Estimates of smoking-attributable mortality and hospitalization in BC, 2002-2007.

Tu, Andrew W. ; Buxton, Jane A. ; Stockwell, Tim 等

Tobacco use is a growing worldwide epidemic; it is estimated that tobacco use will kill over 8 million people annually by 2030. (1,2) Today, it is a risk factor of six of the eight leading causes of death worldwide and is estimated to cause 1 in 10 deaths. (1) In Canada, results from the Canadian Tobacco Use Monitoring Survey (CTUMS) have shown a decline in the prevalence of current smokers from 25% in 1999 to 18% in 2008. (3) However, this decline varied by region, age group, and sex. It is unclear how this trend has affected the burden of smoking on mortality and morbidity.

A number of countries, including Canada, have used attributable fractions (AFs) to estimate the number of deaths caused by smoking, more often referred to as smoking-attributable mortality (SAM). (4-9) AFs describe the fraction of deaths an exposure is responsible for or alternatively, the proportion of disease that would not occur if the exposure were removed. Some of these countries have extended this methodology to hospitalization data to estimate smoking-attributable hospitalization (SAH). (4,8,10) The Cost of Substance Abuse in Canada report (CSAC) estimated that smoking was responsible for over 37 thousand deaths and 339 thousand hospitalizations in 2002 (estimates for British Columbia: SAM 4,616; SAH 34,501). (4) In comparison, it was estimated that alcohol was responsible for over 4 thousand deaths and almost 200 thousand hospitalizations, and illicit drug use accounted for 1,700 deaths and over 60 thousand hospitalizations.

In British Columbia (BC), a comprehensive substance use monitoring project (www.aodmonitoring.ca) is being undertaken with the goal of building a system to provide timely data on risky patterns of substance use and related harms as a means of supporting more effective policy-making, facilitating research on substance use and informing public debate. (11) One component of the project is tracking alcohol-, drug-, and smoking-attributable mortality and hospitalization. This component is currently using the AFs that were used in the CSAC report. This allows for estimates by health region, age, gender, and disease condition.

Given that the AFs used in the CSAC report were based on 2003 smoking prevalence estimates of the Canadian population (4) (current smoker prevalence = 23%), the availability of annual BC smoking prevalence from the Canadian Tobacco Use Monitoring Survey (CTUMS) presents an opportunity to more accurately estimate SAM and SAH in BC. BC has the lowest prevalence of current smokers among all provinces in Canada, so use of the Canadian prevalence would likely overestimate SAM and SAH. In addition, using a static set of AFs to estimate SAM and SAH over time does not take into account changes in prevalence over time. The prevalence of current smokers in BC has decreased from 20% in 2000 to 14% in 2007. (3)

Using current smoking data for BC, the objectives of this study are to:

1) adjust the AFs used in the CSAC report for each year of BC smoking prevalence data from 2002 to 2007;

2) apply the adjusted AFs to the respective annual BC mortality and hospitalization data to calculate the number of SAM and SAH;

3) and calculate standardized rates of SAM and SAH by gender, health region and disease category.

METHODS

Data sources

Mortality and Hospital Data

Mortality and hospital data by 5-year age group, sex, geographic health region, and ICD-10 code were obtained from BC Vital Statistics and the BC Ministry of Health, respectively, for 2002-2007.

Attributable Fractions and Calculation of SAM

Smoking AFs for chronic diseases were calculated using the formula:

AF= [[[summation].sup.k.sub.i=1] [P.sub.i]([RR.sub.i]-1)]/[[summation].sup.k.sub.i=0] [P.sub.i]([RR.sub.i]-1) + 1]

Where:

k = total levels of exposure

i = exposure category with baseline exposure or no exposure i=0

RR(i) = relative risk at exposure level i compared to no consumption

P(i) = prevalence of the ith category of exposure.

AFs were calculated by age groups and sex. The age- and sex-specific AFs were then multiplied with the mortality and hospitalization data to estimate SAM and SAH.

The AF for fire injury was calculated using direct estimates of smoking involvement. (12) This number was used for all estimates.

Prevalence of Smoking

BC smoking prevalence was taken from Canadian Tobacco Use Monitoring Survey (CTUMS) by gender and age groups. In BC, about 2,000 adults 15 years of age and older were sampled per year from 1999 onwards. The sample was weighted prior to calculating prevalence. Because of the low number of survey participants in the higher age groups, a 3-year moving average was calculated for 2002 to 2007 (e.g., to calculate the smoking prevalence in 2002, the weighted average of the 2001, 2002, and 2003 smoking prevalence was used). Categories of smoking prevalence included current, former and never-smokers. Current smoker was further broken down into occasional smoking or daily smoking categories. Detailed smoking definitions can be found in the CSAC report.

[FIGURE 1 OMITTED]

Relative Risks

Relative risks were taken from the CSAC report. Briefly, a list of causal health conditions attributable to smoking was compiled (Table 1). A comprehensive search of meta-analyses was performed for each disease category and its risk relationship with smoking. Where possible, the most detailed dose-response relative risks were used.

Rate Calculation

Rates were age- and sex-standardized using the direct method and the 2001 age 15 and over BC population as the standard.

RESULTS

Among active smoking adults 15 years of age and older, there were an estimated 4,851 deaths and 25,314 hospitalizations attributed to smoking in BC in 2007 (Tables 2 and 3). Males accounted for over 60% of those deaths and hospitalizations. In 2007, 15.6% of all deaths and 3.4% of all hospitalizations in BC were attributed to smoking.

The mortality rates attributable to smoking have hovered around 120 deaths per 100,000 adults over the six-year period from 2002-2007. In the most recent two years of that period, the rates have dropped below that benchmark. From 2002 to 2005, hospitalization rates attributable to smoking increased from 640 to 663 hospitalizations per 100,000, respectively; but like mortality, SAH declined in the subsequent two years to 632 per 100,000 in 2007.

Geographically, the Northern health authority (HA) has consistently held the highest rates of SAM and SAH over the course of the study period, while Vancouver Coastal HA had the lowest. With the largest population, Fraser HA has the largest number of SAM and SAH, with about 5 times more deaths and 4 times more hospitalizations than the Northern HA. Figure 1 displays the SAM and SAH rates by BC health services delivery area (HSDA) in 2006. The areas of high SAM rates generally correspond with areas of high SAH rates. The metro areas of Vancouver and Victoria have among the lowest rates of SAM and SAH.

Almost half of the deaths in 2007 were from cancer (47.8%), with lung cancer accounting for 75% of all smoking-related cancer deaths. Mortality rates by condition can be found in Table 1. Respiratory diseases accounted for 26.6% and cardiovascular disease accounted for 24.8% of smoking-related deaths. The top causes of SAM in 2007 were lung cancer (1,727 deaths), chronic obstructive pulmonary disease (COPD) (1,079), ischemic heart disease (526), and cerebrovascular disease (312). Together, these conditions accounted for 75% of all SAM.

In 2007, smoking caused 263 cardiovascular, 194 respiratory, and 146 cancer hospitalizations per 100,000 adults (Table 1). COPD, ischemic heart disease, bladder cancer, lung cancer and cardiac arrhythmia were the top five causes of SAH, accounting for 70% of all SAH.

DISCUSSION

Despite the decrease of current smokers from 2002 to 2007 in BC, the harms associated with tobacco smoking have not shown the same dramatic decrease. In fact, only since 2006 have there been signs that the rates of SAM and SAH have started to decrease. This is likely due to the latency between smoking and health outcome. (13) In Canada, smoking prevalence has declined since the late 1960s; however, lung cancer rates did not start declining until about 1990. (14) The timeframe of this study is too short to determine which stage of the epidemic curve BC is currently at; however, with continuous monitoring, this will become clear over several years. Another contributing factor to the discrepancy was the increase in the number of 'experimental' users, those who tried a few cigarettes and then for whatever reason stopped. These users moved from being never-smokers to former smokers, artificially increasing their risk of smoking-related harms. A study found that redefining former smokers as having smoked at least 100 cigarettes in a lifetime could decrease SAM estimates by 5%. (15) It has not been determined which definition would produce the most accurate measurement.

In comparison with the SAM estimates produced by BC Vital Statistics in their annual reports, (16) our SAM estimates were 21-26% lower each year over the six-year period for the same age range. The relative risks used by BC Vital Statistics were taken from the American Cancer Society's Cancer Prevention Study II (CPS II). The CPS II has been criticized for not having a nationally representative sample and for not adjusting for potential confounding factors. (17,18) Studies that estimated SAM using relative risks derived from a more representative US sample found that their estimates were between 16% and 40% lower than those derived from using the CPS II. (17) Our estimates are also based on more detailed exposure and relative risk categorization.

There were considerable differences in SAM and SAH rates between health regions in BC, with the Northern HA having over 40% more deaths and 70% more hospitalizations than Vancouver Coastal HA. Northern HA has the highest prevalence of current smokers compared with the other HAs. (19) The high rates in Northern HA are not confined to one HSDA, nor are the low rates in Vancouver Coastal HA. Thompson Cariboo Shuswap HSDA of Interior HA and Fraser East of Fraser HA both have noticeably higher SAM and SAH rates than the other HSDA areas in their respective region.

Limitations

Although AFs can be used to theoretically determine the number of deaths and hospitalizations caused by smoking, there is currently no way to determine whether these estimates hold in real life. We cannot directly link smoking as a causal factor solely on the basis of diagnostic codes. There are other methodological issues that can impact SAM and SAH estimates, such as changes in exposure measurement, exposure and relative risk categorization, or relative risk estimates. (15) However, our methodology is consistent with past Canadian studies of the same nature. The study does not include the harms related to second-hand smoke (passive smokers) or to maternal smoking. Both make up a relatively small fraction of SAM and SAH. Although AFs were adjusted using BC smoking prevalence, there are regional variations of smoking prevalence within BC. The SAM and SAH estimates would be an overestimate in low smoking prevalence regions and an underestimate in high smoking prevalence regions.

Despite potential issues with the methodology, these estimates play an important role in public health. They inform the public of the harms associated with smoking, help researchers identify high-risk areas and evaluate smoking reduction programs, and provide policy-makers with evidence of the effectiveness of policies. The methods are adaptable to other provinces and only require administrative data. Using attributable fractions adjusted for annual smoking prevalence to estimate SAM has been shown to be comparable to estimates derived from physician reports of tobacco-contributing deaths. (20)

CONCLUSION

Smoking still presents a substantial human and economic burden in BC and in Canada. There is some indication that the recent trend is downwards, but with the long latency period between smoking and health outcome, long-term and ongoing follow-up is needed to see if this trend is sustained. The presence of the BC Alcohol and Other Drug Monitoring Project will allow for continued surveillance of emerging trends. Areas in BC with higher rates should be targeted for appropriate tobacco prevention and cessation programs. This methodology can be used by other provinces to allow comparisons between provinces and to observe trends over time by demographic and geographic characteristics.

Acknowledgements: The authors thank Kate Vallance and Gina Martin for their work on the BC Alcohol and Other Drug Monitoring Project.

Sources of Support: This study is funded by the Alcohol and Other Drug Monitoring Project.

Conflict of Interest: None to declare.

Received: October 25, 2011

Accepted: January 21, 2012

REFERENCES

(1.) Mathers CD, Loncar D. Projections of global mortality and burden of disease from 2002 to 2030. PLoS Med 2006;3(11):e442.

(2.) Murray CJ, Lopez AD. Alternative projections of mortality and disability by cause 1990-2020: Global Burden of Disease Study. Lancet 1997;349(9064):1498-504.

(3.) Health Canada. Canadian Tobacco Use Monitoring Survey. Available at http://www.hc-sc.gc.ca/hc-ps/tobac-tabac/research-recherche/stat/ index-eng.php (Accessed October 20, 2011).

(4.) Rehm J, Baliunas D, Brochu S, Fischer B, Gnam W, Patra J, et al. The cost of substance abuse in Canada 2002. Ottawa, ON: Canadian Centre on Substance Abuse, 2006.

(5.) Groenewald P, Vos T, Norman R, Laubscher R, van Walbeek C, Saloojee Y, et al. Estimating the burden of disease attributable to smoking in South Africa in 2000. S Afr Med J 2007;97(8 Pt 2):674-81.

(6.) Wen CP, Tsai SP, Chen CJ, Cheng TY, Tsai MC, Levy DT. Smoking attributable mortality for Taiwan and its projection to 2020 under different smoking scenarios. Tob Control 2005;14(Suppl 1):i76-i80.

(7.) Zorrilla-Torras B, Garcia-Marin N, Galan-Labaca I, Gandarillas-Grande A. Smoking attributable mortality in the community of Madrid: 1992-1998. Eur J Public Health 2005;15(1):43-50.

(8.) English DR, Holman CDJ, Milne E, Winter MJ, Hulse GK, Codde JP, et al. The quantification of drug caused morbidity and mortality in Australia 1995. Canberra, Australia: Commonwealth Department of Human Services and Health, 1995.

(9.) State-specific smoking-attributable mortality and years of potential life lost--United States, 2000-2004. MMWR Morb Mortal Wkly Rep 2009;58(2):29-33.

(10.) Rodriguez TR, Bueno CA, Pueyos SA, Espigares GM, Martinez Gonzalez MA, Galvez VR. [Morbidity, mortality and the potential years of life lost attributable to tobacco]. Med Clin (Barc) 1997;108(4):121-27.

(11.) Stockwell T, Buxton J, Duff C, Marsh D, Macdonald S, Michelow W, et al. The British Columbia Alcohol and Other Drug Monitoring System: Overview and early progress. Contemporary Drug Problems 2009;36(3-4):459-84.

(12.) Kashaninia Z. Fire deaths in British Columbia, 1986 to 1998. Victoria, BC: British Columbia Vital Statistics, 1999.

(13.) Lopez AD, Collishaw NE, Piha T. A descriptive model of the cigarette epidemic in developed countries. Tob Control 1994;3:242-47.

(14.) National Cancer Institute of Canada. Canadian Cancer Statistics 1991. Toronto, ON: NCIC, 1991.

(15.) Tanuseputro P, Manuel DG, Schultz SE, Johansen H, Mustard CA. Improving population attributable fraction methods: Examining smoking-attributable mortality for 87 geographic regions in Canada. Am J Epidemiol 2005;161(8):787-98.

(16.) British Columbia Vital Statistics Agency. Selected vital statistics and health status indicators: One hundred and thirty-sixth annual report 2007. Victoria, BC: British Columbia Vital Statistics Agency, 2007.

(17.) Malarcher AM, Schulman J, Epstein LA, Thun MJ, Mowery P, Pierce B, et al. Methodological issues in estimating smoking-attributable mortality in the United States. Am J Epidemiol 2000;152(6):573-84.

(18.) Sterling TD, Rosenbaum WL, Weinkam JJ. Risk attribution and tobacco related deaths. Am J Epidemiol 1993;138(2):128-39.

(19.) Ipsos Reid, BC Ministry of Health. Smoking prevalence in British Columbia: Final Report. Victoria, BC: BC Ministry of Health, 2003.

(20.) Thomas AR, Hedberg K, Fleming DW. Comparison of physician based reporting of tobacco attributable deaths and computer derived estimates of smoking attributable deaths, Oregon, 1989 to 1996. Tob Control 2001;10(2):161-64.

Andrew W. Tu, MSc, [1] Jane A. Buxton, MBBS, [1] Tim Stockwell, PhD [2]

Author Affiliations

[1.] British Columbia Centre for Disease Control, Vancouver, BC

[2.] Centre for Addictions Research of BC, University of Victoria, Victoria, BC

Correspondence: Andrew Tu, British Columbia Centre for Disease Control, 655 West 12th Ave., Vancouver, BC V5Z 4R4, Tel: 604-707-2557, Fax: 604-707-2516, E-mail: andrew.tu@bccdc.ca

Table 1. Health Conditions and the Corresponding ICD-10 Codes
Attributable to Smoking

Condition
                                            2007            2007
                                         Mortality    Hospitalization
                       ICD-10             Rates *         Rates *

Active Smoking

Malignant neoplasms                         56.5           146.2

  Oropharyngeal        C00-C14, D00.0        2.1             6.8
  cancer

  Oesophageal cancer   C15, D00.1            2.8             5.7

  Stomach cancer       C16, D00.2            0.7             2.2

  Pancreatic cancer    C25, D01.9            1.8             2.2

  Laryngeal cancer     C32, D02.0            0.7             3.4

  Trachea, bronchus    C33, C34             42.3            52.8
  and lung cancers

  Cervical cancer      C53, D06              0.4             7.5

  Urinary tract        C64-C68               0.3             2.0
  cancer

  Renal cell           C64                   0.7             2.2
  carcinoma

  Bladder cancer       C67, D09.0            4.2            60.2

  Acute myeloid        C92.0                 0.5             1.3
  leukaemia

Cardiovascular                              29.9           263.1
diseases

  Ischaemic heart      I20-I25              12.8           125.7
  disease

  Pulmonary            I26-I28               2.2            22.2
  circulatory disease

  Cardiac              I47-I49               1.0            36.1
  arrhythmias

  Heart failure;       I50-I51               1.7            19.0
  complications and
  ill-defined
  descriptions of
  heart disease

  Cerebrovascular      I60-I69               8.0            27.3
  diseases

  Atherosclerosis      I70-I79               4.3            32.8

Respiratory diseases                        31.3           194.1

  Pneumonia and        J10-J18               5.1            27.1
  influenza

  Chronic              J40-J44              26.2           167.0
  obstructive
  pulmonary disease

Other diseases                               0.9            29.3

  Mental and           F17                   0.1             0
  behavioural
  disorders due to
  use of tobacco

  Toxic effects of     T65.2                 0               0
  tobacco

  Ulcers               K25-K28               0.6            28.4

  Fires                X00-X09               0.2             0.9

Total                                      118.7           632.8

* Rates per 100,000 adults age 15+; numbers may not add up due to
rounding.

Table 2. Smoking-attributable Mortality Rates (per 100,000),
2002-2007, and Smoking-attributable Mortality, 2007 for BC
Population Age 15+

                                       Mortality Rate (per 100,000)

                                2002   2003   2004   2005   2006   2007

Sex *
  Male                          152    152    149    147    142    146
  Female                         94     92     98     95     90     95

Health Authority ([dagger])
  Interior                      129    134    140    130    120    128
  Fraser                        116    121    123    122    115    119
  Vancouver Coastal             110    104    102    105    100     99
  Vancouver Island              127    117    122    120    115    122
  Northern                      164    149    151    135    155    163
  British Columbia ([dagger])   121    120    122    120    114    119

                                Smoking-attributable
                                  Mortality, 2007.
                                  ([double dagger])

Sex *
  Male                                 2948
  Female                               1903

Health Authority ([dagger])
  Interior                             1049
  Fraser                               1479
  Vancouver Coastal                     954
  Vancouver Island                     1052
  Northern                              317
  British Columbia ([dagger])          4851

* Rates are age-standardized.

([dagger]) Rates are age- and sex-standardized.

(double dagger]) Calculated using relative risks found in the Cost of
Substance Abuse in Canada, 2002 report.

Table 3. Smoking-attributable Hospitalization Rate (per 100,000),
2002-2007, and Smoking-attributable Morbidity, 2007 for BC Population
Age 15+

                                  Hospitalization Rate (per 100,000)

                                2002   2003   2004   2005   2006   2007
Sex *
  Male                          834    855    851    863    847    819
  Female                        464    459    463    480    457    462

Health Authority ([dagger])
  Interior                      729    741    749    763    733    735
  Fraser                        677    689    682    693    676    675
  Vancouver Coastal             499    500    513    531    519    483
  Vancouver Island              609    589    584    599    566    555
  Northern                      872    911    901    935    922    909
  British Columbia ([dagger])   640    646    648    663    643    632

                                Smoking-attributable
                                  Hospitalization,
                                2007 ([double dagger])

Sex *
  Male                               16,208
  Female                               9106

Health Authority ([dagger])
  Interior                             5735
  Fraser                               8450
  Vancouver Coastal                    4575
  Vancouver Island                     4439
  Northern                             1982
  British Columbia ([dagger])        25,314

* Rates are age-standardized.

([dagger]) Rates are age- and sex-standardized.

([double dagger]) Calculated using relative risks found in the Cost
of Substance Abuse in Canada, 2002 report.