Article Text

Economic evaluation of the recent French tobacco control policy: a model-based approach
  1. Marion Devaux1,
  2. Marina Dorfmuller Ciampi1,
  3. Romain Guignard2,
  4. Aliénor Lerouge1,
  5. Alexandra Aldea1,
  6. Viêt Nguyen-Thanh2,
  7. François Beck2,
  8. Pierre Arwidson2,
  9. Michele Cecchini1
  1. 1OECD, Paris, France
  2. 2Santé Publique France, The French Public Health Agency, Saint-Maurice, France
  1. Correspondence to Dr Marion Devaux, OECD, Paris 75016, France; marion.devaux{at}oecd.org

Abstract

Background One in four French adults smoked daily in 2021, compared with one in six in Organisation for Economic Co-operation and Development (OECD) countries. To strengthen its tobacco control policy, in 2016, France has started implementing a policy package that includes a 3-year gradual price increase, plain packaging, an annual social marketing campaign promoting cessation and the reimbursement of nicotine replacement products. This study aims to evaluate the health and economic impact of this policy package.

Methods The long-term policy impact on disease cases, healthcare expenditure and gains in labour participation and productivity was evaluated by using the OECD microsimulation model for Strategic Public Health Planning for Non-Communicable Diseases. The model was fed with historical and projected trends on tobacco smoking prevalence as produced by the policy package.

Results Over the period 2023–2050, the policy package is estimated to avoid about 4.03 million (2.09–11.84 million) cases of chronic diseases, save €578 million (365–1848 million) per year in health expenditure and increase employment and workforce productivity by the equivalent to 19 800 (9100–59 900) additional full-time workers per year, compared with a scenario in which the intervention package is not implemented. The intervention cost is estimated at about €148 million per year. For each euro invested in the policy package, €4 will be returned in long-term savings in healthcare expenditure.

Conclusions The tobacco control policy package implemented by France, targeting smoking initiation and promoting tobacco cessation is an effective intervention with an excellent return on investment.

  • Prevention
  • Public policy
  • Taxation
  • Social marketing
  • Economics

Data availability statement

Data are available in a public, open access repository. All data relevant to the study are included in the article or uploaded as supplementary information. Input data for the model are publicly available. Data produced by the model are presented in the article and the supplementary file.

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

WHAT IS ALREADY KNOWN ON THIS TOPIC

  • A comprehensive tobacco control policy package has been implemented by France since 2016.

  • Previous studies have evaluated the effectiveness of single interventions on tobacco consumption, but no study has yet evaluated the impact of the whole policy package.

  • We searched PubMed and Scopus for studies that aimed to assess tobacco control policies in France, published in English or French between 2013 and 2023.

  • We were not able to identify any study evaluating the whole tobacco control policy package recently implemented in France and its potential impact on future smoking trends, disease prevalence and economic outcomes.

WHAT THIS STUDY ADDS

  • Based on a simulation model, results show that the long-term benefits of the tobacco control policy package in France will outpace the implementation costs, producing a significant return on investment if the policy is continued.

  • Specifically, the policy package is expected to avoid about 4 million cases of chronic diseases by 2050, to save €578 million per year in health expenditure - corresponding to a return of €4 for each euro invested in the programme - and to increase workforce employment and productivity by the equivalent to 19 800 additional workers per year, compared to a scenario in which the policy package is not implemented.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

  • These findings bring the economic argument in support of the long-standing WHO Framework Convention on Tobacco Control (FCTC) recommendations.

  • This study supports national and European-level policy makers to implement strengthened tobacco control policies with the objective to reduce smoking prevalence, limit the consequences of tobacco consumption, and to achieve a tobacco-free generation.

Introduction

Tobacco smoking was responsible for 7.7 million deaths worldwide in 2019 and accounted for 20% of deaths among men.1 Smoking prevalence in France is high compared with other high-income countries, with 32% of adults who currently smoke and 25% of adults who smoke daily in 2021.2 3 In France, tobacco was responsible for 75 000 deaths in 2015, with a heavier burden among men.4

To tackle the burden of tobacco smoking and to respond to increases in smoking rates observed between 2005 and 2010, France launched its first national smoking reduction programme in 2014, followed by a national tobacco control programme in 2018. Under these programmes, France implemented a policy package between 2016 and 2019, which counts four main pillars. The first pillar is a 3-year gradual tax increase for tobacco products, corresponding to a 41% price increase of the most sold cigarette packs. The second one is mandatory plain packaging as of January 2017. The third one is the reimbursement of nicotine replacement therapies (NRT) as for other medicines in 2019; while the final pillar is an annual social marketing campaign to promote smoking cessation (Mois Sans Tabac) that started in 2016 and was adapted from the British Stoptober campaign.

During the implementation of the policy package, between 2016 and 2019, smoking prevalence decreased from 35% to 31% and quit attempts among people who smoked daily increased from 25% to 33%.5 In addition, sales of NRT increased by 28% between 2019 and 2020, supported by the tobacco control policy package.6 However, smoking prevalence has stabilised since 2020, which could be partly attributed to the COVID-19 crisis.3 While the majority of people who smoked did not change their habits during the pandemic, some increased or decreased their consumption, with various profiles observed according to gender, age, socioeconomic status, working condition or mental health status.7 8 A slowing down in the decline of smoking prevalence rates was also observed in other European countries such as Italy and Ireland.2

Previous studies have assessed the effectiveness of single components of the policy package in France. For instance, the hike in tobacco taxation was associated with reduced consumption with a price elasticity of −0.4 over the period 2000–2015.9 The full reimbursement of the medical management for smoking cessation, including cessation treatments and medical consultations, is found to be highly cost-effective in France.10 The first edition of the Mois Sans Tabac campaign was also shown to be effective in increasing quit attempts in the general population.11 Lastly, the introduction of plain packaging in 2017 and the increase in the size of graphic health warnings were found to potentially contribute to changes in people’s beliefs and attitudes, such as concerns about the consequences of smoking.12

Building on the existing evidence, this study complements the available evidence by carrying out, for the first time, an economic evaluation of the whole package of tobacco control policies implemented by France. Specifically, this study estimates the long-term effects of the policy package over the period 2023–2050 using the Organisation for Economic Co-operation and Development (OECD) microsimulation model for Strategic Public Health Planning for Non-Communicable Diseases (SPHeP-NCD). The model simulates the policy package’s impact on individual’s behaviours, disease incidence and remission, life expectancy, as well as on healthcare expenditure, and on employment and labour productivity. This study adds to the existing body of evidence on modelling economic studies on tobacco control policy in other countries.13–16

Methods

Microsimulation model

The OECD SPHeP-NCD model is an advanced systems modelling tool for public health policy and strategic planning. The model projects future population health and economic outcomes and can be used to assess the long-term impacts of policies. It estimates outcomes over the 2023–2050 period and covers a total of 52 countries including France. The OECD SPHeP-NCD model, described in the online supplemental file 1, has been used in other analyses that can be found elsewhere.17–19

Supplemental material

To obtain results on health and economic indicators, inputs combine data on behavioural and physiological risk factors such as smoking, as well as demographics and population health data. The model integrates a sophisticated smoking module, in which the relationship between smoking and disease is quantified through age-specific and sex-specific relative risks. Data sources include international databases,20–22 as well as national reports and surveys on smoking behaviours from Santé Publique France3 and data on smoking initiation from the EnClass survey administered by the French Observatory for Drugs and Addictive Tendencies.23

The microsimulation model assesses the risk of developing a non-communicable disease (NCD) on an individual basis. The model incorporates 11 disease categories, and six of them are linked to smoking. These include stroke, ischaemic heart disease, cancer, chronic obstructive pulmonary disease (COPD), musculoskeletal disorders (MSD) and diabetes. The risk profile of individuals is built by considering their age and smoking status (never smoked, currently smoke or previously smoked). The quantity of cigarettes smoked is also considered for those who currently smoke, while time since quitting is considered for those who quit smoking. Finally, individuals’ risk profiles are also affected by other risk factors.

The data to model the relative risks of developing an NCD are collected from the IHME database.1 The model uses a competing event framework, meaning that for any individual in the model, diseases and causes of death compete against one another to determine the death of an individual.

Model calibration and assumptions for long-term projections of smoking prevalence

The model is built to replicate the historical and projected epidemiology and demographic of France. The disease incidence and prevalence in the French population have been calibrated to match estimates from national and international databases.3 20 21 Calibration methodology and results are made available in the online supplemental file.

The model produces outputs for two scenarios, including the policy scenario and the no-policy scenario. Results of the study present the comparison of these two scenarios, allowing for the assessment of the policy impact produced by the intervention, with future outcomes that are discounted at a 3% rate.

In the policy scenario, the model is calibrated to replicate the trends in tobacco use and disease incidence experienced by France following the introduction of the tobacco control policy in 2016. In this scenario, tobacco smoking prevalence follows historical trends for the period 2016–2019 as found in survey-based data. For the period between 2020 and 2050, smoking prevalence is projected using an ensemble model that incorporates 15 model estimates. Future smoking prevalence rates are projected from historical trends using four different functional forms. Specifically, linear, exponential, power and constant regression models were used to fit average tobacco prevalence rates by age and sex, based on the 2000–2019 data. The exponential function was also used on the 2016–2019 data, assuming a more optimistic trend in smoking prevalence. More information on the ensemble model is available in the online supplemental file 1. This resulted in a set of 15 predicted values with variations used to create a projection band. This approach was selected as it reduces the generalisation error of the prediction.24 Outputs from the ensemble model produced a distribution of smoking prevalence estimates for each year of the forecast of which the average was used as central value and the values at the 10th and 90th percentiles as the minimum (low end) and the maximum (high end). The central value of the policy scenario assumes a decrease in smoking rates in the general population from 24% in 2019 down to 17% by 2050 (see the online supplemental file). Smoking prevalence rates in the high-end scenario display a continuous reduction down to 3%, while the rates in the low-end scenario are projected to stagnate (around 25%).

The no-policy scenario is calibrated under the assumption that the policy package was never implemented. In this scenario, the levels of tobacco smoking remain higher causing a worsening of population’s health outcomes. Specifically, the predicted trend for the period 2016–2050 is based on the projections of the 2000–2015 data using a power regression and stratified by sex and age group.

Model outputs

Model outputs include health outcomes such as life expectancy, disease prevalence and disability-adjusted life years (DALY), for which standard weights were used.25 Disease prevalence was measured for the following categories of diseases: MSDs, lower respiratory infection (LRI), COPD, cancers related to smoking, cardiovascular diseases (CVDs), dementia and diabetes. The cancers for the following organs are included: lung, breast, colon and rectum, liver and oesophagus. Individuals who smoke are also at higher risk of MSD such as rheumatoid arthritis, low back pain and hip fracture.1

Economic outcomes include healthcare costs extrapolated from the French health-related expenditure data and costs related to reduction in workforce employment and productivity. Healthcare costs were estimated based on a per-case annual cost using the French database Echantillon généraliste de bénéficiaires.26 Other costs include lost labour market output, such as reduced likelihood of employment and higher productivity losses (eg, presenteeism, absenteeism and early retirement). These costs were estimated in number of full-time equivalent workers (FTEWs) and calculated by using a human capital approach and national average wages, which follows normal practice for gauging the economic burden of smoking.27 Relative risks to relate disease status to labour market outputs were retrieved from the literature.28 29 Additional information on modelling costs is available in the online supplemental file.

Sensitivity analysis

The primary analysis of the policy evaluation is based on the effects of the policy observed before the COVID-19 pandemic, hence in the period 2016–2019. As the COVID-19 crisis may have contributed to the stagnation in the long-standing smoking prevalence decline,3 a sensitivity analysis was carried out to assess the potential impact produced by the pandemic. Concretely, the sensitivity analysis extends the period for which historical data are used to carry out the projections of smoking prevalence rates by including data for years 2020 and 2021.

Results

Health outcomes

Findings suggest that the policy package implementation is likely to prevent about 4.03 million cases of NCDs (2.09 and 11.84 million cases under the high-end and the low-end scenarios) by 2050. Under the central scenario, the French population avoids developing 1.87 million cases of MSDs, 1.54 million cases of LRIs, 275 000 cases of COPD, 170 000 cases of cancers, 104 000 cases of CVDs, 40 000 cases of dementia and 32 000 cases of diabetes in cumulative numbers over 2023–2050. In relative terms, the policy is estimated to avoid 5.5% of new cases of COPD, 3.4% of new cases of cancers, nearly 2% of new cases of MSDs, 1.5% of new cases of LRI, 1.2% of new cases of diabetes, 0.7% of new cases of dementia and 0.5% of new cases of CVDs (figure 1).

Figure 1

Impact of the policy package on disease cases, relative numbers, average per year over 2023–2050. The whiskers represent high-end and low-end scenarios. Source: Organisation for Economic Co-operation and Development Strategic Public Health Planning for Non-Communicable Diseases (OECD SPHeP-NCD) model, 2023. COPD, chronic obstructive pulmonary disease; CVD, cardiovascular disease; LRI, lower respiratory infection; MSD, musculoskeletal disorder.

The policy was estimated to increase both life years (LYs) and healthy LYs (DALYs). By 2050, more than 644 000 LYs (436 000–2 168 000) and 1 040 000 DALYs (636 000–3 389 000) will be added at the population level (figure 2). When converted to relative terms, this corresponds to 62 LYs (41–206) per 100 000 population and 99 DALYs (60–318) per 100 000 gained every year.

Figure 2

Impact of the policy package on life years (LYs) and disability-adjusted life years (DALYs), cumulative number over 2023–2050, discounted. The shaded bands represent high-end and low-end scenarios. Source: Organisation for Economic Co-operation and Development Strategic Public Health Planning for Non-Communicable Diseases (OECD SPHeP-NCD) model, 2023.

Healthcare expenditure savings

The policy package is estimated to reduce health expenditure by €578 million (€365–€1848 million) annually over 2023–2050. In relative terms, annual health expenditure savings are estimated at €8.7 per capita per year, which corresponds to 0.32% of total health spending. These estimates range from €5.5 (0.20% of total health spending) to €27.8 (1.02%) in the low-end and high-end scenarios. Figure 3 shows the predicted cumulative health expenditure savings over 2023 and 2050. By 2050, up to €143 (€91–€466) per capita would be saved in healthcare expenditure as a result of the tobacco control policy package.

Figure 3

Impact of the policy package on health expenditure saving, cumulative number over 2023–2050, discounted. The shaded bands represent high-end and low-end scenarios. Source: Organisation for Economic Co-operation and Development Strategic Public Health Planning for Non-Communicable Diseases (OECD SPHeP-NCD) model, 2023.

The policy package is cost saving with a return on investment (ROI) of €4 (€2.5–€12.5) for each euro invested in the programme. Total costs of the policy package were estimated at €148 million per year, which includes the cost of enforcing the tax increase, the cost of running the Mois Sans Tabac campaign and the reimbursement of NRT (see the online supplemental file). The cost for implementing tobacco plain packaging is deemed negligible for the government, as the cost is mainly borne by the industry. Revenues from increased taxation were not included in the analysis, following the WHO CHOICE (CHOosing Interventions that are Cost-Effective) approach for analyses on taxation.30

Effect on employment and productivity

As people are less likely to develop tobacco-related diseases, they are more likely to actively engage in the labour market. They are also likely to be more productive if they are employed. The policy package is thus estimated to increase employment and productivity by the equivalent of 19 800 (9100–59 900) additional FTEWs per year. Increased employment is the largest contributor to the increase (68%), followed by reduced presenteeism (19%), reduced absenteeism (12%) and reduced early retirement (1%). The proportions remain unchanged across the high-end and low-end scenarios.

When expressed in monetary terms using average wages, gains in improved workforce participation and productivity were estimated at €18 per capita per year (€7.9–€54.7), corresponding to €715 million per year at the population level (€307–€2129 million).

Sensitivity analysis

The sensitivity analysis shows that trends in smoking prevalence are predicted to decrease at a slower pace, and that the impact of the policy package is reduced by 10–15% due to the COVID-19 pandemic. Figure 4 depicts the policy impact on disease cases in the primary and the sensitivity analyses. In the sensitivity analysis integrating the impact of the COVID-19 crisis, the policy is estimated to prevent 4.8% (3.0%; 8.1%) of newly diagnosed COPD cases, compared with 5.5% (3.3%; 16.7%) in the primary analysis. Similarly, other outcome results show consistency with the primary analysis with reduced impact (see the online supplemental file).

Figure 4

Impact of the policy package on disease cases in the primary and the sensitivity analyses, relative numbers, average per year over 2023–2050. The whiskers represent high-end and low-end scenarios. Source: Organisation for Economic Co-operation and Development Strategic Public Health Planning for Non-Communicable Diseases (OECD SPHeP-NCD) model, 2023. COPD, chronic obstructive pulmonary disease; CVD, cardiovascular disease; LRI, lower respiratory infection; MSD, musculoskeletal disorder.

Discussion

This study provides novel and unique quantitative estimates of the potential impact of the tobacco control policy package implemented by France since 2016. It assesses the impact of the policy package in terms of both health and economic outcomes, including savings in healthcare expenditure and increased workforce productivity. In the central scenario, and over the period 2023–2050, the policy is estimated to avoid about 4 million NCD cases and to produce savings in health expenditure in the order of €578 million, compared with a scenario in which the intervention package was not implemented. Workforce productivity would be also boosted by €715 million. For each euro invested, the policy package is estimated to return €4 in healthcare expenditure savings.

Two recent studies evaluated the ROI of tobacco control policy in Canada and Florida, USA. In Canada, for each dollar invested in the tobacco control policy implemented between 2001 and 2016, C$19.8 are returned in wider economic benefits (healthcare costs, productivity costs and monetised LYs lost, as well as tax revenues).14 When narrowing to healthcare savings only, the ROI is estimated at 8:1. In the US study, for each dollar invested in the tobacco control programme between 1999 and 2015, smoking-attributable healthcare expenditures decreased by US$10, corresponding to an ROI of 10:1.15 Our study found an ROI (4:1) that is lower than in prior studies. This can be explained by the fact that our dynamic simulation model accounts for healthcare expenses for individuals who are prevented from smoking but who may develop chronic diseases that are unrelated to smoking and incur related healthcare costs.

Results presented in this study are based on dynamic simulations that capture changes in demographics and population age structure by 2050, allowing a realistic assessment of the policy impact on the burden of disease in future years. The model uses a competing event framework, according to which diseases and causes of death compete one against the others to determine the death of individuals. Thus, the model assumes that people who do not develop (or die from) a chronic disease caused by tobacco will live longer and may develop chronic diseases (such as CVDs or dementia highly prevalent in older population) due to other reasons than tobacco. Consequently, in our findings, the gain in the number of NCD cases avoided by the policy package should be considered at the net of any increase in the number of NCD cases caused by demographic change and population ageing. For instance, the expected reduction in CVD cases presented in figure 1 (estimated at 0.5% of new cases) is the net impact caused by a reduction produced by the policy, and an increase caused by the ageing of people who would have died in absence of the policy but who are alive under the policy scenario.

This study has at least three limitations. First, there is no official estimate for the cost of NRT, which is the costliest component of the policy package. Our estimation of the cost of NRT is conservative as it includes all NRT purchases while it could be possible that only a share of all NRTs, which are medically prescribed, are reimbursed by the State. Generally, the French State reimburses 65% of NRT cost. However, there are exceptions for people suffering from a long-standing disease or with low incomes (beneficiaries of Complémentaire santé solidaire, who represent about 11% of beneficiaries of the social security scheme)—for whom the State covers 100% of the cost. Due to data availability, the analysis assumes a global reimbursement rate of 65%. Besides, other costs could have been incorporated in the analysis, including the cost for other regulatory measures such as the ban on flavourings, local actions and the cost of medical consultations related to smoking cessation, but data are not available. Overall, we think that, even if accounted for, these costs are unlikely to significantly change the overall results of the assessment given that these are likely to be considered small costs. For example, in 2016, only 19% of people who attempted to stop smoking in relation with the Mois Sans Tabac campaign (that corresponds to 72 200 individuals) reported to reach out to a healthcare professional, such as a general practitioner (GP) or a pharmacist.31 However, the data do not permit to disentangle visits to GP or pharmacist. For GP visits, the time of a medical consultation spent on promoting smoking cessation is generally low (2–10 min),32 suggesting that the cost related to the time of medical consultation spent on smoking cessation is relatively contained.

A second limitation relates to the impact of the COVID-19 pandemic on smoking behaviour and on the policy impact. The COVID-19 crisis is unprecedented, so it is difficult to predict its long-term impact. The long-lasting impacts of COVID-19 on the trends in smoking prevalence are unknown, and could result in various outcomes, including a return to the previous declining trend, a further stabilisation of smoking prevalence rates or a resurgence in the trend. To overcome the uncertainty associated with the COVID-19 impact, we conducted a sensitivity analysis to evaluate the potential COVID-19 impact on our assessment. The sensitivity analysis incorporates the short-term effects of COVID-19 on smoking behaviour by considering the most recent data available (2020 and 2021) that show a slowdown in the decline in smoking prevalence rates.

A third limitation is that the model does not consider government revenues and payments which merely transfer financial resources from one individual to another. While tobacco tax revenues are not included in the study, we reckon that these revenues are likely to be significant and could serve to fund and pursue public health purposes. For instance, in 2022, the national revenue from tobacco taxation was estimated at €13.8 billion.33 Similarly, the study does not consider the retirement benefits or survivors’ benefits paid by the government. Our model shows that people’s lives are lengthened, and these additional years are mainly spent in good health allowing people to be active and productive at older ages. To adapt to these structural changes and to control expenditure on retirement benefits, most countries have already implemented plans to raise the normal retirement age, and the future retirement age will increase from 64 to 66 years between 2020 and 2064 on average in OECD countries.34 While our analysis does not incorporate any possible future change in retirement age, the model takes into account the effects on healthy ageing on the labour market and, in particular, on early retirement. People with better health are less likely to retire early—which can generate savings in retirement benefit payments for the government.

Finally, it should be noted that the analysis assumes that the impact of the policy package remains constant over time. This assumption is standard practice for this type of study13 35 and seems plausible given the policies in the package being evaluated. In fact, French legislation indexes tobacco taxes to inflation to ensure a sustained effect over time. For the same reason, the graphic warnings on tobacco products are changed every year.36 In addition, to account for possible variability in the impact of the policies, the results are presented with an uncertainty projection band that includes optimistic and pessimistic scenarios.

Despite these limitations, this study is the first to carry out an economic evaluation of the French tobacco control policy and the first to estimate the potential expected returns if the effect of the policy package is maintained until 2050. It also presents an evaluation of the policy package as a whole, enabling a global evaluation of the combined impact of the actions included in the package. Findings show that a comprehensive and extended tobacco control policy package is effective and produces significant returns on investment. This supports the importance of the WHO Framework Convention on Tobacco Control recommendations by providing a strong economic argument. In addition, this analysis advocates for scaling up investments for even more comprehensive and ambitious tobacco control policies, subject to review and, when needed, revision over time. Finally, findings from the study provide a more general contribution to assist other countries in implementing similar measures.

Future research and studies are needed to evaluate the impact of different tobacco control policy packages in different contexts and to measure their effectiveness. As the COVID-19 crisis has possibly affected the effectiveness of tobacco control policies, further evaluation of the long-term impacts of the pandemic would be necessary, as well as the study of approaches to optimise the policy implementation in the context of future disruptions and crises. As the French government is working on further strengthening the national tobacco control plan, policy modifications would benefit from an ex ante economic evaluation.

Data availability statement

Data are available in a public, open access repository. All data relevant to the study are included in the article or uploaded as supplementary information. Input data for the model are publicly available. Data produced by the model are presented in the article and the supplementary file.

Ethics statements

Patient consent for publication

Ethics approval

Not applicable.

Acknowledgments

The OECD programme of work for public health is supported by a number of voluntary contributions by Ministries of Health, national governmental institutions, key partners of OECD member countries and other intergovernmental organisations. Santé Publique France supported the OECD work on public health and provided data.

References

Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

Footnotes

  • Contributors MD, AL, MC, RG, VN-T and PA conceived the study. AA, AL, MD, MDC, RG and VN-T contributed to the investigation and data collection. AA and AL led the programming, data analysis and results visualisation. AA, AL, MC and MD contributed to the validation of results. MD, MDC and RG wrote the first draft of the manuscript. MD and MC are acting as guarantors. All authors contributed to revising and editing of the final manuscript and approved it. MC, FB and PA supervised the research.

  • Funding This study was funded by Santé Publique France.

  • Disclaimer The opinions expressed and arguments employed herein are solely those of the authors and do not necessarily reflect the official views of the OECD or its member countries.

  • Competing interests None declared.

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

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.