enriched quality and quantity of life."³⁷ This was written in the US at a time when it was believed that smokers actually cost society more than they contributed in taxes. Back then, many of the calculations did not include future costs but did include lost income costs. For this reason, economists argued that smoking-attributable costs were too high.³⁷

In Canada, the situation is very different. In 1991, smokers spent $10.45 billion on tobacco products,³⁸ of which about 75% went to governments in the form of taxes. Excluding lost income (which costs individuals and not governments) and disability (which incurs costs for the employer), the cost of smoking for governments was about $2.3 billion in health care costs (including residential care) and an additional $96 million in lost income taxes from smokers who died in 1991. The latter number is estimated by assuming that, on average, those smokers who died in 1991 earned one half of their income at a tax rate of 19.8%.39 Even with this latter amount included, the result is that smokers paid in more than they took out by about $5.4 billion. After the reduction in excise taxes on February 8, 1994, however, smokers saw a substantial decrease in their smoking-related contributions.

Was this excess payment a worthwhile investment for smokers? Raynauld and Vidal claim that smokers are aware that smoking is detrimental to their health and are willing to accept this risk.³⁵ If this is true and if the $5.4 billion extra paid to government by smokers represents their total investment in their future health care costs, then smokers paid in an individual average of $800 in 1991 to help defray their future health care costs. This represents less than two nights in a hospital or about 26 visits to the doctor or about 34 prescriptions for medication. In contrast, it was estimated that the median cost per life saved from cancer was about $750,000 (US) in 1994.⁴⁰

Despite the fact that Canadian smokers paid society more than they cost it in 1991, this argument confuses the debate on smoking. The issue is, as Warner states, health.³⁷ During 1991 in Canada, over 41,000 men, women and children died prematurely as a result of smoking at a cost to society and their families of at least $15 billion. The elimination of smoking might not ultimately save one penny of that cost. Rather, it would save the 41,000 plus lives that ended prematurely. This, together with a significantly enriched quality of life, makes a smoke-free society that much more desirable.

_{Author Reference

Murray J Kaiserman, Office of Tobacco Control, Health Protection Branch,
Health Canada, Tunney's Pasture, Address Locator: 0907D1, Ottawa, Ontario
K1A 0K9}

References

1. Governor General of Canada. Speech from the Throne to open the Second Session, Thirty-fifth Parliament of Canada. House of Commons Debates 1996 Feb 27;133(001):2-6.
2. Medical cost reimbursement roundup. Tobacco on Trial 1996 Nov.
3. Makomaski-Illing EM, Kaiserman MJ. Mortality attributable to tobacco use in Canada and its regions, 1991. Can J Public Health 1995;86:257-65.
4. Ellison LF, Mao Y, Gibbons L. Projected smoking-attributable mortality in Canada, 1991-2000. Chronic Dis Can 1995;16(2):84-9.
5. Minister of Health. Tobacco control: a blueprint to protect the health of Canadians. Ottawa: Health Canada, 1995; Cat H49-100.
6. Statistics Canada. 1991 General Social Survey, Cycle 6: health. Public use microdata file documentation and user's guide. Ottawa, 1992.
7. Statistics Canada (Health Statistics Division). National Population Health Survey, 1994-1995. Public use microdata file documentation and user's guide. Ottawa, 1995.
8. Rice DP, Hodgson TA, Sinsheimer P, Browner WS, Kopstein AN. The economic costs of the health effects of smoking-1984. Milbank Q 1986;64(4):489-547.
9. Shultz JM, Novotny TE, Rice DP. Quantifying the disease impact of cigarette smoking with SAMMEC II software. Public Health Rep 1991;101:148-53.
10. MacMahon B, Cole P. Attributable risk percent in case-control studies. Br J Prev Soc Med 1971;25:242-4.
11. Collishaw NE, Myers G. Dollar estimates of the consequences of tobacco use in Canada, 1979. Can J Public Health 1984;75:192-8.
12. Bertera L. The effects of behavioural risks on absenteeism and health-care costs in the workplace. J Occup Med 1991;33(11):1119-23.
13. Dumas J, ed. Population aging and the elderly: current demographic analysis. Ottawa: Statistics Canada, 1993; Cat 91-533E.
14. American Cancer Society. Cancer Prevention Study II, 1982-1988 [unpublished tabulations]. Atlanta (GA).
15. Statistics Canada. 1991 General Social Survey, Cycle 6: health [unpublished tabulations]. Ottawa (Ont).
16. Statistics Canada (Canadian Centre for Health Information) Mortality summary list of causes, 1991. Ottawa, 1993; Cat 84-209.
17. Statistics Canada (Health Statistics Division). Hospital morbidity, 1991/92 [unpublished tabulations]. Ottawa, 1994.
18. Statistics Canada (Canadian Centre for Health Information). Hospital statistics: preliminary annual report, 1991/92. Health Reports 1993;5(3):250. (Cat 82-003).
19. Statistics Canada (Health Statistics Division). Hospital morbidity, 1991/92. Ottawa, 1994; Cat 82-216.
20. Statistics Canada (Canadian Centre for Health Information). Hospital statistics, 1991/92. Health Reports 1993;6(2):313-4. (Cat 82-003).
21. Randhawa J. Hospital morbidity, 1991/92. Health Reports 1993;5(4):355-61. (Statistics Canada Cat 82-003).
22. Health and Welfare Canada. Canada's Health Promotion Survey 1990 [unpublished tabulations]. 1992 Mar.
23. Tully P, Mohl C. Older residents of health care institutions. Health Reports 1995;7(3):27-30. (Statistics Canada Cat 82-003).
24. Statistics Canada. National Population Health Survey, 1994-1995 [unpublished tabulations]. Ottawa, 1996 Feb.
25. Statistics Canada (Canadian Centre for Health Information). Residential care facilities, 1992/93. Health Reports 1993;6(2):315. (Statistics Canada Cat 82-003).
26. Ghalam NZ. Women in the workplace. 1993:35-42; Statistics Canada Cat 71-534E.
27. Association of Canadian Fire Marshals and Fire Commissioners. Fire losses in Canada: annual report, 1991. Ottawa, 1991:25.
28. Health Canada. National health expenditures in Canada, 1975-1994. Summary report. Ottawa, 1996; Cat H21-99/1994-2.
29. Rice DP, Max W. The cost of smoking in California, 1989. Sacramento (CA): California State Department of Health Services, 1992.
30. Centers for Disease Control. Medical-care expenditures attributable to cigarette smoking-United States, 1993. Morb Mortal Wkly Rep 1994;43(25):469-72.
31. Smith A. The estimated cost of certain identifiable consequences of cigarette smoking upon health, longevity, and property in Canada in 1966. Ottawa: Department of National Health and Welfare, 1968.
32. Shillington ER. Selected economic consequences of cigarette smoking, Ottawa: Health and Welfare Canada; 1977.
33. Luce BR, Schweitzer SO. The economic costs of smoking induced illness. In: Research on smoking behavior. Jarvik ME, et al, eds. Rockville (MD): National Institute on Drug Abuse, US Public Health Service, 1977:221-7; Research Monograph 17.
34. Thompson ME, Forbes WF. Costs and "benefits" of cigarette smoking in Canada. Can Med Assoc J 1982;127(11):831-2.
35. Raynauld A, Vidal J-P. Smokers' burden on society: myth and reality in Canada. Can Public Policy 1992;XVII(3):300-17.
36. Minister of Finance. Budget in brief. Ottawa: Department of Finance, 1996 Mar 6.
37. Warner KE. Health and economic implications of a tobacco-free society. JAMA 1987;258(15):2080-6.
38. Statistics Canada. Market research handbook, 1995. Ottawa, 1995; Cat 63-224.
39. Statistics Canada. Income after tax, distributions by size in Canada, 1991. Ottawa, 1993:21; Cat 13-210.
40. Tengs TO, Adams M, Pliskin JS, Safran DG, Siegel JE, Weinstein MC, et al. Five-hundred life saving interventions and their cost-effectiveness. Risk Anal 1995;15(3):369-90.

[Previous][Table of Contents] [Next]


	Last Updated: 2002-10-29		Important Notices