Within the smoking model, individuals are classified as never smokers from birth until they initiate smoking or die. They may evolve from current to former smokers through cessation or may return to current smokers through relapse. The extent of relapse depends on the number of years since quitting.
Smoking prevalence data for the period 1999–2009 were available from the Multipurpose Surveys “Aspects of daily living”, carried out from 1993 onwards by the Institute of Health Information and Statistics of Italy (ISTAT) [20
]. These surveys were carried out almost every year on representative samples of the Italian population (about 24,000 families and 54,000 persons distributed in about 850 Italian municipalities of different population sizes). Smoking status was based on participant self-report as never, former, or current smoker [20
]. Smoking prevalence and initiation rate data were linearly interpolated to obtain data for the missing years. Figures on the subdivision of former smokers by duration since quitting were derived from the ISTAT Multipurpose Surveys “Aspects of daily living” [20
]. Because former-smokers were not distinguished by years-quit data after the first year, we used data from the Netherlands to estimate the proportions in the years-quit categories of 1
year or more.
Due to empirical challenges in measuring initiation and cessation and in order to ensure stability and internal consistency of the model, initiation rates at each age were measured as the difference between the smoking rate at that age year and the rate at the previous age year. We examined data from ISTAT regarding when smokers and former smokers stated that they first started smoking. Based on that information, we allowed initiation up through age 30.
To measure the annual cessation rate, we used data available for those who quit in the last year. Using the baseline smoking prevalence data, we constructed a cessation rate by age and gender, measured as those who quit in the last year as percent of “those currently smoking plus those who quit in the last year”. Since that rate does not allow for relapse of those who quit less than one year ago, we applied a 50% relapse rate to that measure. These rates are consistent with estimates suggested by West [21
] and those found in studies of quitting behaviors in the Netherlands [23
]. Since data were not available for Italy, we used U.S. relapse rates [8
], but we compared ex-smoker rates and made adjustments accordingly to calibrate the model. After calibrating the model, we increased the relapse rate to 75% for ages 30–34, 70% for ages 35–44, 65% for ages 45–54, and 60% for ages 55–64
includes a separate policy module for each of the major policies. Policy effect sizes are in terms of percentage reductions applied to smoking prevalence in the year when the policy was implemented, and, unless otherwise specified, applied to initiation and cessation rates in future years. Policies and effect sizes are summarized in Table
. The effect of implementing a policy depends on the prior level of that policy (e.g., the incremental effect of a complete work site ban is less when a country already has a partial worksite ban than if no smoking ban is implemented). We input data on policy levels for each year from the baseline year through 2009, based on the WHO MPOWER reports [1
] and on data provided by tobacco control staff in Italy.
Policies, description and effect sizes of the SimSmoke model
uses cigarette prices adjusted for inflation as the policy input through 2010, which changes in price are translated into changes in smoking prevalence through elasticities (i.e., the percent change in smoking prevalence for a one percent change in price). The model uses the total cigarette excise tax (including ad valorem and specific) to adjust future prices. Gallus and colleagues [36
] obtained price elasticities for Italy comparable to those in the U.S., with minimal tendencies for smokers to substitute to smuggled cigarettes and hand rolled cigarettes [37
]. The price effects are assumed to be the same as in U.S. model, where the elasticity for smoking prevalence is −0.3 for those through age 24, -0.2 for those aged 25–34, and −0.1 for those aged 35 and above. Since the vast majority of tobacco users smoke tobacco, the price was based on the tobacco price index deflated by the consumer price index from 1999 through 2009 (stats.oecd.org/index.aspx? querytype
221). We examine the effect of an increase in future tobacco prices by 25% in the first year (i.e. 2011), an additional 20% the second (relative to the preceding year), 15% the third, 10% the fourth and 5% the fifth year. This would imply a total 99.3% price increase over the next 5
years, which would be similar to increasing the current ad valorem and specific excise taxes from its current level by around 20%.
Smoke-free air policy
considers smoking restrictions in: 1) worksites, 2) restaurants and bars, and 3) other places, with their effect dependent on enforcement and publicity (based on the level of tobacco control campaigns). With strong enforcement and publicity, the effect of a ban in restaurants is 2%, in pubs and bars is 1%, in worksites is 6%, and in other public places is 1%, with the full effects dependent on full enforcement, based on recent reviews [4
] and studies from Italy [37
]. Italy now has strong smoke-free air laws. Dating back to 1975, there were bans on smoking in many public places, including hospital wards, in all the school classes, libraries, cinemas and museums, and in public transport (http://www.salute.gov.it/stiliVita/paginaInternaMenuStiliVitaFaq.jsp?id=44&menu=strumentieservizi&label=faq
), which is designated as a full ban in other public places. A ban was not implemented until 2005 for all worksites, bars and restaurants. While there is an exclusion for ventilated areas, it is very costly to comply with this regulation. Italy is considered to have weak worksite laws (ban in government and health facilities), and no ban in restaurant or pubs and bars through 2005, and then complete smoke-free air laws (worksites, bars and restaurants), and other public places since 2005. Based on information in the MPOWER Report [1
] for 2007, the enforcement level is set to 8 out of 10.
] distinguishes enforcement and 4 levels of direct (advertising and point of sale) and indirect (sponsorships, branding, or promotional discounts) marketing
as: 1) none, 2) minimal (up to 3 direct or 1 indirect ban), 3) moderate (4–6 direct or 1 indirect), and 4) complete (direct and indirect). With a complete direct and indirect ban, prevalence is reduced by 5% and initiation by 6%, and cessation is increased by 3% [4
]. Since 1983, there was a ban on direct advertising in newspapers, radio, and television. In 1991, the ban was extended to cover most indirect advertising, with some minor changes in 2004 related to Formula One racing sponsorship and free distribution practices. According to the MPOWER Reports, there is a ban on all types of indirect advertising. On that basis, we categorize Italy as having a complete (level 4) marketing ban from the baseline year to the present. Based on the 2009 MPOWER report [2
], however, enforcement is 10 for direct and 3 for indirect, which were averaged to set enforcement to 7 out of 10 for all years.
MPOWER provides 4 levels for health warnings
on cigarette packs: none, minimal (< 30% of the principal display area), moderate (covering at least 30% of the display area and includes seven pack warning criteria) and complete (covers at least 50% of the display area and includes all seven warning criteria and a ban on deceitful terms). Based on Levy et al. [4
] and more recent studies [44
] strong health warnings reduce prevalence by 2% and initiation by 1%, and increase cessation by 5%. When warning labels are moderate (low), prevalence is reduced by 0.75% (0.5%), cessation is increased by 2.5% (1.0%) and initiation is reduced by 0.5% (0.5%). When warning labels are moderate or high, a synergy from publicity through tobacco control campaigns reduces prevalence by an additional 1% and increases cessation by an additional 2% if tobacco control funding is high and half of that amount if it is medium. Health warnings were minimal in Italy until 2003. Following EU directives, warnings were increased to cover at least 30% of both sides of the pack in 2003, but have not included pictorial images. Thus, we assigned a moderate warning from 2003 to 2010.
specifies three levels for tobacco control campaigns
: 1) low: a national agency and minimal funding or employees, 2) medium: a national agency plus 10+ employees or per capita expenditures over $0.10 (USD) per capita, and 3) high: a national agency and expenditures over $0.50 per capita, and incorporates synergies arising from publicity surrounding other tobacco control policies. Based on reviews [4
] and several recent studies [50
], a well-funded tobacco control campaign in conjunction with other policies yields an effect size of 6.5%. A low campaign yields an effect size of 1%. Without other policies in place, the effects are reduced by half. Anti-tobacco media campaigns have been implemented in Italy in 2002–2003, 2004, 2005, 2009, and 2010 (http://www.salute.gov.it/servizio/campagna.jsp
). The 2010 media campaign was a 35
second television spot delivered on 15
days in summer, and cinema and radio spots delivered from November 15 to December 15, 2010. The MPOWER Report [1
] indicates health expenditures of less than $1 USD per person. Because of the sporadic and generally low level of he campaign, we assign a low intensity media campaign to Italy since 1999.
A strongly enforced policy restricting the purchase of cigarettes by youth
and with bans on self-service and vending machines reduces smoking prevalence by those under the age of 18 by as much as 25% [54
]. While retailers have a duty to ensure tobacco products are not sold to anyone under the age of 16, youth are rarely denied purchase of cigarettes in Italy [55
]. Enforcement is set to a zero level since the baseline year of the model, with no bans on vending machines or self-service displays.
includes four sub-policies. Treatment coverage is based on the places providing cessation treatments (physician offices, hospitals, community centres, provider offices and other). With a high-level media campaign, prevalence is reduced by 2.25%, and the cessation rate is increased by 12% in future years [48
]. The effect of a quitline also depends on publicity, with prevalence reduced by 0.5% and cessation increased by 5% in all future years [56
]. An index for pharmacotherapy availability was developed, which was given full weight if: i) both nicotine replacement therapy (NRT) and buproprion were available, and ii) NRT is available in a pharmacy without prescription or general store. When fully implemented, prevalence is reduced by 1.0%, and the cessation rate is increased by 6% in all future years [57
]. Brief interventions would involve at minimum that health care providers advise and assist in cessation. When fully implemented, prevalence is reduced by 0.5% in the first year, and the cessation rate is increased by 10%, based on evidence provided in Levy et al. [4
]. When all sub-policies are implemented (from a scenario of no policy), smoking prevalence is reduced by 4.75% and the cessation rate is increased by 39.3%.
Since 2003, NRT has been available OTC in pharmacies, and was only provided by prescription in previous years. Buproprion is available with a prescription for all years. The National Health Service provides no reimbursement to smokers for pharmacotherapy or behavioral treatments [59
] and there is little use, although heavy smokers are more frequent compared with other high income nations [60
]. Several Smoking Cessation Services are not adequately funded and their activities are only periodically implemented [62
]. According to the MPOWER Report [2
] and based on the characteristics described above, financial coverage of treatments is provided only in some places for primary care facilities, hospitals, and offices of health professionals, and we assume those values for all years. MPOWER [2
] indicates that there is a quitline, which Italian sources indicate has been in existence since 1999, but is available for limited hours and serves only about 0.6-0.7% of smokers [61
], and is considered to be passive. A recent survey found that only 22% of smokers reported having only received advice to quit by their general practitioners [63
], but another recent survey (
PASSI; see http://www.epicentro.iss.it/passi/english.asp
) found that health professionals delivered smoking cessation counseling to 60% of smokers. We set brief interventions at a level of 30% for all years.
The model outcomes
As described above, the model estimates the effects over time for two primary outcomes: smoking prevalence and SADs. Smoking prevalence is provided for the population aged 15 and above, but the model also has the capability to provide breakdowns by age. Separate results are provided for males and females and overall. The model estimates these outcomes for the tracking period, which is from 1999 to 2010, and projects future outcomes for 2011 through 2040.
To calibrate the model, we compared trends in the model over the period from 1999 to 2002 to data gleaned from ISTAT surveys over the period 1993–2002. To validate the model, we compared trends from the model to data from ISTAT for more recent years.
In examining the potential effect of future policies consistent with the FCTC, we first present the status quo case, where tobacco control policies are maintained at their 2010 level. We then consider the effect of varying levels of tobacco control policies in isolation and through a comprehensive tobacco control strategy involving all policies being simultaneously implemented in 2011. In comparing the effect of policies to the status quo, we focus on the relative change in smoking prevalence, i.e., the change in smoking prevalence from the status quo in response to the future policy scenario divided by the status quo smoking prevalence. For SADs, we calculate lives saved as the difference between the number of deaths under the new policy and the number of deaths under the status quo.
For the present study, no ethical approval is required since only aggregate data obtained from published databases were used