A longitudinal analysis of MPOWER implementation, 2008-2018

In 2008, the World Health Organization (WHO) introduced MPOWER-a package of evidence-based, high-impact policy measures to help countries reduce tobacco use. These measures align with selected articles within the World Health Organization’s Framework Convention on Tobacco Control (FCTC), a global public health treaty to prevent and reduce tobacco use. Observational longitudinal study involving 195 countries that submitted biannual national reports between 2008 and 2018. To report changes in MPOWER scores, countries were assessed using a validated checklist of the seven MPOWER indicators and a maximum possible unweighted composite score of 34 points. Covariate analysis was conducted among selected health, sociopolitical, and economic indicators. 176 out of 195 countries improved their MPOWER scores between 2008 and 2018, with two achieving full implementation (Brazil and Turkey). Twenty-three (23) countries representing 11.2% of the global population recorded an MPOWER scoring increase of at least 10 points between 2008 and 2018. The overall mean 10-year scoring increase was 5.1 points or a relative improvement of 27.1%. In 2018, 20 countries representing 10.4% of the global population excelled in MPOWER implementation by receiving a total composite score of at least 30 of 34 possible points. The MPOWER elements with the highest degree of implementation in 2018 include Warn (package), Protect (smoking ban) and Enforce (ad ban). Several covariates were positively associated with higher levels of MPOWER implementation, including cigarette affordability, the existence of national tobacco control objectives, the human development index score, the national corruption index score, national literacy rates and the political regime index score. MPOWER implementation increased markedly between 2008 and 2018 in all seven WHO regions and countries representing all four World Bank income classifications. However, only two countries achieved full implementation by 2018. More work is needed to improve MPOWER implementation. Countries with low-income levels, compromised human development, reduced literacy rates, higher rates of corruption, and autocratic political regimes appear to struggle more with MPOWER implementation.

5][6][7] To achieve this objective, the WHO negotiated the Framework Convention on Tobacco Control (FCTC) in 2003-representing the world's first public health treaty.As of July 2021, 182 countries representing 90% of the global population have ratified the FCTC. 7The implementation of the FCTC is credited with preventing an estimated 49 million smoking-attributable deaths within the first ten years of its implementation. 5,6In 2008, a policy and program measures package based on critical articles of the FCTC was developed to help promote and guide FCTC implementation.This package includes six main components, namely: Monitor tobacco use and prevention policies, Protect people from tobacco smoke, Offer help to quit tobacco use, Warn people about the dangers of tobacco, Enforce bans on tobacco advertising, promotion and sponsorship, and Raise taxes on tobacco (MPOWER; to empower). 8,9][16][17][18] The objectives of this study are to use results from the 2008 to 2018 MPOWER scores to (i) Report national MPOWER scores to allow countries to benchmark, measure, monitor, compare and stimulate MPOWER implementation; and (ii) Identify independent factors that influence MPOWER implementation.

METHODS STUDY DESIGN
5][16][17][18] The data for all MPOWER reporting periods from 2008 to 2018 was retrieved from the WHO Global Health Observatory. 19he data for each country was represented using seven measurable variables, reflecting all MPOWER components, including two indicators for Warn (package warnings and mass media campaigns).All variables have a gradient range of 0-4 (5 gradients) except for Monitor, which has a range of 0-3 (4 gradients) for a maximum possible score of 34 (4 + (5×6) = 34).The scoring is based on previously published methodology. 13RIABLES AND DATA SOURCES The primary data source was the WHO Global Health Observatory (GHO) 20 which includes all the MPOWER national results collected between 2008 and 2018.The GHO also includes national datasets for (1) tobacco smoking prevalence, 21 (2) government objectives on tobacco control, 22 (3) national cigarette affordability, 23  (4) the annual tobacco control budget of each country, 24 (5) the existence of a government tobacco control agency, 25 and (6) the number of tobacco control staff employed by national governments. 25ll variables, except smoking prevalence, were used for the covariate analysis.Smoking prevalence and its relationship to MPOWER scores have been examined and reported by others.
The secondary data source was Our World in Data (OWID), 26 an online repository of numerous health, social, political and economic indices and datasets.The OWID indices selected for this study included (1) literacy rates on a scale of 0 to 100 27 ; (2) the political regime scale from a full democracy to full autocracy 28 ; (3) the United Nations human development index which includes life expectancy, education, and gross national income with an aggregate score between 0 and 1 29 ; and (4) the World Bank's four national income classifications including low income, low-middle, upper-middle, and high income. 30er capita tobacco control budgets for each country were calculated by dividing a country's annual budget for tobacco control in equivalent U.S. dollars by its total population size.The standard currency reported by the U.S. Department of Treasury was used for any country without its currency reported in the GHO database.The total population for all countries was obtained from the World Development Indicator Databank (WDI) of the World Bank.For all country currencies without a standard average equivalent of 1 USD in the U.S. Department of Treasury database, an average rate (March 31 st , June 30 th , September 30 th , and December 31 st ) was calculated from historical currency tables. 31

STATISTICAL ANALYSIS
The results were organized by the six WHO regions and by the four World Bank national income classifications to allow for geopolitical and economic analysis.Missing or incomplete data in the WHO report received a score of zero in our analysis.
The association between MPOWER scores and potential covariates was analysed using chi-square for categorical variables and student t-test and repeated measure ANOVA for continuous variables to check between and within subjects' differences.Paired tests were used for the difference in the mean between years, and predictive modelling was performed using traditional forecasting methods.Analysis was performed using the Statistical Package for the Social Sciences (SPSS) version 27.0 (IBM SPSS, Armonk, NY).

RESULTS
The WHO   1 , scoring changes by income level appear in Figure 2 , and MPOWER rankings in 2018 appear in Figure 3 .
The highest rates of implementation for the MPOWER variables in 2018 were (1) Warn (package) with a mean score of 3.86 out of a possible 5 points; (2) Enforce (advertising bans) with a mean score of 3.79 out of a possible 5 points; (3) Offer with a mean score of 3.70 out of a possible 5 points and (4) Raise with a mean score of 3.48 out of a possible 5 points.The lowest rates of implementation were (1) Warn (mass media) with a mean score of 2.84 out of a possible 5 points; (2) Protect with a mean score of 3.40 out of a possible 5 points; and (3) Monitor with a mean score of 2.81 out of a possible 4 points.These results can be found in Table 2.
Repeated measure analysis showed that the estimated marginal means of MPOWER score has been increasing since 2008.This increase is observed across different levels of income and different WHO regions except for the bottom 20 countries in the 2018 MPOWER rankings.These lowperforming countries reported declining scores after an initial increase between 2008 and 2012.However, paired test analysis revealed that the largest increase in MPOWER scoring occurred in the first reporting period following the introduction of the MPOWER measures in 2008.Figure 5 shows projected dates for achieving full MPOWER implementation score stratified by countries' income levels.At the current implementation rate, high-and middle-income countries are projected to achieve full MPOWER implementation by 2040, led by high-income countries, which are forecast to reach the 34-point target by 2036.The projections also reveal that low-income countries are not expected to achieve full MPOWER implementation in the foreseeable future.While these trends are encouraging for most countries, they suggest that full MPOWER implementation is unlikely in many low-income countries.

IMPLEMENTATION OF MPOWER ELEMENTS
The highest degree of implementation of MPOWER elements between 2008 and 2018 was Warn (packaging), with a 1.17-point increase, reflecting a 10-year improvement of 43.7%.The second highest increase was Protect, with a 0.82-point rise, representing a relative change of 31.6%.MPOWER implementation was Monitor, with a relative 10-year increase of 0.9%, reflecting a 0.03-point increase.
The second lowest improvement was Warn (mass media), reflecting a 0.10-point rise, or a relative change of 3.8%.The third lowest change was Raise, with a 0.25-point increase, representing a relative improvement of 7.6%.These results can be found in Table 2 .

IMPACT OF COVARIATES
Several influential covariates were identified through selective analysis of health, social, political and economic indicators in the WHO Global Health Observatory and the Our World in Data online dataset repository.This study examined these indicators between the 20 best performing MPOWER countries and the 20 worst performing nations (Table 4).Covariates that appear to positively influence MPOWER implementation include lower cigarette affordability (P=0.03;t=-2.30), the overall tax score (P<0.001;t=4.67) the presence of national tobacco control objectives (P=0.001,X 2 =22.7), the human development index score (P<0.001;t=5.1), the national corruption index score (P<0.001;t=3.96), the political regime index score (P<0.001;t=5.76), and the national literacy rate (P=0.003;t=3.18).Covariates that did not influence MPOWER implementation included a government tobacco control agency (P=0.12;X 2 =4.28) and the number of tobacco control staff employed by the government (P=0.78;t=1.85).

DISCUSSION
For the first time since MPOWER was launched in 2008, two countries achieved a perfect score of 34 in the 2018 MPOWER rankings: Brazil and Turkey.Notably, neither country is ranked as a high-income country, which should motivate middle-income countries and high-income countries to implement MPOWER fully.Of the 20 top performing countries in 2018, 10 were ranked as high-income, seven were rated as upper-middle income, 3 were lower-middle income, and none were low-income nations.Together, these countries represent 10.4% of the world population.This modest degree of high MPOWER implementation is concerning, considering the large number of countries (182) that have ratified the FCTC.Of the 20 countries with the lowest MPOWER scores in 2018, 9 were ranked as lowincome, two were rated as low-middle income, 5 were highmiddle income, and 4 were high-income countries.
Of the 23 countries that achieved a minimum 10-point increase between 2008 and 2018, 3 were high-income countries, 12 were upper-middle income countries, 7 were lower-middle income, and 1 was a low-income nation.Nineteen (19) middle-income countries were among the 23 most improved nations for MPOWER implementation between 2008 and 2018.All WHO regions realised 10-year improvements in MPOWER scores, with the greatest progress achieved by South Asia, East Asia & Pacific, and Europe & Central Asia (Figure 2).Although high-income countries had the highest MPOWER rankings in 2018, middle-income countries are the top performers in achieving the most improved 10-year scores.MPOWER scores improved among all income gradients.However, lower-and upper-middleincome nations achieved more relative progress than highand low-income countries.Unfortunately, the gap in MPOWER implementation between low-income and nonlow-income countries is growing wider over time, perhaps due to external influences on MPOWER scores, such as compromised human development, more oppressive government regimes, and higher rates of corruption.Given these challenges, more work is justified to assist low-income countries with MPOWER implementation.
Regarding the 10-year implementation of MPOWER elements, Warn (package) was by far the most well-implemented policy measure with a 15.3 point scoring increase reflecting a relative rise of 44%, followed by Protect (smoking ban) (9.7 points = 32% increase) with and Enforce (advertising ban) (9.1 points = 21% increase).The most disappointing results were Monitor with a 1% increase, Warn (mass media) with a 4% increase and Raise (taxes) with    an 8% increase in implementation over ten years.Tobacco taxes are the single most effective means of reducing tobacco use.This vital tobacco control measure's relatively low implementation rate is disturbing and warrants further attention.

COVARIATES
The relationship between MPOWER and selected covariates was somewhat inconsistent and counter-intuitive.For example, tobacco control objectives and cigarette affordability are positively associated with MPOWER implementation, but the existence of a government tobacco control agency and government funding for tobacco control is not correlated.Similarly, socio-economic indicators such as the human development index, the national corruption index and the political regime index are positively associated with MPOWER implementation.High-and middle-income countries report higher rates of MPOWER implementation, and low-income countries report lower implementation rates.Further analysis of country demographics and social, political, and economic indicators and their relationship to MPOWER implementation is warranted.

LIMITATIONS
The MPOWER implementation results were self-reported to the WHO by participating nations and were validated by WHO regional offices in consultation with tobacco control experts in each country. 8,9The scope of this study did not allow for any additional auditing or scrutiny of the national self-reports.Secondly, the MPOWER numerical scoring tables were obtained from the WHO Global Health Observatory.The numerical data does not appear in the MPOWER biannual reports, and there is potential for inconsistencies.Thirdly, MPOWER does not address all the possible influences on tobacco control and smoking behaviour, such as tobacco industry interference, which has been identified as the largest barrier to implementing FCTC globally. 32Arti-cle 5.3 of the FCTC addresses tobacco industry interference, and another independent global index measures national adherence to this critical indicator. 33In addition, MPOWER does not address any supply-side tobacco control measures nor the four established elements of the tobacco marketing mix, specifically place (point of sale) and product (design and production). 34Finally, the authors selected the covariates subjectively based on their perceived potential impact on MPOWER scores.

CONCLUSIONS
Twenty-three (23)   35 The relative ranking of nations' implementation of WHO MPOWER tobacco control measures provides a global performance index to benchmark, monitor, measure, compare and stimulate MPOWER implementation and to promote tobacco use reduction among all countries.
Although this analysis is not exhaustive, it helps to identify common characteristics of countries that have achieved high levels of MPOWER implementation.The analysis also sheds light on independent covariates that influence MPOWER implementation.We encourage others to conduct further analyses.The increasing availability of public domain datasets provides further opportunities to enhance

Figure 4
reveals a significant increase in MPOWER score in 2008-2010 (range, 3.5 in the Southeast Asia region to 1.15 in the bottom 20 countries).This increase declined sharply from 2010-2012 (ranging from 1.7 in the top 20 countries to 0.41 in low-income countries).The 2014-2016 and 2016-2018 reveal largely insignificant increases in MPOWER scoring (ranging from 1.25 in the top 20 countries to -0.75 in Southeast Asia) (Figures 4, panels A-C).

Figure 1 .
Figure 1.MPOWER score and country income level.

Figure 3 .
Figure 3. MPOWER score and top 20 and bottom 20 countries.

A
longitudinal analysis of MPOWER implementation, 2008-2018 Journal of Global Health Reports

Figure 4 .
Figure 4. A: Trend in difference in mean MPOWER score by country income level.B: The difference in mean MPOWER score by WHO region.C: The difference in mean MPOWER score for top and bottom 20 countries.

Figure 5 .
Figure 5.Estimated dates of full MPOWER implementation by country income level.

Table 1 . MPOWER scores and rankings of 195 countries in 2008 and 2018
*Rank is based on 2018 score and relative change to 2008 score

Table 3 . Countries with notable MPOWER scoring change since 2008
Although MPOWER rates are increasing among lowincome countries, the projected date of full implementation is uncertain.The approval of a new global strategy to accelerate FCTC implementation at the 8th Conference of the Parties held in Geneva in 2018 holds promise for further progress, especially if it is adequately resourced.

Table 4 . MPOWER scoring comparisons between highest 20 and lowest 20 performing countries
*data from 2 countries, §data from 7 countries the understanding of MPOWER and tobacco control implementation.